Keithley 171 vintage nixie tube 4.5 digit DMM

[PfannenHans]

This is my first real post, so I hope I get the formatting right...
I am also not quite sure if Test Equipment or Repairs is the right topic, but since it is not realy a repair I chose Test Equipment.

A few weeks ago I got myself my first bit of vintage test gear, namely a Keithley 171 DMM from the early 70s. Since I could not find a thread about it in the forum I thought I make some photos and share my experiences.
I got it from Ebay for about 80€. It was described as "turns on and the display is working". When it arrived I couldn't get it to measure anything, so I got it open and had a look around:
The culprit was a blown input fuse, which was sandwiched between the analog circuitry board and the range switch board:
After replacing it with a new one i got the DC voltage, ampere and ohm ranges working, but no luck with the AC ranges.
The manual mentions (see https://download.tek.com/manual/171(Model171_Multimeter).pdf) states that both AC ranges use the same AC/DC converter circuit before the signal goes in the common siginal path (see PDF p.17).
An impedance transforming JFET (TO-72 2N4220) is right at the beginning of the circuitry (PDF p.84), which I suspect to be dead.
Unfortunately it is on the range switch board, right under the range switch, which is soldered down and blocks any attempt to measure it in circuit. Q1101 is the Can right in the middle of the picture above the lowest bar:
I tried to get the range selection board out of the unit, but I already fail at the knob on the front panel, which is screwed in with what I think is a weird imperial hex screw for which I do not have a fitting bit (1.5mm slips, 2mm is to big)
But since I will anyway only use the DC and Ohm range it is not so big of a problem.
Last but not least a shot of the Nixie tube display, which uses NL-841s from National Electronics. I can not compare them to other Nixies, but I think they are in a pretty good shape for their age:
Component wise the DMM is full of carbon film resistors which could become a problem in the future. The electrolytic capacitors look good but have no vents. I am not sure if they are solid ones, or they just did not have vents back in the 70s?
I hope you enjoyed my little post.
Edit: Didn't quite get the images right, without having them to large, so i leave them at the end for better readability.

[Kleinstein]

For the imperial hex screws there is a good chance a torx bit can work. It can still be tricky to find a suitable size.

The input fets mainly works as a source follower. So one should be able to see if the signal comes through.

[TimFox]

The hex is probably 1/16 inch, a typical US size (in conventional units), probably for a 6-32 UNC screw (maybe a 5-40 thread).

[Gyro]

Nice vintage meter, definitely worth the effort of restoration.

It's well worth getting a set of ordinary right angle imperial hex keys [Edit: Not ball ended!] and keeping them to hand. They're easy to find in Europe and sometimes you only get one chance with a tight grub screw.

[AVGresponding]

Those brown bodied resistors are carbon composition rather than carbon film, though you are correct they can drift wildly out of spec over time. Film resistors tend not to.

The electros are almost certainly wet, with the venting being out of the base through the lead holes...   

At least it won't be an impossible-to-repair multi-layer board if one/more of them do leak electrolyte.

[PfannenHans]

Thank you for all the feedback. I will definitely look into buying imperial hex keys, so i can disassemble the whole DMM and give it an overhaul.
Unfortunately i can not measure the signals of the AC-DC signal converter circuit, since it is right under the soldered down range switch assembly, but it should at least be possible to measure the components when I manage to get the board out.

[Kleinstein]

The link on the axel for the range switch look like that the card may slide up, if the swtich is at the right position. So one may not even have to remove the knob.

[Cubdriver]

Nice!  Welcome to the nut house, and the addiction that is old test gear.  Cool old meter.

-Pat

[Martian Tech]

The link on the axel for the range switch look like that the card may slide up, if the swtich is at the right position. So one may not even have to remove the knob.

This is correct.  It's not necessary to remove the knob.  That coupling will slide apart.

Old thread, I know, but I'm wondering how things are going/went on this repair?  I've been working on my own repair of the same model (documented here, if anyone's interested: https://www.youtube.com/playlist?list=PL0gjXJVpVUR2Yf5saNvqK6YLmZgSOooF_ )

[PfannenHans]

The project is currently on hold till i finish my master thesis and move out of the student's dormitory, which is soon. Currently I have only one desk and have to move everything away when i want to cook/eat.
Your videos look good and seem to be the first on  youtube for the meter.

