[time-nuts] Terrestrial GPS

[Stewart Cobb]

Peter Reilley suggests a backup to GPS using terrestrial transmitters. This
idea has been around since the early days of GPS. The terrestrial
transmitters were called "pseudo-satellites", or "pseudolites" for short.
The big problem with this idea is that the GPS signal format has a narrow
dynamic range. The signal strength from a terrestrial transmitter varies
widely (inverse square law) from positions near the transmitter to
positions far away. The variation in any practical system is larger than
the GPS signal format can handle. This is called the "near-far problem".
For an extensive discussion of the pseudolite concept, including the
near-far problem, see my dissertation. You can find it with a web search
for my full name and the word "pseudolites".

Cheers!
--Stu
___
time-nuts mailing list -- time-nuts@febo.com
To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
and follow the instructions there.

[time-nuts] "Super Rubidium" filter mod for PRS-10?

[Stewart Cobb]

Corby has said more than once that it's not worth doing the "super 5065"
optical filter mod to small telecon rubidiums, because they have too many
other compromises in their designs.

What about the SRS PRS-10? By some indications, it's the best
current-production rubidium around. It doesn't have separate filter and
resonance cells, but other than that it seems like a good design. And the
time constants and such are digitally tunable. Is the PRS-10 worth trying
the filter mod on?

Cheers!
--Stu
___
time-nuts mailing list -- time-nuts@febo.com
To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
and follow the instructions there.

[time-nuts] eBay GPS antennas

[Stewart Cobb]

FYI, Antcom and AeroAntenna are two manufacturers of high-quality
American-made GPS / GNSS antennas. Antcom's datasheets are confusing to
read, and AeroAntenna hides theirs behind a customer-login barrier, so most
of eBay doesn't know exactly what they are. This means that good antennas
are often available cheap.

For example, any AeroAntenna whose part number starts with "AT2775" is a
dual-frequency L1/L2 GPS antenna. AT2775-42 is a near-survey-grade antenna,
and one of them is available right now (item 263500294711) at a starting
bid of $20, after expiring yesterday with no bids at $25.

Several other AT2775 antennas are on eBay for under $200. Any of these
would be excellent for time-nut purposes.

IMHO, Glonass coverage is not useful for precision timing. If you don't
have a good enough antenna installation to get continuous GPS coverage,
you're not really doing precision timing. If you are getting continuous GPS
coverage, Glonass doesn't add anything.

Disclaimer: I have no relationship to either company mentioned, other than
previously being a satisfied customer of both companies. I have no
connection to any sales currently listed on eBay.

Cheers!
--Stu
___
time-nuts mailing list -- time-nuts@febo.com
To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
and follow the instructions there.

[time-nuts] theoretical Allan Variance question

[Stewart Cobb]

What's the expected value of ADEV at tau = 1 s for time-interval
measurements quantized at 1 ns?

This question can probably be answered from pure theory (by someone more
mathematical than me), but it arises from a very practical situation. I
have several HP5334B counters comparing PPS pulses from various devices.
The HP5334B readout is quantized at 1 ns, and the spec sheet (IIRC) also
gives the instrument accuracy as 1 ns.

The devices under test are relatively stable. Their PPS pulses are all
within a few microseconds of each other but uncorrelated.  They are stable
enough that the dominant error source on the ADEV plot out to several
hundred seconds is the 1 ns quantization of the counter. The plots all
start near 1 ns and follow a -1 slope down to the point where the
individual device characteristics start to dominate the counter
quantization error.

One might expect that the actual ADEV value in this situation would be
exactly 1 ns at tau = 1 second.  Values of 0.5 ns or sqrt(2)/2 ns might not
be surprising. My actual measured value is about 0.65 ns, which does not
seem to have an obvious explanation.  This brings to mind various questions:

What is the theoretical ADEV value of a perfect time-interval measurement
quantized at 1 ns? What's the effect of an imperfect measurement
(instrument errors)? Can one use this technique in reverse to sort
instruments by their error contributions, or to tune up an instrument
calibration?

I'd be grateful for answers to any of these questions.

BTW, thanks to whichever time-nuts recommended the HP5334B, back in the
archives; they're perfect for what I'm doing. And thanks to fellow time-nut
Rick Karlquist for his part in designing them.

Cheers!
--Stu
___
time-nuts mailing list -- time-nuts@febo.com
To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
and follow the instructions there.

[time-nuts] Timing from Iridium satellites

[Stewart Cobb]

Some time-nuts had expressed interest in the work I do to obtain timing
information from Iridium satellites, as an independent backup to GNSS. A
presentation we gave in May is now available at

<
http://www.atis.org/wsts/docs/2016/5-03_Satelles_Cobb_Test%20Results%20Iridium.pdf
>

We are scheduled to deliver a more technical paper at the ION NTM meeting
co-located with PTTI in January. If any other time-nuts are coming to PTTI,
it would be fun to meet up.

I hope this is not too blatantly commercial for this list. If it is, please
allow TVB to correct me privately, rather than filling the list with
complaints.

Cheers!
--Stu
___
time-nuts mailing list -- time-nuts@febo.com
To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
and follow the instructions there.

Re: [time-nuts] PRS10 rubidium lamp module failure

[Stewart Cobb]

Follow-up: as several time-nuts noted, there was a lot of corrosion on
parts of the lamp module. Oddly, there was no corrosion on the rest of the
PRS10, or on the TS2500 which it came out of.  The resistor lead was
severely oxidized and partly melted. It's possible that the inductance of
the wire-wound resistor helped sustain an arc. Alternatively, the grounded
lead may have arced to the hot end of the lamp drive coil, although the
coil itself showed very little damage. Perhaps ozone from an arc
accelerated the corrosion. Dunno, it's a mystery.

The resistor is specified at 2W, but it's much smaller than a typical 2W
resistor. I removed the damaged lead and replaced it with a bit of 16 AWG
magnet wire. This gave mechanical support as well as an electrical
connection. Buttoned it all back up and applied power, and the lamp lit
almost immediately (it wasn't lit before) and it locked quickly.  Tuned it
to match a GPSDO, and it seems to be working fine.

A note on tuning: the documentation says that the "SF" tuning command has a
resolution of 1E-12. That's true, but the actual available tuning
resolution seems to be more like 1E-11.

The actual tuning is done by adjusting the C-field with a 12-bit DAC. The
"MR?" command gives the present value of that DAC.  Unfortunately, the full
range of the SF command (4000 counts) only moves the DAC about 600 counts.
You can increment the SF command several times before seeing a response.
Anyone planning to use a PRS10 inside a control loop should be aware of
this quirk.

Cheers!
--Stu
___
time-nuts mailing list -- time-nuts@febo.com
To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
and follow the instructions there.

[time-nuts] PRS10 rubidium lamp module failure

[Stewart Cobb]

R901, on the far right, is shown on the schematic as "air heater". It
apparently got hot enough to create a localized thermal runaway in its lead
wire. There was ample evidence of arcing. A half-melted bit of wire fell
away before I could take the picture.  Without that heater working, the
lamp is apparently too cold to start.

Cheers!
--Stu
___
time-nuts mailing list -- time-nuts@febo.com
To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
and follow the instructions there.

[time-nuts] Convert ADEV specs to MDEV?

[Stewart Cobb]

TimeLab can show "mask" overlays on various plots. It comes with a set of
ADEV masks for cesium clocks, derived from the manufacturer's
specifications for those clocks.

Those masks don't show up on MDEV plots, because the manufacturers don't
provide MDEV specs.

For the mathematicians out there: is there a way to convert ADEV specs into
equivalent MDEV specs?

If not, does anyone have good measured MDEV data on any of the popular
cesium clocks?

Cheers!
--Stu
___
time-nuts mailing list -- time-nuts@febo.com
To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
and follow the instructions there.

Re: [time-nuts] 1 PPS 50-ohm driver

[Stewart Cobb]

Sorry, newer scopes annotate a little better. Settings were 0.5 V/div and 2
ns/div. Trigger level is 1.4 V. Scope is HP 54720D with 54712A plugin
(built in 50-ohm termination). The installed bandwidth of the plugin is
somewhere between 1.0 and 1.5 GHz. I should have mentioned all that in the
previous post.

Cheers!
--Stu
___
time-nuts mailing list -- time-nuts@febo.com
To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
and follow the instructions there.

[time-nuts] 1 PPS 50-ohm driver

[Stewart Cobb]

Here's a scope photo from a PPS driver built exactly to the description in
my earlier post.  It's a 74ACT04 in a TSSOP package, with five parallel
outputs driving five 220-ohm resistors (0402 SMT) to form a 50-ohm output.

The photo shows 2 ns rise time for the leading edge of the pulse. The scope
bandwidth is around 1 GHz, so this measurement is pushing its limits.  The
pulse looks fairly clean, but with a bit more care in layout and cabling,
it might get even better.

Cheers!
--Stu
___
time-nuts mailing list -- time-nuts@febo.com
To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
and follow the instructions there.

[time-nuts] 50-ohm PPS driver

[Stewart Cobb]

The "standard cheap" way to make a 1 PPS 50-ohm output is to use a 74ACT04
hex inverter chip powered by 5V. The chip is specified for 24 mA from each
output pin. Connect 5 inputs together to your PPS signal. Run each of the 5
outputs through a 220 ohm series resistor, then tie them together as your
PPS output. This will drive about 2.4 volts into a 50-ohm load, with crisp
edges. It will easily trigger most instruments.

