Keithley 197 Repair

[robrenz]

Nice paper on shielding and guarding alm, Thanks.

[alm]

I must confess that that was mostly lucky. I just searched for driven guard on Google and picked the first hit that showed what I wanted to show .

[staze]

A quick glance at the schematic suggested to me that they were driving the guard (see Fig 23 a here): driving the guard trace with a buffered version of the input signal. Could have been wrong, though.

Why do you think that you need split supplies to drive an op-amp? An op-amp does not have a ground pin. It can't tell whether you connected it to +15 V and ground, ground and -15 V or +7.5 V and -7.5 V. You can just short the inputs or configure it as a buffer and feed it a voltage from a divider between the power rails (eg. pot). If it's really broken, then I expect it to draw several mA regardless of the input and output connections.

It's mainly the outputs that are interesting, since any current flowing from the output will directly affect current draw.

Ah, that makes more sense. I guess I just always figured guard rails were more passive than that.

okay, I can do that. I'll just feed it with a single rail (assuming it works with a single rail). I would assume, then, with no input, the output should sit at 7.5V (assuming a +15V rail)? Then I guess I'll short the rails and see what I get. Guess I'll also take a close look at the traces around there... but after I put in the DIP socket, I checked all the pins and they all seemed to be reasonable. Also checked that chip out of circuit and there didn't seem to be any shorts between pins either. But, given the two chips I've replaced were both marked as "static sensitive". I also know that when I had U107 out for replacement I turned the unit on... I wonder if U101 didn't like having a +10V rail, and a floating gnd/neg rail. e.g. U107 was the fault the unit had to start with, and in the process of replacing/testing it, I killed U101. =/

[staze]

I must confess that that was mostly lucky. I just searched for driven guard on Google and picked the first hit that showed what I wanted to show .

Huh? Was there a link I missed?

Update... nevermind. Cool!

[alm]

Ah, that makes more sense. I guess I just always figured guard rails were more passive than that.

The advantage of active guards is that the voltage between guard and signal is very tiny, so very little current will flow, resulting in very low leakage and very high impedance. Very useful if you're aiming for Gohm input impedances . The AD paper discusses this in more detail.

okay, I can do that. I'll just feed it with a single rail (assuming it works with a single rail).

There is no such thing as a single supply vs. dual supply op-amp. The difference is just how close the inputs and output can get to those rails. An op-amp that can't go lower than V- + 2V is not very useful with a single +5V rail, for example.

I would assume, then, with no input, the output should sit at 7.5V (assuming a +15V rail)?

Unlikely. There's always a tiny offset between the input, and since the amplifier is operating open loop, the output will likely be close to one of the rails.

Then I guess I'll short the rails and see what I get.

Short the rails?

Guess I'll also take a close look at the traces around there... but after I put in the DIP socket, I checked all the pins and they all seemed to be reasonable. Also checked that chip out of circuit and there didn't seem to be any shorts between pins either. But, given the two chips I've replaced were both marked as "static sensitive". I also know that when I had U107 out for replacement I turned the unit on... I wonder if U101 didn't like having a +10V rail, and a floating gnd/neg rail. e.g. U107 was the fault the unit had to start with, and in the process of replacing/testing it, I killed U101. =/

I'm not discounting the possibility that U101 is dead, but I'm not completely convinced that it's dead either. So it's possible that replacing U101 won't fix the problem. I wouldn't really expect a floating V- pin to kill the chip, although I guess it's possible that it would blow up one of the ESD clamping diodes between the inputs/outputs and the V- pin. Does one of the pins on the U101 that you desoldered have a low resistance to V-? If the voltage from your meter is high enough to forward bias the ESD clamping diodes, then reverse the leads so they're reverse biased.

[staze]

Short the rails?

Sorry, inputs. But I would assume that if I short the inputs, I should see an output that either oscillates (bad), or gives me 0V (good), but draws a lot of current in it's rails (bad).

