Re: [Flexradio] A question on Frequency stability vs. Temperature

[Eric Ellison]

Bob (and anyone else into this bullseye at microwave type thing!)

Anyone interested I will be around on Teamspeak at 0100 tomorrow night to
brainstorm this neat stuff! Forum is It's about Time


Eric


-Original Message-
From: Bob Tracy [mailto:[EMAIL PROTECTED] 
Sent: Sunday, January 01, 2006 11:05 AM
To: Eric Ellison; [EMAIL PROTECTED]; FlexRadio@flex-radio.biz
Subject: RE: [Flexradio] A question on Frequency stability vs. Temperature

Eric et al,

Thought this tread might be a good place to post an update on the oscillator
project.

I assembled the PLL using the card I designed from Joe Ruggieri's (the eBay
source) schematic.  I got the GPS antenna up and got the Jupiter receiver
working.  I initially tried to use my Kenwood IC-232 level converter on the
Jupiter data interface but that was unsuccessful.  I resolved the problem
with a very simple circuit using a MAX232.

Using some free software (GPS Diagnostic
http://www.commlinx.com.au/gps_diag.htm) I verified I was getting good GPS
signals.  At my location and with the antenna mounted just outside the shack
window, I am seeing from 3 to 8 satellites at any given time.  Hooking all
the various components together appears to produce a well-locked 10 mHz
signal.  I say appears because I have nothing better than WWV and a DVM on
the OCXO control line to check the accuracy.

Now the bad news.  I picked up an external clock kit from Flex and tried to
install it in my SDR-1000.  First thing I found was that there have been
hardware changes that affect how the kit is installed.  I don't know
precisely when the changes were made but my two year old unit has a PA
circuit board connector exactly lined up with and adjacent to the mounting
hole for the external clock BNC connector.  Gerald said to move it to one of
the transverter BNC holes (which is OK but I really plan to install the 2
meter transverter at some point in time).  Next, I found that the RG-174
coax furnished with the kit was about an inch too short to reach the new BNC
location so I lost another day finding some RG-174.

Connecting the SDR-1000 to the external clock resulted in severe signal
distortion.  The OCXO supplies a 1.7 V P-P sine wave into a 50 ohm load
which is within the Flex specification.  All the internal jumpers were
correctly installed and the software setup was modified as required.
Converting the SDR-1000 back to internal oscillator gets rid of the
distortion so I'm confident the problem is somewhere in the oscillator
control circuit.

Today I'm going to go back to square one.  I am going to eliminate all the
circuitry in the GPSDO and start with just the 10 mHz oscillator to see if I
can isolate the source of the distortion.  I suspect it is some sort of
phase noise coming out to the PLL but I am going to start with just the OCXO
driving the SDR-1000 and go from there.  My second step will be to decrease
some ground lengths and add some better shielding to the 10 mHz and 25 kHz
circuit wiring.

I'll post my results as they occur.

Bob K5KDN



-Original Message-
From: [EMAIL PROTECTED]
[mailto:[EMAIL PROTECTED] Behalf Of Eric Ellison
Sent: Sunday, January 01, 2006 8:18 AM
To: [EMAIL PROTECTED]; FlexRadio@flex-radio.biz
Subject: Re: [Flexradio] A question on Frequency stability vs.
Temperature


Larry

The last I heard from Bill - KD5TFD on Teamspeak yesterday, was that his
weekend goal was to interface the Xylo to his GPS 1 pps, so he may beat us
all! Also I am interested in where he is going to pull the sigs out of the
SDR. No body commented on that message, but it it seemed like a better idea
than messing with the 200 mhz LO.

In the FPGA - SDR iteration, if we can do a long term average of the GPS
sigs then we might not even need a stable slaved 10 mhz standard.

In any case, I have ordered what I think is a pretty good deal from the guy
Bob - K5KDN is working with. He is selling these Oscillators on E-Bay for 75
bux and has a simple kit of parts for slaving to the 10 khz GPS sig.

http://cgi.ebay.com/New-Isotemp-10MHz-OCXO-1PPB-EFC-For-GPS-Discipline-12V_W
0QQitemZ5847210309QQcategoryZ48702QQssPageNameZWDVWQQrdZ1QQcmdZViewItem

On way or tother this type of circuit and osc is what non-fpga SDR-1000
owners could use for good stability with the existing modification kit from
FlexRadio.

Brainstorming session on Teamspeak next Wed night at 0100.

Eric2


-Original Message-
From: [EMAIL PROTECTED]
[mailto:[EMAIL PROTECTED] On Behalf Of [EMAIL PROTECTED]
Sent: Saturday, December 31, 2005 11:02 PM
To: FlexRadio@flex-radio.biz
Subject: [Flexradio] A question on Frequency stability vs. Temperature

Xyloonians,

Build a selectable output frequency standard referenced to GPS or WWV.  M/N
type thing.  Have its frequency output within the range of the soundcard by
looking at but offset from the frequency tuned to at the moment.  Maybe have
it outside the pan adapter window so it doesn't distract the operator.  Or
put it in the window with a red X to show

Re: [Flexradio] A question on Frequency stability vs. Temperature

[Jim Lux]

At 05:17 PM 1/1/2006, [EMAIL PROTECTED] wrote:



 Actually, you need to synthesize TWO signals in the passband, because 
errors in the A/D sample clock can't be calibrated out with a single marker.


Jim,

Well, I have to ask why two markers are needed or can you give a reference 
text I could study to see why?  Warning, KISS please as I can follow 
y=mx+b but after that I'm lost.


A change in the DDS frequency (or the reference) shifts the entire RF band 
around by the same amount (that is, if you're tuned to 14.100 and the DDS 
is (erroneously) at 14.101, then everything will be the same 1 kHz off, 
marker and desired signal).   Let's call that an additive frequency error


For a conventional receiver, additive errors are the only kind you'll see.


A change in the sampling clock changes the scale factor into the 
filtering.  A 10 kHz signal coming into the A/D and being sampled at 40 kHz 
has 4 samples per cycle.  If the sampling clock is 50 kHz, you'll have 5 
samples/cycle, which, if you're expecting 40 kHz sampling rate, would 
correspond to 8kHz.  Likewise, a 5 kHz signal would be digitized at 10 
samples/cycle instead of 8, so it would look like 4 kHz.  Here, the error 
is multiplicative (that is, the received signal appears to be at a 
frequency which is the real frequency multiplied by a constant;4/5 in this 
example)



In a system where you run data in and out at the same sample rate, small 
sample rate changes might not be noticeable (since the errors would 
counteract each other..).  However, if you are depending on the sample rate 
to be accurate: as in a digital demodulator, or in a self calibration 
system, you need to deal with it.


Errors in the sampling rate are kind of like having the marker spacing 
being wrong, i.e. the markers are every 26 kHz instead of every 25 kHz.


So, you can put in two tones (fa and fb) a kilohertz or so apart and 
measure them.  What shows up in the sampled output is two signals (f1 and f2):


f1 = k1*(fa+k2)
f2 = k1*(fb+k2)

where k2 is the error in the DDS frequency and k1 is the sampling rate 
error (actual rate/expected rate).  You know fa and fb (because you put 
them in) and you measure f1 and f2, and solve for k1 and k2.  It's actually 
pretty easy.


[ f1= k1*fa + k1*k2;  f2 =k1*fb + k1*k2
  f1-f2 = k1*(fa-fb)
  k1 = (f1-f2)/(fa-fb)


Say your radio's DDS is tuned to 14.100 MHz.  You put in a tone at 14.101 
and a tone at 14.102 (i.e. at 1000 and 2000 Hz relative to the DDS).


You measure the two tones in the audio stream and one tone shows up at 900 
Hz and the other shows up at 2100Hz (that would be the frequencies where 
they appear in the fft, assuming the sample rate is correct). The tones 
SHOULD be 1000 Hz apart, but they're not, they're 1200 Hz apart, which 
means that the sampling clock is running slow by 20%.  (k1 = 1.2 in the 
example above)


Now, apply that correction.. and the tones appear at 750 and 1750 (the 
proper 1 kHz apart), but are now both low by 250 Hz, which implies that the 
DDS frequency is high by 250 Hz.





