First of all, thanks to everyone for the valuable input.? I'm not by any
means done with this yet, but I thought I'd post an update before
proceeding.? Prelude: yes, I have checked the precision resistors and
they're all spot on according to my HP3490A as well as my Fluke 8800A.?
I knew about the trick to open all the switches.

My parts arrived today, so I figured - why not just try replacing Q11
with the (better) FET before I do any more tests?? It only takes a few
minutes and then I'll have more data on which to continue.? I made sure
of the pinout and tacked it in place.? Lo!? The voltage on the resistor
chain is now 0.995V and the meter reads 0.995 for a 1K 1% resistor!? I
then quickly checked it with my MC-7 resistance 'standard'.? It is a
tiny bit low on all ranges, but I suspect some of the numbers may have
shifted due to the many parts which have been replaced during
troubleshooting (plus CR13 is still out of circuit), so I'm going to put
the originals back one at a time and keep testing to make sure one of
them is not defective as well.? But I now have some confidence that Q11
was a major part of the problem and I can proceed.

George, based on your comments I think it's a good idea to check CR10-11
as well.? I saw some zero offset on DCV when testing it the other day -
is it possible that these may affect the DC offset since they're right
at the ADC input?? I see that they're Fairchild low leakage parts
according to the BOM, so I'll have to track down a suitable replacement
if they turn out to be degraded.

Yes, this version has the 10K pot and no R69.? When I get to the point
that I can calibrate the instrument I'll see how touchy it is and
consider updating that to a? 5K model.

Anyway, I have plenty left to do but I thought I'd provide an update to
"reward" you all for your assistance.? I think I understand this puppy a
lot better now.? ;-)

One additional thought - if the 4V Vgp FET works OK and the unit can be
adjusted to tolerance, will I see any benefit from tracking down a 3V
component, or have I just shown that it's unnecessary?? :-)

First of all, thanks to everyone for the valuable input.? I'm not by any means done with this yet, but I thought I'd post an update before proceeding.? Prelude: yes, I have checked the precision resistors and they're all spot on according to my HP3490A as well as my Fluke 8800A.? I knew about the trick to open all the switches.

My parts arrived today, so I figured - why not just try replacing Q11 with the (better) FET before I do any more tests?? It only takes a few minutes and then I'll have more data on which to continue.? I made sure of the pinout and tacked it in place.? Lo!? The voltage on the resistor chain is now 0.995V and the meter reads 0.995 for a 1K 1% resistor!? I then quickly checked it with my MC-7 resistance 'standard'.? It is a tiny bit low on all ranges, but I suspect some of the numbers may have shifted due to the many parts which have been replaced during troubleshooting (plus CR13 is still out of circuit), so I'm going to put the originals back one at a time and keep testing to make sure one of them is not defective as well.? But I now have some confidence that Q11 was a major part of the problem and I can proceed.

George, based on your comments I think it's a good idea to check CR10-11 as well.? I saw some zero offset on DCV when testing it the other day - is it possible that these may affect the DC offset since they're right at the ADC input?? I see that they're Fairchild low leakage parts according to the BOM, so I'll have to track down a suitable replacement if they turn out to be degraded.

Yes, this version has the 10K pot and no R69.? When I get to the point that I can calibrate the instrument I'll see how touchy it is and consider updating that to a? 5K model.

Anyway, I have plenty left to do but I thought I'd provide an update to "reward" you all for your assistance.? I think I understand this puppy a lot better now.? ;-)

One additional thought - if the 4V Vgp FET works OK and the unit can be adjusted to tolerance, will I see any benefit from tracking down a 3V component, or have I just shown that it's unnecessary?? :-)

Can you give me the HP part numbers you need? It is possible we have them in stock, but if HP bought them I have a copy of the “purchased parts” database and it generally lists most of the specs for the parts

Regards,

?

Stephen Hanselman

Datagate Systems, LLC

3107 North Deer Run Road #24

Carson City, Nevada, 89701

(775) 882-5117?office

(775) 720-6020?mobile

s.hanselman@...

a Service Disabled, Veteran Owned Small Business

DISCLAIMER:

This e-mail and any attachments are intended only for use by the addressee(s) named herein and may contain legally privileged and/or proprietary information. If you are not the intended recipient, any dissemination, distribution or copying of this e-mail and any attachments is strictly prohibited. If you have received this e-mail in error, please immediately notify me and permanently delete the original and all copies and printouts of this e-mail and any attachments.

