Keithley 199 -- FIXED!!!!

[PedroDaGr8]

okay, then I'm just baffled.

Because looking further up, it says that chunk of the counter ÷5, but I can't make heads or tails of the part you sent since it seems to contradict the previous page's IEC logic symbol (unless I'm completely misreading that).

On your busted one, what do you get out of U17?

I'm going to pull the filter caps tomorrow and do ESR tests. Guessing their fine, but who knows.

Well on my working one, I think my multimeter has a hard time reading the frequency or something. It read the input fine 3.842Mhz for both, but the others gave weird results. I'll bust out my scope tomorrow and see what I get. That beign said pins 6 and 7 showed the same frequency on my multimeter (just the number was ~1/2 what it should be) but their duty cycles were different.

On my broken one only pin 1 showed ANYTHING. The rest all showed zero volts output and zero frequency. So on my device it appears this IC is dead! Not sure what killed it but I need to start tracing back from here. At least SOMETHING is dead (not at my own hands).

[staze]

Well on my working one, I think my multimeter has a hard time reading the frequency or something. It read the input fine 3.842Mhz for both, but the others gave weird results. I'll bust out my scope tomorrow and see what I get. That beign said pins 6 and 7 showed the same frequency on my multimeter (just the number was ~1/2 what it should be) but their duty cycles were different.

On my broken one only pin 1 showed ANYTHING. The rest all showed zero volts output and zero frequency. So on my device it appears this IC is dead! Not sure what killed it but I need to start tracing back from here. At least SOMETHING is dead (not at my own hands).

cool, thanks. certainly let me know.

I pulled the main filter caps today, and they all check out ESR/Capacitance wise. So that's not it (didn't think it was). So, still searching. Keep looking at the top and bottom of the board hoping something will jump out at me. =/

[staze]

Since I've gotten really no-where on this, I think I'm actually going to start from scratch. I'm still assuming it's something rather obvious, and probably a single point of failure, but at this point, I'm just jumping all over the place hoping to find a smoking gun (or at least a bullet hole).

I'll admit some of the signal pathway is a bit confusing (like, what the hell is the point of Q5 and Q6), but it must be that way for a reason. =P I'll report back if I find anything odd.

Any further input from alm or codeboy2k, or PedroDaGr8 would be much appreciated.

Thanks!

[PedroDaGr8]

Since I've gotten really no-where on this, I think I'm actually going to start from scratch. I'm still assuming it's something rather obvious, and probably a single point of failure, but at this point, I'm just jumping all over the place hoping to find a smoking gun (or at least a bullet hole).

I'll admit some of the signal pathway is a bit confusing (like, what the hell is the point of Q5 and Q6), but it must be that way for a reason. =P I'll report back if I find anything odd.

Any further input from alm or codeboy2k, or PedroDaGr8 would be much appreciated.

Thanks!

A few observations.
So yesterday I picked up a LeCroy 9410A with two Tek 10x probes for $125 locally, the thing is huge lol. Anyways, that I started tracing the clock signal from the GPIB board (which actually also controls the display). So the signal looks good going from U8 through AT1 (the optoisolator on the mainboard) but at or after that the clock signal starts looking REALLY messy. I don't know if its my fault but the signal goes from being a clean 0-5V to 2V-5V and with a lot of high frequency noise superimposed on top of it. I am not sure if U49 does it or if it is injected after U49. I'll try and figure it out tonight if I can. Also, after running the meter for a while I noticed the filter caps on the Vreg get quite warm.

[staze]

Anyways, that I started tracing the clock signal from the GPIB board (which actually also controls the display). So the signal looks good going from U8 through AT1 (the optoisolator on the mainboard) but at or after that the clock signal starts looking REALLY messy. I don't know if its my fault but the signal goes from being a clean 0-5V to 2V-5V and with a lot of high frequency noise superimposed on top of it. I am not sure if U49 does it or if it is injected after U49. I'll try and figure it out tonight if I can. Also, after running the meter for a while I noticed the filter caps on the Vreg get quite warm.