Since two are needed then generate markers spaced every 25 KHz and just 
move the markers around so they aren't sitting on the center 
frequency.  Looks like with just a little more thought on the hardware we 
can push it off the end of the bench and let the softheads do the trench 
work.


Golly, product out the door next week.

73, Larry  K2LT


James Lux, P.E.
Spacecraft Radio Frequency Subsystems Group
Flight Communications Systems Section
Jet Propulsion Laboratory, Mail Stop 161-213
4800 Oak Grove Drive
Pasadena CA 91109
tel: (818)354-2075
fax: (818)393-6875

Re: [Flexradio] A question on Frequency stability vs. Temperature

[Jim Lux]

At 06:25 PM 1/1/2006, Tom Clark, W3IWI wrote:

KD5NWA wrote:

The project being discussed for the Xylo is;

1.  Thermal isolation with a accurate heater to reduce the drift.
2.  Use the Xylo to connect to a GPS and have a accurate 1 pps clock
3.  Use the 1 pps clock to measure the DDS 200MHz clock, integrate 
multiple reading to make for finer resolution.
4.  Use the 1 pps clock to measure the A/D clock, take multiple 
readings also.
5.  Feed the results to the PC who will proceed to correct the 
internal math so there is no drift.
Let me suggest a different approach at step 3. Connect the DDS reference 
to a long counter (more than one second) -- it can be straight binary, and 
32 bits is overkill. Use the 1PPS to strobe the counter into a register 
without stopping the counter (i.e. like lap timing with a stop watch). 
Then make your corrections based on accumulated phase, which can then be 
averaged over many seconds.


73, Tom



Doesn't even need to be a long counter.  you know approximately what the 
MSBs are going to be, all you're interested in is the LSBs. (i.e. whether 
there were 200,000,001 counts last second, or 199,999,999).  Figure out 
your worst case variation, and that's how long the counter needs to be.  Of 
course, in a modern FPGA, you're probably not short of counter stages, but 
you still have to pump all those bits back to the CPU, and most of them are 
superfluous.


The other thing to watch out for is the type of counter.  It needs to be 
synchronous, so that when you latch it, you aren't capturing it in mid 
ripple carry.  Sometimes, it's easier to set up some fixed dividers with 
relatively prime divisors and just latch the single bit of output from each 
divider, so you aren't as concerned about the state of ALL the flipflops in 
the divider. non power of 2 Ripple counter/dividers have jitter on their 
MSB output bit, so you might not gain much here, but for small divisors, 
you can use things like Johnson counters.  You can also clock a short PN 
generator and latch the state of the shift register (it's sometimes easier 
to make a synchronous PN generator than a synchronous counter).



Jim, W6RMK 

Re: [Flexradio] A question on Frequency stability vs. Temperature

[ltaft]

Jim,

Thank you for an excellent explanation of the two markers required.  I'll be 
back with the next question after I digest the information. 

What I need is an accurate measure of frequency in the AM broadcast band, 500 
to 1700 KHz, so I would think 1-10^7 is good enough for me.  It needs to be 
mobile, I can run it from my vehicle so the power requirement is trivial.  I 
can monitor GPS on the move, need it for navigation anyways.

Looks like my requirement has minimal intrusion into the SDR1000 box.  Just 
need the digital data and control for the marker freqs to the PC.  USB type 
link should work.  Stuff the signals into the QRP BNC.

The second requirement for the weak signal/microwave crowd use would be to get 
a lower phase noise source into the SDR1000.  Perhaps this would also be a good 
time to move to the new AD9950 series DDS as those specs seem to be a major 
improvement.  Lots to ponder.

Again, Thanks for your lesson.

73, Larry  K2LT
 
 From: Jim Lux [EMAIL PROTECTED]
 Date: 2006/01/02 Mon PM 02:45:09 WET

Re: [Flexradio] A question on Frequency stability vs. Temperature

[ltaft]

You and I are headed in the same direction.  I'm saying that the Xyloonia world 
should creat a known stable frequency within the passband of the soundcard that 
the PowerSDR can measure continuously to creat the offset needed to display the 
correct frequency.  Something like the old 100 kc xtal oscillator with a 25 kc 
divider such as Drake used in the 4 and the 7 series.  Only do the math as 
needed to creat a single frequency rather than fixed comb that would show up 
every 25 kc.  It really doesn't have to look at each oscillator, just the 
resultant signal.

Wow, Its 2006 now and I need to get my beauty rest.

73, Larry  K2LT  now a Numbhead
 
 From: KD5NWA [EMAIL PROTECTED]
 Date: 2006/01/01 Sun AM 04:27:55 WET
 To: [EMAIL PROTECTED]
 Subject: Re: [Flexradio]  A question on Frequency stability vs. 
   Temperature
 
 The scheme outline below in steps 1 through 5 does not correct the 
 frequency by changing the hardware oscillator, it does not require a 
 VCO to replace the current oscillator, it measures the errors and 
 feeds it digitally through a USB port to the PC who will make 
 software corrections.
 
 By measuring the 200MHz signal it will more accurately detect errors.
 
 
 At 10:02 PM 12/31/2005, you wrote:
 Xyloonians,
 
 Build a selectable output frequency standard referenced to GPS or 
 WWV.  M/N type thing.  Have its frequency output within the range of 
 the soundcard by looking at but offset from the frequency tuned to 
 at the moment.  Maybe have it outside the pan adapter window so it 
 doesn't distract the operator.  Or put it in the window with a red X 
 to show it is working.
 
 The PowerSDR software measures this frequency and compares it to the 
 Xylo frequency number, and adjusts the SDR display to correct 
 it.  This now takes into account all the oscillators in the SDR1000 
 and the soundcard and is continuously compensating.
 No need for VCO stabilizing the oscillators.
 
 All to You Year of the New Happy!
 
 Hardware designed by hardheads
 Software designed by softheads
 
 Larry - K2LT - I design hardware!
 
   From: KD5NWA [EMAIL PROTECTED]
   
   The project being discussed for the Xylo is;
  
   1.  Thermal isolation with a accurate heater to reduce the drift.
   2.  Use the Xylo to connect to a GPS and have a accurate 1 pps clock
   3.  Use the 1 pps clock to measure the DDS 200MHz clock,
   integrate multiple reading to make for finer resolution.
   4.  Use the 1 pps clock to measure the A/D clock, take multiple
   readings also.
   5.  Feed the results to the PC who will proceed to correct the
   internal math so there is no drift.
  
   No attempt will be made to change the physical 200 MHz clock to
   something else that is changeable, so phase noise is not
   affected.  The Xylo will measure, the PowerSDR software will use the
   data to correct it's internals, resulting in very accurate, and
   stable frequencies with minimal changes to the existing hardware
  
   That is my understanding on what the game plan is.
  
  
 
 
 ___
 FlexRadio mailing list
 FlexRadio@flex-radio.biz
 http://mail.flex-radio.biz/mailman/listinfo/flexradio_flex-radio.biz
 
 
 Cecil Bayona
 KD5NWA
 www.qrpradio.com
 
 I fail to see why doing the same thing over and over and getting the 
 same results every time is insanity: I've almost proved it isn't; 
 only a few more tests now and I'm sure results will differ this time ...  
 

Re: [Flexradio] A question on Frequency stability vs. Temperature

[Eric Ellison]

Larry

The last I heard from Bill - KD5TFD on Teamspeak yesterday, was that his
weekend goal was to interface the Xylo to his GPS 1 pps, so he may beat us
all! Also I am interested in where he is going to pull the sigs out of the
SDR. No body commented on that message, but it it seemed like a better idea
than messing with the 200 mhz LO.

In the FPGA - SDR iteration, if we can do a long term average of the GPS
sigs then we might not even need a stable slaved 10 mhz standard.

