Re: [volt-nuts] How can I make a 2000 V DC meter with an input resistance of at least 100 T ohms?

[ed breya]

Regarding making your own extreme high-value resistors - any object that 
has insulators and leads but with nothing connected inside will have 
some high R that can be perhaps be measured, but won't be stable against 
environment effects on the outer surfaces. There's not much point to 
carbonizing things for home-made ones, except for curiosity.


You can, however, use existing things that are fairly stable internally, 
have hermetic seals, and can be treated externally to reduce environment 
issues. I mentioned that reed relay capsule that I used as an unknown, 
but very high, yet not infinite R. Burned out light bulbs, vacuum tubes 
(especially something like a 5642 HV rectifier - fairly small, lots of 
glass), and xenon flashtubes are other examples of common hermetic 
glass/metal parts that can be used. But, the R is what it is, and can't 
readily be adjusted, only measured and maybe used in circuits that can 
accommodate the value. Also, along with the R, there will be some C that 
depends on the structure of whatever is used. The C can be good or bad, 
depending on the application.


At extreme values, the surface characteristics will dominate, so the 
glass envelope would have to be silicone treated. Then the measured R of 
the device will be almost all intrinsic. So, you can measure it, but you 
won't know how stable it may be with temperature and voltage and time, 
for example, so don't expect much precision.


Regarding over-voltaging electrolytic caps - you can reform caps to 
somewhat higher voltage, given enough time. They are formed 
electrolyitically to begin with, so the dielectric layer thickness is 
right for the rated voltage. If you gradually up the voltage, the 
thickness will increase and the C will go down over time. It's best to 
just use them only up to the design rating though, or the leakage will 
become unpredictable.


A good way to do voltage splitting/protecting on medium-high voltage 
series connected electrolytic caps with low leakage, is with an 
appropriate high voltage "Zener" (actually an avalanche device, not 
truly Zener) across each one. The Zeners will prevent over-voltage of 
the caps in the normal direction, and reverse protection in the diode's 
forward region. Look for transient voltage suppressors (TVS or TVSS) 
devices to get into the hundreds of volts region, and of course they can 
be stacked for more. Unipolar ones will provide intrinsic reverse 
protection for the cap, while bipolar ones will not. They are usually 
specified fairly loosely in terms of leakage current, but it should be 
possible to find ones in the low nA region at applied V reasonably below 
the knee, at room temperature. That sounds like a lot in a High-Z 
context, but it's almost certainly much less than the leakage of a 
typical electrolytic cap.


Ed

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

Re: [volt-nuts] How can I make a 2000 V DC meter with an input resistance of at least 100 T ohms?

[ed breya]

I'm guessing the application relates back to your leaf electrometer 
project discussed earlier - trying to assess how the bias charge on the 
capacitor holds up from leakage and use of the instrument. If this is 
the case, then it's for a one-time use for design of the item, so 
shouldn't be too fancy or expensive. I believe the original goal was to 
have the cap get charged up and then not need any electric support for 
the leaf electrometer, appearing totally passive, for some amount of 
operating time. If built-in monitoring of the cap voltage is now 
desired, that's a different story.


If the measurement is just for design, to roughly see the cap 
charge-holding time situation, then I'd recommend using methods that 
Chris described, comparing to a variable HV supply at various times and 
settings - all manual iterations, but doable. You can always say, 
recharge the cap, then guess what the voltage may be after so much time, 
then set the test supply and compare - over and over and over.


If continuous, long-term, fairly accurate monitoring is desired, then 
you'd have to go with some sort of non-contact electrostatic voltmeter 
or such, as others have mentioned.


Relating back to recent discussions, it's pretty clear that you're not 
going to find an actual specified resistor in the hundred T-ohm region. 
You can certainly make your own from T-ohms to infinite, but you won't 
be able to know the "exact" value. The commercial instruments that have 
say "200 T-ohms" input R don't actually have that resistor value inside 
- it's an "effective" or "equivalent" derived value that depends on a 
real resistance of maybe E11-E12, multiplied by system gain.


Some electrometers like the old Keithleys have a voltage mode where the 
high-Z input amplifier is bootstrapped up as a voltage follower, but 
have less range than you want. It's conceivable that you could build the 
same thing, but with a HV amplifier follower that can reach the desired 
level. This would not be trivial.


Again, if the purpose is just to measure the droop in bias voltage of 
the charged cap over certain time intervals, there may be another 
option. Since this is a dv/dt rather than DC measurement, you could 
possibly set up an electrometer to view the change of the bias voltage 
via current through another capacitor, and conceivably even rig it up to 
directly measure the total change in cap voltage over a given time.


Let's say the charge storage cap is 1 uF, and you put a much smaller, 
less leaky, test cap plus some protective series R from the HV node to 
the input of the electrometer, and also clamp the input with a low 
leakage diode circuit. The test cap could be say 100 or 1000 times 
smaller than the main cap, so its effect will be small. This could be in 
the 10 nF or less range, where it should be fairly easy to find 3 kV or 
so rated metalized film plastic capacitors with suitably low leakage. 
Any constant DC leakage from the cap could be zeroed out or accounted 
for, at least for short-term measurements.