Interesting! I started with the analog board last night, and managed to trace everything successfully to U46. What's interesting is at the output of U46, there's an offset added by U45, which I'm guessing is due to what amounts to a resistor divider where U45's and U46's output come together. So where I get a nice 2.5VDC into U46 (which is my input signal), the output pin shows about 2.9V out. U45 is where the multiplier is done for some of the ranges...

Also, what's VERY interesting, is C38 is missing. Looking at the board, it was never there (the solder pads/holes look to have the factory "fill"). PedroDaGr8: could you look and see if C38 is there? It's not listed on the diagram as being purposefully omitted. Also, what revision does your digital board say it is (mine says A4). Doubt it matters that much, but it would be curious to figure that out.

As for the optoisolator... the offset MIGHT make sense (looks like R83 coming from the +5V rail may be the offset) (page 5 of the schematic), but the noise SHOULD be shunted to ground through C61. I guess my first thought would be, did you move your ground lead to the analog side when you measured AT1's output (I say this because I forgot to do this, and things looked wonky (60Hz hum) until I realized it, and moved the ground lead). But I need to recheck all that once I confirm my signal path is clean.

All that said, I do wonder if the issue is clock related.

Just some thoughts. =)

[PedroDaGr8]

sorry I meant after the op-amp  U49 following the optoisolator.  I'm going to double check tonight.

[staze]

ah. Well, wouldn't that be because U49 sits halfway between its +5V and Gnd rails... ? So zero input would get shifted up to 2.5V? I'm guessing here, so please, call me wrong if that's just complete BS.

[PedroDaGr8]

ah. Well, wouldn't that be because U49 sits halfway between its +5V and Gnd rails... ? So zero input would get shifted up to 2.5V? I'm guessing here, so please, call me wrong if that's just complete BS.

Actually, looking at the schematic again. U49 isn't an opamp, its a hex inverter. So it should be just fine with 0-5V output. The thing is, since the noise follows the hex inverter, I'm not sure if this noise comes from the hex inverter or is being injected backwards from somewhere else.

[staze]

ah. Well, wouldn't that be because U49 sits halfway between its +5V and Gnd rails... ? So zero input would get shifted up to 2.5V? I'm guessing here, so please, call me wrong if that's just complete BS.

Actually, looking at the schematic again. U49 isn't an opamp, its a hex inverter. So it should be just fine with 0-5V output. The thing is, since the noise follows the hex inverter, I'm not sure if this noise comes from the hex inverter or is being injected backwards from somewhere else.

Ah, okay. Sure looks like an Op-amp from the drawings. =P

hmm, I'll have to check that out, see if mine sees the same thing. What's the noise look like?

[PedroDaGr8]

Anyways, that I started tracing the clock signal from the GPIB board (which actually also controls the display). So the signal looks good going from U8 through AT1 (the optoisolator on the mainboard) but at or after that the clock signal starts looking REALLY messy. I don't know if its my fault but the signal goes from being a clean 0-5V to 2V-5V and with a lot of high frequency noise superimposed on top of it. I am not sure if U49 does it or if it is injected after U49. I'll try and figure it out tonight if I can. Also, after running the meter for a while I noticed the filter caps on the Vreg get quite warm.

Interesting! I started with the analog board last night, and managed to trace everything successfully to U46. What's interesting is at the output of U46, there's an offset added by U45, which I'm guessing is due to what amounts to a resistor divider where U45's and U46's output come together. So where I get a nice 2.5VDC into U46 (which is my input signal), the output pin shows about 2.9V out. U45 is where the multiplier is done for some of the ranges...

Also, what's VERY interesting, is C38 is missing. Looking at the board, it was never there (the solder pads/holes look to have the factory "fill"). PedroDaGr8: could you look and see if C38 is there? It's not listed on the diagram as being purposefully omitted. Also, what revision does your digital board say it is (mine says A4). Doubt it matters that much, but it would be curious to figure that out.