In any case, I have ordered what I think is a pretty good deal from the guy
Bob - K5KDN is working with. He is selling these Oscillators on E-Bay for 75
bux and has a simple kit of parts for slaving to the 10 khz GPS sig. 

http://cgi.ebay.com/New-Isotemp-10MHz-OCXO-1PPB-EFC-For-GPS-Discipline-12V_W
0QQitemZ5847210309QQcategoryZ48702QQssPageNameZWDVWQQrdZ1QQcmdZViewItem

On way or tother this type of circuit and osc is what non-fpga SDR-1000
owners could use for good stability with the existing modification kit from
FlexRadio.

Brainstorming session on Teamspeak next Wed night at 0100.

Eric2


-Original Message-
From: [EMAIL PROTECTED]
[mailto:[EMAIL PROTECTED] On Behalf Of [EMAIL PROTECTED]
Sent: Saturday, December 31, 2005 11:02 PM
To: FlexRadio@flex-radio.biz
Subject: [Flexradio] A question on Frequency stability vs. Temperature

Xyloonians,

Build a selectable output frequency standard referenced to GPS or WWV.  M/N
type thing.  Have its frequency output within the range of the soundcard by
looking at but offset from the frequency tuned to at the moment.  Maybe have
it outside the pan adapter window so it doesn't distract the operator.  Or
put it in the window with a red X to show it is working.  

The PowerSDR software measures this frequency and compares it to the Xylo
frequency number, and adjusts the SDR display to correct it.  This now takes
into account all the oscillators in the SDR1000 and the soundcard and is
continuously compensating.
No need for VCO stabilizing the oscillators. 

All to You Year of the New Happy!

Hardware designed by hardheads
Software designed by softheads

Larry - K2LT - I design hardware!
 
 From: KD5NWA [EMAIL PROTECTED]
  
 The project being discussed for the Xylo is;
 
 1.  Thermal isolation with a accurate heater to reduce the drift.
 2.  Use the Xylo to connect to a GPS and have a accurate 1 pps clock
 3.  Use the 1 pps clock to measure the DDS 200MHz clock, 
 integrate multiple reading to make for finer resolution.
 4.  Use the 1 pps clock to measure the A/D clock, take multiple 
 readings also.
 5.  Feed the results to the PC who will proceed to correct the 
 internal math so there is no drift.
 
 No attempt will be made to change the physical 200 MHz clock to 
 something else that is changeable, so phase noise is not 
 affected.  The Xylo will measure, the PowerSDR software will use the 
 data to correct it's internals, resulting in very accurate, and 
 stable frequencies with minimal changes to the existing hardware
 
 That is my understanding on what the game plan is.
 
 


___
FlexRadio mailing list
FlexRadio@flex-radio.biz
http://mail.flex-radio.biz/mailman/listinfo/flexradio_flex-radio.biz

Re: [Flexradio] A question on Frequency stability vs. Temperature

[Eric Ellison]

Bob

Excellent work! Sorry to hear that you are having the problems. Looks like
we might have to drill a hole in the case somewhere! Would a MAX 233 (?) be
better choice, can it go on a revised circuit board you designed? We are
definitely going to want to talk to the GPS board. Are you using the OCXO
that Joe has for sale on E-Bay?

Thanks
Eric


-Original Message-
From: Bob Tracy [mailto:[EMAIL PROTECTED] 
Sent: Sunday, January 01, 2006 11:05 AM
To: Eric Ellison; [EMAIL PROTECTED]; FlexRadio@flex-radio.biz
Subject: RE: [Flexradio] A question on Frequency stability vs. Temperature

Eric et al,

Thought this tread might be a good place to post an update on the oscillator
project.

I assembled the PLL using the card I designed from Joe Ruggieri's (the eBay
source) schematic.  I got the GPS antenna up and got the Jupiter receiver
working.  I initially tried to use my Kenwood IC-232 level converter on the
Jupiter data interface but that was unsuccessful.  I resolved the problem
with a very simple circuit using a MAX232.

Using some free software (GPS Diagnostic
http://www.commlinx.com.au/gps_diag.htm) I verified I was getting good GPS
signals.  At my location and with the antenna mounted just outside the shack
window, I am seeing from 3 to 8 satellites at any given time.  Hooking all
the various components together appears to produce a well-locked 10 mHz
signal.  I say appears because I have nothing better than WWV and a DVM on
the OCXO control line to check the accuracy.

Now the bad news.  I picked up an external clock kit from Flex and tried to
install it in my SDR-1000.  First thing I found was that there have been
hardware changes that affect how the kit is installed.  I don't know
precisely when the changes were made but my two year old unit has a PA
circuit board connector exactly lined up with and adjacent to the mounting
hole for the external clock BNC connector.  Gerald said to move it to one of
the transverter BNC holes (which is OK but I really plan to install the 2
meter transverter at some point in time).  Next, I found that the RG-174
coax furnished with the kit was about an inch too short to reach the new BNC
location so I lost another day finding some RG-174.

Connecting the SDR-1000 to the external clock resulted in severe signal
distortion.  The OCXO supplies a 1.7 V P-P sine wave into a 50 ohm load
which is within the Flex specification.  All the internal jumpers were
correctly installed and the software setup was modified as required.
Converting the SDR-1000 back to internal oscillator gets rid of the
distortion so I'm confident the problem is somewhere in the oscillator
control circuit.

Today I'm going to go back to square one.  I am going to eliminate all the
circuitry in the GPSDO and start with just the 10 mHz oscillator to see if I
can isolate the source of the distortion.  I suspect it is some sort of
phase noise coming out to the PLL but I am going to start with just the OCXO
driving the SDR-1000 and go from there.  My second step will be to decrease
some ground lengths and add some better shielding to the 10 mHz and 25 kHz
circuit wiring.

I'll post my results as they occur.

Bob K5KDN



-Original Message-
From: [EMAIL PROTECTED]
[mailto:[EMAIL PROTECTED] Behalf Of Eric Ellison
Sent: Sunday, January 01, 2006 8:18 AM
To: [EMAIL PROTECTED]; FlexRadio@flex-radio.biz
Subject: Re: [Flexradio] A question on Frequency stability vs.
Temperature


Larry

The last I heard from Bill - KD5TFD on Teamspeak yesterday, was that his
weekend goal was to interface the Xylo to his GPS 1 pps, so he may beat us
all! Also I am interested in where he is going to pull the sigs out of the
SDR. No body commented on that message, but it it seemed like a better idea
than messing with the 200 mhz LO.

In the FPGA - SDR iteration, if we can do a long term average of the GPS
sigs then we might not even need a stable slaved 10 mhz standard.

In any case, I have ordered what I think is a pretty good deal from the guy
Bob - K5KDN is working with. He is selling these Oscillators on E-Bay for 75
bux and has a simple kit of parts for slaving to the 10 khz GPS sig.

http://cgi.ebay.com/New-Isotemp-10MHz-OCXO-1PPB-EFC-For-GPS-Discipline-12V_W
0QQitemZ5847210309QQcategoryZ48702QQssPageNameZWDVWQQrdZ1QQcmdZViewItem

On way or tother this type of circuit and osc is what non-fpga SDR-1000
owners could use for good stability with the existing modification kit from
FlexRadio.

Brainstorming session on Teamspeak next Wed night at 0100.

Eric2


-Original Message-
From: [EMAIL PROTECTED]
[mailto:[EMAIL PROTECTED] On Behalf Of [EMAIL PROTECTED]
Sent: Saturday, December 31, 2005 11:02 PM
To: FlexRadio@flex-radio.biz
Subject: [Flexradio] A question on Frequency stability vs. Temperature

Xyloonians,

Build a selectable output frequency standard referenced to GPS or WWV.  M/N
type thing.  Have its frequency output within the range of the soundcard by
looking at but offset from the frequency tuned

Re: [Flexradio] A question on Frequency stability vs. Temperature

[Eric Ellison]

Bob

That is a slick little program. Nice get! I wonder if using VCOM you could
parse this output to do correction directly in software from the GPS as is
being suggested? Course we don't have the source for this program.

With a few mods to your board it looks like it could be the basis for the
Xylo frequency standard. Just need pick-off points on the board to go to the
xylo.

