This follows on from my question about cleaning a 400 k ohm resistor in a
34970A data acquisition unit, but I thought the tasks were sufficiently
general to warrant another thread.

I've replaced a 400 k ohm resistor in what is effectively a 6.5 digit
multimeter, as a service note indicates that the resistor can drift,
putting the instrument out of specification on the 10 M ohm and 100 M ohm
ranges. I bought the resistor from Keysight, as it was the only source of
this odd value. As far as I can see, the specifications are 400 k ohm, 1%
tolerance, 2 ppm/deg C temperature coefficient. Andreas Jahn mentioned that
simply soldering a metal film resistor will change its value by several
tens of ppm. I don't know what type of resistor this is. Poul-Henning Kamp
wrote

"You can bring that down both in PPM and time by a "degauss"
temperature-cycling sequence [+N, -N, +(N-2), -(N-2), +(N-4), -(N-4) ...]"

I'm not sure how to interpret that equation, or how best to do this with
the limited equipment I have available. I don't have any environmental
chamber where I can carefully control the temperature. What I do have
available to me is

1. Large domestic chest freezer that's not in use, and could be pressed
   into service to cool something. I believe that cools to about -18 deg C.

The operating temperature of the instrument is 0 to 55 deg C, and the
storage temperature is -40 deg to + 70 deg C, although storing above 40 deg
C decreases the battery life.

2. I could put a heater in the chest freezer, and hook up some electronics
   so the freezer and heater acts to heat and cool.

3. Switching the unit on makes the internal temperature rise about 5 deg C
   hotter than ambient, so power cycling the unit will cause a change of
   internal temperature, but not very rapid and not very much.

Since changing the case on this unit, for one that's ventilated, the
instrument runs a lot cooler than it used to. Previously the side panels
got uncomfortably warm to hold for a long period, but now its almost
impossible to see from the side panels the unit is on. So I'm guessing that
changing the case, as detailed on a service note, will have reduced the
temperature of a lot of internal components quite a bit.

--
Dr David Kirkby Ph.D C.Eng MIET
Kirkby Microwave Ltd
Registered office: Stokes Hall Lodge, Burnham Rd, Althorne, CHELMSFORD,
Essex, CM3 6DT, United Kingdom.
Registered in England and Wales as company number 08914892
https://www.kirkbymicrowave.co.uk/
Tel 01621-680100 / +44 1621-680100

I don't think you need to take that temperature profile terribly seriously.  The intent is to relieve the stresses by alternately cooling and heating.  The amounts aren't critical, but the idea is to reduce gradually the excursions from room temperature bit by bit.  Eventually the internal temperature will stabilitze and, if you let it sit for a long enough time, will also be pretty constant throughout the resistor.
Then you can use the resistor and expect it to be in tolerance.
If the replacement resistor is of 1% tolerance, that is not terribly tight, and I wouldn't worry much about its drift.  If you want to go tighter, you can make up an assembly of a few resistors to equal the desired value.  Wirewound resistors probably have better long term stability than other types.  A 400k wirewound isn't too common but I have some in that range.  They are old, which is an advantage in that probably most of the aging has already occurred.
Yes the temperature rise within the unit directly affects accuracy, although not in any really predictable manner.  When measuring high resistance you don't have need for a 4 wire setup but you do have to think about using a guard.  These will help confine the current path to the desired route.  A 400 k resistor will move about 1% when shunted with a path of resistance of 40 Megohms.  That's not likely in practice so it's not really an issue.  Make sure the board is clean, if it's board mounted.
You can measure a resistor of that value by use of a calibrated voltage source and a current shunt.  The shunt should be of a value you can measure very accurately, as well as the voltage source.  Be cautious regarding self heating of the device under test; it should dissipate at least an order of magnitude less than its rating.  And the derating chart will tell you how much its internal temperature is likely to change for a given power dissipation.
Bob
On Saturday, June 29, 2019, 02:04:00 PM PDT, Dr. David Kirkby drkirkby@kirkbymicrowave.co.uk wrote:

This follows on from my question about cleaning a 400 k ohm resistor in a
34970A data acquisition unit, but I thought the tasks were sufficiently
general to warrant another thread.

I've replaced a 400 k ohm resistor in what is effectively a 6.5 digit
multimeter, as a service note indicates that the resistor can drift,
putting the instrument out of specification on the 10 M ohm and 100 M ohm
ranges. I bought the resistor from Keysight, as it was the only source of
this odd value. As far as I can see, the specifications are 400 k ohm, 1%
tolerance, 2 ppm/deg C temperature coefficient. Andreas Jahn mentioned that
simply soldering a metal film resistor will change its value by several
tens of ppm. I don't know what type of resistor this is. Poul-Henning Kamp
wrote

"You can bring that down both in PPM and time by a "degauss"
temperature-cycling sequence [+N, -N, +(N-2), -(N-2), +(N-4), -(N-4) ...]"

I'm not sure how to interpret that equation, or how best to do this with
the limited equipment I have available. I don't have any environmental
chamber where I can carefully control the temperature. What I do have
available to me is

1. Large domestic chest freezer that's not in use, and could be pressed
   into service to cool something. I believe that cools to about -18 deg C.

The operating temperature of the instrument is 0 to 55 deg C, and the
storage temperature is -40 deg to + 70 deg C, although storing above 40 deg
C decreases the battery life.

2. I could put a heater in the chest freezer, and hook up some electronics
   so the freezer and heater acts to heat and cool.

3. Switching the unit on makes the internal temperature rise about 5 deg C
   hotter than ambient, so power cycling the unit will cause a change of
   internal temperature, but not very rapid and not very much.

Since changing the case on this unit, for one that's ventilated, the
instrument runs a lot cooler than it used to. Previously the side panels
got uncomfortably warm to hold for a long period, but now its almost
impossible to see from the side panels the unit is on. So I'm guessing that
changing the case, as detailed on a service note, will have reduced the
temperature of a lot of internal components quite a bit.

--
Dr David Kirkby Ph.D C.Eng MIET
Kirkby Microwave Ltd
Registered office: Stokes Hall Lodge, Burnham Rd, Althorne, CHELMSFORD,
Essex, CM3 6DT, United Kingdom.
Registered in England and Wales as company number 08914892
https://www.kirkbymicrowave.co.uk/
Tel 01621-680100 / +44 1621-680100

volt-nuts mailing list -- volt-nuts@lists.febo.com
To unsubscribe, go to http://lists.febo.com/mailman/listinfo/volt-nuts_lists.febo.com
and follow the instructions there.