The chip is specified up to 6V supply voltage, so you can get an output
pulse of almost 3V if you have a 6V rail in your design.

You can use the sixth inverter as your PPS input, with two paralleled
100-ohm resistors as the load (and ESD guard) at its input pin. The
switching point for the 'ACT version of the chip is around 1.4V, so the
2.4V output signal will easily drive it.

This circuit may not be perfect, but it shows up in a lot of professional
gear.

Cheers!
--Stu
___
time-nuts mailing list -- time-nuts@febo.com
To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
and follow the instructions there.

[time-nuts] HP 5065A flaws and improvements

[Stewart Cobb]

Bob Camp wrote:

> The 5065 has great ADEV numbers.
> In “as delivered” condition it has horrid
> TC and pressure sensitivity.

That's the first I've heard of any drawbacks to the 5065.  Can you give
more details on this?

The "as delivered" implies that there may be improvements available? Any
details on improvements?

I'm aware of Corby's optical filter trick, which improves SNR and hence
ADEV, but that would not affect sensitivity to temperature or pressure.

Cheers!
--Stu
___
time-nuts mailing list -- time-nuts@febo.com
To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
and follow the instructions there.

Re: [time-nuts] Iridium source?

[Stewart Cobb]

The use of Iridium satellite signals for timing, location, and
position-based authentication is being commercialized by a startup called
Satelles. The technology is known as Satelles Timing and Location, or STL.



Prototype STL timing systems are providing PPS with accuracy better than 1
microsecond. This is expected to improve as the technology evolves.

Full disclosure: Satelles is my employer. We have submitted a paper on STL
timing to PTTI 2016. Don't know yet whether it was accepted.

Cheers!
--Stu
___
time-nuts mailing list -- time-nuts@febo.com
To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
and follow the instructions there.

Re: [time-nuts] Unique TBolt GPS characteristics

[Stewart Cobb]

The Tbolt computes phase error by doing a position fix. It computes
frequency error by doing a velocity fix. These two fixes use almost the
same algorithm, a least-squares fit that incorporates the line-of-sight
vectors to each satellite.

The position fix starts with the pseudorange (distance) to each satellite,
and computes X, Y, Z, and T. XYZ are the position vector from the center of
the earth, and T is the clock bias (phase error).

The velocity fix starts with the pseudorange-rate (aka Doppler) measurement
for each satellite, and computes X-dot, Y-dot, Z-dot, and T-dot. XYZ-dot
are the velocity vector of the receiver (usually zero for time-nuts), and
T-dot is the clock rate (frequency error).

Every GPS receiver has to measure Doppler on each signal as part of the
tracking loop. Since the measurements are already there, computing velocity
is just an extra math step. Most receivers do this. However, that doesn't
help us in a PPS/TDC system, because the resulting frequency error
measurement is a measurement of the GPS receiver's clock, not the OCXO
we're trying to steer.

The trick with the Tbolt is that the GPS receiver clock /is/ the OCXO we're
trying to steer, so the frequency error measurement from each velocity fix
applies directly to the OCXO.

The error on each satellite Doppler measurement is typically a small number
of cm / sec. Scale by the speed of light (3E10 cm / sec) to estimate
instantaneous frequency measurement errors around 100 ppt. If N satellites
contribute to the velocity fix, that should reduce the resultant error by
sqrt(N) at each fix. That handwaving calculation agrees fairly well with
the 25 ppt that you're seeing.

Cheers!
--Stu

PS: Are you saying that the hardware 10 MHz output from a Tbolt is 3E-12
off compared to other GPSDO (or cesium, or whatever)? That's the first I've
heard of that. Is the Tbolt low or high? I can imagine an explanation
involving truncation in the carrier-phase NCO, or in the carrier
feed-forward to the code tracking loop. But it would be hard to prove
without detailed info on the guts of the Tbolt. Hey, if it's true, that's
another way we could improve on a Tbolt.
On Jan 27, 2015 9:32 PM, WS warrens...@netzero.com wrote:

 Stewart

 The part that I do not understand is how the TBolt is able to calculate
 such
 high resolution frequency offset answers for its Osc ppt output so fast.
 The frequency offset answers seem to have way too high of absolute
 accuracy compared to what is possible using the phase data.

 Example:
 It is not unusual for the raw 1 second phase data to have some 1 ns jumps
 in
 it due to GPS signal noise.
 If that phase data where used to calculate the frequency offset directly,
 it
 would cause some 1 ppb (1e-9) frequency jumps to occur.
 More interesting is that the frequency offset output and its polarity are
 not a function of the slope of the displayed phase noise when viewed
 over a time period of a few seconds.

 The peak one second frequency jumps on the osc-ppt output are more like
 2.5e-11 ppt, that's 25ps per second of phase noise and this is using the
 GPS
 as the reference.
 Note that any frequency change of the 10 MHz oscillator is measured and
 fully settled and displayed in the next one second readout, accurate to
 within the limits of its noise.
 If the low noise stability of the delta frequency output was achieved by
 using multi phase points, then any external frequency change would
 take multiple seconds to fully respond, and the polarity of the frequency
 offset would be a function of the phase slope.

 One experiment I did was to compare my TBolt's PP freq noise errors on it's
 Phase output to the reported PPT Osc output.
 The phase error noise, short term over say most 10 second periods was
 around
 +- 1ns relative, and  10 ns absolute.
 The PPT-Osc output was providing 1e-11 peak frequency error, that's
 absolute not relative, when using LH's display filter set for 3 second.

 One way the TBolt may be doing this is by somehow averaging lots of high
 speed and very high resolution internal delta raw Phase data points,
 such as using a one plus KHz sample rate with sub ps resolution, possibly
 as some have suggested, with the use of some sort of additional carrier
 phase data.
 One of the tradeoffs/limitation of the TBolt's Osc-Freq output is that it
 typical  has a ~3e-12 frequency offset error.

 If this was better understood, it should be useful in improving the
 TBolt GPSDO.
 This is likely because of another one of the unique TBolt characteristics,
 which is that it is possible to do an external  S/W control algorithm with
 inputs from various multiple new sources such as WAAS, or remote
 TBolts using common view.


 The SwiftNav, even though a bit expensive for me,  sounds like a
 interesting
 and powerful GPS engine with its cm  50Hz solution updates.
 I wonder how good of frequency resolution/accuracy  vs. time it can do?

 ws
 

  Stewart Cobb posted:


 Every GPS receiver

[time-nuts] Unique TBolt GPS characteristics

[Stewart Cobb]

Every GPS receiver calculates its clock offset (phase error, in time-nut
terms) as part of its four-dimensional position fix. It can apply almost
the same math to calculate its clock rate (frequency error) as part of a
four-dimensional velocity fix.

In position hold mode, the Thunderbolt calculates its timing errors (both
phase and frequency) using similar one-dimensional algorithms. The accuracy
of these measurements is determined by all the factors that affect GPS, but
the precision (resolution) of these measurements is effectively infinite,
limited only by IEEE double-precision floating-point math.

Almost every other GPSDO uses a hardware time-to-digital converter (TDC,
interpolator, etc) to compare the OCXO timebase to the PPS output of a
separate GPS receiver.

The PPS/TDC scheme has four disadvantages relative to the Trimble scheme:

A) The resolution of the phase error measurement is limited by the TDC
hardware.  For example, the HP Z38xx units appear to have 10ns resolution.

B) The accuracy of the phase error measurement may be degraded by analog
effects in the PPS connection.

C) The phase error measurement must be compensated by a software sawtooth
correction for best accuracy, because the PPS output is quantized by the
receiver clock.

D) Frequency error cannot be measured directly, but must be derived from
successive phase measurements. The derivative process introduces noise, so
the derived frequency error must be heavily filtered.

Unfortunately, the Trimble scheme is only available to GPSDO builders who
have access to the internal architecture of their GPS receiver.
Historically, among the major players, only Trimble and perhaps Zyfer did.
Even mighty HP did not.

Fortunately, SwiftNav is now selling a (mostly) open-source GPS receiver.
Only the FPGA correlator chip is closed-source, and that can be ignored for
timing purposes. One would need to build a VCXO-based synthesizer to create
the SwiftNav receiver clock frequency from a good OCXO, add a DAC to
control the OCXO, and do some software work to add timing functions to the
receiver firmware. Adding WAAS corrections to this hypothetical open-source
GPSDO could make it noticeably more accurate than a Thunderbolt.

Cheers!
--Stu
___
time-nuts mailing list -- time-nuts@febo.com
To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
and follow the instructions there.

Re: [time-nuts] Errors on HP 58503A GPS time frequency reference receiver bought from yixunhk.

[Stewart Cobb]

I have bought several items from this seller and left positive feedback
because the items arrived promptly and appeared to work. Problems only
became apparent days, weeks, or months later. But by then it was too late
to change feedback.

This seller admits to changing the firmware on a GPSDO to upgrade it to a
more valuable part number. I bought a device from him for which included a
newly fabricated sheet-metal case complete with new false trademarked
labels.  It's a very good fake, but it was not in fact produced by the
manufacturer whose name is on the label, and it did not meet the
requirements of the device inside. I have bought other devices from him
which had other problems.

It is possible that the seller is obtaining his stock from other sources
inside China, and that he is not directly involved in changing firmware or
creating fakes. It is possible that he himself is an innocent victim of
sharp practice by others. I make no direct accusations. However, the
equipment I have obtained from this seller in the past has not proved
satisfactory to me over the long term, and I have chosen not to do business
with this seller in the future.