Thanks Again.
Eric


-Original Message-
From: Bob Tracy [mailto:[EMAIL PROTECTED] 
Sent: Sunday, January 01, 2006 11:05 AM
To: Eric Ellison; [EMAIL PROTECTED]; FlexRadio@flex-radio.biz
Subject: RE: [Flexradio] A question on Frequency stability vs. Temperature

Eric et al,

Thought this tread might be a good place to post an update on the oscillator
project.

I assembled the PLL using the card I designed from Joe Ruggieri's (the eBay
source) schematic.  I got the GPS antenna up and got the Jupiter receiver
working.  I initially tried to use my Kenwood IC-232 level converter on the
Jupiter data interface but that was unsuccessful.  I resolved the problem
with a very simple circuit using a MAX232.

Using some free software (GPS Diagnostic
http://www.commlinx.com.au/gps_diag.htm) I verified I was getting good GPS
signals.  At my location and with the antenna mounted just outside the shack
window, I am seeing from 3 to 8 satellites at any given time.  Hooking all
the various components together appears to produce a well-locked 10 mHz
signal.  I say appears because I have nothing better than WWV and a DVM on
the OCXO control line to check the accuracy.

Now the bad news.  I picked up an external clock kit from Flex and tried to
install it in my SDR-1000.  First thing I found was that there have been
hardware changes that affect how the kit is installed.  I don't know
precisely when the changes were made but my two year old unit has a PA
circuit board connector exactly lined up with and adjacent to the mounting
hole for the external clock BNC connector.  Gerald said to move it to one of
the transverter BNC holes (which is OK but I really plan to install the 2
meter transverter at some point in time).  Next, I found that the RG-174
coax furnished with the kit was about an inch too short to reach the new BNC
location so I lost another day finding some RG-174.

Connecting the SDR-1000 to the external clock resulted in severe signal
distortion.  The OCXO supplies a 1.7 V P-P sine wave into a 50 ohm load
which is within the Flex specification.  All the internal jumpers were
correctly installed and the software setup was modified as required.
Converting the SDR-1000 back to internal oscillator gets rid of the
distortion so I'm confident the problem is somewhere in the oscillator
control circuit.

Today I'm going to go back to square one.  I am going to eliminate all the
circuitry in the GPSDO and start with just the 10 mHz oscillator to see if I
can isolate the source of the distortion.  I suspect it is some sort of
phase noise coming out to the PLL but I am going to start with just the OCXO
driving the SDR-1000 and go from there.  My second step will be to decrease
some ground lengths and add some better shielding to the 10 mHz and 25 kHz
circuit wiring.

I'll post my results as they occur.

Bob K5KDN



-Original Message-
From: [EMAIL PROTECTED]
[mailto:[EMAIL PROTECTED] Behalf Of Eric Ellison
Sent: Sunday, January 01, 2006 8:18 AM
To: [EMAIL PROTECTED]; FlexRadio@flex-radio.biz
Subject: Re: [Flexradio] A question on Frequency stability vs.
Temperature


Larry

The last I heard from Bill - KD5TFD on Teamspeak yesterday, was that his
weekend goal was to interface the Xylo to his GPS 1 pps, so he may beat us
all! Also I am interested in where he is going to pull the sigs out of the
SDR. No body commented on that message, but it it seemed like a better idea
than messing with the 200 mhz LO.

In the FPGA - SDR iteration, if we can do a long term average of the GPS
sigs then we might not even need a stable slaved 10 mhz standard.

In any case, I have ordered what I think is a pretty good deal from the guy
Bob - K5KDN is working with. He is selling these Oscillators on E-Bay for 75
bux and has a simple kit of parts for slaving to the 10 khz GPS sig.

http://cgi.ebay.com/New-Isotemp-10MHz-OCXO-1PPB-EFC-For-GPS-Discipline-12V_W
0QQitemZ5847210309QQcategoryZ48702QQssPageNameZWDVWQQrdZ1QQcmdZViewItem

On way or tother this type of circuit and osc is what non-fpga SDR-1000
owners could use for good stability with the existing modification kit from
FlexRadio.

Brainstorming session on Teamspeak next Wed night at 0100.

Eric2


-Original Message-
From: [EMAIL PROTECTED]
[mailto:[EMAIL PROTECTED] On Behalf Of [EMAIL PROTECTED]
Sent: Saturday, December 31, 2005 11:02 PM
To: FlexRadio@flex-radio.biz
Subject: [Flexradio] A question on Frequency stability vs. Temperature

Xyloonians,

Build a selectable output frequency standard referenced to GPS or WWV.  M/N
type thing.  Have its frequency output within the range of the soundcard

Re: [Flexradio] A question on Frequency stability vs. Temperature

[Jim Lux]

At 10:15 PM 12/31/2005, [EMAIL PROTECTED] wrote:
You and I are headed in the same direction.  I'm saying that the Xyloonia 
world should creat a known stable frequency within the passband of the 
soundcard that the PowerSDR can measure continuously to creat the offset 
needed to display the correct frequency.  Something like the old 100 kc 
xtal oscillator with a 25 kc divider such as Drake used in the 4 and the 7 
series.  Only do the math as needed to creat a single frequency rather 
than fixed comb that would show up every 25 kc.  It really doesn't have to 
look at each oscillator, just the resultant signal.


Actually, you need to synthesize TWO signals in the passband, because 
errors in the A/D sample clock can't be calibrated out with a single 
marker.  The other thing is that this marker tone needs to be spectrally 
fairly clean, so that its noise doesn't wind up in the desired signal 
passband.  Presumably, you'd adjust the marker(s) so that it is (they are) 
comparable in level to the desired receive signal (at least as far as front 
end gain, etc. goes).  Nothing too tricky, but things that need to be dealt 
with in the design.


The nice thing about this approach is that the external marker generator 
only needs to be commanded (by the SDR1000 SW), you don't need to get any 
information back from the generator.
All the measurement/calibration is done within the context of the signals 
coming in on the audio from the SDR1000 hardware.


I used an approach very similar to this to calibrate multiple 1260 MHz 
receivers which use an SDR1000 as the backend where each receiver has its 
own 10 MHz OCXO.


Jim, W6RMK 

Re: [Flexradio] A question on Frequency stability vs. Temperature

[ltaft]

 
 Actually, you need to synthesize TWO signals in the passband, because errors 
 in the A/D sample clock can't be calibrated out with a single marker.  

Jim,

Well, I have to ask why two markers are needed or can you give a reference text 
I could study to see why?  Warning, KISS please as I can follow y=mx+b but 
after that I'm lost.

Since two are needed then generate markers spaced every 25 KHz and just move 
the markers around so they aren't sitting on the center frequency.  Looks like 
with just a little more thought on the hardware we can push it off the end of 
the bench and let the softheads do the trench work.  

Golly, product out the door next week.

73, Larry  K2LT

Re: [Flexradio] A question on Frequency stability vs. Temperature

[Tom Clark, W3IWI]

KD5NWA wrote:
The project being discussed for the Xylo is;
  
1.Thermal isolation
with a accurate heater to reduce the drift.
2.Use the Xylo to
connect to a GPS and have a accurate 1 pps clock
3.Use the 1 pps clock
to measure the DDS 200MHz clock, integrate multiple reading to make for
finer resolution.
4.Use the 1 pps clock
to measure the A/D clock, take multiple readings also.
5.Feed the results to
the PC who will proceed to correct the internal math so there is no
drift.
  

Let me suggest a different approach at step 3. Connect the DDS
reference to a long counter (more than one second) -- it can be
straight binary, and 32 bits is overkill. Use the 1PPS to strobe the
counter into a register without stopping the counter (i.e. like lap
timing with a stop watch). Then make your corrections based on
accumulated phase, which can then be averaged over many seconds.

73, Tom

Re: [Flexradio] A question on Frequency stability vs. Temperature

[Eric Ellison]

Tom



Your suggestion will most definitely be
implemented in any design. It really is the only way to go if we are to use GPS
and want to have really high accuracy. If we are going to spend the money on a
solution, then we might as well make it as good as it gets with these tools.
They really are not all that expensive. Also want to make it generic
enough that it will provide the very accurate 10 mhz in a stand alone
mode as the SDR and other targets require. I REALLY liked your PIC solution for
the divider chain. That really is not all that expensive, and can keep the long
term averages, as well as yield the divisors, if they are needed. The nice
thing about our Xylo (FPGA) solution for the SDR is that we have it under software
control on both the clock side, the reference side, and the PowerSDR side.