The electrometer could then read the test cap current directly 
proportional to dv/dt, or integrate it back up to delta V in the charge 
mode. There are limits to the reasonable measuring ranges, of course. 
For example, 1 nF would provide 1 nA at 1V/sec - a fairly easy 
measurement. But 1V/1000 seconds could be tricky - only 1 pA to work with.


Ed


On 3/22/2018 7:12 PM, kc9ieq via volt-nuts wrote:

I guess I don't see what the issue is.  No, impedance is not infinate when not 
nulled, but this is why V supply #2 Is adjustable by whatever convenient means. 
 Rough adjust, connect, adjust for null, measure.  Rinse and repeat.  If it 
were my project, I'd just run up an HV transformer on a variac, with a 
rectifier, cap, and probably some series R thrown at it to limit current 
through the meter.  Curious to know what the application is, if this will not 
work.
Good luck with whatever solution you choose.
Regards, Chris


Sent from my SMRTphone
 Original message From: "Dr. David Kirkby"  
Date: 3/22/18  8:58 PM  (GMT-06:00) To: kc9ieq , Discussion of precise voltage 
measurement  Subject: Re: [volt-nuts] How can I make a 2000 V DC meter with an 
input resistance of at least 100 T ohms?
On 23 March 2018 at 01:49, kc9ieq via volt-nuts  wrote:
How about using (or building) an additional 2kV power supply and a sensitive 
meter movement like a differential voltmeter, adjusting for/measuring the null? 
 Impedance at null will be theoretically infinate, current will be 
theoretically zero, and you can measure/monitor the voltage of your second 
supply directly with the probe/meter of your choice.

Regards,Chris

No, that will not work for me, as while the impedance at null is infinite, it 
is not when not nulled, and that will mess up the measurements.

Absolute accuracy is not important. +/- 10% or even 20% would be 

Re: [volt-nuts] Bohnenberger electrometer

[ed breya]

Yes Hendrik, same principle as the butterfly disk style, but mine use 
cylinders - the field exposure is radial instead of axial.  Ed


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

Re: [volt-nuts] Bohnenberger electrometer DANGER

[ed breya]

Here's a simplistic view that may be sufficient. Some energy (in the 
form of charge redistribution, which includes current flow) has to come 
from the capacitor, and some from the input signal, to do the work 
needed to push the leaf against gravity. When the input signal is 
removed, some of the energy (charge) stored on the leaf is returned to 
the cap as gravity restores the initial position - roughly the same 
amount of work, depending on leakage and mechanical loss, and heating of 
the protective series resistor.


With a quick review of electrostatics, you could estimate up a simple 
model and the field equations to get a more satisfying, detailed answer. 
It may be more straightforward to look at it from a circuit perspective. 
Picture it as as a very small, non-linear capacitor (the leaf structure) 
in series with a much much larger regular capacitor charged up to a 
constant DC voltage. Any actual series resistance is just resistance, 
and the mechanical loss can be represented as more resistance added in 
series. Presuming the leaf never actually touches or emits particles* or 
arcs to the reference capacitor node, it's basically a capacitive 
voltage divider, and the applied signal may be considered to be 
transient, or even AC - it steps to the applied voltage, then returns to 
zero (or open), then the cycle may be repeated. Each experiment is 
adding or subtracting charge, then reversing the process. Ideally, the 
cap would never lose its DC bias if there were no losses.


The problem is that figuring out all these details may not be trivial. 
It may be more fun to just try some experiments and see how it goes - 
you'll get some idea of how the real thing holds up, and figure what 
amount of C is OK.


*You shouldn't have to worry too much about corona discharge if the 
maximum voltage on anything is below 3 kV or so. Beyond that, it could 
cause problems.


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

Re: [volt-nuts] Bohnenberger electrometer

[ed breya]

I looked at that link that Brooke put up about Bohnenberger's 
Electroscope. I don't know what your specific arrangement needs to be, 
but it appears you need a plus and a minus HV wrt ground in the most 
general form. If so, then this would mean having to split the voltage of 
a single cap, or have two caps, one for each polarity.  Then I'd 
recommend using good old microwave oven caps. You could charge them both 
to say 2 kV from one HV source, then switch them around so they're 
stacked and grounded at the midpoint.


Ed

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

Re: [volt-nuts] Vibrating reed electrometers

[ed breya]

Another thing I noticed in these instruments - the highest R value used 
is E12, even though decades higher would have been appropriate in 
certain ranges. It shows that was about the practical limit for somewhat 
decent precision and cost. Filling in the desired higher ranges had to 
be done by adding complexity like more gain elsewhere, and writing the 
specs accordingly.


Ed

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

Re: [volt-nuts] Bohnenberger electrometer

[ed breya]

Oops - forgot to mention a detail about microwave oven caps. Sometimes 
they have built-in bleeder resistors, which would of course spoil this 
kind of application.  Ed


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

Re: [volt-nuts] Vibrating reed electrometers

[ed breya]

Yup - Keithley 640 - that must be the one. This stirred my memory 
somewhat, and I just located info on the model 642 also, which was 
apparently newer. The 642 went to (or back to) ultra-low bias MOSFETs, 
while keeping sapphire insulation and a separate input head. The MOSFETs 
need all kinds of offset, temperature, and bias current compensation. 
The 642 also uses a few digits of DVM that obscure the real 
capabilities, as I mentioned previously. The specs apparently show the 
most sensitive range as 1 pA FS, so all stuff below E-12 A depends on 
those digits to resolve. The manual recommends that extremely small 
currents below 1 fA (the third digit down) be measured in charge mode. I 
think this is to compensate for bias current and to average out some of 
the 1/f noise.