Cheers!
--Stu
___
time-nuts mailing list -- time-nuts@febo.com
To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
and follow the instructions there.

Re: [time-nuts] Lucent KS-24361, HP/Symmetricom Z3809A, Z3810A, Z3811A, Z3812A GPSDO system

[Stewart Cobb]

A wiring diagram of the Z3809A cable interconnect cable was published
earlier on this list.  That information appears to be incorrect.  The
cable is actually wired pin 1 to pin 15, pin 2 to pin 14, etc.
Another way to describe it is that for each wire in the cable, the pin
numbers on each end of the cable add up to 16.

A mated pair of these units is running in my lab with a scratch-built
interconnect cable following the above rules.  This scratch-built
cable allowed access to the interconnect signals while the system was
operating happily.  No lights were lit except the green ON light on
the Ref-0 unit (Z3812A, no GPS) and the yellow STBY light on the Ref-1
unit (Z3911A with GPS receiver).  The following signals were observed
on the interconnect (pin numbers given for the J5 interconnect socket
on the Ref-1 unit):

Pin 1:  9600 baud serial data (described below)

Pin 2:  logic low (0.11V)

Pin 3:  Ground (0.00V)  Presence detect? (see below)

Pin 4:  logic high (4.79V)

Pin 5:  inverted Motorola PPS, high (5V) for 800ms, low for 200ms

Pin 6: 17 / 23 dBm signal from Ref-0 unit (see below)

Pin 7:  logic high (4.48V)

Pin 8:  Ground (0.00V)

Pin 9:  logic low (0.11V)

Pin 10: 17 / 23 dBm signal from Ref-1 unit (see below)

Pin 11:  inverted PPS, low 400us, high (5V) otherwise

Pin 12:  logic low (0.12V)

Pin 13:  Ground (0.00V)

Pin 14:  logic low (0.08V)

Pin 15:  logic high (4.78V)

Pins 3, 8, and 13 appear to be firmly connected to Ground.  (Note that
these are the three pins which are clipped short on the HP
interconnect cable.)  On an unpowered, disconnected box (either Ref-0
or Ref-1), pins 8 and 13 are connected to Ground (low resistance) and
pin 3 is high impedance.  Presumably pin 3 on each box (connected to
the grounded pin 13 on the other box) is used to sense the presence of
the other box and/or the interconnect cable.

The timing of the PPS signal on pin 11 matches precisely the timing of
the PPS signal available on pins 1 and 6 of J6 (RS422/PPS) on the
active Ref-0 unit.  Presumably this signal is coming across the cable
from the Ref-0 unit.

Note: when the system is coming up from a cold start, SatStat on the
unit with the GPS receiver (Ref-1) will show [Ext 1PPS valid] in the
space where it shows [GPS 1PPS valid] after the survey is complete.
It appears that the Ref-1 unit timing system is locking its oscillator
to the PPS coming from the Ref-0 unit during this time.

The timing of the PPS signal on pin 5 matches the timing of the PPS
output described in the Motorola OnCore manual.  Presumably this
signal is sourced by the Ref-1 unit to allow the Ref-0 unit to lock to
GPS.  The edges of this PPS signal look very dirty compared to the
signal on pin 11.  This may be an artifact of the homemade cable used
for this experiment.  The HP cable clearly has an overall shield
(visible through the cable sheath) and may have internal coax or
twisted pair for these PPS signals.

When pin 5 and pin 11 are observed together, the usual GPS sawtooth
pattern is evident.

Someone discovered earlier that the both units will blink their green
ON lights if the front-panel switch on either unit is set to 23 dBm
vice the normal 17.  Obviously each unit can communicate its switch
status to the other unit.  They use pins 6 and 10 to do that.  Pin 10
(on the Ref-1 unit) is high (~5V)  if the switch on the Ref-1 unit is
in the 17 dBm position, and low in the 23 dBm position. Pin 6 (on the
Ref-1 unit) gives the same indications for the switch on the Ref-0
unit.

The serial data on pin 1 is transmitted at 9600 baud, with a burst of
data every second.  The signal idles at logic low (near 0V) and rises
to logic high (near 5V) during the burst.  This may be the standard
for TTL (not RS-232) transmission of serial data, or it may be
inverted.  The first few characters of one burst were hand-decoded
from a scope trace as 0x40, 0x40, 0x45, 0x61, 0x0B, or ASCII @@Ea.
This appears to be the Motorola Oncore binary data format, although
Ea does not appear to be a valid Motorola command or response.
Perhaps the hand-decoding was in error.

One can use SatStat, talking to the Ref-0 (non-GPS) box, to issue
queries and commands to the GPS receiver.  The results are
inconsistent, but it seems that at least some of the queries get
through and trigger responses.  If the Ref-0 box is actually talking
to the GPS receiver, it must be doing so through the interconnect
cable.  The specific wire in the cable used for this (if any) has not
yet been identified.

An earlier post speculated that the computer in each unit only had two
UARTs.  This does not seem possible.  Clearly each unit uses one UART
to communicate with the J8 diagnostic port.  The Ref-1 unit needs
another UART to communicate with the GPS receiver. And both units need
to be able to transmit the legacy Lucent timecode message out the J6
(RS422/1PPS) port.  Perhaps there is a transmit-only UART coded into
the FPGA, or perhaps one of the UARTs is timeshared with the Lucent

Re: [time-nuts] Lucent KS-24361, HP/Symmetricom Z3809A, Z3810A, Z3811A, Z3812A GPSDO system

[Stewart Cobb]

I have now tested 3 pairs of these units (full Z3810AS system), always
in pairs.  One Z3811 box was DOA.  It drew power at the appropriate
level, but the lights did not light up and there was no response on
the SatStat port.  AECI quickly swapped it for a working unit, and
even paid return shipping for the dud.  Can't complain about the
service.  One pair was built in the US in approximately March 1999; it
has serial numbers in the old HP style: 3844A41xxx.  The others were
built later, in Korea, and have new-style serial numbers like
KR92840xxx.  They all seem to have the same software.

My lab uses a dual USB-to-serial converter to talk to both boxes at
once.  This does not use FTDI chips.  One side seems to talk perfectly
both ways through the RS-422 hack wiring; the other side seems to
receive perfectly but the GPSDO complains about erroneous commands
about once or twice an hour.  If anyone wants to duplicate my setup,
the converter comes from Microconnectors and can be found here:

http://www.amazon.com/Micro-Connectors-Serial-Adapter-E07-162/dp/B0032325LE

(Link provided for information only; I have no connection with any of
these people.)

It is NOT TRUE that only the box with the green light on will talk
over its diagnostic port.  Both boxes can be interrogated
simultaneously with SatStat, starting immediately after power-up.  The
J8 diagnostic ports on both units speak only when spoken to.  If
you're using a scope to look for activity on the diagnostic port, you
won't see any unless a PC is connected.  If you're switching one
serial cable back and forth between two boxes, be sure to close the
port (menu CommPort/PortOpen unchecked) before removing the cable from
the first box, and open the port again (menu CommPort/PortOpen
checked) after connecting the cable to the second box.  The PortOpen
command seems to do more than merely opening the serial port; it seems
to be necessary to get communications going with the box.  If you
watch the window title, you can see that it's sending a *IDN? command
and interpreting the results, but it may be doing other things as
well.

On the Z3812A, the front panel 10 MHz test point J1 connects to a
group of three 100-ohm SMT (0805?) resistors and one SMT (0805 again?)
ceramic capacitor.  These are next to U206, a 74ACT14 in a SO-14
package, on the bottom of the board behind J6.  On the Z3811A, these
three resistors and one capacitor are missing.  It is quite probable
that adding them would send a 10 MHz signal to the SMA jack footprint
buried under the GPS antenna input TNC jack.  One could presumably
solder a small coax cable to that footprint and route it around to a
convenient location on the front panel.  (The capacitor is not marked
with a value, but it's probably small.  It connects from the 10 MHz
output to ground, and its purpose is probably just to slow down the
edge rates of the 10 MHz output to reduce EMI.)

The 10 MHz output seems to have a duty cycle of about 55% high / 45%
low.  This may be related to the synthesis technique, or it may simply
be due to asymmetrical thresholds in the 74ACT14 schmitt-trigger
driver.

The firmware in the Z3811 and Z3812 boxes appears to be exactly the
same.  The PCBs are stuffed slightly differently, and of course the
Z3812 has no GPS receiver.  It is possible that one could add a GPS
receiver speaking the correct protocol to a Z3812, and turn it into a
GPSDO.  It would be necessary to find out how the firmware detects
what sort of box it is.  Perhaps it simply looks for a GPS receiver,
or perhaps there is a pullup resistor somewhere on the board that is
only stuffed on one version of the box.

The GPS receivers in these boxes use fairly old technology.  In
particular, they can only track 8 channels at a time, and they cannot
make use of the WAAS signals.  WAAS transmits corrections for
satellite ephemeris errors and for the changing ionosphere.  These are
two of the most prominent error sources for GPS timing.  (The others
are antenna position error and multipath, both of which are
potentially under the control of the user, and troposphere delay,
which is small but can't easily be measured or removed.)  It is
theoretically possible to build a plug-compatible board carrying a
modern GPS timing receiver and a small microcontroller to translate
its communications into Motorola Oncore language.  With a well-sited
antenna, such a system might display noticeably improved performance.

