Restoration : Fluke 720A KVD/Ratio standard

[TiN]

Intro

Something strange happened today. I bought F720A's resistor decades half a year ago, and they were sitting in the lonely box quiet some time.
Today they transformed themselves into complete Fluke 720A Kelvin-Varley Divider box, just like in those sci-fi movies!! 



After transformation:



Exterior



Nothing on the rear, as this box does not have a single semiconductor component in it

Old school Fluke logo, yellowed switches and dirty connectors:



And to make it more exciting and fun learning experience, this poor KVD is...



I don't want to disappoint Dr.Frank with more unverified calibrations, so I had to have KVD for transfers and nV-comparisons between various standards, such as LTZ modules and sources. I had preamp and card for K2002 for quiet a while, thanks to chuckb, but had no divider to calibrate that setup to 2mV,200uV and 20uV levels properly. Hope now with 720A I'll be able to do so eventually. The calibration reference for 720A will be of course 3458A.

Hope we can fix this toy and see what she's got for us.

[Vgkid]

I will be looking forward to this.

[TiN]

I already hear shouting - tear it apart, what you waiting for...

Alright... remove some screws and we're in like Flynn...



CAUTIONs everywhere. Keep children and TiN's away....



Bottom view with glorious oil-filled resistor chamber for most critical decade A.

All components seem to be in place, nothing stolen or missing. Maybe except screw or two...

Internal switch for calibration:



Little board with beautiful Fluke-made teal resistors to do something:



What's with that solder spec near screw?



If I'm reading that correct, datecode on on decade A switch is 6th week 1979, and 37th week 1978 on 720A-3001 board resistors. Older than me!

Function switch:



Take a look on decades more closely. A,B,C:



And F (D and G are same)



I had shown separate decade photos in my previous article, including individual resistor photos.

Bottom PCB have long array of low-resistance trimmer wirewounds, variable 1-turn potentiometers and multiturn trimpots accessible from front panel to implement calibration functionality.



Also connection to input and output binding post caught my eye, seem to be dodgy soldering. Seems like failed repair attempt by previous owner?



Let's see more oil-filled chamber:



Hermetic pass-thru terminals, color coded wires, filling port. Isn't it beautiful?



And some more wiring:





Little 5K KVD-pot and view on front panel trimpots location.

[TiN]

Input resistance measurements shows >1Gohm on any A-decade setting, but sometimes shows 108KOhm if I wiggle it here and there.
But even then 10V applied on input and setting ratio to 0.9xxxxxx gives not 9V but some arbitrary value. No good.

I checked output terminals resistance with HPAK 34970A..



Not sure if values are what supposed to be, yet, need dig that manual.

[Jay_Diddy_B]

TiN,

I am not sure what you are measuring. But, I am quite happy to make some measurements on my 720A for you.

Here is a link to mine: https://www.eevblog.com/forum/testgear/list-your-test-equipment-score-here!/msg946456/#msg946456

Regards,

Jay_Diddy_B

[TiN]

Just output resistance on output HI/LO terminals, each dial (rest of dials set to zero).

Thanks for link, I remember there was thread about 720, but could not find it quick enough.

What is your resistance on input 1.0/LO terminals of 720, with dials set at 1.0000000, in OPER mode?

[Jay_Diddy_B]

Hi TiN and the Group,

My Fluke 720A isn't in perfect condition there are one or two dirty contacts in a couple of places.

From the 1.1 input terminal to Lo the resistance is 110.00 kOhms with dials set to 1.00000000

From the 1.0 input to the Lo the resistance is 99.999 52 k Ohms, with the dials set 0.999 999 9

The 1.0 input with the dials set to 1.000 000 0 is not a valid operating condition but mine measures about 107.5k Ohms.


The input resistance in the manual is

100.00 K +/- 0.005% at 1.0 Input

110.00 k +/- 0.005% at the 1.1 Input

So mine is correct.

I checked the linearity very quickly. I applied 10V from a Fluke 720A to the 1.0 input and Lo. I measured the output with a Datron 1281.