The 640 on the other hand, can apparently reach 1 fA FS (1000x lower) 
with 100 aA p-p (+/- 5%) noise on analog readout. Given a choice between 
the two, I think I'd pick the 640, and hook a DVM to the output, and 
average a whole bunch if necessary. I think the 640 uses a superior 
front-end technology that maybe could be even further improved in the 
middle and back end, while the 642 probably is as good as it can ever be 
already.


Ed

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

Re: [volt-nuts] Bohnenberger electrometer

[ed breya]

For static bias, look up "electret" for ideas on some other possible 
options.


I would recommend against your option 2 capacitor - that's a dangerous 
amount of energy to store in something that may be fooled around with 
experimentally. Also, even though it's a lot of C, being electrolytic, 
the charge will eventually leak off anyway - probably faster than any 
charge loss from using the machine.


The option 2 (2 nF at 4.2 kV) seems more appropriate for this use, 
because of the much higher sensitivity attainable. It's charge will leak 
off too, but since it's likely a plastic or oil capacitor, the retention 
time will hopefully be OK overall.


I wouldn't want to take a jolt from either one. In the ultimate design, 
be sure to use some sort of series current limiting resistance to 
isolate the capacitor from the outside world. The R can be quite high 
(megohms, and of course suitable for the maximum voltage) since not much 
current is needed for operation, so the contact/fault hazard would be 
reduced from dangerous to a tingle. It would be good to also have a safe 
discharging method - another R - that can be switched or jammed in, to 
quickly clear the charge for safe keeping when not in use, or during 
design.


In the old days, optical methods were used for "gain," as in a mirror 
galvanometer, for instance. Putting some simple magnification and 
illumination (sun light if electricity is a no-no) in the system can 
increase the visibility of any deflection.


Lastly, regarding capacitors, a good option if available, is to use the 
nice HV oil caps that can be salvaged from older-era (before they went 
to switching supplies) microwave ovens. These are typically rated around 
1 uF, 2 kV AC. Two in series would do for up to 4-5 kV service. Since 
you don't want bleeder/balancing Rs in this application, it would be 
best to use identical caps, or slightly more complicated charging 
circuitry. They can bought new, but may be pretty spendy, depending on 
the project budget. I have dozens of them - saved from every microwave 
oven I've junked out over the years.


At 1 uF, these would have much better retention time, with hazard energy 
between the original options.


Ed

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

Re: [volt-nuts] Precision high resistance measurements / calibration of HP 4339B high-resistance meter.

[ed breya]

Oops - I think I didn't send this message properly yesterday - here goes 
again. Ed


Yes, David, unless you go to very extreme measures, you won't see real R 
values that have any practical meaning beyond E12 ohms or so. Most 
practical insulation Rs may be around E12-E14 tops, unless you go to 
sapphire. Up in that region, the R may be all within a material, or 
include surface components like a film of dirt or moisture, or a 
fingerprint.


E11 resistors can be made to fairly high precision, and maybe E12 
nowadays. In the old days, higher values were made by stacking E11s - 
like ten in series to get E12 with decent precision. The glass packaging 
also limits how high it can go, due to leakage within and on the 
surface. I once used a glass reed relay capsule as an ultra-high 
resistance in a circuit. There was no precision or stability at all, but 
it made a nice high resistor (probably E14-ish dry) even though there 
was no element in there, and the circuit didn't care, as long as it was 
very high, but not infinite.


The specs on this HP unit are likely just the most extreme capability 
taking maximum voltage over minimum current resolution, but any 
measurements would tend to be very noisy and unstable anyway. Also, 
testing at the extreme 1 kV makes the numbers seem more impressive, but 
the voltage coefficient of resistance will pretty much be unpredictable.


If this is a digital meter, then the other spec trick that tends to 
obscure the real performance limit is that the ultimate resolution and 
noise is that last digit - or even last two or three - that may may be 
pretty jumpy, unless very long averaging time is used.


There may be newer, fancier electrometers nowadays, but Keithley used to 
be the standard for these in the old days, before several digits of DVM 
resolution complicated the specs. They had a vibrating capacitor 
electrometer with all-sapphire input structure back in the 1970s/80s I 
think, that was the epitome of electrometers. I forget the model number, 
but vaguely recall that it could reach the aA region full scale - not 
that last digit of resolution thing. It's long obsolete, and I don't 
think they ever made anything actually better - only added DVM digits to 
less capable, conventional semiconductor amplifier techniques. If you 
can find info on it, it's an interesting read. I found a pdf of the 
manual years ago, but have no idea where it is now, or what info may 
still be around.