The current GPS receivers in these boxes take a remarkably long time
to settle down after power-up.  The GPS satellites transmit their full
almanac every 12.5 minutes, which is normally all the time required
for a GPS receiver to become completely happy.  These receivers seem
to require over an hour to become happy enough to START their
self-survey, and several more hours to complete it.  The old rubidium
RFTG units could take 24 hours or more to become fully operational;
these units seem to take 4 to 12 hours, depending on GPS antenna
quality 

[time-nuts] HP 5370, 5371, 5372, 5373

[Stewart Cobb]

This list has a lot of discussion of HP 5370 time-interval counters,
including the BeagleBone CPU upgrade. The usual sources seem to have HP
5371, 5372, and 5373 units available as well. From the profession of model
numbers, and the cosmetics of the front panels, one might assume that these
are successor units to the 5370.

Are they actually successor units? Are they upward-compatible with the
5370?   Are they more accurate, or less accurate?  Is there anything that a
5370 can do that the later units can't?

Cheers!
--Stu
___
time-nuts mailing list -- time-nuts@febo.com
To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
and follow the instructions there.

Re: [time-nuts] Lucent KS-24361, HP/Symmetricom Z3809A, Z3810A, Z3811A, Z3812...

[Stewart Cobb]

 Message: 7
 Date: Tue, 21 Oct 2014 06:59:04 -0400
 From: gandal...@aol.com
 To: time-nuts@febo.com
 Subject: Re: [time-nuts] Lucent KS-24361, HP/Symmetricom Z3809A,
 Z3810A, Z3811A, Z3812...
 Message-ID: eac2.4de53503.41779...@aol.com
 Content-Type: text/plain; charset=US-ASCII

 It seems from the auction revision table that this seller has been offering
  these for some time, so perhaps another hidden gem:-), but it's  perhaps
 also worth noting that if this system functions on similar principles to
 earlier RFTG kit then the GPS conditioning is only applied to the unit
 actually  containing the GPS module, with the other unit intended as  a 
 standby
 should the first one fail.

 In other words, unless the system redundancy is really required most users
 would probably only need the GPS based unit, or would at least be  better
 off buying two of those for the same money that the matched pair  would
 cost.

 The only advantage, as far as I'm aware anyway, of the non-GPS unit is that
  it contains a 10MHz output.
 However, Skip Withrow published modification details in January 2013
 showing how straightforward it was to add the the 10MHz output, to the
 RFTGm-II-XO module, the PCB location for the socket was already available, so 
 I
 would suspect it wouldn't be too difficult on these either.

 Regards

 Nigel
 GM8PZR


I'm sorry, but most of this is inaccurate.  The earlier RFTG units
(built by Datum and its successors, I think) had one rubidium and one
OCXO.  Both were disciplined by the GPS receiver.  In the set that I
have (RFTGm-II Rb and XO), the diagnostic software can actually
display a list of the last ten or so frequency and time corrections to
both the Rb and the OCXO (two separate lists).  The OCXO is indeed a
backup in these units, but it is disciplined by the GPS receiver so
that it is constantly ready to take over if needed.

The current HP units appear to function the same way.  Both units
contain the equivalent of a Z3805A, and both steer their OCXOs to lock
to GPS time and frequency.  Both can be interrogated independently to
observe their steering corrections and statistics.  I assume that the
PPS and timetag data is fed across the interconnect cable from the
unit with a GPS receiver to the one without.

Finally, the current HP units are completely different internally from
the older RFTG units.  The 10 MHz modification mentioned above does
not apply to the HP units.  I believe there is an equivalent
modification, involving several surface-mount resistors, one
surface-mount capacitor, and an output cable, that can add a 10 MHz
output to the unit which lacks it.  However, I have not yet completed
or tested this mod.  If I get it to work reliably, I will post it to
the list.

A better solution for time-nuts would be to repurpose the 15 MHz
outputs on both units and set them up to output 10 MHz instead.
However, that mod would require much more detailed tracing of the
circuitry than I have done.

Cheers!
-Stu
___
time-nuts mailing list -- time-nuts@febo.com
To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
and follow the instructions there.

[time-nuts] Lucent KS-24361, HP/Symmetricom Z3809A, Z3810A, Z3811A, Z3812A GPSDO system

[Stewart Cobb]

Fellow time-nuts,

This (long) post is a review of the HP/Symmetricom Z3810A (or Z3810AS)
GPSDO system built for Lucent circa 2000.  I wrote it because I looked
for more information before I bought one, and couldn't find much.
It's relevant because (as of this writing), you can buy a full system
on the usual auction site for about $150 plus shipping.  For those of
you lamenting the dearth of cheap Thunderbolts, this looks like one of
the best deals going.  The description of these objects does not
include GPSDO, so time-nuts may have missed it.  Search for one of
the part numbers in the subject line and you should find it.

So what is it?  It's a dual GPSDO built by HP as a reference
(Redundant Frequency and Time Generator, or RFTG) for a Lucent
cell-phone base station, built to Lucent's spec KS-24361. Internally,
it's a close cousin of a later-model Z3805A.  Externally, it looks to
be almost a drop-in replacement for the earlier RFTG system built to
Lucent's spec KS-24019.  That was a redundant system containing one
rubidium (LPRO, in the one I have) and one OCXO in two
almost-identical boxes.  That spec went through several revisions with
slightly different nameplates and presumably slightly different
internals.  You can generally find one or two examples on the auction
site (search for RFTG or KS-24019).

This system is similar, but the two boxes each contain a Milliren
(MTI) 260-0624-C 5.000MHz DOCXO, and neither contains a rubidium.  The
Milliren DOXCO is the same one used in the later models of the HP
Z3805A / 58503A.  It's a very high-performance DOCXO, in the same
class as the legendary HP 10811, and better than the one in most
surplus Thunderbolts.  The 5 MHz output is multiplied up to 10 MHz in
at least one unit, and 15 MHz in both units.  I don't have the ability
to measure phase noise on these outputs, but I'd be interested to see
the results if someone could.

Nomenclature:  The Z3810AS (there always seems to be an S at the
end) is a system consisting of the Z3811A (the unit containing a GPS
receiver), the Z3812A (the unit with no GPS receiver), and the Z3809A
(a stupid little interconnect cable).  The GPS receiver inside the
Z3811A is a Motorola device, presumably some version of an OnCore.
Where the Z3811A has a TNC GPS antenna input, the Z3812A has an SMA
connector labeled 10MHz TP.  That is indeed a 10 MHz output.  It
comes active as soon as power is applied to the unit, and its
frequency follows the warmup curve of the OCXO.  The two units have
identical PCBs (stuffed slightly differently), and I have no doubt
that someone can figure out how to add a 10 MHz output to the Z3811A
as well.

Operation:  From the outside, these units are broadly similar to
earlier units in the Lucent RFTG series. The (extremely valuable)
website run by Didier, KO4BB, has a lot of information on those
earlier units, much of which still applies here.  The purpose of these
units was to provide a reliable source of frequency and timing
information to the cell-site electronics.  The 15 MHz outputs from
both units were connected to a power combiner/splitter and directed to
various parts of the transmitter.  The units negotiate with each other
so that only one 15 MHz output is active at a time.  The outputs
labeled RS422/1PPS contained a 4800 baud (?) serial time code as
well as the PPS signal, which were sent to the control computer.

Power is applied to the connector labeled +24VDC and P1, in
exactly the same way as the earlier RFTG units. Apply +24V to pin 1
and the other side of the power supply (GND or RTN) to pin 2.  In
these units, that power supply goes directly to an isolated Lucent
DC/DC converter brick labeled IN: DC 18-36, 1.9A.  Presumably you
can run both units with a 4-amp supply.

Once you have applied power, connect the Z3809A cable between the
jacks labeled INTERFACE J5 on each unit.  The earlier RFTG units
used a special cable between two DE-9 connectors, and it mattered
which end of the cable connected to which unit.  The interconnect for
these units is a high-density DE-15 connector (like a VGA plug).  The
Z3809A cable is so short that the two units need to be stacked one
above the other, or the cable won't reach.  It doesn't seem to matter
which end of the cable goes to which unit.  I don't know whether it's
a straight-through cable, or whether you could use a VGA cable as a
substitute.

When you apply power, all the LEDs on the front panel will flash.  The
NO GPS light will continue flashing until you connect a GPS antenna.
Once it sees a satellite, the light will stop flashing and remain on.
The unit will conduct a self-survey for several hours.  Eventually, if
all is well, the Z3812A (REF 0 on its front panel) will show one
green ON light and the Z3811A (REF 1) will show one yellow STBY
light.  This means that the Z3812A is actually transmitting its 15MHz
output, and the other one is silently waiting to take over if it
fails.

Most time-nuts want to see more than a pretty green light.  The 

Re: [time-nuts] Homemade GPS Receiver

[Stewart Cobb]

 Date: Fri, 26 Sep 2014 10:07:56 +0100
 From: Dr. David Kirkby (Kirkby Microwave Ltd)
 drkir...@kirkbymicrowave.co.uk
 To: Peter Putnam pe...@ni6e.com,  Discussion of precise time and
 frequency measurement time-nuts@febo.com
 Subject: Re: [time-nuts] Homemade GPS Receiver
 Message-ID:
 canx10hbonpsnop5sjtvrzcdw4rhz_cytfcbtdp55ovytess...@mail.gmail.com
 Content-Type: text/plain; charset=UTF-8

 I don't understand the units of signal strength

 The L1 carrier is spread over a 2 MHz bandwidth and its strength at the
 Earth's surface is -130 dBm. Thermal noise power in the same bandwidth is
 -111 dBm

 Then goes on to talk about the signal being 20 dB below the noise.