I went through each dial, setting the other dials to zero, and the Datron displayed the correct voltage.

I did have an offset of 6uV (0.6ppm of the 10V input) This is probably thermal EMFs in the leads. I just used random leads.

 The combination of the Fluke 720A and Datron 1281 was near perfect.

I have not adjusted my F 720A at all. There are some built in bridge circuits to adjust the divider.

The output resistance that I measured on the first three decades are very similar to yours.


I would check your output voltage with your (famous) HP3458A using a 10V reference on the input.

Regards,

Jay_Diddy_B

[Macbeth]

I'm thinking this is nearly as easy a fix as the 3245A

Waggle some leads and get open circuit or ~ 100 - 110 kohm? 

[quarks]

Looking forward for more, thanks for sharing

[Dr. Frank]

I don't want to disappoint Dr.Frank with more unverified calibrations, so I had to have KVD for transfers and nV-comparisons between various standards, such as LTZ modules and sources. ...

Illya,

I really apologize, being a square, sometimes. I know, I might have been too critical..

Anyhow, congrats to your nice catch of this 720A, seems to be repairable..


Looking at your pictures, somebody soldered on the switch of  decade B, all these paper numberings do not belong into the instrument.
I assume, this decade has been exchanged.

Your output measurement clearly indicates, that the chain of Decade A resistors is unbroken, all resistors seems to be fine, as a few ppm variation only is visible, (@ 2W measurement!), so it's still very linear.
Decade A switch seems to be ok in first instance, as well as Function Switch.


The whole first decade ( 12 x 10k) is normally assembled in a fixed manner between the "1.1 INPUT " and "Low Input", so you must always measure constant 110k, disregarding any setting of Decade A switch and the Function switch. You must always measure constant 100k between "1 Input"  and "Low Input".

Therefore, if you measure open, or unstable, then there are two possibilities only (in order of probability):

1) R 1047, that is sort of contact resistance compensation, is defect, or not assembled correctly
2) Somebody unsoldered connection from "1.1 Input", "1.0 Input" and "/ or "Low Input" and did not re-assemble that correctly.
3) the cables, involved in 2), may be broken.

So check at first, if "Low Input" has a low impedance connection to pin 1 of oil bath assembly, or to "Low Output".
These junctions (and maybe the cables) are suspect, so check these, and measure connection from there to resistor chain of Decade A:



Check at first the fat black and red cables from the input jacks, if they are broken.
 
I think, if you find that bug, the divider will work principally, directly after repair.
Proper precision operation will be another point.. a lot of debris is visible, which might require careful cleaning.

Frank

[TiN]

I really apologize, being a square, sometimes. I know, I might have been too critical..

Nothing to apologize about, it's always a good thing to get some fresh cold vision on things. We all easy to get overexited of our own work or buying 8.5-digit meter, thinking "oh, NOW i can measure stuff", and proceed with horrid mistakes, only measuring errors to 8.5-digit precision. That's what forums are for, design reviews, eh?   

all these paper numberings do not belong into the instrument.

That was my thinking as well. I did not see much 720A photos on web, but those I saw never had any paper labels anywhere.

I'll have to draw a partial schematics, which wires go where, to make clear vision how all interconnected. After that isolation of the issue would be matter of minutes.

Not sure if that will be done these days, as there is something bigger going on with my homelab (no, not getting second rusty 3458A , so this project might suffer little lack of attention next few weeks.
I don't want to rush into fixing this, as 720A not kind of instrument to accept ungraceful handling and user mistakes.

[Dr. Frank]

I really apologize, being a square, sometimes. I know, I might have been too critical..

Nothing to apologize about, it's always a good thing to get some fresh cold vision on things. We all easy to get overexited of our own work or buying 8.5-digit meter, thinking "oh, NOW i can measure stuff", and proceed with horrid mistakes, only measuring errors to 8.5-digit precision. That's what forums are for, design reviews, eh?   

Thank you for your kind words 

I'll have to draw a partial schematics, which wires go where, to make clear vision how all interconnected. After that isolation of the issue would be matter of minutes.