On Feb 12, 2019, at 18:37, gregdunn@... wrote:

That's weird - I thought I saw another response to this thread but when I refreshed the page, it disappeared!? :-(

That's weird - I thought I saw another response to this thread but when I refreshed the page, it disappeared!? :-(

Greg,
I agree, it is time to stop pulling parts.? ?This meter was designed before electronic calibration where gain and offset constant are stored in memory to correct for circuit imperfections.? ?The first DMM Hp produced with electronic calibration was the Hp3468A which I was the principle analog designer.? So not having electronic cal, the current source will need to have the converter currents very close to cardinal values hence the need for adjustment pots.? ? While designing DMMs at Hp, I never?gave this instrument serious consideration so never studied this particular design but upon looking over the schematic you posted, this converter is unusual for it SINKS rather than sources current to develop a voltage drop across the unknown which is measured with the voltmeter section of the meter.??

Before you go further, make sure the voltage across the range resistors is not shifting.? I am assuming the range resistors (R52-R56) have not changed in value which you might want to eventually check by pulling the left lead of R52 up? (lead connected to?gate of Q15A)?and put the meter in the 10M range when checking the resistors.? These should be precision matched resistors - possibly the old custom made wire wound Red Devils they used back then.? That is just a guess for I have never seen the inside of one of these meters.

First connect a very high impedance voltmeter between the gate of Q15A and the output of U3 and step through the ranges.? This voltage needs to be solid, not changing, at close to 1.000 volt except on the 10Meg range for that voltage drops to 0.100 V on the 10M range.? The importance that this voltage be solid as you step thru ranges cannot be overemphasized.? If this voltage is solid and the range resistors have not shifted in value, then the reference currents are correct and the problem has to be an unacceptable leakage path between this part of the circuit and the output terminals.? ?A leakage path anywhere from the gate of Q15A to the front panel terminals will affect the lower value sink currents used on the higher ranges more than the higher sink currents used on the lower ranges.? Such a leakage path will cause an increasing larger error as you step up in range as the sink current pulled from the unknown becomes less by a factor of 10 as you change ranges.?

A critical component that could cause a significant leakage path is the AC feedback capacitor C1 that is used to insure proper operation of the differential FET pair Q15.??This capacitor MUST have a quality dielectric.? Any leakage due to this capacitor's equivalent shunt resistance will significantly compromise the operation of the reference currents.? You might also check the PCB to make sure it is clean and clean with alcohol around the ohm's converter and DC input.? in the protection circuit (Q11- Q14, CR12 - CR14, R39 - R43) or elsewhere?There might also be a leakage path in the voltmeter circuit since it is connected to the unknown to measure the voltage across it and any leakage there will also degraded the current source.

The differential pair needs to maintain a solid voltage at the gate of Q15A equal to Zener voltage of CR16 (approx -6.3v). The voltage at the gate of Q15A should be solid and not change as you step through ALL the ranges including the 10M range and match that of the Zener CR16 at pin 3 of U3.? These two voltages should be within 10ths of a volt of each other indicating the differential pair is balanced and operating properly.? The FET pair needs to be well matched since there is no DC feedback to insure balance of the differential circuit -- diff pair is operating open loop at DC.? Any abnormal gate current into Q15A (JFET) will also degrade the reference currents of the converter just as any other leakage current will.? The gate current should be on the order of pA.? Ideally one would like there to be no gate current at all.

When the meter range is changed from 1 Meg to the 10 Meg, the voltage at the output of U3 (pin 6) should drop from approx.? -7.3v? to? -6.4v.? That is to say that the voltage at the output of U3 needs to be? 0.1v lower than at the gate of Q15A when the instrument is properly adjusted when calibrating the ohms converter.??

There seems to be potential problem with the circuit involving U3 that I can see.? The feedback network of R58, R59, R60, R61 is fine as shown, allowing for plenty of adjustment on the 10 MEG range and also sets the amplifier gain on the lower ranges.? There should be no interaction when adjusting the R59 and R64 pots during calibration.? The adjustments are indeed independent since any wiper arm setting on R59 for the 10 MEG range does not change the parallel resistance of R59 and R60 in this part of the feedback network when a lower range (1MEG and lower) is selected given the pot wipe arm is switched out of the circuit on the lower ranges.? However, the values as given in the schematic for the gain setting resistors in lower leg of the feeback voltage divider?around U3 will be problematic when trying to calibrate the instrument if the Zener?voltage of CR16 is lower than 6.373 volts.? The ohms converter will have a larger reference current than required even with the R64 pot set to zero when R69 is 3480 ohms as indicated.? If R69 is reduced in value to 2200 ohms, then the adjustment range of the 1K pot,? R64,? will accommodate the Zener,? CR16,? having a nominal Zener voltage of? 6.2v? ( 2% tolerance or better required).?