Ed

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

Re: [volt-nuts] Best way to measure micro Ohms

[ed breya]

Coincidental to all this micro-ohms and nanovolt talk, I've been doing 
some severe large scale garage cleaning to thin stuff out. I found that 
audio amplifier that I mentioned earlier, that is good for some LIA 
reference driver applications. I also found my low-level measurement 
notebooks, including the datasheet and my notes about its operation and 
modifications. I found that the exact same datasheet I got online years 
ago is still available, so here it is, if anyone is interested.


http://www.toacanada.com/assets/files/BG-10_IM.pdf

I also dug out the old Keithley 148 nanovoltmeters, etc, and couldn't 
resist fooling around with them for a bit. I'll have more to say later, 
but in new threads.


Ed

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

Re: [volt-nuts] Best way to measure micro Ohms

[ed breya]

Hmm. Alternating the direction of the current repeatedly and processing 
the results - sure seems like that is fundamentally an AC measurement 
too, despite using DC measurement equipment.


Ed

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

Re: [volt-nuts] Best way to measure micro Ohms

[ed breya]

I just noticed this discussion recently, so I'm late to the party, but 
that never stops me from adding my one-cent's worth.


David, regardless of the aluminum and other material issues, I think 
your initial idea of using a lock-in analyzer is definitely the way to 
go. I'm very fond of LIAs, although I seldom need or use them, so my 
opinion is somewhat biased. I have five - two Ithaco 391A orange-band, a 
PAR 5204, an SR830 and SR850.


If you use an audio power amplifier for driving the experiment, you can 
rig it up so that the LIA can be used to measure the drive current as 
well as the resulting voltage drop. Let's say the amplifier is for 8 
ohms, so you put a few ohms in series with the output, then from there 
into a precision one-ohm sampling resistor, then into the RUT, forming a 
voltage divider. The RUT is expected to be in the micro-ohm region, 
which is many thousands of times smaller than the sampling R, so its 
tiny voltage drop will be negligible, allowing the sample voltage to be 
a good representation of the test current. You could also just treat the 
whole thing as a voltage divider and calculate the "exact" results.


The voltage on the RUT is measured at whatever gain is needed. The 
voltage on the sample resistor will be plentiful at 1V/A, and both 
signals will have very low source R, and minimal noise. Since both 
signals can be measured by the LIA, the uncertainties in assessing each 
part with different equipment are much reduced.


The reference frequency should be as low as possible, limited by the 
amplifier's low-end capability, and selected so it and its harmonics 
land as far as possible from the power line frequency and its harmonics, 
for say up to n=15, or whatever is practical. This will help to reduce 
line interference from nearby sources, and ground loops, and from the 
amplifier. Especially at low frequencies, the amplifier may show a lot 
of line harmonics when driven to high levels - the filter capacitors in 
its power supply can only do so much, and audio PAs are likely not all 
that great in terms of PSRR. Turning on the LIA's line notch filter will 
also help, at least with the fundamental.


The frequency needs to be very low in order to minimize the parasitic 
currents that will cause errors, especially considering that this setup 
is dividing on the order of a million in a single stage. If this appears 
to cause problems, you can reduce the large division ratio by using a 
much smaller sample resistor, and treating it as a divider for 
calculation purposes. Alternatively, adding some appropriate shielding, 
or splitting the division into isolated sections can greatly reduce the 
effects. To avoid signal ground loops, measuring the drive current and 
the RUT voltage should be separate operations, each carried on its own 
BNC cable to the LIA, while the other is completely disconnected and out 
of the way - having no common-grounding or cable bundling or fancy 
signal routing/switching is best. The weakest link ground-loop-wise may 
be the necessity of carrying the reference drive between the LIA and 
power amplifier input, likely sharing the same ground as the output. 
This could force you to set it up for differential measurement of the 
RUT signal. The special audio PA that I have for such purposes has its 
ins and outs transformer-coupled, which helps a lot.


This could be fun and interesting. There are plenty of pieces and 
variables involved to experiment with to optimize the measurement, and 
lots of other tricks available to enhance it if necessary.


Ed

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

Re: [volt-nuts] low emf solder

[ed breya]

I would recommend against trying to use cadmium - it's very toxic, which 
is why Cd-based solders are rare nowadays. They are probably still made, 
but for lab or industrial use with proper handling. If you try to alloy 
it with Sn yourself without proper handling, you could get poisoned. You 
can't just throw the ingredients in a solder pot and expect good results.


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

Re: [volt-nuts] Anyone know how to make stable inductors?

[Ed Breya]

Haha, so it is legit - just a poorly decribed knock-off of the H-P unit. 
I had never heard of this unit, but it looks like good info to have, to 
replicate some equivalent reference inductors. Thanks for finding this 
document.


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

Re: [volt-nuts] Anyone know how to make stable inductors?

[Ed Breya]

Sorry about the previous double-posting - had email problems.

There is another option that occurred to me last night, to get an 
equivalent inductor from an accurate reference capacitor by using an 
active circuit gyrator. The problem of course is that the circuit will 
add errors too, diminishing performance. Also, for four-port use, the 
circuit would have to be battery-powered, and float within the fixture, 
adding more parasitics.


I could experiment with such a circuit fairly easily since I use my 
ground-converter for nearly all measurements, so the gyrator would not 
have to float. All in all though, it still seems that just using a 
capacitor and resulting negative inductance readout is the simplest and 
most accurate approach.