As for the optoisolator... the offset MIGHT make sense (looks like R83 coming from the +5V rail may be the offset) (page 5 of the schematic), but the noise SHOULD be shunted to ground through C61. I guess my first thought would be, did you move your ground lead to the analog side when you measured AT1's output (I say this because I forgot to do this, and things looked wonky (60Hz hum) until I realized it, and moved the ground lead). But I need to recheck all that once I confirm my signal path is clean.

All that said, I do wonder if the issue is clock related.

Just some thoughts. =)

Also, yes C38 is filled on BOTH of my devices. It is one of the ones that I thought was defective actually. Going back and measuring again I can't get the same results oddly enough.

[staze]

Also, yes C38 is filled on BOTH of my devices. It is one of the ones that I thought was defective actually. Going back and measuring again I can't get the same results oddly enough.

Well crap. I'll check again, but it's not in the space it shows on the drawing. Is it in a different space for you? Shows it as being right below R17 and R13, behind Q7.

[PedroDaGr8]

Also, yes C38 is filled on BOTH of my devices. It is one of the ones that I thought was defective actually. Going back and measuring again I can't get the same results oddly enough.

Well crap. I'll check again, but it's not in the space it shows on the drawing. Is it in a different space for you? Shows it as being right below R17 and R13, behind Q7.

yep it looks identical to C53 except its larger. I'll take a pic when I get home. I'm not entirely sure it is necessary

[staze]

yep it looks identical to C53 except its larger. I'll take a pic when I get home. I'm not entirely sure it is necessary

Here's hoping it's not. Guessing it must not be if the factory left it off. If it IS, hopefully a Mica cap will be a suitable replacement, as finding a 500V polystyrene 680pF cap looks to be rather troublesome.

[staze]

Okay, just checked pin 10 of U49, and I'm guessing this is what you mean by "noisy". Mind you, that's just me stopping the scope. Maybe things like this don't matter with a clock signal, but it certainly doesn't seem clean. I've confirmed this signal remains crappy looking on the CLK inputs for U29, U30, U31, and U32. Note that my channel 2 is my source signal. And while the optocoupler output is a positive going clock signal, U49's is negative going. Looks like it's a base 5V signal, that clocks at 1V. Second picture is the large gaps in the clock signal.

Indeed, C38 is missing. Guessing it must not be necessary.

The output of the AT1 for the clock signal, btw, has gaps. guessing whatever is generating the clock is being reset every X cycles. So it's not a steady lock signal, but goes for a bit, has a gap, goes a bit more, gap, etc. That what you see? =/

Also interestingly, the clock on U49 pin 10 (and AT1 input and output) shuts down in diagnostic mode (any N level). So it must be driving the MUX. Which I guess makes sense.

[PedroDaGr8]

Actually, i think mine may have been a grounding issue because on the other side of the inverter, I am reading non-stop 5V solid. So iether U49 is blown or one of the other ones is blown. I may start lifting ICs to see if I can isolate it.

[staze]

Actually, i think mine may have been a grounding issue because on the other side of the inverter, I am reading non-stop 5V solid. So iether U49 is blown or one of the other ones is blown. I may start lifting ICs to see if I can isolate it.

This is on the broken one I would assume.

Do you get a clock between AT1 and U49? I just get a solid 5V when I'm in diagnostic mode as well... the clock seems to only run in normal operating mode.

I've pretty much traced my signal as far as I can... since once I get to Q17, which no matter the N mode, the gate is at -10V, and my signal disappears. There's about 30R of resistance between the Source and Drain (with the device off). Does this mean that FET is toast? If so, anyone know what model FET it is? The parts list just says "Trans, N-CHANNEL FET" and "TG-128". If it's supposed to dump my signal to ground, then how the heck does the integrator work?! I see that Q4 will switch on and off... and I can see some noise (740mVpp 48khz triangle wave centered on -220mV or so in N1 and N2, then it shifts down to -800mV for N3, and I get nothing on N4). But that's certainly not my signal. My signal seems to die on the Q17 side of R58, and there it sits. I did lift R58 and confirm it is a working 15.8kR resistor. Though checking the other "TG-128"'s on the board, they seem the same... about 30R between drain and source with no power applied. =/ But I would think with a -10V bias on the gate, it should be completely closed. =/

So, in short, I'm stuck at this point. The only thing I can think is if I can lift the drain leg on Q17 and see if that gets my signal further down the line... but I'm not sure if that's safe to do. I'm not even sure what the point of Q17 is, other than for some other measurement where you don't need voltage on this signal path.

