Keithley 178 DMM repair - advice needed

[rich]

I'm part way through restoring this meter and need to replace a feedback resistor in the input attenuation stage. I'm not sure what type of resistor I should use to keep the original accuracy - I guess "MtF" is metal film but beginning to doubt myself as the original 1970s resistor is large compared to what my searches are returning    

From the service manual:
998k 0.1% 0.25W MtF;   Mfr: TRW;   Mfr desig: Type: MAR-7 T13;  Keithley# R264-998k

Not sure of TC ppm as google returns nothing useful for me, and the 998k value seems quite non-standard so I wondered whether it would be sacrilege to series up a 487k and 511k if it came to it?

My searching hasn't turned up anything I translate into a purchasing decision, please could someone with the knowledge point me in the right direction of how to go about choosing a replacement? Thanks.

Also, I've made notes and taken photos along the way, so will post them up if there is interest.

[Cloud]

I asume, that it is high voltage, so thats why its so big. And as far as I know the bigger resistors usually have better temperature coefficient

[Dr. Frank]

You might answer these questions yourself, simply have a look into the AC specification, 2V range..

It's 0.4% only, and a T.C. of about 400ppm/K. So a 50ppm/K metal film resistor, 250mW is ok.
998k is a special type, 0.2% low from 1M. It's value has been chosen to have a +/- 0.25% trimming possibility. You might also get some 1M 1% resistors and select for these 998k.
Any serial combination is ok, but you also have to select 2 resistors for the correct total value to 0.1%, as you want to trim this range properly.
Only problem might be the frequency compensation, so a single resistor is preferred.
Use well reputed brand.

This resistor is loaded with about 2Veff max, as it's used  in the feedback loop.

So it is nonsense, that it should be intended for high voltage, also nonsense, that bigger sized resistors have lower T.C.

In the very past, i.e. 60ties and 70ties, precision resistors used to be bigger, but technology advanced, and they got smaller, with same or better specification.

Frank

[rich]

Thanks Dr Frank, are you getting the T.C of 4ppm/K directly from the published meter specs T.C of 0.04% on the AC 2V range?  What may not have been obvious because I chopped the lower half of the schematic is that this resistor is also used for the 20V DC range, so would it be more correct to target the 0.006% T.C from 20V DC  range?  (Yes, I know 50ppm < 0.006%, just trying make sure I understand where 400ppm/K came from)

[Dr. Frank]

Thanks Dr Frank, are you getting the T.C of 4ppm/K directly from the published meter specs T.C of 0.04% on the AC 2V range?  What may not have been obvious because I chopped the lower half of the schematic is that this resistor is also used for the 20V DC range, so would it be more correct to target the 0.006% T.C from 20V DC  range?  (Yes, I know 50ppm < 0.006%, just trying make sure I understand where 400ppm/K came from)

You're right. I concluded from the AC spec only.
Now, I downloaded the complete schematics.
Therefore you need a 25ppm/K resistor, or better, as additional T.C.s are added (e.g. reference).

Btw, what's wrong with this resistor?

Frank

[rich]

Thanks for working through the other T.C.s I shall have a good hunt around for suitable resistors tomorrow

Btw, what's wrong with this resistor?

The resistor in my meter isn't the original - somebody has been inside before me and swapped it out for two resistors totalling 978k (the picture in post #1 was a library pic so not my meter).  The soldering suggests it may have been intended as a temporary fix but not sure why they chose to make 978k instead of 998k. I did find that Vref was under spec so perhaps the 978k was chosen to attempt to compensate for an incorrect Vref    

The Vref problem is the divider uses a 2.15k resistor (0.1%) in the high side to produce 1V from 6.3V, unfortunately today this resistor measures 2.25k so Vref could have been 4-5% under spec (I can only trim it up to 0.98C with the 2.25k still in circuit).