I'm not discounting the possibility that U101 is dead, but I'm not completely convinced that it's dead either. So it's possible that replacing U101 won't fix the problem. I wouldn't really expect a floating V- pin to kill the chip, although I guess it's possible that it would blow up one of the ESD clamping diodes between the inputs/outputs and the V- pin. Does one of the pins on the U101 that you desoldered have a low resistance to V-? If the voltage from your meter is high enough to forward bias the ESD clamping diodes, then reverse the leads so they're reverse biased.

okay. I'll take a look. I would assume "low resistance" would be less than 1K, or more like, 100R or less?

[alm]

Well under 1k.

[staze]

Wow, this IS a great reference. huh.

[staze]

okay, so something definitely happened testing U101 out of circuit.

At first, I just fired it up with a 9V rail, shorting the inputs. As you said, the output sit close to the rail at 8V. So I ramped up the voltage and at about 12V, it seems to have latched, destructively, because it's now drawing about 35mA, the PSU is in CC mode at 2.4V (too low to power it, and it's outputting all of 68mV). Turning the power to it off and on doesn't change anything, so it would seem to be toasted at this point. Guess I should have limited the PSU to only supply about 10mA or so. =/ So, chip could have been dead before, or I killed it, but given it's behavior, it seems reasonable that it dragged that -6.4V rail, and therefore the -9V rail down due to current draw.

She's dead jim. Checking resistances on the chip:

2-3: OL
5-6: OL
4-8: 120R. BINGO!

Back to the board of the 197, in the U101 socket.

4-8: 13kR
2-3: 4MR
5-6: 1MR
1-7: 4MR
5-7: 1MR
1-3: 4MR

Obviously 1-2, and 5-6 are 0R.

Now I just wait for the replacement. Hopefully by Saturday (don't want to wait until Tuesday). I got other stuff to work on! =)

[alm]

12 V is unlikely to have killed the op-amp, since it was running at about 15 V (10 V + 6.4 V) in circuit. The higher current may have been the final blow, but the op-amp is probably rated for something like 30 V.

[staze]

12 V is unlikely to have killed the op-amp, since it was running at about 15 V (10 V + 6.4 V) in circuit. The higher current may have been the final blow, but the op-amp is probably rated for something like 30 V.

The replacement is only to 18V max... but, yeah, 12V shouldn't have killed it.

Looking online, it sounds like if you short the rails together, it can saturate the output at one of the rails... but my meter is 10MR input, so it shouldn't have been loading the amp down.

So, think the op-amp was failing, and I just threw it over the cliff? Replacement should be here Saturday.

[alm]

Yep, the combination of the circuit suddenly starting to work when you removed it, no obvious faults in the traces connecting to it and it self destructing when exposed to less than 18 V suggests certainly a failing U101. Guess we'll know in a few days .

I just checked the datasheet of the TLC27L2. 3 - 16 V supply voltage and 18 V absolute max, that's kind of tight on a 16.4V (nominal) supply. If the original had similar specs than I could certainly see something weird with the supply rails (eg. U107 developing a short between one of the inputs and V-) killing it. Hope those zeners are accurate .

[staze]

Yeah, that's what I saw for that op-amp too.

But I can't find any details on the original one I pulled, so I can't find specs.

Attached is the image for it. If you could find specs, that would be awesome... but I keep turning up empty. I guess for all I know, Keithley had TI build it for them... I only used the TLC27L2CP because I saw that other image, and yeah, the absolute max was 18V... obviously in the meter it's going to be running at about 16.4V, which is just above it's nominal rating. =/

[alm]

Most likely an (selected) off-the-shelf part stamped with a Keithley p/n by TI. The only way to figure that out is if you find the same Keithley p/n in another manual together with the TI p/n. I'm not aware of any cross references for custom Keithley parts like those available for HP and Tektronix parts.

[staze]

Most likely an (selected) off-the-shelf part stamped with a Keithley p/n by TI. The only way to figure that out is if you find the same Keithley p/n in another manual together with the TI p/n. I'm not aware of any cross references for custom Keithley parts like those available for HP and Tektronix parts.