Thoughts?

[staze]

I guess there could also be a short in Q4 between the drain and gate as well... but it would seem SOMETHING is just dumping my signal to ground around that point.

Would this seem like a reasonable failure due to abuse (over-volting the meter)? Can I just lift a leg on either of these and see what happens?

[staze]

I didn't pull the jfet (yet) but checking it with the diode test of my fluke, I get about .75V DC between gate-source/drain one direction, and about 2.2V the other direction. My understanding is you shouldn't get ANYTHING in one direction, is that not true, or is it because I didn't pull it to test?

Thanks!

[PedroDaGr8]

I didn't pull the jfet (yet) but checking it with the diode test of my fluke, I get about .75V DC between gate-source/drain one direction, and about 2.2V the other direction. My understanding is you shouldn't get ANYTHING in one direction, is that not true, or is it because I didn't pull it to test?

Thanks!

http://www.allaboutcircuits.com/vol_3/chpt_5/3.html

[PedroDaGr8]

I didn't pull the jfet (yet) but checking it with the diode test of my fluke, I get about .75V DC between gate-source/drain one direction, and about 2.2V the other direction. My understanding is you shouldn't get ANYTHING in one direction, is that not true, or is it because I didn't pull it to test?

Thanks!

http://www.allaboutcircuits.com/vol_3/chpt_5/3.html

This also explains why I got shorts across the JFETs sometimes but not others.

[staze]

well, then I have no clue. somewhere around that area my signal is completely disappearing... so something is dumping to ground. It's there on the mux side of R58, but missing on the A/D side... just a nice 0V. Will have to check again tomorrow evening... but right now, I'm guessing this is the general area of fault. Everything LOOKS fine... but something has seemingly shorted/failed.

Any suggestions? =/

[staze]

Okay, on this again. Signal still disappears after R58, but looking at the mux circuit, I notice some overshoot on it.

I've found that it's coming from the output of U45A. My understanding is the mux is switching rapidly from 0, 0, -2.8V, and my input signal (2.5V). But there's a blip now and then, which would seem to be coming from U45A (unless it's coming from somewhere else further up the chain, but it's not on TE1/TE2. I don't see any noise issues on the rails to anything, so that's not it.

Any thoughts? For all I know, it's the same cause as the weird runt pulses on the hex inverter as well...

[staze]

Taking another approach, with the unit off, I get 30R of resistance from ground to the A/D side of R58. PedroDaGr8, or alm, can you please check this on yours? wait, nevermind. With Q17 not getting a -10V on it's gate, it just goes to ground. Crap. How the hell do I figure out where my signal is being grounded out then? Lift Q17 (or at least, the source/drain pin) and test then?

With the unit on, in diagnostic mode, with any N, I confirmed it's NOT Q17, as I get no conductivity between source and drain (because the gate is at -10V). So it's somewhere else things are going to hell.

Help!

[staze]

Looking at the integrator in diag-mode, I see pin 2 at 9.928, 9.928, 9.915, and 10.205V on N's 1-4 respectively. Pin 3 is at 10.527, 10.527, 10.583, 10.208V on N's 1-4. So Q1 must be taking input from SOMEWHERE, and creating an offset from something, and using that to drive the integrator?

And at least in N4, it's doing some type of integration. so something is getting through. maybe there's something I'm missing from the schematics with regards to where the signal goes and what drives the integrator. Maybe this is all down to the hex inverter runts or the overshoot on the mux... but unless I start taking some peyote, I'm not sure I'm going to have any epiphanies. Anyone got anything? Bueller? 

[PedroDaGr8]

I just realized you asked me what the model of the JFETs are. On my non-working one they are ALL 2N part numbers (the working one has the weird custom parts) if you give me a few minutes I'll list the part numbres of all of the TO-92 devices.