Keithley 197 Repair

[staze]

All,

After my success repairing a Keithley 199, I hopped on eBay and found a 197 for a song. It too is damaged (all readings show as oL). After some basic voltage probing, I THINK I have figured out the issue.

It would appear the negative voltage zener is toast, but I'd love a second opinion.

The manual can be found here: http://www.ko4bb.com/Manuals/09)_Misc_Test_Equipment/Keithley/Keithley_197.pdf

Schematics are included. On the first page of schematics for the board, it appears that out of the bridge I get a filter cap, then through VR104 a positive rail, and VR105 a negative rail. On one side of VR105, I get 0V (makes sense, it's grounded), on the other I get about 0.8V. I confirmed off the bridge rectifier I'm getting a +15V, and a -18V.

Everything I've checked that has a negative rail, all gets that 0.8V instead of the -15V ±10% it should be getting. Is it most likely that zener is toast, or am I overlooking something? Looking at the board, it appears it's a 1N47 15V zener (1N4744, or NTE145A). Is there a good way to test it short of what I've done? It does show 0.53V in one direction, and 1.96V in the other... so... guessing that's bad.

Thanks!

[alm]

This copy lacks schematics. In general, it helps to be specific: at least give a page / figure number, and preferably post the relevant part of the schematic as an inline image. The easier you make it for people to help you, the more likely you are to get a response.

Any semiconductor on the bus might be pulling the negative rail down (up). Desoldering the suspect component, measuring it out of circuit, and if necessary replacing it is a good strategy. Sometimes it may be necessary to isolate parts of the circuit (cutting a pin / track) if there's a lot attached to that rail. Some manufacturers will even include wire links that you can cut and replace for that exact purpose.

[staze]

Shoot. There are two version of the manual out there... guess I linked to the wrong one. =/

Okay, schematics attached. This copy here: http://www.utwente.nl/tnw/onderwijs/Practica_TNW/doc/apparatuur/multimeters/keithley197.pdf has the full schematics as well. The voltage rails are all on the first page, in the lower right quadrant. I see where the positive and negative rails are "created", as well as the +5V analog and digital rails, and the +10V rail (for many of the opamps on the board running on +10V and -9V), but I must be blind because I can't see where the -9V rail is created from the -15V that the zener in question should be providing. 

The digital stuff all seems to be working, since obviously it just runs on +5V. 

So is your thought that it's unlikely the zener has died? I figured something could be dragging it down, but given I can't find where the -15V actually goes to be converted to all the other rails...

If I pull the zener, what's the best way to test? Just run 18V (or whatever) through it, with a 100R resistor in there and make sure it breaks down at ~15V?

As always, thanks alm.

[staze]

Ah, looks like the -9V rail is created in F2, on page 1 of the schematics. It looks like VR101 may be the answer... but I can't read a model number, and keithley seem intent on not listing actual off the shelf part numbers in their manuals... =/

[alm]

The parts list from the first manual you linked to gives a description and part number for VR101: DIODE, ZENER 8.2V, IN756A [sic] (DO-7).

I'm not saying it's unlikely, just that anything connected to V- and a positive rail could be responsible for this. Were those diode measurements you gave for VR105 in circuit? That's not very reliable. You can indeed desolder one leg and test it with a resistor + voltage source.

If you suspect a component, then cutting or desoldering its connection to V- should restore the rail to -15 V (this of course does not apply to VR105). I'm guessing R132 probably gets quite toasty with ~16 V across it?

I haven't looked at the schematic in detail. If a lot is powered via the -9 V / -6.4 V rails, then you could try isolating the -9 V section and see if the V- rail comes up (goes down) to -15 V. For example by desoldering one leg of VR101.

If there are tantalum caps in there, then a shorted tantalum cap would definitely be one of the candidates.

[staze]

Okay, I must be blind. Where did you find the part number? The manual with schematics have hand written notes with keithley numbers. Oh... the manual without the schematic... weird. It's like one is pre-production.

I'll check the diode this evening, but yeah, seems possible something is pulling it down. Would I see a short between the rails if something like an opamp was bad? There are a couple tantalems, but not sure if they are on that rail.

[staze]

okay, it's not that diode. so, I'll have to check for what might be dragging it down. should be able to, like you say, disconnect VR101, see if that helps. Do some devices have their rails buffered so you don't get stuff like this happening (one device pulling the whole rail down)?

[alm]

You can of course have multiple power rails, like the analog and digital rails, but in something like a DMM there's not very useful to keep supplying the uC with power if the input stage is not working. Like I said, some equipment (I recall seeing it in some Tek scopes) feature jumper links to make it easy to isolate sections.