 Unless the -130 dBm is over the whole surface area of the earth,  which I
 doubt, the units make no sense to me. The units of signal strength should
 be V/m, A/m or W/square metre.

 The noise power should be in Watts or dBm. So taking the difference (19 dB,
 which is approximately 20 dB) between these figures seems odd to me.

 Dave

The units come from the official GPS system specification, which is
available here:

http://www.navcen.uscg.gov/pdf/IS-GPS-200F.pdf

Quoting section 3.3.1.6:

3.3.1.6 User-Received Signal Levels. The SV shall provide L1 and L2
navigation signal strength at end-of-life (EOL), worst-case, in order
to meet the minimum levels specified in Table 3-V. Any combining
operation done by the SV and associated loss is compensated by an
increase in SV transmitted power and thus transparent to the user
segment. The minimum received power is measured at the output of a 3
dBi linearly polarized user receiving antenna (located near ground) at
worst normal orientation, when the SV is above a 5-degree elevation
angle. . . .

Table 3-V gives the minimum received power level for L1 C/A code as
-158.5 dBW, equivalent to -128.5 dBm.  The old spec (some 20 years
ago) was -160 dBW, but the actual satellites were always a bit hotter
than spec, and they finally decided to just bump the spec up a bit.
Of course, the new satellites are hotter than the new spec, and the
spec actually describes a relatively poor receiving antenna.  Actual
received power levels from a well-sited high-quality antenna (which
all time-nuts should have) can be several dB higher than the spec.

So to answer your question, the original source was using somewhat
sloppy wording.  However, the actual system spec is indeed written in
terms of signal power (dBm) at the antenna feedpoint, not in terms of
field units like V/m or W/m^2.

Cheers!
--Stu
___
time-nuts mailing list -- time-nuts@febo.com
To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
and follow the instructions there.

[time-nuts] How to open solder-sealed OCXOs?

[Stewart Cobb]

What's the best way to open an OCXO in the typical solder-sealed tinned
steel can?  I don't mind destroying the can itself, as long as the innards
are not harmed. The goal is to run some experiments with thermal impedance
as discussed here last week, and to ovenize parts of the EFC controller for
better stability.

Cheers!
--Stu
___
time-nuts mailing list -- time-nuts@febo.com
To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
and follow the instructions there.

[time-nuts] Thunderbolt tuning DAC theory of operation

[Stewart Cobb]

While poking around the Thunderbolt to determine whether -5V could be
used in  place of -12V, I discovered how the OCXO tuning DAC works.
Apologies if this is old news, but I haven't seen it documented
before.

The 10MHz sine wave from the OCXO  is squared up and used to clock the
Xilinx 5200 CPLD (U22) and a 74AC174 hex D flip-flop (U14).  Inside
the CPLD (apparently) the 10 MHz clock is divided by 1024, giving a
square wave with a period of 102.4 us (about 9.7 kHz).  The duty cycle
of that square wave is modulated by the 10 MSBs of the commanded DAC
value.  The LSBs are used to offset the falling edge of the square
wave one clock cycle (100 ns) later, during a fraction of the 9.7 kHz
square waves proportional to the LSBs value.  On a modern digital
scope, you can zoom in on the falling edge of the square wave, set the
display to average, and see that the averaged height of that clock
cycle is proportional to the DAC LSBs.  There appear to be at least 8
LSBs, perhaps as many as 10, giving a total DAC resolution of 18 to 20
bits.  (If the DAC value is averaged over one second, there are 10^7
clock cycles which can be controlled, giving a theoretical maximum
resolution of 23+ bits.  Trimble may have chosen a shorter averaging
time and fewer bits.)

The PWM square wave travels from pin 13 of the CPLD (U22) to pin 4,
the D1 input of the 74AC174 (U14).  The flip-flops in this chip are
also clocked by the squared-up 10 MHz from the OCXO.  The Q1 output,
pin 5 of U14, goes to one side of R83 in the circuitry around the
LT1014 op-amp.  The other five inputs and outputs of U14 are
constantly high or low.  They may also be fed to the op-amp circuits,
to help it handle the square wave in a purely ratiometric manner.

The inputs and outputs of the Xilinx CPLD can be programmed for many
different I/O standards.  Unfortunately, this makes their output pin
drivers far from ideal.  The purpose of the 74AC174 is presumably to
drive the analog circuitry with a input that is as close as possible
to a mathematically ideal digital signal.  Outputs in the 74AC logic
family can source or sink 24 mA and have relatively balanced raise and
fall times.  This was probably the most ideal digital output available
to the Thunderbolt's designers in the late '90s.

This DAC implementation is guaranteed monotonic, an important
consideration.  There is exactly one rising edge and one falling edge
per cycle, so that any difference between rise and fall times will
have a constant effect which can be tuned out.  Unlike a sigma-delta
DAC, this PWM DAC produces strong spectral lines at multiples of the
9.7 kHz square wave frequency.  On the one hand, it is comparatively
easy to design filters to remove a single frequency (and its
harmonics).  On the other hand, this signal is strong enough that it
may appear in phase noise plots anyway.

If you want to view the 9.7 kHz square wave for yourself, it appears
on a small square test point next to the silkscreen designator for
C78, very close to the 6-pin power input jack.  This test point is
part of the connection from the Xilinx CPLD to the hex D flip-flop.
Probing it does not affect the OCXO tuning.

Hope this helps.

Cheers!
--Stu
___
time-nuts mailing list -- time-nuts@febo.com
To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
and follow the instructions there.

[time-nuts] powering Trimble Thunderbolt with -5V rather than -12V

[Stewart Cobb]

Executive summary:  you can power a surplus gold Thunderbolt using a
-5V supply in place of a -12V supply, and it will probably work just
fine.

Details:  The manual for Trimble Thunderbolts specifies power supplies
of +5V, +12V, and -12V.  It turns out that power supplies that provide
+5V, +12V, and -5V are easier to obtain locally.  I began to wonder
what circuitry in the Thunderbolt required -12V, and whether it would
run just as well on -5V.  So I took one apart and started probing.

As far as I can tell, the -12V supply goes to only two places.  One is
the negative supply pin for the quad op-amp (LT1014) in the DAC
circuit for the OCXO.  The other is a strange little circuit involving
a 2N3904 (SOT-23 marked 1A) near the 232 driver chip, right next to
the serial port.  This circuit seems to be comparing the -12V input
with one of the charge-pump pins on the 232 chip.  Its output (?)
connects to a test point labeled MON.  I assumed this was
non-critical and decided to ignore it.

The LT1014 op-amp is rated for operation on supply voltage as little
as 5V and as much as 30V (+/- 15V).  The spec sheet says the output
saturates about (1V typical / 3.5V max over temperature) above the
negative supply.  Presumably, if the op-amp is not asked to generate
output voltages lower than -1.5V, it should run fine with a -5V
negative supply.
The only negative voltages I could find, probing around the op-amp
circuit, were generated by AC-coupling digital square waves.  None of
the op-amp outputs were negative.  (My DAC steady-state value was
around +300mV, which appeared many places in the circuit.  Presumably
a slightly negative DAC value would also appear in many places, but as
long as it's greater than -1500mV, it won't matter.)

Armed with theoretical and practical confirmation that this should
work, I tried it.  And, oddly enough, it appears to be working.  Two
different Thunderbolts have been powered by +5/+12/-5 supplies, and
both have settled down and started tracking exactly as one would
expect.  For one, the settled DAC voltage was within a few
millivolts of the value it had on the specified power supplies,
shortly before the change.  The other had not been powered on for a
while and is still settling, but it seems happy.

There is a subtle possibility for concern, in that the sensitive DAC
signals near ground are now about 3.5V away from the center of the
op-amp supply range.  This could theoretically cause increased
distortion, offset, or offset drift due to the larger common-mode
voltage on the op-amp inputs.  In practice, it does not appear to be
an issue.

This note applies to the common surplus Thunderbolts in the
gold-anodized box, with the Trimble-branded OCXO.  All of those I've
tried seem to settle with DAC voltages near zero.  If you try this
with another style of Thunderbolt, you're on your own.

Cheers!
--Stu
___
time-nuts mailing list -- time-nuts@febo.com
To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
and follow the instructions there.

[time-nuts] Trading stocks faster than light

[Stewart Cobb]

www.nanex.net/aqck2/4436.html

Financial markets in Chicago and New York reacted to news from Washington
-- milliseconds faster than the news could have reached them.

Includes at the end some discussion of news organizations that aren't so
good about synchronizing their master clock ...  Historically, these news
services have shown a time range of about 30 milliseconds ...  Sounds like
they need time-nuts on staff.

Cheers!
--Stu
___
time-nuts mailing list -- time-nuts@febo.com
To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
and follow the instructions there.

[time-nuts] [OT] a different sort of time-nut

[Stewart Cobb]

http://www.lawoftime.org

The 13 Moon / 28-day calendar (synchronometer) is the key that allows us
access to the vast realm of the synchronic order ... This is a Galactic
timing program that also serves as a master synchronization matrix that any
other calendar or numerical system can be plugged into.

Hmm. What are the pinouts on those plugs? Do they use leap seconds to
synchronize galaxies, or do they have a more elegant method?

