Failure calibrating Fluke 720A

[perdrix]

I'm working on a Fluke 720A which isn't co-operating  .

Here's a set of resistance readings taken at the output terminals with the device in operate mode.

      Fluke 720A Output Resistance Values in Ohms               

Dial setting    Decade A   
0                    0.38   
1                10000.83
2                19998.90
3                29999.50
4                40000.00
5                50000.60
6                60001.20
7                70001.66
8                80002.10
9                90099.28
1.0             100003.45                  
CAL             110019.40


I am trying to follow the instructions on page 2.7 of the manual - with main function switch set to "Cal A", I set the dials to zero and balanced the bridge according to step e.

I then turned the A decade to 0.1 and the null detector was way out of balance - I reduced the range on the detector to 3mV full scale and it was 2mV out of balance.

No way can the A.1 trimmer accommodate that much unbalance  .

So there seems to be a big problem here ...  Any suggestions on how to determine what is wrong?

I'm trying to read the schematic to determine exactly what the bridge is comparing the lower left leg of the bridge with when the function switch is set CAL A and the A decade set to .0,
 and what it is comparing that to when the A decade is set to .1 but I'm failing miserably.

Many thanks, David

[perdrix]

My bafflement increases:  I measured the resistance between Bridge Power + and Bridge Balance - with the unit set to operate mode.

I measured a resistance of about 9992.6 ohms - that was after doing the bridge balance step.   That seems very low - I would have expected to see very close to 10000 ohms.   So what is the bridge comparing against when doing the bridge balance adjustment at step e. of the procedure?  That seems crucial to resolving this problem...

[Bill158]

So what is the bridge comparing against when doing the bridge balance adjustment at step e. of the procedure?  That seems crucial to resolving this problem...

The resistor that is the "reference" for all of the others is R312 and it's associated "trim" factory selected, and fixed, resistors R1045, R1046 and R1047.  When the 1st decade switch is set to "0" , SW1B is connected to the ".1" tap of the resistor string and SW1C is connected to the "0" tap, most of which is inside the "OIL TANK" along with the various trim resistors on the outside of the tank.  Observe that there is no variable trim resistor associated with R312.  Once you set the "bridge balance" at decade A switch set to "0" then you adjust the other trim resistors R1001, 1005, 1009, 1013, 1018, 1021, 1025, 1029, 1033, 1037, and R1041 as well as those other as directed by the CAL procedure.  To sum up R312 is the critical "reference" resistor.  If this one is way off the others can't be set to match it's value.
In the schematic SW1B and SW1C is shown connected across the R1005, 1007, 1006, 1008 and RR302 series/parallel combination.  This is when the decade A SW is set to "1.0".  This makes it a little difficult to understand what is happening in my opinion.
The first thing I would do is get a list of the "mv" offsets from the "0" setting of each of the decade A switch settings from ".1" to "1.0.  This can help you determine what is wrong.  If the offsets are all "+" or all "-" in relation to the "0" position then R312 or associated resistors may have an issue?
Also observe that there is R203 in the upper right hand schematic of the BRIDGE DETECTOR circuit.  This is to allow the BRIDGE BALANCE front panel control to be centered around R312.  See section 4-25 of the manual "Bridge Calibration" procedure.
Report back if still confused.
Bill

[perdrix]

Thank you Bill,

That all makes sense now!  The problem is now a lot clearer - R311 has drifted low by around 4.3 ohms so I need to insert an additional trim resistance between R1044 and R311 to get that set of resistors to right value.   I'll try two 2.2\$\Omega\$ in series ... I think that should be close enough given the adjustment range of the trimmer pot of 98.83\$\Omega\$ to 99.26\$\Omega\$

Thank you - "a little difficult to understand" the schematic pretty much sums it up!

Cheers, David

[perdrix]

Hmmm... I couldn't get a stable result with different values of added padding resistance.

So I re-measured R312 and R311.