[staze]

alm,

if you take a look at the schematics, could you please explain how I could separate the negative rails? I can disconnect VR101, or VR102, but that seems like it would just raise the negative rail up, rather than potentially isolate them from each other. I'm slowly looking at everything attached to the negative rail to see if there's a short between ground, and the negative rail... so far nothing. I don't see the tantalem's being connected to the negative rails at all, so that seems unlikely. One thing I note is there's a tantalem cap on the board that doesn't exist, according to the schematic. It's right below R103, and is listed as COK 1.5C 1KV. =/

[PedroDaGr8]

alm,

if you take a look at the schematics, could you please explain how I could separate the negative rails? I can disconnect VR101, or VR102, but that seems like it would just raise the negative rail up, rather than potentially isolate them from each other. I'm slowly looking at everything attached to the negative rail to see if there's a short between ground, and the negative rail... so far nothing. I don't see the tantalem's being connected to the negative rails at all, so that seems unlikely. One thing I note is there's a tantalem cap on the board that doesn't exist, according to the schematic. It's right below R103, and is listed as COK 1.5C 1KV. =/

pretty sure that's NOT a tantalum as IIRC they don't go up to 1KV. Likely a ceramic instead.

[staze]

ah, okay. it LOOKS like a tantalem (orange blob), but guess it could be.

Pedro, How's the 199 repair? Get the replacement zener?

[alm]

if you take a look at the schematics, could you please explain how I could separate the negative rails? I can disconnect VR101, or VR102, but that seems like it would just raise the negative rail up, rather than potentially isolate them from each other.

Unless I'm missing a connection in the schematic, it seems to me that if you disconnect VR101 the -6.4 and -9 V rails should go to ground and should be disconnected from V-. If something on the -9 V rail is shorted, then this should allow V- to return to -15 V.

I'm slowly looking at everything attached to the negative rail to see if there's a short between ground, and the negative rail... so far nothing.

Maybe I misinterpreted something, but I thought you mentioned the V- rail was +0.8 V. How would you get +0.8 V from a short between ground and V-?

[staze]

if you take a look at the schematics, could you please explain how I could separate the negative rails? I can disconnect VR101, or VR102, but that seems like it would just raise the negative rail up, rather than potentially isolate them from each other.

Unless I'm missing a connection in the schematic, it seems to me that if you disconnect VR101 the -6.4 and -9 V rails should go to ground and should be disconnected from V-. If something on the -9 V rail is shorted, then this should allow V- to return to -15 V.

I'm slowly looking at everything attached to the negative rail to see if there's a short between ground, and the negative rail... so far nothing.

Maybe I misinterpreted something, but I thought you mentioned the V- rail was +0.8 V. How would you get +0.8 V from a short between ground and V-?

True. It seems likely it's more like a -9V and +10V short... but I'm not seeing anything that would indicate that's the case. I'll try unsoldering one leg of VR101 and see if that brings the negative supply back up. If not, I'll try VR102 (with VR101 reattached), which should give me my -9V, but kill the -6.4V rail. Really need a board size panavise for this type of thing... this board is a pain in the but to work on. =P

Also of note, and I still can't figure out why, I'm seeing some (what looks like 60Hz) ripple on the cathode side of VR104. I checked C124 (out of circuit), and it seems fine for both ESR and capacitance. So not sure why it's not filtering that out. Could it be due to the issue with the negative rail? Anyway... one thing at a time.

[alm]

If lifting VR101 does not help (then lifting VR102 won't do anything either), then I would check CR107 and S101. I wonder if it's something stupid like bad switch contacts or an open rectifier. Do the signals on S101 pins 11 and 5 (or any convenient location connected to the same net) look reasonable with the power turned off? Pin 5 should give a smooth ~ 15 VDC signal, pin 11 a rectified sine with ~ -18 V amplitude relative to ground. What about resistance with the cord unplugged and the power switched on?

Yours doesn't have the battery option, does it?

[staze]

If lifting VR101 does not help (then lifting VR102 won't do anything either), then I would check CR107 and S101. I wonder if it's something stupid like bad switch contacts or an open rectifier. Do the signals on S101 pins 11 and 5 (or any convenient location connected to the same net) look reasonable with the power turned off? Pin 5 should give a smooth ~ 15 VDC signal, pin 11 a rectified sine with ~ -18 V amplitude relative to ground. What about resistance with the cord unplugged and the power switched on?

Yours doesn't have the battery option, does it?

It does not have the battery. I was wondering about the switch, actually. My main question is, how the hell are the pins numbered? There's one tall pin, which I gather is pin 6, but does it continue around clockwise, or does it return to the top and go down again? I would have liked to see pin numberings on the switch itself, but alas, there are none, and the drawing doesn't show any either. =/

[alm]

You can also measure at CR107 or some other convenient point.

[staze]

Ah, sorry, I forgot CR107 was the rectifier. Yes, I get +15V rectified on the positive side, and -18V rectified on the negative side. So that is working. Like I mentioned, I'm seeing that -18V on R132, so it's definitely coming out of the rectifier.

From my looking around, it seems like the pinout of the switch is:

top
7     1
8     2
9     3
10   4
11   5
12   6

Where the "top" is the pushbutton, and this is looking down at it on the component side, rather than the solder side. A bit later I'll disconnect VR101 and see what we get (if the negative rail comes back up/down or not).

[staze]

Disconnected VR101, no change. -15V rail is still being pulled down. So it's not anything on the -9V or -6.4V rail, it's got to be something on the -V rail.