Ah, could be. Weird. Stupid unobtainium parts.

[staze]

Replacement op-amp worked great, and meter is up and running. I'll get some pictures here in the next day or so comparing readings to my 199. Certainly don't like the display as much (non-backlit, LCD). Guess they fixed that with the 197A (it has a backlight).

[alm]

I'm glad it's now fixed.

Yep, those low contrast LCDs without backlight are not great. HP also used a bunch a lot of them, including in some fairly expensive equipment like the 3468a, 3478a and even the 3457a. Although I still prefer LED to even LCD with backlight. Some of the earlier 160/170 series meters used LED, but battery power was an option for many of these meters (including the 197), and LED hurts your battery life.

[staze]

I'm glad it's now fixed.

Yep, those low contrast LCDs without backlight are not great. HP also used a bunch a lot of them, including in some fairly expensive equipment like the 3468a, 3478a and even the 3457a. Although I still prefer LED to even LCD with backlight. Some of the earlier 160/170 series meters used LED, but battery power was an option for many of these meters (including the 197), and LED hurts your battery life.

Yeah, I love the LED as well now. Before I found out about the 199, I was looking at the venerable Fluke 8840, and the HP 3478a, both of which have their issues. The 199 is awesome. The 197 would be a great, and cheaper alternative to the HP.

So, comparing the 197 to my 199 and my Fluke 189, I'd say pretty darn good. Let both the Keithley's warm up for an hour before doing the testing.

Sadly, I think I'm going to sell the 197. I don't have much use for it. =/ Also sucks that it doesn't have an IEC socket (just an attached power cord).

[SeanB]

What opamp did you put in there to replace the house part?

[staze]

What opamp did you put in there to replace the house part?

A TLC27L2CP. See post here: https://www.eevblog.com/forum/testgear/keithley-197-repair/msg282942/#msg282942

[Tepe]

Also sucks that it doesn't have an IEC socket (just an attached power cord).

Would there be room for adding a socket? (If it doesn't have a GPIB board)

[staze]

alm,

Okay, one thing still isn't right, and I'm not sure why. Pin 12 of S101 (power switch) reads -19V rather than the supposed -15V ± 10% it's supposed to. The diode reads ~-14.8V on the correct side, as does R132.

Could this mean the rectifier is bad, or that the transformer is bad? I mean, VR105 is doing it's job, but I worry about it's, and R132's longevity if they're dropping 4V rather than a few hundred millivolts they're supposed to.

Thoughts? And thanks! 

[staze]

So, the rectifier does seem a bit odd.

It's 4 pins (normal) in a DIP type layout.

Left inputs are AC, right outputs are negative on top, positive on bottom.

So basically:

UL: AC
LL: AC
UR: -Rectified
LR: +Rectified

Measuring with black probe on UL (diode check), I get:
UR: 0.51V
LR: 1.25V

black probe on LL:
UR: 0.52V
LR: 1.25V

Hmm...

Red probe on UL:
UR: 2.25V
LR: 0.52V

Red probe LL:
UR: 2.25V
LR: 0.52V

And, just because I should: Red UR, Black LR: 1.05V. Reversed: OL.

Seems to me the testing should show the same between the neg and positive.

Thoughts? (this was all done in circuit)

[staze]

Here's what my scope has to say... this is on the AC side of the rectifier... I have... no explanation for the clipping at the top, or the artifact on the waveform, but I'm guessing something has failed in the rectifier in some weird way (not a short, and not an open).

[staze]

so, it's not the rectifier (surprise). Pulled it, and it tested fine out of circuit.

I'm still getting the weird waveform on the AC though. And still seeing the weird integration looking waveform on the negative rail.

What worries me though is the negative rail is at -19V at pin 12 of the power switch, which means R132, and VR105 are dropping ~5V, rather than the 1.5V or so they're supposed to.

What I don't understand is how could that supply be LOWER than it should be. I could see closer to ground... but being 4-5V BELOW where it should be defies logic in my mind.

So, I got nothing. The unit works...