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

Re: [volt-nuts] How are femto-amps measured?

[Ed Breya]

Yes, the input DC bias current can be compensated for, as long as it 
doesn't change too quickly, or isn't too big to make the input amplifier 
circuits run out of dynamic range. In a digital meter environment, all 
sorts of auto-zeroing and multi-sloping things and digital signal 
processing can be going on to correct for various device and circuit 
limitations, which would have been very difficult or impossible in the 
old analog machines.


Even in analog, some correction can be implemented fairly easily. A lot 
of the old Keithley electrometers have controls called zero 
suppression, which can effectively offset the bias current, or even 
much larger amounts up to hundreds of full-spans, for certain 
applications. It's not quite the same as having true zero bias current - 
you still have to be aware of the effects and the compensation settings, 
depending on the measurement.


I modified my Keithley 417 by replacing the original zero suppression 
last digit fine adjust pot with a ten turn helipot and kilodial knob, 
which provides higher resolution. I can dial in the 17 fA bias current 
offset directly. When I eventually get around to washing and silicone 
coating the electrometer tube, I expect to get it down to around 2-5 fA.


My Kethley 410 electrometer doesn't have zero suppression - just a 
zero knob to set the input offset voltage over a small range to zero 
the meter. I added a bias nulling circuit which adds a small variable 
offset voltage superimposed on the output signal to the highest feedback 
resistor (E11 ohms) , with an optocoupler in PV mode. The effective bias 
current can be set to zero on the last two or three most sensitive 
ranges. On the other ranges, it doesn't matter since the effect is so small.


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

Re: [volt-nuts] How are femto-amps measured?

[Ed Breya]

With modern digital readout meters, this can be very misleading in terms 
of actual useable capability. In this case, using the specified highest 
sensitivity 100 pA range, and six digits of digital resolution, gives 
E-10 A/E6 or E-16 A, which is 100 aA for the last digit. But, looking at 
the noise and accuracy specs shows that it is swamped by these numbers. 
The noise can be reduced greatly by sufficient averaging over long 
enough time. The same issues are encountered in voltage measurement.


For good info on making very small current measurements, investigate 
electrometer technology of the good old days, especially from Keithley. 
Back in the 1960s - 1980s, electrometers typically used special tubes or 
MOSFETs that had bias currents in the fA region, along with high 
resistances that topped out at around E11 to E12 ohms (which is still 
about the limit of practicality for R).


Nowadays there are opamps available with bias current in the fA region 
at room temperature, but noise is still a limiting factor. For better 
absolute accuracy and drift performance needed for modern digital 
meters, the ranging resistors can be lower - probably E10 ohms or less, 
since the meters can measure and resolve much smaller voltages.


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

Re: [volt-nuts] HP 3456A Input Impedance Check

[ed breya]

You should include the effects of input bias current - the maximum 
should be specified, and likely in the pA range at room temperature. 
Just put a very high resistance from input to common, and read the 
voltage to calculate the current at zero input. Likewise, you can 
connect the resistor to various voltages to see how it changes. This 
can only be done in ranges that don't have attenuation, where the 
input goes directly to the DC amplifier.


If you do your original experiment again, but with the DC source at 
opposite polarity, I think you'll find the readings will be quite 
different - the voltage may even read higher than what you apply. The 
bias current isn't necessarily constant - it can change with input 
level and especially with temperature, and it can change polarity, 
depending on design.


Ed

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

Re: [volt-nuts] OCD About My HP419A Attenuator Switch Gold Plated Contacts

[ed breya]

Yes, you are being a little too OCD about this. Instrument washing 
issues come up often, and there are plenty of opinions available - 
here are some of mine:


In the 419s that I have, the battery leakage crud has not gone beyond 
the circuit boards or maybe the edge connectors. Cleaning the boards 
should take care of it, and it's not that complicated. For stubborn 
alkaline deposits, a vinegar wash (if necessary) should neutralize 
and descale, followed by scrubbing with liquid dish detergent and a 
toothbrush, and then lots of rinsing with hot tap water and finally 
thorough drying - preferably with an oven, but air drying for a 
couple of days should suffice. Blasting with compressed air helps to 
get most of the water out.


Don't bother with pH indicators - I doubt they would show much unless 
there's so much crud that you can see it anyway. If KOH has gotten 
onto any critical circuits, it could cause symptoms such as excessive 
leakage currents, since it's hygroscopic and would tend to be ionized 
and conductive on the surfaces. If you are concerned about the 
switches, you can wash the whole thing with tap water and liquid 
detergent. First, remove or protect the meter movement and any 
batteries, and maybe the pilot light. As I recall, the rest is pretty 
much open so easy to flush out. There are different schools of 
thought about items such as power transformers and pots - you have to 
apply judgement on whether they will wash out and dry OK. If in 
doubt, protect them from the washing process.


When washing electronics, always finish with an alkaline (like liquid 
detergent) to neutral (water or alcohol) step before rinsing - you 
don't want any acidic residue left anywhere. Lots of rinsing and 
thorough drying is always good - several days of air drying for a 
whole instrument. You may have to relube some of the switches - I 
don't recall if they have grease on the contacts or mechanisms.


