| Electronics > Repair |
| Keithley 197 - jitter/jumping last 2 digits |
| << < (7/7) |
| corehbola:
Very well noticed that U117, even though it's on the digital "realm" it's purposedely fed from the +5V_A... It's because its output is used to feed the charge-Balance current (through R178 7.78k). It's even curious they went all the way to connect another 7.78k resistor to the Q output of the flip-flop... Possibly to minimize fluctuation of the ground? Not really sure. And well noticed (again) it would have been more sensible to generate the -6.4V using the +5V_A reference, because both are used to source currents to the A/D input so, minor drifts from one would be "tracked" by the other and possibly cancel out. But, as you said, being "just" 5.5 digits, they may have been able to get away with that. I`m also stumped by the single slope run-down could be completely inoperative and still the meter would somehow get along with it... It looks like the A/D "times-out" at the end of the cycle and the uC doesn't care about it. And I know it's not a matter of firmware revision, because I just got 2 newer firmwares (C1 and C2), and they work exactly the same. I got them because, the documentation of Keithley, being so "hand made" don't provide any information, in terms of what versions of firmware were released and which ones accompanied each of the electrical changes. And when you get only the latest version of the schematics (because it's the only one you can find on the Interwebs), anyone trying to make the meter "free of bugs" only have the option to implement everything that's on the later drawing... which is exactly what I ended up doing. Regardng the positive saturation of the integrator... I agree with you... it must be either leakage (for which I did already some tests of the FETs Q119 and Q120 and they seem to be fine), or U105 may be acting up. I recall having read on some datasheet that those CMOS analog switches are rather sensitive to order of power-up, that the negative power coming on before the positive may damage them in the long run. And the power supplies of the 197... Let's say there haven't been much design effort in start-up / shut-down sequencing. If much, the only thing I noticed that resembles this kind of effort is VR101, bet even so, I never got to fully understand its purpose. Anyway, in a couple of days more (tops) I will know what is going on with that A/D. |
| Kleinstein:
VR101 is likely there for the start up. The -9 V supply is used to provider the zener current to VR102 and the voltage at VR102 is used to generate the -9 V. Depending on the offset of the OP-amp the circuit without VR101 may get stuck with 0 V instead of -9 V. The resistor to ground at the inverted output of the FF is there to keep the current loading to +5V_A constant, at the cost of a variable ground current. The ADC is actually relative similar to the 6 digit bigger brother K193 with not that much extra refinements. The 2nd slope for the rundown is more a faster one, not so much one for higher resolution. So the ADC itself may be capable of more resolution, though maybe noise and accuracy limited. |
| corehbola:
Hello Kleinstein, Bingo! The analog switch IC (U105, CD4053BE) was shot. U105B (pin 5) was not pulling down the gate of U120 and therefore, U120 was conduting all the time, making it impossible for R129 to compete. Therefore, instead of a rundown, the integrator was doing a run-up all the way to +V saturation. Indeed, the Single-Slope comparator was never firing, and somehow, the uC / firmware didn't care about that. After replacing U105, when I ground the input leads I get rock solid 0.00000V in the 2V scale, and 000.000 ±1 LSD (which is in accordance to the 197 specs 1.5uV of noise). The ± 43 LSD I was getting, regardless of scale, I think it was exactly the "remainder" of the charge-balance conversion which was missing the single slope run down to "zero-in" to the exact reading. I hope the issue with Trobbins is exactly the same. It's very easy to spot (now that I know). You only need to look at the output of the inegrator (U110, pin 6) and if it never goes to 0V and instead, goes every so often to positive rail saturation (± 9V), it's probably the same culprit. For records sake, I will post pictures of the waveforms (both before and after the repair). Merry Christmas and Happy New Year! |
| indy:
Hi all, I hope can find some help here. I've a Keithley 197 which shows an always positive drift on the last 2~3 digits, It's seems to stop after ~1 hour if I don't touch any of the front panel controls. This happens on any range. I've started to do check of internal nodes and the only weird thing I've found is -V node at -14.5V, it's showing ~200mv of 100hz rectifier cap ripple. This node, from schematic, should be -15V see R132 and VR105 (which should be a 15V zener) R132 is at nominal value and VR105 was tested working, so I guess there is some unexpected current request from -15V that is able to bring the node too low to make VR105 work. all other voltages seems clean and at nominal value (-9V, -6.4V, +5V A and D, +10V) Moreover I've was not able to spot any issue in the measure path or A/D waveform. Please let me know if -V is expected to be -15V & clean, and any of your thoughts on this issue. Regards, Fabio I. |
| Kleinstein:
Under which conditions is the drift observed ? (open / shorted input, Az mode enabled). In the non AZ mode some drift is normal. Even with the AZ mode a little warm up drift in the first hour or so (for a meter without a fan also longer) can also be normal. Is there an offset - so what is the reading with a shorted input. With an open input drift is normal, especially in high Z mode - there 3 LSB steps would even be surprisingly little. In the 10 M impedance more the input bias can also caus additional drift and a little offset can be normal. |
| Navigation |
| Message Index |
| Previous page |