I think I have asked this question before
and perhaps Jim Lux answered:



Can we just count for one second and roll
the carry out to the bit bucket, say at 12 or 16 bits. The low order byte or
bytes should be accurate enough for averaging the phase, where we are pretty
accurate on the 1 second count in the first place? The high order bits will
probably be the same every count in any case.



Bill  KD5TFD is working on counting
from the DDS in the SDR (I think) so we dont have to mess with the 200
mhz LO.



Bob  K5KDN is working on a board
with phase locked voltage control from the 10 khz osc on the Jupiter, and
hopefully will make mods to mount the board directly to the Jupiter DIP pins,
and also offer off board connects to the Xylo.



This is finally getting to some design
phase! Should be slick!



Drat, missed the leap second again!



Thanks

Eric

















From: [EMAIL PROTECTED]
[mailto:[EMAIL PROTECTED] On
Behalf Of Tom Clark, W3IWI
Sent: Sunday, January 01, 2006
9:25 PM
To: KD5NWA
Cc: FlexRadio@flex-radio.biz
Subject: Re: [Flexradio] A
question on Frequency stability vs. Temperature





KD5NWA wrote: 

The project being
discussed for the Xylo is;

1.Thermal isolation with a
accurate heater to reduce the drift.
2.Use the Xylo to connect to
a GPS and have a accurate 1 pps clock
3.Use the 1 pps clock to
measure the DDS 200MHz clock, integrate multiple reading to make for finer
resolution.
4.Use the 1 pps clock to
measure the A/D clock, take multiple readings also.
5.Feed the results to the PC
who will proceed to correct the internal math so there is no drift.

Let me suggest a different approach at step 3. Connect
the DDS reference to a long counter (more than one second) -- it can be
straight binary, and 32 bits is overkill. Use the 1PPS to strobe the counter
into a register without stopping the counter (i.e. like lap timing with a stop
watch). Then make your corrections based on accumulated phase, which can then
be averaged over many seconds.

73, Tom

Re: [Flexradio] A question on Frequency stability vs. Temperature

[Bill Guyger]

Wouldn't this constitute a GROUND loop? Ok that was cruel and unusual 
punnishment.

In all seriousness, Marti RPU transmitters have a small styrofoam box around 
the crystals (and heat element) on the modulator board. Would taking a small 
styro block, hollowing out one side and placing it over the oscillator be a 
quick and dirty (there's that word again) fix? It sure wouldn't be as stable as 
the more complex hardware and software tricks that have been tossed out for 
consideration lately, but it might be a KISS attack plan.

Bill AD5OL

Almost true -- what you really want to do is to heat-sink the oscillator 
to a large thermal mass. In addition to a temperature controlled heater, 
another good heat sink is DIRT. If you were to dig a post-hole a few 
feet down into the earth and then drop the entire oscillator down the 
hole, and fill it back up, you would be amazed at the stability you 
achieve. We learned this trick in interferometry when we want to have 
stable microwave LOs out at the feed of several dishes. Even in the 
desert with a huge night/day cycling, we got good stability by burying 
the coax 1-2 feet.

Tom

___
FlexRadio mailing list
FlexRadio@flex-radio.biz 
http://mail.flex-radio.biz/mailman/listinfo/flexradio_flex-radio.biz

Re: [Flexradio] A question on Frequency stability vs. Temperature

[Tom Clark, W3IWI]

Bill Guyger wrote:

Wouldn't this constitute a GROUND loop? Ok that was cruel and unusual 
punnishment.
  

Pun ignored ;-)

... Would taking a small styro block, hollowing out one side and placing it 
over the oscillator be a quick and dirty (there's that word again) fix? It sure 
wouldn't be as stable as the more complex hardware and software tricks that 
have been tossed out for consideration lately, but it might be a KISS attack 
plan.
  
Using a thermal insulator will certainly help to buffer you from ambient 
temperature variation, and might well stretch the oscillator's time 
constant from ~10 seconds to a few minutes. Personally, I have had very 
good luck using wide-mouthed thermos bottles. I have also found that 
wrapping an oscillator in bubble wrap and putting it in a cardboard box, 
or a cardboard box filled with plastic peanuts works as well as a 
block of Styrofoam and only takes a minute to build.


But even then, your heat sink is the ambient air supply which probably 
cycles 10-20 degrees on a daily basis. My hole in the ground 
suggestion was simply a way to find a more stable heat sink. N8UR's 
statement about using a heated box also is a way to solve the problem. 
One thing to remember, which may not be totally obvious, is you use a 
heater element and a temp sensor, put the heat source and the sensor in 
close contact. Do not  put the sensor at the oscillator inside some 
thermal insulation away from the heater -- this is guaranteed to give 
thermal overshoot with a cycle time related to the time constant of the 
insulation.


73, Tom

Re: [Flexradio] A question on Frequency stability vs. Temperature

[KD5NWA]

That is precisely what will occur, the reading will occur over 
decades of seconds.


At 08:25 PM 1/1/2006, you wrote:

KD5NWA wrote:

The project being discussed for the Xylo is;

1.  Thermal isolation with a accurate heater to reduce the drift.
2.  Use the Xylo to connect to a GPS and have a accurate 1 pps clock
3.  Use the 1 pps clock to measure the DDS 200MHz clock, 
integrate multiple reading to make for finer resolution.
4.  Use the 1 pps clock to measure the A/D clock, take multiple 
readings also.
5.  Feed the results to the PC who will proceed to correct the 
internal math so there is no drift.
Let me suggest a different approach at step 3. Connect the DDS 
reference to a long counter (more than one second) -- it can be 
straight binary, and 32 bits is overkill. Use the 1PPS to strobe the 
counter into a register without stopping the counter (i.e. like lap 
timing with a stop watch). Then make your corrections based on 
accumulated phase, which can then be averaged over many seconds.


73, Tom



Cecil Bayona
KD5NWA
www.qrpradio.com

I fail to see why doing the same thing over and over and getting the 
same results every time is insanity: I've almost proved it isn't; 
only a few more tests now and I'm sure results will differ this time ...   

[Flexradio] A question on Frequency stability vs. Temperature

[Tim Ellison]

I have a question about frequency stability of the SDR1K in regard to
variable temperatures inside the SDR1K.  The reason for concern is
frequency drift while working digital modes.  For the sake of this
argument and staying on point, let's assume that the frequency delta is
significant, although in reality it probably is insignificant.


Some initial background information.  I am seriously thinking about
installing a temperature-proportional cooling fan speed controller in
the SDR1K to further reduce fan noise.  The controller is not designed
to provide a stable temperature (thermostatically controlled), it just
increases the fan's RPMs as the temperature increases until it reaches
max RPMs @ 105 F.  I assume that since the fan is not providing constant
CFM air flow, under high duty cycle operation, there will be a more
rapid rise in temperature until the fan is running at full speed, at
which some level of temperature equilibrium is reached, as opposed to
having the fan run at full speed all the time resulting in a temperature
change that would be less drastic.  The net effect is that the internal
temperature will vary more with the fan controller therefore resulting
in more drift in the XO.

I am aware of the ability to use external precision clock sources to
more precisely drive the DDS, but in my current configuration, the
Valpey-Fisher VF-161 XO is utilized and is sensitive to frequency drift
with changes in temperature.  From the specs, the max temperature for
the XO is 85 C and the stability is +/- 20 ppm.
http://www.valpeyfisher.com/PDFs/vf161_E.pdf

Assuming temperature stabilization is warranted, my question is this -
what advantages / disadvantages would there be for trying to minimize
temperature variations of the OX by adding some insulating material
(Styrofoam maybe) around the XO or by adding an external precision
crystal heater ( @ 40 C) to *stabilize* the temperature at a fixed value
and hence the frequency drift?