When you think about it, though, many of us are doing our best to
synchronize the vibrations of carefully prepared quartz crystals to beams
of pure light attuned to the energy levels of specially selected atoms
flung from the hearts of dying stars. That's like, cosmic, man.

So maybe those 13-moon guys would fit right in, after all.

Cheers!
--Stu
___
time-nuts mailing list -- time-nuts@febo.com
To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
and follow the instructions there.

[time-nuts] GPS antenna splitter buyers' guide

[Stewart Cobb]

(If this shows up twice, I apologize.  Gmail seems to have switched me to
sending HTML-formatted mail by default, and the time-nuts server doesn't
like that. Anyone know how to switch back permanently?)

It's possible to use standard microwave power dividers as GPS antenna
splitters, but they have several drawbacks.  Management of DC to the
antenna LNA creates one set of problems. Insertion loss and lack of RF
isolation between outputs can present additional problems.  In professional
settings, it's common to use purpose-built GPS antenna splitters.

HP/Agilent/Symmetricom GPS antenna splitters are fairly common on the
auction site.  Unfortunately, they were designed and specialized for
all-Agilent systems and they don't necessarily play well with others. You
can get GPS splitters that work better with a mix of GPSDOs, and generally
pay less as well because you're not buying the Aglient name.

The two most common (in the US) makers of purpose-built GPS antenna
splitters are GPS Source and GPS Networking. Their devices are more or less
interchangeable. One of these (the latter, I think) was formerly known as
WR Incorporated.  Googling WR no longer gets you any useful information, so
stuff with that name tends to be cheaper at auction than the others.  (Full
disclosure: I know several of the principals in this business.  But I'm
sure they'd want me to tell you to buy their stuff new rather than used.)

Purpose-built GPS splitters generally consist of an RF power divider and a
method of managing the DC bias sources for the antenna LNA.  They often
contain an RF amplifier to make up the loss in the power divider itself (10
dB for an 8-way splitter) and the loss in a potentially long cable run from
the antenna.

Some models are built with high RF isolation between ports.  This is done
by adding (say) 20dB to the amplifier gain, and then inserting a 20dB
attenuator between the power divider and each port.  This adds 40 dB of
isolation between ports, which keeps the various receivers from interfering
with each other.  (Some common receivers through the 1990's would leak high
levels of local oscillator signals out their antenna ports, which would jam
other receivers connected to the same antenna.)

If the splitter says a receiver port is DC blocked, that port is
generally connected to ground through an RF choke and a resistor of 200 to
500 ohms.  This keeps the antenna fault detector on many receivers happy.
If the port says DC thru, that port is DC-connected to the antenna port
to power the antenna LNA. That port will generally also power the
splitter's internal amplifier, if it has one. Sometimes the combined load
can overload the receiver's DC output current limit.

It's common for larger splitters to have a separate power source
(wall-wart) to power the antenna and the internal amplifier.  In this case,
all the receiver ports will be DC blocked.  This configuration is most
convenient for a time-nut lab, so that you can swap out any receiver you
want without interrupting the others.

These devices are fairly simple internally, and the various ordering
options (amplifiers, isolation, DC blocking, etc) were typically done by
stuffing the correct set of parts onto a common PC board. The ones I've
worked on all had a two-layer PCB, with parts and microstrip on one side
and mostly ground-plane on the other side.  If you find one for sale that's
not quite configured the way you want, you can probably reconfigure it with
a bit of thought and some surface-mount soldering.  The cases come apart
easily, with conventional screws along the sides, sometimes hidden under
the labels.

These splitters were available new with SMA, TNC, and N connectors, and
sometimes BNC as well.  BNC connectors can often be replaced with TNC
connectors without too much trouble (and should be replaced; BNC connectors
are not suitable for precision GPS installations).

If possible, get a splitter with one or two more ports than you think you
will need.  There's always another GPSDO to test, and you may be able to
borrow some professional survey equipment and connect it to your spare
antenna feed to get an accurate antenna position.

Hope this helps.

PS: The irony of doing a post like this is that more people will bid on
these things now that they know more about them, and prices will rise.  But
I already have all the splitters I need :)
___
time-nuts mailing list -- time-nuts@febo.com
To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
and follow the instructions there.

Re: [time-nuts] Austron 2200 Y2K?

[Stewart Cobb]

The ST2010 GPS receiver chip took 10 MHz and a GPS L1 signal in, and
produced a 35.42 MHz second IF to drive a SAW filter. (It also digitized
the SAW filter output, but you wouldn't need that part.)

These chips were designed by Plessey, which sold the business to Mitel,
which sold the business to Zarlink, which finally discontinued them a few
years ago.  But you may still be able to find the receiver chips.

One of those chips, plus the SAW filter, plus a fast op-amp to convert the
differential filter output to single-ended and drive it back down the
cable, would be all you'd need.

The CMC All-Star receiver board used those parts, and I think the SuperStar
did, too. I could probably dig up a few of those boards from my collection,
if anyone wants to strip the parts off to build an antenna converter.

Cheers!
--Stu

 Date: Mon, 10 Jun 2013 08:28:55 -0400
 From: paul swed paulsw...@gmail.com
 To: Discussion of precise time and frequency measurement
 time-nuts@febo.com
 Subject: Re: [time-nuts] Austron 2200 Y2K?
 Message-ID:
 
cad2jfaiafdhaqp+-k1eo6hvwixbdj74gigazmekmtp5atyw...@mail.gmail.com
 Content-Type: text/plain; charset=ISO-8859-1

 I do not have an antenna for my unit. But in fact homebrewed an adapter
for
 the Odetics antenna and then an alternate from a starlink gps rcvr that I
 use today. Not pretty but does work and frankly the method could be used
 with any gps rcvr that allows a 10 Mhz drive and has a 35.42 Mhz IF. So
 there is an answer.
 Regards
 Paul
 WB8TSL
___
time-nuts mailing list -- time-nuts@febo.com
To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
and follow the instructions there.

[time-nuts] HP 53132A time interval measurement quantized?

[Stewart Cobb]

Guys,

My HP 53132A counter works fine as a frequency counter.  When I try to use
it to measure time intervals (say, between two PPS pulses), the
measurements are quantized in steps of 100ns.  For example, if the two PPS
are nearly aligned, the measurements will be either 0.0980us or -0.0020us.
If the PPS are drifting with respect to each other, there will be a long
string of (say) 0.100 000 0980 sec, and then another long string of 0.100
000 1980 sec, with no intermediate values.  (I can send a TimeLab .TIM file
showing this behavior to anyone interested.)  This quantization appears on
measurements using both front-panel inputs (1 and 2) and on measurements
using only input 1 and the COMMON: ON setting enabled.

According to the manual, the time-interval measurement resolution spec is
150ps.  The quantization effect on my counter is about 3.5 orders of
magnitude worse than that.  That made me wonder whether there was a problem
with the counter.

The 100ns quantization is the reciprocal of the 10 MHz timebase frequency,
which seems like a clue.  One might suspect that the internal interpolators
were not working correctly.  However, the counter passes all its self-tests
including the specific interpolator self-test, and it seems to interpolate
correctly in frequency mode.  I ran the internal time-interval calibration
procedure CAL: TI QUICK with no errors, using a fast CMOS logic output
for the required square-wave input.  (Before my QUICK cal, the
measurements were reported as 0.0965us and -0.0035us, so the cal procedure
did actually change something, but it didn't fix the quantization
problem.)  I can't run the CAL: TI FULL procedure because it requires a
specific HP calibration fixture.  The procedure in the operators manual
procedure for time-iinterval measurements doesn't seem to contain any
tricks.  Making TI measurements with and without an external timebase
connected produces the same quantized results.

I'm trying to figure out whether this quantization effect is expected
behavior, operator error, or a problem with the counter.  Any thoughts?

Cheers!
--Stu
___
time-nuts mailing list -- time-nuts@febo.com
To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
and follow the instructions there.

Re: [time-nuts] Z3815A

[Stewart Cobb]

Your Z3815A may need more cooling than it's getting, especially if you have
it resting horizontally as it looks like it should.  I bought one of
those kits from China about a year and a half ago.  Powered it up, waited
for it to lock ... and it was dead within a week, with the unmistakable
smell of overheated electronics.  One of the Vicor power bricks inside
(probably) overheated and shorted out.  Here's what I learned:

The Z3815A board was designed to go into a VXI-like mainframe, with a
carefully specified amount of cooling airflow.  I think that particular
board was designed to require that airflow, and overheats without it.
There's a group in Australia which has experience with these boxes, and I
got the impression from my contacts with them that they see the Vicor
bricks fail pretty regularly.  That would imply that they're not getting
the cooling they need, because Vicor bricks in other applications are
pretty reliable in my experience.

You can see a photo of the original Z3815A on TVB's website here:

http://www.leapsecond.com/museum/z3815a/

The Z3815A I got from China was in a different case, just two bent pieces
of sheet aluminum.  The case _looks_ official, with the right label on the
front and silkscreen on the back.  But the board inside had a lot more crud
and corrosion than the nice clean case did, and parts of the plastic edge
connector on the back of the board were broken.  Worst of all, the coaxial
cable from the antenna connector ended in a one-inch flying lead soldered
to the board.  The shield of the coax cable ended in another flying lead,
soldered to ground somewhere else.  (Any RF engineers reading this are
probably cringing now.)  I'm pretty sure that no one at HP designed or
approved that connection.  Once I saw it, I understood why the GPS receiver
appeared to be deaf.  Even connected to a very good antenna, it never saw
more than 4 satellites, and even those had weak signals.