 R312 - pretty stable reading of 9,895.298 +/- 0.004
 R311 - readings varying between 9,891.55 and 9,893.3

That's not good

The remaining resistors in the oil bath give stable readings.

On the principle of "nothing ventured, nothing gained" I hit R311 with 100V (1W dissipation) which is in spec as the allowed input is 1kV across all ten A decade resistors.  R311 promptly fell to 9,714.xx ohms, and still a bit unstable (though not as bad as before). so it looks like there's some shorted turns in there...

Is there any hope that this can be fixed, or should I "fold" and sell this unit for for spares?  Is it known what resistors were used in the oil bath?

David

[Bill158]

Is there any hope that this can be fixed, or should I "fold" and sell this unit for for spares?  Is it known what resistors were used in the oil bath?
David

The issue with R311 sounds pretty bad.  You are right that there are some shorted turns in that resistor.  See the picture of the insides of one 720a oil tank after tear down.

https://www.eevblog.com/forum/metrology/beatiful-precision-from-the-golden-age/

At least, in this unit, there were wire wound resistors.  But these could be simply that FLUKE used their normal high accuracy/high stability resistors before encapsulation?  This would make sense because most standard resistor elements, the kind NIST uses, are not encapsulated and are designed to be in an temperature controlled oil bath.
I would say that this unit is toast.
Bill

[TheDefpom]

I’m starting to think I should try doing a cal on my unit… haven’t done it since I got it a couple of years ago, but not really used it for anything precise.

I really hope I’m not going to find the same kind of problem.

[guenthert]

I would say that this unit is toast.

    That's harsh and a wee bit discouraging.  It's clearly out of spec and it seems it would be difficult to bring it back to original specifications (including the unwritten ones regarding temperature, power and long-time stability).  But since it's a KVD, it might still be possible to restore it to fulfill that role, even if just for a short duration every time (and in a lower input voltage range).  Can't call it a Fluke 720A anymore, but still a self-calibrating KVD, no?

[guenthert]

Hmmm... I couldn't get a stable result with different values of added padding resistance.

So I re-measured R312 and R311.

 R312 - pretty stable reading of 9,895.298 +/- 0.004
 R311 - readings varying between 9,891.55 and 9,893.3

That's not good

The remaining resistors in the oil bath give stable readings.

On the principle of "nothing ventured, nothing gained" I hit R311 with 100V (1W dissipation) which is in spec as the allowed input is 1kV across all ten A decade resistors.

Eleven.  If it is a KVD, then there will be eleven resistors to a decade.  I know, that was pretty pedantic   
And yes, if that resistor was made to withstand 91V indefinitely, it won't mind 100V for a short while.

 R311 promptly fell to 9,714.xx ohms, and still a bit unstable (though not as bad as before). so it looks like there's some shorted turns in there...

   Hmmh, I didn't expect the resistors value to change that much under load, but then, I didn't know what to expect.  Did R312's value change by a comparable amount?

    Shorted turns could be caused by a severe overload in the past and themselves cause a lower overall resistance, just as you noted.  I don't however see how they could cause the instability you mention.  What is meant by that?  How exactly are you measuring those resistances?

[perdrix]

R312 was unmoved by that! 

I was measuring them using a Solartron 7081 initially but switched to a 3458A - same behaviour.  I suspect that the shorts aren't consistently 0 ohms, and that the ohms checking didn't "punch through" that inter-turn resistance, but that 100V across the resistor did do so (to a large extent).   But even then I suspect that its still not 0 ohms and varies somewhat (as much as any poor contact does).

David

[Kleinstein]

The KVD chain has 11 resistors, but 2 of them are in parallel with the next decade and this way see only half the voltage. So the 9 other resistors would still see 1/10 the input voltage.

The instability could be from failing insulation between turns of a wire wound resistor. It may be from over-load, maybe lack of oil or corrosion or vibrations.

[bastl_r]

The instability could be from failing insulation between turns of a wire wound resistor.

I would agree with you if the resistances are lower than expected...