Measure the Zener voltage of CR16 and if it is lower than 6.372 volts, R69 will need to be reduced from 3480 ohms for the calibration adjustment to work.
The optimum value for? ?R69? =? 2200 ohms? ? ?----? see below

CONCLUSION:
This discussion should allow you to better understand the ohm's converter circuit and troubleshoot any problems you are having.? ?First check your reference currents for proper values by measuring the voltage across the range resistors to make sure it is stable at 1.000 volt on the lower ranges and 0.100 volt on the 10 MEG range.? If this voltage is stable and the range resistors are the correct value, then you certainly have a current leakage issue.? ?Check capacitor C1 first for leakage and the input gate current on Q15A.? Next check the protection circuit for leakage specifically around Q11 and CR13.?

------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
? ? ? ? ? ? ? ? ? ? ? DESIGN /? SELECTION? ?of? ?ADJUSTMENT? ?CIRCUIT? ?RESISTOR? ?VALUES? for? FEEDBACK? NETWORK? around? U3

? R58? =? 350? ? ? ? ? ? ? ? ? ? ? ?R60? =? 110? ? ? ? ? ? ? ? ?R63? =? 24.3 K? ? ? ? ? ? R69? to be determined? ( schematic gives 3480 ohms but value too high )
? R59? =? 10 K pot? ? ? ? ? ? ? ? R61? = 3.9 K? ? ? ? ? ? ? ?R64? ?=? 1 K pot

? R59 // R60? =? 109 ohms

In order to adjust the reference voltage across the range resistors to be? 1.000 volt? on the lower ranges so that the reference current will be a cardinal value of milliAmps ( 10mA, 1mA, 0.1mA? etc.), the relationship between the Zener voltage, Vz, the?feedback resistance, Rf, and the combined resistance of the lower leg of the feedback network, R1,? is? ? ? [[? ?Vz *? Rf? =? R1? ? ]]? ? ? ? ? ? since? ?1v? ?=? Vz * ( 1? +? Rf / R1 )? -? Vz

? ? ? ? ? ? ? ? ? ? ? ? ? ?Rf? =? R58? +? R59 // R60? +? R61? =? ?4359 ohms? ? ? ? ? ? --0--? ? ? ? ? ? ? ?R1? = ( R69? +? R64 ) +? R63? =? ( R69? +? R64)? +? 24.3K?

? ? ? ? ? ? ? ? ? ? ? ? ? ? ?Vz * (4359)? =??( R69? +? R64)? +? 24.3K?
? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? Rf? ? ? ? ? -----------? ? R1? ?-----------

? ? ? ? ? ? ? ? ? ? ? ? ? ? ?( R69? +? R64 )? ?=? ? Vz * 4359? -? 24300

? ? ? ?so if? R69 = 2200 ohm,? ? then? ? ? ?( 2200? +? R64 )? ?=? ? Vz * 4359? -? 24300? ? ? ? or? ? ? solving for Vz:

? ? ? ? ? ? ? ? ? ? ? Vz? ?=? ?(? 26500? ?+? ?R64? )? /? 4359?

? ? ? ? ? ?R64 is a 1K pot:? ? ?0? <? ?R64? <? 1000? ? ? ? ? ? ? ? ?add? ?26500? and divide by? 4359? throughout the inequality:

? ? ? ? ? ? ? ? ?26500 / 4359? <? ?[( 26500? +? ?R64 ) / 4359]? ?<? ?(1000? +? 26500) / 4359??
? ? ? ? ? ? ? ? ? ? ? ? ?6.08? ? ? ?<? ? ?[ Vz ]? ? <? ? ? 6.31? ? ?


?For a 2% Zener with nominal Zener voltage of? 6.2v,? ?Vz? =? 6.2v? +/-? 0.12v? rewritten as? 6.08? <? Vz? <? 6.32? ?which matches the designed adjustment range
?above of? ? 6.08? ?<? ? [ Vz ]? ?<? ?6.31? ? ?

? ? ? ? ? ? ? ? ? ?The proper choice for? ? R69? ?is? ?2200 ohms? if R64? is a 1 K ohm adjustment pot