keithley 182-M teardown

[acbern]

Selling my 182-M nanovoltmeter in the for-sale-section here, and suggestion was to do a teardown before it goes. So here it is:

Background is, this is not the normal 182 meter, but the M-version (whatever that means). I did not know this exists when I got it (together with a standard 182), so when I opened it it was quite a surprise to see a battery inside. This is obviously used (although I have not analyzed the circuitry or seen any documentation, if anybody has, let us know) to power some input amplifier section to reduce noise. I did actually do some tests, not statistically exhaustive, but compared to the standard 182 I had back then, the peak to peak noise was about half of the standard one, when doing a measurement at the same source.
Pictures show the factory changes to hold the battery (not there in the standard 182), the area of the input amp under the plastic lid (avoiding airflows) with the battery cables, and the lid removed, inluding the add-on PCB that also is there only with the M version (probably a switch-off of te battery if the unit is off, the battery is about a year old and still at 9V). Finally a test at 19mV against a recently calibrated Fluke 5440, pretty precise as can be seen.

[Zucca]

Thanks acbern, very interesting.

[VintageNut]

The purpose of the battery is to keep the front end energized thereby making sure that the input impedance does not drop when the power is turned off. If you don't care about this behavior, the battery is not needed.

[MadTux]

My broken ebay Keithley 182 arrived today. Already mostly done cleaning and fixing it. (milled replacement power knob from black POM plastic).

30V range had much too low input impedance and totally wrong, low readings (like 1.4V at 30V on input). Since there are no schematics available, I just applied 30V (max allowed input voltage, input current was around 30mA) onto input and started measuring and feeling around for hot components. All opamps around the frontend had more or less zero input offset voltage, good sign.
After a while, I noticed that one of the TO92 components below the big 500ohm resistor right after the input, got a little hot. Junction was toast, too.  After unsoldering it, the readings at 30V range are normal. Lower readings are now wrong, though. Thereby I guess the toasted part is a JFET used for range switching.

Unfortunately, it's labeled "IFC TG128", some cryptic Keithley part?
Pinout is typical Source/Drain/Gate config that most JFETs have.
If I can't find any info on that, I'll try generic J310 or something likewise. But I guess doing so will increase input noise


Sweet, googling "TG128 JFET" revealed, it's apparently a 2N4392. Nice

Replaced it with a TO18 2N4393 from the "old man's box" (some might know , nearly every imaginable old component can be found in there) and all ranges are back into working order, yeah

[TiN]

I'm not sure TG128 is standard 2N4392. If I remember right, it is also used in K2001 ADC and analog board in few spots.

BTW, 182-M depicted in this thread will be for sale soon  . I decided to upgrade to more modern nanovoltmeter.