[Martian Tech]

Hmmm...  Looking back at your photos, I see yours has a .33 MFD capacitor for C312.  Mine had .22 with R342 and R305=118K - schematic calls for .1 with R342 and R305=392K.  The result of this apparent factory mod was that it wouldn't do a full 19999 count.  I'm thinking that it was a workaround because they couldn't get 392K resistors, but now seeing yours makes me wonder if there was some kind of running improvement or something (or whether they were just using whatever parts they could get their hands on at the time).  I can't see any date codes to tell if yours was built before or after mine though...  Mine appears to be late 1973 from the date codes.

[Kleinstein]

I see changing the capacitor an resistor at the integrator more like a shift in the balance between noise and linearity.
The time constant of R342*C312 determines how long the integration can be and the maximum count (could be a difference between US and European models be cause of the mains frequency).
Larger resistors and smaller cap results in more noise and for some of the effect better linearity.
A smaller resistor and larger cap can improve on the noise, but may result in larger INL errors, though some parts (effect of leakage currents) can also improve.

C312 should be a PP or PS type cap and .33 MFD is already relatively large for this technology. Resistor values should not be a problem with availabity - if at all it is the capacitors that can be a bit tricky.

[Martian Tech]

Certainly today it's true that resistor availability is not a problem.  I think though that when these things were built in 1973 or 1974, 1% precision resistors were rather rare beasts so there could have been availability issues.  I'm putting mine back to the original spec (with a PP cap - PS caps, especially in these large values, are very hard to come by today).

[Kleinstein]

For the resistor accuracy does not matter, even 10% off would not be a problem. The point is having at least reasonable low TC to keep ther nonlinearity from selfheating low. With a 1 V range and 390 K this would mean something like 100 ppm/ K (maybe 200 pmm/K) or better and thus nothing really demanding.  Low excess noise could be a factor too - so better not a carbon cap.  The difference in performance of PP or PS caps is relatively small, more within the the variations between brands models.

[Martian Tech]

OK, yes.  That is why they are using RN55D resistors there.  And in 1973 there probably weren't any low-tempco resistors that were available as anything other than 1% tolerance...

But I still lean toward a resistor availability explanation, as they used the wrong value in mine when changing the capacitor to 2.2 MFD.  Input voltage range is 2V and they are aiming for a maximum output voltage of 10V.  Integration time is 200msec.  So Vout=(Vin/RC)*t, which reduces to R=.04/C
With C=0.1 MFD, R is 400K and they chose the closest available value of 392K.  That's fine.  But if C is .22 MFD, R should be 182K and in my unit they used 118K, completely eliminating all headroom in the integrator.  118K would be a reasonable resistor value to use with .33 MFD, as the formula works out to 121K.  SO maybe mine was meant to have a .33 MFD cap installed but they ran out of those?  Or whoever was doing the install read the substitution chart wrong?  It's all just speculation at this point, but it's fun to speculate 

[Kleinstein]

.22 MFD and 118 K seems to really odd and some mistake. Using a larger resistor (e.g. 220 K) would have been OK, with only a little higher noise and more head room.

I doubt they ran outout of 0.33 MFD capacitors, more like a wrong pic at the cap or the resistor or maybe just a meter around a product change, mixing old and new version parts.

It is quite normal that the low TC resistors come in low tolerances and this same resistor type may be used already for a different purpose. For the current compensation it is good to have the same resitor value once more. With more high end intruments it is also not uncommon to find parts way better than actually needed in some places. The old meters were not plagued by bean counters so much as many modern meters are.
A property of the dual slope ADC is that they don't need an accurate values for the resistor and integration cap.

[PfannenHans]

I can't see any date codes to tell if yours was built before or after mine though...  Mine appears to be late 1973 from the date codes.

According to the date codes my unit seems to be mid-74.

[Martian Tech]

OK, it's looking like I have maybe a Rev D card and the schematics show Rev F.  R303, 304 and 369 have different values on my card than shown in the schematic.  I probably should have left it alone, but now I guess I'll just try to modify it to match Rev F.

[cncjerry]

I have one of these beasts as well. Works perfectly but I'll admit I haven't power it one since the last time we were all taking pictures in "show us your nixie" thread.  I have a bunch of Keithley brown meters in addition, all work fine and hold cal.  Definitely worth restoring it.