Ed

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

Re: [volt-nuts] HP 3456A -3 Fault

[ed breya]

I agree these caps are the most failure-prone parts in the 3456A, and 
also the 3455A. Sometimes they also cause damage to other parts like 
the associated three-terminal regulators. Don't be surprised if 
there's still more to fix after cap replacement - but usually easy to find.


I had one where an LM339 status comparator was shot - I suspected 
over-voltage or reversal of an input polarity during the cap failure, 
even though it wasn't obvious how it could have done that. The 
diagnostics instructions in the manual worked very well and led 
quickly to the faulty part.


Ed

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

Re: [volt-nuts] current-nut question .. total waste of ones time type question

[ed breya]

I agree with John on this one - put in another meter socket 
downstream if you really want to experiment, and be sure to have a 
main disconnect besides the utility's meter.


If you are changing the service entrance, you will likely have issues 
with the local permits for it, and with the utility company - they 
don't like to see anything unusual or confusing.


A couple of years ago I put in a new service at our vacation house, 
that had an ancient mish-mosh of crap that was ready to catch on 
fire, stuffed into too small a panel, and with no main disconnect 
except by pulling the utility meter outside. I collected a nice new 
meter box/load center and other needed parts, assuming I could just 
have PGE pull the meter outside, replace the old load center with the 
new, and have my own meter and disconnect inside. When I applied for 
the permit from the county, this caused nothing but grief and 
confusion - they didn't understand, and neither did PGE, why anyone 
would want a second meter. They get suspicious I think, because they 
don't want multiple residences set up on single-residence zoning, so 
another meter is a red flag, even though I explained many times that 
it was just in series, for monitoring and alternative energy 
experiments such as grid-tie stuff to come later. Furthermore, I had 
to make it meet all current codes, which meant I needed to have a 
separate disconnect on the outside, and updated grounding. 
Ultimately, I ended up replacing the whole works with my new 
meter/load center on the outside, with PGE's meter installed, and 
another new meterless load center on the inside. It all worked out 
OK, but it became a week-long project involving a big hole in the 
wall and reframing, instead of a one day quick-changeout as I had planned.


I can now change it and add meters and CTs and whatever, downstream, 
but I kind of lost interest after going through all that. The novelty 
wears off quickly when you have to first get all the basic stuff 
properly working.


Also, whatever you add should be done in ways that are safe common 
sense-wise, and meet electrical codes. A mistake in fooling with a 
typical 200A 240V main may cause much more than a spark and a pop.


Ed

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

Re: [volt-nuts] 34401A Why 10M ohm default i/p resistance?

[ed breya]

Only specialized meters can provide virtually infinite input R at 
voltages above the 10 to 20 V or so native range of conventional 
amplifiers, so you have to use some kind of attenuator to cover the 
higher ranges anyway. 10 megs and 1 meg (and sometimes 11) are the 
traditional values used, with 10 of course providing less loading on 
the signal source. It is difficult to get good resistor precision, 
stability, and voltage coefficient at higher values, so 10 megs is a 
good compromise.


As far as I know, no DVM uses an actual 10 or 1 gigaohm resistor for 
its input termination - that's just an equivalent input R range 
(sometimes just a part spec right from the datasheet) for the high 
impedance opamps and JFET circuits typically used to amplify DC at 
reasonably high accuracy and low noise. All it means is that it is 
nearly non-loading to conventional DC circuits. If there is an actual 
resistor this large in there, then it is just to get the input near 
zero when it's disconnected - it will read the bias current times the 
resistance, which can be quite large. If the applied input voltage 
exceeds the native range, the protection circuitry will take over.


For ultra-high Z applications, the equivalent input R would need to 
be in the teraohm range instead, using electrometer-class opamps, 
with much lower bias current, but higher offset voltage and noise.


If you put DVMs in the low ranges below 10 or 20 V, ones without 
actual termination R will tend to drift off due to input bias 
current. Once it's connected, the effect is much smaller (but not 
zero) since the source R is usually comparatively very small. One way 
to always assure a zero reading is to have a definite and fairly low 
(to not show bias current too obviously) input R, so there's the 10 
megs option. It's also possible to make the actual value of the input 
R (and not just the dividing ratios) very precise - or measure it - 
so that its effect on measurements at known source resistances can be 
figured out.


As you have already figured out, in auto-ranging, a non-terminated 
DVM left disconnected and unattended will form a relaxation 
oscillator and tend to wear out its front-end relays. Seeing no 
signal in the higher ranges, the system will switch down to the lower 
ranges and be OK until the input drifts off to a range limit, then it 
will up-range until it reaches one with an attenuator, then the 
signal goes back to zero, and the process repeats.


Ed

At 07:23 AM 4/10/2014, you wrote:
There is no suggestion in the specifications for the 34401A that the 
accuracy suffers by selecting 10G ohm input resistance on the .1 to 
10V range so why would they make 10M ohm the default? I can think of 
very few cases where having the 10M ohm i/p resistor switched  in is 
better for accuracy than not.