An example of the precision crystal heater is found here:
http://www.kuhne-electronic.de/english/special/crystalheater.htm

Any and all comments welcome.  Thanks!

-Tim
---
Tim Ellison mailto:[EMAIL PROTECTED] 
Integrated Technical Services http://www.itsco.com/  
Apex, NC USA
919.674.0044 Ext. 25 / 919.674.0045 (FAX)
919.215.6375 - cell
 PGP public key available at all public KeyServers 

Re: [Flexradio] A question on Frequency stability vs. Temperature

[KD5NWA]

I've heard it on Teamspeak that when a scheme like that has been 
tried it worked very well.


Thermal stability like you mentioned, along with electronic 
compensation of the clock by comparing it to a GPS clock is one of 
the projects the Xylo will be used for.


At 12:49 AM 12/31/2005, Tim Ellison wrote:

I have a question about frequency stability of the SDR1K in regard to
variable temperatures inside the SDR1K.  The reason for concern is
frequency drift while working digital modes.  For the sake of this
argument and staying on point, let's assume that the frequency delta is
significant, although in reality it probably is insignificant.


Some initial background information.  I am seriously thinking about
installing a temperature-proportional cooling fan speed controller in
the SDR1K to further reduce fan noise.  The controller is not designed
to provide a stable temperature (thermostatically controlled), it just
increases the fan's RPMs as the temperature increases until it reaches
max RPMs @ 105 F.  I assume that since the fan is not providing constant
CFM air flow, under high duty cycle operation, there will be a more
rapid rise in temperature until the fan is running at full speed, at
which some level of temperature equilibrium is reached, as opposed to
having the fan run at full speed all the time resulting in a temperature
change that would be less drastic.  The net effect is that the internal
temperature will vary more with the fan controller therefore resulting
in more drift in the XO.

I am aware of the ability to use external precision clock sources to
more precisely drive the DDS, but in my current configuration, the
Valpey-Fisher VF-161 XO is utilized and is sensitive to frequency drift
with changes in temperature.  From the specs, the max temperature for
the XO is 85 C and the stability is +/- 20 ppm.
http://www.valpeyfisher.com/PDFs/vf161_E.pdf

Assuming temperature stabilization is warranted, my question is this -
what advantages / disadvantages would there be for trying to minimize
temperature variations of the OX by adding some insulating material
(Styrofoam maybe) around the XO or by adding an external precision
crystal heater ( @ 40 C) to *stabilize* the temperature at a fixed value
and hence the frequency drift?

An example of the precision crystal heater is found here:
http://www.kuhne-electronic.de/english/special/crystalheater.htm

Any and all comments welcome.  Thanks!

-Tim
---
Tim Ellison mailto:[EMAIL PROTECTED]
Integrated Technical Services http://www.itsco.com/
Apex, NC USA
919.674.0044 Ext. 25 / 919.674.0045 (FAX)
919.215.6375 - cell
 PGP public key available at all public KeyServers 




___
FlexRadio mailing list
FlexRadio@flex-radio.biz
http://mail.flex-radio.biz/mailman/listinfo/flexradio_flex-radio.biz



Cecil Bayona
KD5NWA
www.qrpradio.com

I fail to see why doing the same thing over and over and getting the 
same results every time is insanity: I've almost proved it isn't; 
only a few more tests now and I'm sure results will differ this time ...  

Re: [Flexradio] A question on Frequency stability vs. Temperature

[ecellison]

Folks

Yes, Bob N4HY mentioned on Teamspeak that he had 'insulated' the osc with some material which cut down on drift. Perhaps he could elaborate a bit more on the scheme he used.

Eric2

-- Original message -- From: KD5NWA [EMAIL PROTECTED]  I've heard it on Teamspeak that when a scheme like that has been  tried it worked very well.   Thermal stability like you mentioned, along with electronic  compensation of the clock by comparing it to a GPS clock is one of  the projects the Xylo will be used for.   At 12:49 AM 12/31/2005, Tim Ellison wrote:  I have a question about frequency stability of the SDR1K in regard to  variable temperatures inside the SDR1K. The reason for concern is  frequency drift while working digital modes. For the sake of this  argument and staying on point, let's assume that the frequency delta is  significant, although in reality it probably is insignificant.  Some initial background information. I am seriously thinking about  installing a temperature-proportional cooling fan speed controller in  the SDR1K to further reduce fan noise. The controller is not designed  to provide a stable temperature (thermostatically controlled), it just  increases the fan's RPMs as the temperature increases until it reaches  max RPMs @ 105 F. I assume that since the fan is not providing constant  CFM air flow, under high duty cycle operation, there will be a more  rapid rise in temperature until the fan is running at full speed, at  which some level of temperature equilibrium is reached, as opposed to  having the fan run at full speed all the time resulting in a temperature  change that would be less drastic. The net effect is that the internal  temperature will vary more with the fan controller therefore resulting  in more drift in the XO.I am aware of the ability to use external precision clock sources to  more precisely drive the DDS, but in my current configuration, the  Valpey-Fisher VF-161 XO is utilized and is sensitive to frequency drift  with changes in temperature. From the specs, the max temperature for  the XO is 85 C and the stability is +/- 20 ppm.  http://www.valpeyfisher.com/PDFs/vf161_E.pdfAssuming temperature stabilization is warranted, my question is this -  what advantages / disadvantages would there be for trying to minimize  temperature variations of the OX by adding some insulating material  (Styrofoam maybe) around the XO or by adding an external precision  crystal heater ( @ 40 C) to *stabilize* the temperature at a fixed value  and hence the frequency drift?An example of the precision crystal heater is found here:  http://www.kuhne-electronic.de/english/special/crystalheater.htmAny and all comments welcome. Thanks!-Tim  ---  Tim Ellison  Integrated Technical Services  Apex, NC USA  919.674.0044 Ext. 25 / 919.674.0045 (FAX)  919.215.6375 - cellPGP public key available at all public KeyServers   ___  FlexRadio mailing list  FlexRadio@flex-radio.biz  http://mail.flex-radio.biz/mailman/listinfo/flexradio_flex-radio.bizCecil Bayona  KD5NWA  www.qrpradio.com   "I fail to see why doing the same thing over and over and getting the  same results every time is insanity: I've almost proved it isn't;  only a few more tests now and I'm sure results will differ this time ... "___  FlexRadio mailing list  FlexRadio@flex-radio.biz  http://mail.flex-radio.biz/mailman/listinfo/flexradio_flex-radio.biz 

Re: [Flexradio] A question on Frequency stability vs. Temperature

[John Ackermann N8UR]

Insulation will slow down the response to external temperature
variations, but unless there's a heat source inside the insulated area,
it won't prevent the change (sooner or later, the internal temperature
will match the external).

One very simple answer, that I think Gerald has been playing with, is to
use a heating element to raise the temperature of the oscillator above
the ambient.

In a packet radio project many years ago, I was working with Kantronics
UHF data radios that had horrible temperature stability that was
exacerbated by the fact they were used in non-temperature controlled
environments.

We tried a couple of ways to fake a crystal oven.  The easiest, and most
successful, was a crystal heater that you could buy from the Yaesu parts
department.  It was a power thermistor mounted to a clip that would
slide over the crystal can.  Attached to 12 volts, it would heat the
crystal up to something over 100 degrees F (we never measured the exact
temperature).  Those solved our temperature stability problems, but I
suspect that it cooked some of the crystals (which were spec'd for
room temperature operation) because after a couple of years we found
some rocks that would no longer net to frequency.

Something similar might work with the V-F oscillator can.

John


[EMAIL PROTECTED] said the following on 12/31/2005 07:31 AM:
 Folks
 
 Yes, Bob N4HY mentioned on Teamspeak that he had 'insulated' the osc with 
 some material which cut down on drift. Perhaps he could elaborate a bit more 
 on the scheme he used.
 
 Eric2
 
 -- Original message -- 
 From: KD5NWA [EMAIL PROTECTED] 
 
 
I've heard it on Teamspeak that when a scheme like that has been 
tried it worked very well. 