Not sure if that will be done these days, as there is something bigger going on with my homelab (no, not getting second rusty 3458A , so this project might suffer little lack of attention next few weeks.
I don't want to rush into fixing this, as 720A not kind of instrument to accept ungraceful handling and user mistakes.

That's a pity.. I assume this will really be an easy repair.
The labelled switch is not involved, that may very well do fine.

If I understand your description of the input resistance behaviour correctly, the error definitely has to be a broken solder junction, or this black cable from "Input Low" is broken. That's one simple Ohm measurement, perhaps.

Frank

[TiN]

Wires to decade switch A, R1047 and wires are OK.
But resistance check between oil tank pins 22 and 23 is open.
Resistance across 23 and 24 is 9.896K. 20-19 and so on are 7.2xxK.

Also from 23 to Input 1.0 is open.

[Jay_Diddy_B]

Hi group,

TiN wrote:


Wires to decade switch A, R1047 and wires are OK.
But resistance check between oil tank pins 22 and 23 is open.
Resistance across 23 and 24 is 9.896K. 20-19 and so on are 7.2xxK.

Also from 23 to Input 1.0 is open.



For those following along at home, here is a partial schematic:



The full schematic is included in the 720A manual.

Link: http://us.flukecal.com/literature/product-manuals/720a-instruction-manual

Hopefully it is a wire somewhere.

Regards,

Jay_Diddy_B

[lowimpedance]

Looking at the oil tank filling port 'plug' , the cap screw seems out of place as a plug that fluke would have used  not sure without peeking inside another one !.

[Dr. Frank]

Me and my big mouth.. this section is (of course) not included in your output resistance measurement.

So the error is definitely (again?) in the trimming section of R302, (that is the 9.896k resistor between pin 21 and pin 22).

Either R1008 is open (~5 Ohm), or one of the 3 cables, between pin 23 and R1005/R1007, or R1006/1007 towards R1008, or between R1008 and pin 22.
 
Maybe the 4 fold junction is open, where the cable from "1.0 Input", from pin 23, to R1005/R1007 and to S1B meet.
Frank

[Jay_Diddy_B]

Looking at the oil tank filling port 'plug' , the cap screw seems out of place as a plug that fluke would have used  not sure without peeking inside another one !.

TiN has showed us his oil plug, here is a picture of mine:




I think they look pretty similar.

Here are some more pictures from my unit, serial number 164.

Top View



Cal switch and decades A,B and C






The Oil Tank with matching serial number



The adjustment wire R1047




Regards,
Jay_Diddy_B

[TiN]

Can someone take good pic of dec.A switch from bottom side (where oil resistor visible).
I checked wire from pin 22 to trim network - its ok, R1008 is ok too around 5.2 ohm, but pin 23  does not come to trim network, but just to switch S1B position.
Also Input 1.0 does not go same pin too. Seems wrong wiring on switch S1. As I clearly can tell it was soldered badly.

[ManateeMafia]

Here you go.

[Dr. Frank]

Gotcha, two cables missing, at the brown "1.0 Input" cable.
One of them is pin 23, black and white colored.

Frank

[Jay_Diddy_B]

Hi,

Here are my pictures of the same thing 






(Where else can you get pictures of switch) 

Jay_Diddy_B

[Jay_Diddy_B]

Hi group,

There are actually two black and white wires attached to the brown wire (1.0 input) and the switch:



Jay_Diddy_B

[ManateeMafia]

I recently picked up a second "720A" from an auction recently. I grabbed it for parts so wasn't expecting much.

I wanted to make some comparison measurements for TiN but something is definitely wrong here.... FYI, all resistors have 82 datecodes. No mica ww in this beast. Anybody recognize the sig on the backside of the pcb's?

The pics I uploaded are from 720A #1.

[TiN]

(Where else can you get pictures of switch)

....inside of 20000$USD resistor....

You guys the Best!

Here's what I see here:





Hmm, what is connected that point on to Decade B ?

[TiN]

...

That is no 720A....    I'd expect no 0.1ppm out of it, whatever it is