Did someone in China find a cache of bare Z3815 boards in a scrapyard
somewhere, and fab an official-looking case to match?  I don't know, but it
might be the way to bet.  Meanwhile, take the lid off your Z3815A and feel
the heatsinks on the power bricks.  If they're too hot to touch, they're
too hot; give them some air.

Cheers!
--Stu
___
time-nuts mailing list -- time-nuts@febo.com
To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
and follow the instructions there.

Re: [time-nuts] GPS position survey

[Stewart Cobb]

A subtle point:  The GPS satellite orbits are controlled so that their
(nominal) ground tracks precisely repeat every day.  To make this work,
their day is not a standard 24-hour day but a sidereal day lasting 23
hours, 56 minutes, and about 4 seconds.  The satellites actually go around
the earth twice in that time, so their orbits take 11 hours and 58+
minutes, but the earth is only halfway around after the first orbit, so the
ground tracks only repeat every other orbit.

The ground tracks repeat so that multipath errors can be averaged out by
taking data in multiples of one sidereal day.  If you want to get the most
accurate position by averaging satellite data, do your average in multiples
of 86164 seconds (sidereal day) rather than 86400 seconds (24 hours).  It
also helps to take data during a time when the weather is relatively
constant.  In particular, avoid precipitation.

Cheers!
--Stu
___
time-nuts mailing list -- time-nuts@febo.com
To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
and follow the instructions there.

[time-nuts] Precise positions for GPSDOs

[Stewart Cobb]

A GPSDO typically makes the assumption that the position of its antenna is
fixed and well-known. That removes position uncertainty from the navigation
equations, and allows all the information from the satellite measurements
to be used to improve the time estimate. Errors in this position create
errors in timing, with a magnitude scaled by the speed of light (one ns per
foot, three ns per meter).

Most GPSDOs do some sort of position averaging when they are first turned
on, to come up with a good-enough estimate of antenna position. For a true
time-nut, that might not be good enough.

GPS surveying equipment can easily determine the position of your antenna
to within a few centimeters (~20 ps). Unfortunately, such equipment is
expensive and difficult to borrow.

A high-end GPSDO designed today should have the ability to record phase
data into RINEX files, which could be sent to a service like OPUS to find
the antenna position.

http://www.ngs.noaa.gov/opus/

But few do, so far.

The next best idea is to locate your antenna on Google Maps. Type in the
self-surveyed position to the Google search box, either as decimal degrees
or as DMS, formatted like this but without the quote marks:

37.384542, -122.005526

37 23 4.35, -122 0 19.89

Click on the map and zoom in. Click on the Map box in the upper right and
uncheck the 45 degree view icon. Then right-click on the spot on the
picture where your antenna is actually located, and select What's here?
from the pop-up menu. A green arrow marker will appear, pointing to your
antenna. Left-click on the arrow, and read your latitude and longitude in
both formats. Enter one of them into your GPSDO, replacing the self-survey,
and enjoy increased accuracy.

A true time-nut will take one more step to improve accuracy. (Sorry, but
the rest of this is specific to North America. Similar details apply to
other parts of the world, but I only know the recipe for the place I live.)

Google Maps photos are registered (quite accurately) to the North American
Datum NAD83. Unfortunately, your GPSDO operates in a different datum
known variously as WGS84, ITRF, or IGS (these are all essentially the
same). The difference between these two datums can be a couple of meters,
easily visible on the map photos and worth 5 ns or more of time error.
Fortunately, you can convert NAD83 to ITRF2008 at this website:

http://www.geod.nrcan.gc.ca/apps/tmobs/tmobs_e.php

For ITRF epoch, just enter today's date. For ellipsoidal height, use
the value from your self-survey if you don't have a better one. You might
be able to get a better one from Google Earth, or by finding a nearby
benchmark from this site (US only) and extrapolating to your antenna
location.

http://www.ngs.noaa.gov/cgi-bin/ds_radius.prl

Note that the WGS84 ellipsoid is tens of meters higher than sea level
through most of North America, so if you live near the ocean, your
ellipsoidal height will probably be negative.

Hope someone find this useful.

Cheers!
--Stu
___
time-nuts mailing list -- time-nuts@febo.com
To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
and follow the instructions there.

Re: [time-nuts] FTS-4040A cesium, and STEL-1173 NCO chip

[Stewart Cobb]

Interesting, I have a problem Datum 4040A as well. I purchased it several
years ago and when it fail to work correctly I put it aside.  I wonder if
mine has the same problem ...

You could try powering it up and checking its telemetry with the Monitor3
program available on the Symmetricom website.  If it's like mine, after
it's fully warmed up, the DAC gain and numeric gain will be maxed out,
and there will be a bunch of alarm codes. One of them will be alarm 11. The
Monitor3 program (which was designed for a newer model of cesium) will not
translate that code, but the 4040A manual says it means 12.6 MHz signal
power low.  If you get that error, it's worth probing the lines between
the STEL chip and the DAC. Mine were all stuck low.

The STEL chip, its DAC, and related filters and amplifiers are all just
under the top cover of the unit, on the far right of the PCB as seen from
the front.

Units with serial numbers lower than 2000 apparently used a different
frequency synth scheme that does not include an STEL chip.

Cheers!
--Stu
___
time-nuts mailing list -- time-nuts@febo.com
To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
and follow the instructions there.

[time-nuts] (Datum) FTS-4040A cesium, and STEL-1173 NCO chip

[Stewart Cobb]

Guys,

I'm working on a sick FTS-4040A cesium frequency reference, which is
basically a box and power supply wrapped around the FTS-5045 cesium beam
module.  (Datum bought FTS, then Symmetricom bought Datum, but this 1995
device is too old to appear on the Symmetricom website.

This was one of the first digital cesiums.  It uses a microcontroller to
drive a frequency synthesizer to sample many different points in the cesium
tube's frequency response.  Part of the synthesizer is a 12.6+ MHz DDS,
built from a Stanford Telecom STEL-1173 digital numerically controlled
oscillator (NCO) chip and an Analog Devices fast D/A chip.
In my unit, the STEL chip seems to be broken.  All the inputs seem to be
correct, but the output bits to the D/A are all continuously low, so there
is no 12.6 MHz signal and thus no lock.  The unit also throws a fault code
11, which translates as 12.6 MHz signal power low, confirming this
diagnosis.

Web searching gives the impression that these STEL-1173 chips are known to
be fragile.  There was a reference on time-nuts in 2006 to repairing a
cesium by replacing the STEL chip, In 2009, a Steve Swift who read the 2006
thread offered to sell some STEL chips to the thread originator, but the
deal (if there was one) was conducted off-list and I can't find valid
contact information for Steve Swift.

So it's time to consult the hive mind:

1) Are these STEL-1173 NCO chips known to fail early?  Are there ways to
baby them so they don't fail early?

2) Does anyone know a source for replacement STEL-1173 chips?

3) Is there any detailed documentation (schematics, procedures, test
points, etc) available for the FTS-5045 cesium module?  (I have the 4040A
operators manual, but it doesn't show much detail.

4) Is there a way to tell whether the cesium beam tube itself is good,
without the frequency synth working?  All other telemetry points (ion pump,
electron multiplier, etc) seem to settle to nominal values after warmup, so
I have some hope that the tube is still good.

5) Any other suggestions or hints for repairing this unit?

Cheers!
--Stu
___
time-nuts mailing list -- time-nuts@febo.com
To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
and follow the instructions there.

[time-nuts] DATUM 9390-52054 Grief again...

[Stewart Cobb]

 I do not recognize the GPS receiver module, but it has the following
 number on it:  TNL 22880-B.  I have the schematics for the overall
 DATUM 9390-25054, but the GPS module is just a block.  By the way,
 the GPS block on the DATUM overall schematic is marked, SV6 /
 (TANS).  I suspect this means something noteworthy.

TNL is Trimble Navigation Labs (their original name). TANS is Trimble
Advanced Navigation Sensor, a GPS receiver designed circa 1989 for UAVs and
similar mil/aero applications.  Almost certainly has six GPS channels.
Generally spoke a version of TSIP, like the Thunderbolts. Often RS-422.

SV6 as someone else mentioned, is another Trimble 6-channel receiver from
the same era. I think it was more commercial than military.  Probably also
spoke TSIP.

22880 is the Trimble part number for the board you have, rev B. Trimble
used random 5-digit part numbers until very recently, when they had to go
to six digits because they'd used up all the 5-digit numbers. That only
took about 25 years. :)

You might be able to hook up Lady Heather to snoop the receiver
communications on one of your good Datum units. If that works, you might be
able to hot-wire a Thunderbolt in place of your dead 22880 board (assuming
it really is dead).

Cheers!
--Stu
___
time-nuts mailing list -- time-nuts@febo.com
To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
and follow the instructions there.

Re: [time-nuts] GPS usable for weather forecasting?

[Stewart Cobb]

 I wonder if you cannot do this same work from the ground.  Has anyone
tried
 tracking single GPS satellites from the ground using very high gain
 tracking antenna.

Many times. USAF does this each time they launch a new GPS satellite, to
check out all the kit in a high-res view before they switch it on for
general use. They used to use an antenna at Camp Parks in the California
central valley. When that one was being overhauled a few years ago, they
used Stanford's big dish for a while. It gives about 60 dB gain at L1,
IIRC.