So, looks like the stuff on -V are: Q127, Q128, U107, and S101... that's pretty much all I can see. What's the deal with Q127? It's just a NPN base attached to -V, and a Collector attached to the circuit, with the emitter not connected.

Thoughts?

[alm]

Ah, sorry, I forgot CR107 was the rectifier. Yes, I get +15V rectified on the positive side,

+15 V rectified? This should be pretty much DC, since there's a cap (C124) and no load.

Looks like they use Q103/Q127 as protection diodes to prevent the voltage on the sense line from exceeding +V + 0.6 V and -V - 0.6 V. Remember that a BJT looks like two diodes from B to C/E (the other way around for PNP), although the collector-base diode is reverse-biased during normal operation. No idea why they use this instead of regular diodes, could just be something simple like BOM size.  This could actually be where the +0.8 V is coming from, but that means something else is pulling it high. What's the voltage on the collector of Q127?

What's the resistance between -V and +V? I see nothing obvious between the various positive rails and -V.

[staze]

+15 V rectified? This should be pretty much DC, since there's a cap (C124) and no load.

The positive rail has about 1.2Vpp ripple around 15V. The negative rail has 2.2Vpp ripple. Both of which seem pretty high since that filter cap is in there. I kind of figured the ripple was due to the negative rail being "shorted" so the cap was basically being loaded down.

Looks like they use Q103/Q127 as protection diodes to prevent the voltage on the sense line from exceeding +V + 0.6 V and -V - 0.6 V. Remember that a BJT looks like two diodes from B to C/E (the other way around for PNP), although the collector-base diode is reverse-biased during normal operation. No idea why they use this instead of regular diodes, could just be something simple like BOM size.  This could actually be where the +0.8 V is coming from, but that means something else is pulling it high. What's the voltage on the collector of Q127?

0.08V. Little noise, but otherwise, nothing.

What's the resistance between -V and +V? I see nothing obvious between the various positive rails and -V.

Starts at about 15kR, and ramps up to 25-30kR. Not positive why that would happen? Capacitance? That's measured from the anode of VR105 to the cathode of VR104.

Also of note, on power up, the negative rail is about 0.2V, and it slowly ramps up to 0.6V or something (let it sit about 3 minutes and that's how far it got). I'm guessing it would keep going if I left it long enough. I'd almost wonder if C127 was shorted, but then I wouldn't get the -18V on that side of R132... so, hell if I know.

I've hooked VR101 back up at this point... since it didn't seem to make any difference. Arg...

[staze]

Okay, interesting... was just basically touching random things with my multimeter, and interestingly, the vent on the top of C127 is at -15V.... pretty sure, that shouldn't happen. As C124 is a bit below 0V, but nothing major.

So... ?

Yeah, just tested with a new cap, same thing. So, is the vent supposed to be attached to the negative lead?

[alm]

Yes. Also the case for caps without foil around them. I once drew some sparks and slightly scorched a scope probe because I didn't know the cap was for the negative trails, so the negative terminal was not grounded.

[staze]

Yes. Also the case for caps without foil around them. I once drew some sparks and slightly scorched a scope probe because I didn't know the cap was for the negative trails, so the negative terminal was not grounded.

Okay... that's good to know. So, thoughts were I should check next? At this point, I'm thinking of pulling Q128, or at least the collector. It's a pretty jellybean part (3906), so just cutting it off would work too. Because at this point, it's either that, U107, or I guess Q127. I would think if Q127 was it, I would see a short between it's collector and base... And with Q128... I don't see how it would be grounding out the rail (or bringing it down), since it should only be connected to other negative rails, but I'm just guessing. Is there something I'm missing? Some other component attached to -V?

I don't have a thermocouple (yet) so I can't really measure temperature of components, but C127 and/or R132 do seem warm. Guessing R132 is heating up C127.

[staze]

Well, it's not Q128 or Q127. Disconnected the collector on Q127, no change. And removed, then replaced Q128, no change with either. So, it's down to two options.

1. It's U107, which I don't have a replacement on hand. It's a TCL271CP. Or...
2. Removing VR101 didn't kill the -9V and -6.4V rails since I didn't also remove Q128.

 

[alm]

The positive rail has about 1.2Vpp ripple around 15V. The negative rail has 2.2Vpp ripple. Both of which seem pretty high since that filter cap is in there. I kind of figured the ripple was due to the negative rail being "shorted" so the cap was basically being loaded down.

I originally suggested measuring it with the power turned off. Then C124 is energized, but disconnected from the load.

Starts at about 15kR, and ramps up to 25-30kR. Not positive why that would happen? Capacitance? That's measured from the anode of VR105 to the cathode of VR104.

Capacitors charging up. Nothing interesting, 15k is way too high to pull the rail up to 0.8 V.

I don't really see the point of replacing parts before you've fixed the problem, it's not like putting them back is going to help. What you can do is cut the V- pin of U107 close to the body, but leave a little metal sticking out of the body. You can then easily resolder this connection later on. If this doesn't help, remove VR101, Q127 and Q128 again. If that doesn't help,  go searching the PCB (is easier than the schematic) for other parts connected to -V. For example a short on the PCB, or some other component that we missed in the schematic.