On the other hand 10M is sufficiently low to produce significant 
errors on a 6 1/2 digit DVM for sources with resistances as low as 
10 ohms. Measuring 1V divided by a 100k/100k ohm divider for example 
causes a .5% error - 502.488mV instead of 500.000mV. That might not 
be a problem but I wouldn't be surprised if this catches a lot of 
people out (including me) when not pausing to do the mental 
arithmetic to estimate the error. It's just too easy to be seduced 
by all those digits into thinking you've made an accurate 
measurement even though you discarded those last three digits.


And if it's not a problem then you probably don't need an expensive 
6 1/2 digit meter in the first place.


It's a small point I agree but it can get irritating to have to keep 
going into the measurement menus to change it when the meter is 
turned on when measuring high impedance sources (e.g. capacitor 
leakage testing).


It can't be to improve i/p protection as 10M is too high to make any 
significant difference to ESD and in any case there is plenty of 
other over-voltage protection. OK. it provides a path for the  DC 
amplifier's input bias current, specified to be  30pA at 25 degrees 
C, but I imagine that varies significantly from one meter to the 
next, and with temperature, so not useful for nulling out that error.


So why would they do this? Could it be psychological? By limiting 
the drift caused by the i/p bias current to 300uV max when the meter 
is left unconnected? A voltmeter with a rapidly drifting reading 
(several mV/s) when not connected to anything is a bit disconcerting 
and would probably lead to complaints that the meter is obviously 
faulty to users who are used to DVMs which read 0V when open circuit 
- because they have i/p resistance  10G ohms and don't have the 
resolution to show the offset voltage caused by the i/p bias current.


Personally I'd have though that the default should be the other way 
round - especially given that there is no indication on the front 
panel or display as to which i/p resistance is currently selected.


Any thoughts? What do other meters do?

Tony H

Re: [volt-nuts] monitoring LTZ1000 chip temperature

[ed breya]

 No Message Collected 
___
volt-nuts mailing list -- volt-nuts@febo.com
To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/volt-nuts
and follow the instructions there.

Re: [volt-nuts] Cool looking old stuff

[ed breya]

Sometimes the old stuff is still the best. At room temperature, a 
mechanical chopper combined with a high-ratio step up transformer and 
high impedance follower amplifier is unbeatable for low noise, low 
impedance signal amplification. The solid state devices that replaced 
mechanical choppers in most applications have much higher resistance, 
so more intrinsic noise.


You should save the chopper for possible future use - and if there is 
an input  transformer that goes with it, especially save that. Some 
circuits used a transformer (typically 1:100 ratio) to reach high 
gains for tiny signals into the nV DC region - without it, the noise 
level swamps the signal. The transformer is the key to the noise 
performance, but it wouldn't do much good without the low 
on-resistance of the chopper. The main problem with mechanical 
choppers is that they wear out - I think 10,000 hours is a typical 
life specification, and there's no way to know the history of a 
salvaged part. Transformers should last virtually forever, unless abused.


As an example, I built a transformer-based preamp for my lock-in 
analyzers, that provides 1000X gain with 300 pV/root Hz noise, when 
the source Z is low enough - around a few ohms. I could make this 
unit measure DC by putting a chopper in front of it, but it must have 
very low resistance to preserve the noise performance.


There is no conventional IC or discrete amplifier that I know of that 
can directly amplify with this noise performance, without using a 
transformer. The analyzer front-ends typically use regular low 
noise, broadband IC amplifiers that have around 7 nV/root Hz - about 
twenty times as much. The lock-in analyzer makers used to offer 
special transformers for this purpose.


There are tradeoffs, of course - the transformer circuits only work 
well over a limited frequency range like 1 Hz to maybe a few kHz, 
unlike the ~0 to 100 kHz that an analyzer can run. Also, they only 
make sense in low-Z circuits - higher Z adds lots of noise, and 
dramatically affects the transformer frequency response. As I recall, 
300 pV/root Hz is about the intrinsic noise of 50 ohms at room 
temperature, so in the more normal high impedance realm, conventional 
amplifiers are good enough.


Ed

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

Re: [volt-nuts] Some questions to zeners (1N823-1N829)

[Ed Breya]

Andreas,

I think your expectations are not realistic - even if you could make 
such a reference, you could not transport its voltage to the ADC without 
thermoelectric effects causing error that would swamp the performance. 
To keep everything below the 1 ppm/deg C range you would have to put the 
entire circuit in controlled temperature - the reference, the ADC, and 
the signal connection to the outside world. Constant temperature 
operation would help with the overall tempco and hysteresis, but the 
long term drift and noise will be intrinsic to the devices, and 
unpredictable except in a statistical sense.


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

Re: [volt-nuts] New Life of the Solartron 7081

[Ed Breya]

Yes, very impressive - Solartron should hire you as a consultant to 
design or improve their current products.


Ed

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

[volt-nuts] How long can standard cells last?

[ed breya]

I just junked out a very beat up old Fluke 803 differential 
voltmeter, and found deep within, an old-school Cd/Hg standard cell. 
It was well protected in an aluminum box, and wrapped in foam and 
foil. It looks brand-new, and still measures around 1.018... V. I'd 
like to keep this one as another reference point if it's still good. 
I assume that it just wasn't used much, or that the Fluke circuits 
were very good at not loading it down.