Thermal stability like you mentioned, along with electronic 
compensation of the clock by comparing it to a GPS clock is one of 
the projects the Xylo will be used for. 

At 12:49 AM 12/31/2005, Tim Ellison wrote: 

I have a question about frequency stability of the SDR1K in regard to 
variable temperatures inside the SDR1K. The reason for concern is 
frequency drift while working digital modes. For the sake of this 
argument and staying on point, let's assume that the frequency delta is 
significant, although in reality it probably is insignificant. 


Some initial background information. I am seriously thinking about 
installing a temperature-proportional cooling fan speed controller in 
the SDR1K to further reduce fan noise. The controller is not designed 
to provide a stable temperature (thermostatically controlled), it just 
increases the fan's RPMs as the temperature increases until it reaches 
max RPMs @ 105 F. I assume that since the fan is not providing constant 
CFM air flow, under high duty cycle operation, there will be a more 
rapid rise in temperature until the fan is running at full speed, at 
which some level of temperature equilibrium is reached, as opposed to 
having the fan run at full speed all the time resulting in a temperature 
change that would be less drastic. The net effect is that the internal 
temperature will vary more with the fan controller therefore resulting 
in more drift in the XO. 

I am aware of the ability to use external precision clock sources to 
more precisely drive the DDS, but in my current configuration, the 
Valpey-Fisher VF-161 XO is utilized and is sensitive to frequency drift 
with changes in temperature. From the specs, the max temperature for 
the XO is 85 C and the stability is +/- 20 ppm. 
http://www.valpeyfisher.com/PDFs/vf161_E.pdf 

Assuming temperature stabilization is warranted, my question is this - 
what advantages / disadvantages would there be for trying to minimize 
temperature variations of the OX by adding some insulating material 
(Styrofoam maybe) around the XO or by adding an external precision 
crystal heater ( @ 40 C) to *stabilize* the temperature at a fixed value 
and hence the frequency drift? 

An example of the precision crystal heater is found here: 
http://www.kuhne-electronic.de/english/special/crystalheater.htm 

Any and all comments welcome. Thanks! 

-Tim 
--- 
Tim Ellison 
Integrated Technical Services 
Apex, NC USA 
919.674.0044 Ext. 25 / 919.674.0045 (FAX) 
919.215.6375 - cell 

PGP public key available at all public KeyServers  




___ 
FlexRadio mailing list 
FlexRadio@flex-radio.biz 
http://mail.flex-radio.biz/mailman/listinfo/flexradio_flex-radio.biz 


Cecil Bayona 
KD5NWA 
www.qrpradio.com 

I fail to see why doing the same thing over and over and getting the 
same results every time is insanity: I've almost proved it isn't; 
only a few more tests now and I'm sure results will differ this time ...  


___ 
FlexRadio mailing list 
FlexRadio@flex-radio.biz 
http://mail.flex-radio.biz/mailman/listinfo/flexradio_flex-radio.biz 

Re: [Flexradio] A question on Frequency stability vs. Temperature

[Duane - N9DG]

You can find 60 deg C thermisters at:
http://www.downeastmicrowave.com/Catalog.htm

They are quite inexpensive @ $1.50 a pop.

The DEMI catalog links to:
http://www.thermometrics.com/assets/images/ptcnotes.pdf

I've been adding them to my various VHF transverters that do
not yet have them. They do definitely help for stability.

Duane
N9DG 

--- John Ackermann N8UR [EMAIL PROTECTED] wrote:

 Insulation will slow down the response to external
 temperature
 variations, but unless there's a heat source inside the
 insulated area,
 it won't prevent the change (sooner or later, the internal
 temperature
 will match the external).
 
 One very simple answer, that I think Gerald has been
 playing with, is to
 use a heating element to raise the temperature of the
 oscillator above
 the ambient.
 
 In a packet radio project many years ago, I was working
 with Kantronics
 UHF data radios that had horrible temperature stability
 that was
 exacerbated by the fact they were used in non-temperature
 controlled
 environments.
 
 We tried a couple of ways to fake a crystal oven.  The
 easiest, and most
 successful, was a crystal heater that you could buy from
 the Yaesu parts
 department.  It was a power thermistor mounted to a clip
 that would
 slide over the crystal can.  Attached to 12 volts, it would
 heat the
 crystal up to something over 100 degrees F (we never
 measured the exact
 temperature).  Those solved our temperature stability
 problems, but I
 suspect that it cooked some of the crystals (which were
 spec'd for
 room temperature operation) because after a couple of years
 we found
 some rocks that would no longer net to frequency.
 
 Something similar might work with the V-F oscillator can.
 
 John
 





__ 
Yahoo! for Good - Make a difference this year. 
http://brand.yahoo.com/cybergivingweek2005/

Re: [Flexradio] A question on Frequency stability vs. Temperature

[Philip M. Lanese]

Tim, Eric2, et al

You canenclose the oscillator canwithin a thick block of 
material that hashigh thermal resistance(high thermal 
inertia). You do not want to have airmoving around the 
oscillatorbecause the heatconducted away from the oscillator will 
vary as the air flow varies and that effectwill make the problem 
worse.This technique assumes you can safely heat the crystal to a 
good margin above any maximum ambient. In order to heatthe crystal, 
youneed to knowthe CRYSTALturn-over temperature, a 
specification not usually available for 'canned' oscillators, or you could end 
up frying theoscillator that is also inside the can orover 
stressingthe crystal itself, in which case it willgo 
PERMANENTLY off frequency and ready for the round file. Most likely 
when you are about to make that once-in-a-lifetime DX contact.

You may reduce drift somewhat but the whole question of frequency stability 
and accuracy is: does anything you do matter if the various other ends of the 
communications circuitare not at least as stable as your 
endAND there are no variations due to propagation effects of the 
signals. Take EME at and above 432 MHz as an example. No matter how 
accurate and stable your frequency or that of your QSO partner, you still have 
to useyour RIT to chase the returning echoes because of changing Doppler 
shift. WSJT's EME mode 65C is no longerused much, if at 
all,because it depends on the INCOMING signal, as well as your receiver, 
being stable within 5 to 8 Hz or so overanentire EME receive 
period. Really serious EME operators use Linrad because they can SEE the 
incoming signals' drift in 'real time' and canuse the RIT when necessary 
tomanually correct for frequency change.

You also have to remember that canned oscillators are 'consumer' products 
and the manufacturer can only invest so much in frequency 
stability,accuracy, reduced phase noise, etc.before he prices the 
product out of the market.

Any quickie after-market adjustment would at best be a 'one-off' and what 
are the chances ofreplicating a one-off'fix' many times over.

If you wantextreme accuracy, you have to start from scratch and go 
thru the whole (non trivial, notinexpensive anddefinitely time 
consuming) design process. That's just for the oscillator. Then you 
have to designtemperature stable buffer circuits to insert between the 
oscillator andthe load to effectively isolate the load, as well as 
other problem sources, voltage regulator noise, RF, etc., so the oscillator is 
not affected.

Then, as Jim, W6RMK,and others have pointed out, you turn your 
attention to 'fixing' the PC oscillator, the sound card oscillator, the OS 
software and theapplication software as needed.

Phil, K3IB



  - Original Message - 
  From: 
  [EMAIL PROTECTED] 
  
  Folks
  
  Yes, Bob N4HY mentioned on Teamspeak that he had 'insulated' the osc with 
  some material which cut down on drift. Perhaps he could elaborate a bit more 
  on the scheme he used.
  
  Eric2
  

Re: [Flexradio] A question on Frequency stability vs. Temperature

[KD5NWA]

The project being discussed for the Xylo is;
1.Thermal isolation
with a accurate heater to reduce the drift.
2.Use the Xylo to
connect to a GPS and have a accurate 1 pps clock
3.Use the 1 pps clock
to measure the DDS 200MHz clock, integrate multiple reading to make for
finer resolution.
4.Use the 1 pps clock
to measure the A/D clock, take multiple readings also.
5.Feed the results to
the PC who will proceed to correct the internal math so there is no
drift.
No attempt will be made to change the physical 200 MHz clock to something
else that is changeable, so phase noise is not affected. The Xylo
will measure, the PowerSDR software will use the data to correct it's
internals, resulting in very accurate, and stable frequencies with
minimal changes to the existing hardware
That is my understanding on what the game plan is.