Hardcore GPS researchers have used that dish and a bunch of others over the
years. If you're interested, contact SRI. They used to charge a couple of
hundred bucks an hour for the Stanford dish.

Hard to use for weather forecasting, though, because you can only see one
tiny chunk of atmosphere at a time.

Cheers!
--Stu
___
time-nuts mailing list -- time-nuts@febo.com
To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
and follow the instructions there.

Re: [time-nuts] Need info on Trimble 4000S GPS Surveyor

[Stewart Cobb]

The fundamental frequency in the GPS system is 1.023 MHz, chosen so that a
10-bit maximum-length feedback shift register (MLFSR) clocked at that
frequency would complete one cycle of 1023 states in precisely one
millisecond.  The civilian C/A codes or Gold codes are generated by
combining two such MLFSRs.  This fundamental frequency is often called
f0, pronounced f-naught.  It's just far enough away from 1.000 MHz to
get you into trouble if you mistake one for the other.

The military P/Y-code signal is transmitted at precisely ten times that
rate, or 10.23 MHz.  Some military-focused documentation calls that
frequency f0 instead.

The L1 frequency is 1540 * f0 = 1575.42 MHz.  The L2 frequency is 1200 *
f0, or 1227.6 MHz.  Many civilian receivers use an IF of 4 * f0 = 4.092
MHz.  This requires an LO frequency of 1536 * f0.  I'm not familiar with
the 4000S, but some Trimble receivers of similar vintage (late 80's, early
90's) used a LO at 768 * f0 with a sub-harmonic image-reject mixer to
produce the 4* f0 IF frequency.


 The L1 1575.42 MHz chain uses a 16.368 MHz VCXO locked to the 10 MHz
 reference, running an LO of some integer multiple that results in a
 reference around 38.4 MHz labeled ECL 38.4 F0 on the main board, and
 an unlabeled signal IF called TTL LIMITER. Internal markings 768 and
 384 may indicate PLL IFs of 76.8 and 38.4 (76.8/2) MHz.


38.4 F0  almost certainly  means 38.4 * f0.  This would produce a digital
clock fast enough to sample P-code data.  (Recall that the P-code wasn't
encrypted to Y-code until about 1990, so this receiver was probably built
to track P-code.)


 The L2 1227.6 MHz chain uses a 28.644 MHz VCXO locked to the 16.368 MHz
 reference, and LO that results in another unknown IF that runs through a
 similar TTL limiter. It appears that the LO is an integer multiple of
 the 28.644 MHz, with a PLL IF possibly around 59.2 MHz, marked 592 FO.
 Only the unknown signal IF from this section goes to the processing
 boards - no PLL IF seems to go beyond these modules. The unknown signal
 IF goes only to one of two apparently identical DSP boards, unlike the
 others that all go to the main board.


If 592 F0 were used in a subharmonic mixer as described above, simulating
an LO of 2 * 592 = 1184 * f0, that would produce an L2 IF signal at 16 * f0
= 16.368 MHz, which is a high enough IF frequency to carry the P/Y
modulation (+/- 10 f0 = 10.23 MHz) without aliasing.  (Not sure that's
what's going on, but it's a plausible explanation for your observations.)


 The L1 downconverter appears to use quadrature mixing, but I can't tell
 what happens after that - the I-Q signals go into a bunch of baseband
 circuitry. The L2 one also has a quadrature mixer, but only one output
 goes into its baseband circuits - the other is just terminated.


The L1 signal carried P-code and C/A code in quadrature.  The L2 signal
only carried P-code.  Sounds like the L1 outputs are split between the
P-code and C/A code processors.  The L2, carrying no C/A code, only goes to
the P-code processor.


 As I
 understand, the L2 is always encrypted, so useless for data, but its
 carrier can be used to enhance overall accuracy - I recall studying that
 a few years ago, but forgot the details. So, maybe the L2 portion is
 only for carrier recovery of some sort.


As above, P-code (on L1 and L2) was not encrypted when this box was
designed.


 I'd appreciate any info or ideas on deciphering the rest of the way -
 maybe the modules will be useful for something as a system, rather than
 just parts. I'm especially interested in GPS carrier recovery techniques
 for frequency only - not time.


The GPS signals are spread-spectrum, well below the thermal noise level, so
you can't easily see the carriers without actually tracking the signals.

Hope this helps.

Cheers!
--Stu
___
time-nuts mailing list -- time-nuts@febo.com
To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
and follow the instructions there.

[time-nuts] repairing General Technology (Tracor) 304-B rubidium standard

[Stewart Cobb]

Guys,

I'm repairing a 1960's vintage lab-grade rubidium standard, General
Technology Corporation model 304-B.  Apparently Tracor bought GTC soon
after this unit was made, because references to this as a Tracor 304-B
seem to be more common.  I've made some progress, but now it seems like
time to consult the hive mind.

The unit appears clean, but it doesn't lock.  I've read through old
comments on the list regarding this unit, and I've downloaded a copy of the
manual and schematics available at

*http://sundry.i2phd.com/ServiceManual_304b.pdf*

That file seems to contain a complete copy of the manual text, but some
schematics are missing.  In particular, the schematics for the
sweep/acquisition board (A8) and the three boards inside the physics
package (the lamp oscillator (A13), the SRD driver (A12), and the photocell
preamp (A11)) are not shown.  Does anyone know where to find copies of
those schematics?

The main power supply voltage on my unit seems to have been deliberately
adjusted lower than spec (18.54 V actual, versus 20 +/- 0.1V specified in
the manual).  Replacing a resistor on the regulator board (that had smoked
from overload due to the low voltage) didn't change the voltage much.  I
had to crank the trimmer across half of its range to get the voltage back
within spec.  Nothing in the regulator circuitry seemed to have drifted
enough to change the setpoint that much.  Is there a reason why a tech
would have deliberately set this voltage lower than spec, or did it just
drift down over the years?

A frequency counter (GPSDO reference) shows that the crystal oven warms up
as expected.  The output can be centered on 5 MHz and the sweep circuit
covers a symmetrical range around 5 MHz as expected.  The ovens for the
lamp and filter cell appear to warm up properly as well, judging from test
points available on the A1 oven controller board.  The test point voltages
don't quite match the ones in the PDF manual, but it looks like those
readings were typed into each individual manual after being read off the
particular unit that came with that manual.

The test point on the A5 board shows that 155 Hz resonance detector
modulation is within spec.  The A6 filter-amplifier board test points show
the system attempting (and failing) to detect 155 Hz and 310 Hz resonance
signals coming back from the photocell.

The manual says that the A7 RF pre-driver board (the x14 multiplier) should
be supplying 70 MHz at +13 dBm to the SRD driver inside the physics
package.  That would be about 2.8Vpp, assuming a 50-ohm system.  Instead,
it's supplying a clean 70 MHz at about 100mV into a 50-ohm load.  My best
guess is that the final amplifier transistor on that board is blown,
possibly from being operated with only a scope probe as a load (infinite
VSWR).  Replacement transistors are on order.  Any other thoughts?

Obviously, the box won't lock until the RF input is the right level.  But
it also requires the Rb lamp to light.  Corby Dawson posted to the list
back on 12 November 2009:

Tracor bulbs fail with a different mechanism and last maybe 10 years.

Anyone know what that different failure mechanism is?  Is it repairable
in an ordinary lab, like the heat-gun trick for LPRO bulbs?  If not, is it
feasible to build a Frankenstein replacement using something like an LPRO
or FEI bulb?

Is it possible to tell whether the lamp is lit without opening the physics
package?  If not, are there any tricks to opening the physics package?  Any
precautions to take before doing so?

Any other comments on how to get this box working again?

Cheers!
--Stu

Side note:  This unit was built during the era of elastic seconds
(roughly, the 1960's).  It contains a board (A9) which digitally offsets
the output frequency in increments of roughly 7E-10, without changing the
rubidium resonance frequency or the C-field.  There's also a note in the
manual saying that annual changes to the definition of the second may
require replacing the rubidium resonance cell in the physics package with a
new cell calibrated for the new second in the new year.  Leap seconds bring
their own problems, but compared to dismantling your lab instruments every
year, they're a breeze.
___
time-nuts mailing list -- time-nuts@febo.com
To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
and follow the instructions there.

[time-nuts] Z3805A cooling requirements?

[Stewart Cobb]

This may be a newbie question, but I'm a newbie, so:

Do the HP telecom GPSDOs (Z38xx) require external airflow for cooling?
They don't have built-in fans, but they sorta look like they depend on a
rack-level cooling fan, which a telecom rack would almost certainly have.

I ask because I bought a Z3816 awhile back which worked for about a week
and then failed. I traced the failure to an internal power supply brick,
which had a big finned heat-sink attached but nevertheless smelled
overheated and was shorted internally.

I never found a replacement power brick, and I don't have time to mess with
it right now, so I recently bought a Z3805A. It, too, looks like it's
working, but it started to feel awfully warm after a few hours, so I
unplugged it for now.

It probably wouldn't take much of a fan to bring the internal temperature
down close to ambient, and the fan could be powered easily enough from the
supply rails. But that might create a temperature gradient where the
designers didn't intend one. Or it might cause problems I don't even know
about yet.

At the moment, the Z3805A is in a fan-less 19-inch rack with a bunch of
other equipment, in a lab environment. Should it have its own fan?

Cheers!
--Stu
___
time-nuts mailing list -- time-nuts@febo.com
To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
and follow the instructions there.