I'm sure it is the original unit installed in the instrument - marked 
5/12/1960. It is a Muirhead D-845-C. There's no test voltage tag or 
any other info but a serial number.


So, I'm wondering if a 52 year old standard cell can still be OK, and 
if anyone knows the specs on these, or where to find the info. I 
don't know if it's possible, but I'd like to find what the official 
voltage was supposed to be to a few more digits resolution. I think 
various types and brands each had slightly different nominal voltages 
around that determined by the basic chemistry. I remember in the old 
days, every one I saw included a sticker with the 25 deg C exact 
voltage measured as accurately as possible back then against the NBS. 
I'd like to especially know if this is a saturated or unsaturated cell type.


Ed


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

Re: [volt-nuts] 731A output impedance

[ed breya]

If you do modify the circuit, you will probably want to decrease the 
series R. Even though the external loads are likely to be in the 
megohm range, the feedback load is significant, and likely the 
biggest part, except for fault conditions. For lowest noise and 
suppression of effects due to bias current, the DC feedback network 
needs to be low in resistance, so 1k in series may add too much drop 
and make the amplifier run out of headroom, or change the amount of 
self-heating in the amplifier die, affecting it in unpredictable ways.


If you use series R in the 100 ohm range, it should have less effect 
on the normal amplifier conditions, and still provide some degree of 
fault isolation. You can add clamp diodes if it doesn't have them 
already. For 10 V output, I'd recommend a 12 V, 1W zener to ground 
right on the amplifier output, and a diode to the plus supply.


Ed


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

Re: [volt-nuts] 731A output impedance

[ed breya]

The simplest way to drop the output impedance without adding much 
circuitry is to just change the series R to 100 ohms or so - that 
would still give pretty good isolation from capacitive loading.


If the R is dropped to zero, the DC performance will be best, but 
you'll have to worry about the amount of capacitive loading. If the 
lines are short - say a couple of meters or less of open wire, it 
would probably be OK, but that much coaxial cable may make it oscillate.


The suggestion to get the feedback right from the output terminal,or 
even with external sensing at the load would be best for DC accuracy, 
but would have the same problems as above.


You can also take the DC feedback from the output directly, and the 
AC feedback from the amplifier output, while the series resistor 
isolates the two. This would give good DC accuracy and AC stability, 
but would alter the dynamic response and LF noise shape somewhat.


If you add an amplifier, you'll of course have to consider its offset 
and noise contribution, and it will have the same stability issues to resolve.


Ed


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

Re: [volt-nuts] 731A output impedance

[ed breya]

I'm not sure how much elaboration is needed, but here's some:

If you take all of the feedback from the output terminal, that's 
better for DC accuracy by eliminating the voltage drop of the series 
resistor, while still providing some overload protection to the 
opamp. But, it also decreases phase margin so that it will be more 
prone to oscillate with capacitive load. If the series R becomes 
zero, the voltage drop and the extra loss of phase margin are 
eliminated, but the inability to drive large capacitive loads remains 
- it is a limitation of the amplifier.


Usually a small amount of series R can help a lot with capacitive 
loading stability, but even when small it can drop enough DCV to be a 
problem. A common way to solve both problems is to sense the DC right 
at the output to eliminate the drop in the series R as above, but to 
increase stability by taking some AC ahead of the resistor - usually 
at the output of the amplifier.


If the amplifier has an integrating feedback capacitor, it's usually 
already connected that way, so only the resistive part of the 
feedback needs to go to the terminal. If there is no feedback 
capacitance, then a small amount can be added from the amplifier 
output to the effective inverting input.


I don't know what the output stage of the 731A looks like, but it 
must be an inverting (integrator) amplifier or a buffer, if using an 
opamp. In either case there should be a way to modify the feedback 
network. However, whatever is changed or added may affect the overall 
frequency response and noise.


Ed


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

Re: [volt-nuts] HP 3458A DC current accuracy

[ed breya]

Yes, the effect can be estimated quite easily. Also, keep in mind 
that the 0.13 nV is the RMS noise, so the peak to peak excursions can 
be around six times that, or almost 1 nV p-p. If the FS is 1 mA, then 
it's about 1 ppm - one count on a six digit DVM, or ten times more 
with each additional digit. In reality, one ohm would be commensurate 
with a higher range like hundreds of mA, so the noise voltage would 
be much further down as a percentage of FS.


Also, it's hard to amplify up from a very small FS value up to the 
native range of high resolution DVMs without adding even more noise 
and thermal effects, so it makes sense to use more resistance - and 
undesirable burden - to get a decent FS signal that needs less 
amplification, but the resistor noise would be higher. It's part of 
the tradeoffs needed in design.


When comparing current range performance to that of the voltage 
ranges, you should compare to the voltage range that is closest to 
the FS burden in order to take into account the noise and error 
contribution of the required amplification. Also, the specs should 
include the effect of amplifier bias current, which introduces 
additional error on high sensitivity (low I-FS) ranges.


In the pursuit of ever-higher DC resolution, once you get down to 
where individual to tens of nV are significant to your signal, 
realistic resistance values contribute significant noise. If your 
signals are big, then it's insignificant. That's one of the reasons 
why the native range of the DVM is made as big as possible.


Ed


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