At 10:02 AM 12/31/2005, Philip M. Lanese wrote:
Tim, Eric2, et al

You can enclose the oscillator can within a thick block of material that
has high thermal resistance (high thermal inertia). You do not want
to have air moving around the oscillator because the heat conducted away
from the oscillator will vary as the air flow varies and that effect will
make the problem worse. This technique assumes you can safely heat
the crystal to a good margin above any maximum ambient. In order to
heat the crystal, you need to know the CRYSTAL turn-over temperature, a
specification not usually available for 'canned' oscillators, or you
could end up frying the oscillator that is also inside the can or over
stressing the crystal itself, in which case it will go PERMANENTLY
off frequency and ready for the round file. Most likely when you
are about to make that once-in-a-lifetime DX contact.

You may reduce drift somewhat but the whole question of frequency
stability and accuracy is: does anything you do matter if the various
other ends of the communications circuit are not at least as
stable as your end AND there are no variations due to propagation
effects of the signals. Take EME at and above 432 MHz as an
example. No matter how accurate and stable your frequency or that
of your QSO partner, you still have to use your RIT to chase the
returning echoes because of changing Doppler shift. WSJT's EME mode
65C is no longer used much, if at all, because it depends on the INCOMING
signal, as well as your receiver, being stable within 5 to 8 Hz or so
over an entire EME receive period. Really serious EME operators use
Linrad because they can SEE the incoming signals' drift in 'real time'
and can use the RIT when necessary to manually correct for frequency
change.

You also have to remember that canned oscillators are 'consumer' products
and the manufacturer can only invest so much in frequency stability,
accuracy, reduced phase noise, etc. before he prices the product out of
the market.

Any quickie after-market adjustment would at best be a 'one-off' and what
are the chances of replicating a one-off 'fix' many times over.

If you want extreme accuracy, you have to start from scratch and go thru
the whole (non trivial, not inexpensive and definitely time consuming)
design process. That's just for the oscillator. Then you have
to design temperature stable buffer circuits to insert between the
oscillator and the load to effectively isolate the load, as well
as other problem sources, voltage regulator noise, RF, etc., so the
oscillator is not affected.

Then, as Jim, W6RMK, and others have pointed out, you turn your attention
to 'fixing' the PC oscillator, the sound card oscillator, the OS software
and the application software as needed.

Phil, K3IB




- Original Message - 

From:
[EMAIL PROTECTED] 



Folks



Yes, Bob N4HY mentioned on Teamspeak that he had 'insulated' the osc
with some material which cut down on drift. Perhaps he could elaborate a
bit more on the scheme he used.



Eric2






Cecil Bayona
KD5NWA
www.qrpradio.com

I fail to see why doing the same thing over and over and
getting the same results every time is insanity: I've almost proved it
isn't; only a few more tests now and I'm sure results will differ this
time ... 

Re: [Flexradio] A question on Frequency stability vs. Temperature

[Jim Lux]

At 08:20 AM 12/31/2005, KD5NWA wrote:

The project being discussed for the Xylo is;

1.  Thermal isolation with a accurate heater to reduce the drift.
2.  Use the Xylo to connect to a GPS and have a accurate 1 pps clock
3.  Use the 1 pps clock to measure the DDS 200MHz clock, integrate 
multiple reading to make for finer resolution.
4.  Use the 1 pps clock to measure the A/D clock, take multiple 
readings also.
5.  Feed the results to the PC who will proceed to correct the 
internal math so there is no drift.


Given that you're going to be calibrating the 200MHz oscillator on a 
periodic basis, you don't need a heater for the oscillator, just something 
to make its thermal time constant much longer than the calibration time 
constant.  As others have pointed out, heating that oscillator may make 
things worse, since it's designed for room temp operation.  Why not just 
put a big lump of something with decent insulation and mass around it?


Jim, W6RMK 

[Flexradio] A question on Frequency stability vs. Temperature

[ltaft]

Xyloonians,

Build a selectable output frequency standard referenced to GPS or WWV.  M/N 
type thing.  Have its frequency output within the range of the soundcard by 
looking at but offset from the frequency tuned to at the moment.  Maybe have it 
outside the pan adapter window so it doesn't distract the operator.  Or put it 
in the window with a red X to show it is working.  

The PowerSDR software measures this frequency and compares it to the Xylo 
frequency number, and adjusts the SDR display to correct it.  This now takes 
into account all the oscillators in the SDR1000 and the soundcard and is 
continuously compensating.
No need for VCO stabilizing the oscillators. 

All to You Year of the New Happy!

Hardware designed by hardheads
Software designed by softheads

Larry - K2LT - I design hardware!
 
 From: KD5NWA [EMAIL PROTECTED]
  
 The project being discussed for the Xylo is;
 
 1.  Thermal isolation with a accurate heater to reduce the drift.
 2.  Use the Xylo to connect to a GPS and have a accurate 1 pps clock
 3.  Use the 1 pps clock to measure the DDS 200MHz clock, 
 integrate multiple reading to make for finer resolution.
 4.  Use the 1 pps clock to measure the A/D clock, take multiple 
 readings also.
 5.  Feed the results to the PC who will proceed to correct the 
 internal math so there is no drift.
 
 No attempt will be made to change the physical 200 MHz clock to 
 something else that is changeable, so phase noise is not 
 affected.  The Xylo will measure, the PowerSDR software will use the 
 data to correct it's internals, resulting in very accurate, and 
 stable frequencies with minimal changes to the existing hardware
 
 That is my understanding on what the game plan is.
 
 

Re: [Flexradio] A question on Frequency stability vs. Temperature

[KD5NWA]

The scheme outline below in steps 1 through 5 does not correct the 
frequency by changing the hardware oscillator, it does not require a 
VCO to replace the current oscillator, it measures the errors and 
feeds it digitally through a USB port to the PC who will make 
software corrections.


By measuring the 200MHz signal it will more accurately detect errors.


At 10:02 PM 12/31/2005, you wrote:

Xyloonians,

Build a selectable output frequency standard referenced to GPS or 
WWV.  M/N type thing.  Have its frequency output within the range of 
the soundcard by looking at but offset from the frequency tuned to 
at the moment.  Maybe have it outside the pan adapter window so it 
doesn't distract the operator.  Or put it in the window with a red X 
to show it is working.


The PowerSDR software measures this frequency and compares it to the 
Xylo frequency number, and adjusts the SDR display to correct 
it.  This now takes into account all the oscillators in the SDR1000 
and the soundcard and is continuously compensating.

No need for VCO stabilizing the oscillators.

All to You Year of the New Happy!

Hardware designed by hardheads
Software designed by softheads

Larry - K2LT - I design hardware!

 From: KD5NWA [EMAIL PROTECTED]
 
 The project being discussed for the Xylo is;

 1.  Thermal isolation with a accurate heater to reduce the drift.
 2.  Use the Xylo to connect to a GPS and have a accurate 1 pps clock
 3.  Use the 1 pps clock to measure the DDS 200MHz clock,
 integrate multiple reading to make for finer resolution.
 4.  Use the 1 pps clock to measure the A/D clock, take multiple
 readings also.
 5.  Feed the results to the PC who will proceed to correct the
 internal math so there is no drift.

 No attempt will be made to change the physical 200 MHz clock to
 something else that is changeable, so phase noise is not
 affected.  The Xylo will measure, the PowerSDR software will use the
 data to correct it's internals, resulting in very accurate, and
 stable frequencies with minimal changes to the existing hardware

 That is my understanding on what the game plan is.




___
FlexRadio mailing list
FlexRadio@flex-radio.biz
http://mail.flex-radio.biz/mailman/listinfo/flexradio_flex-radio.biz



Cecil Bayona
KD5NWA
www.qrpradio.com

I fail to see why doing the same thing over and over and getting the 
same results every time is insanity: I've almost proved it isn't; 
only a few more tests now and I'm sure results will differ this time ...