-
Keithley 195 stopped working after a lightning strike near my house.
Posted by jezel123 on 18 Apr, 2024 16:00 -
So about 3 months ago I bought a used Keithley 195 for $60, which was a pretty good deal
. I was pretty happy with it, but then last week there was a storm and a lightning hit a building nearby, and after that it stopped working 
. When I turn it on it doesn't show anything on the display, but if I turn it off and then on again in a short amount of time it will light up all the display elements at once.
I measured the 5V digital rail and got 4,8V, so power supply should be ok. Now, I noticed that the F6808P processor gets warm when I turn it on, so I thought maybe it was damaged somehow. I bought a replacement on ebay, but when I replaced it, I just got the same behavior.
Is it possible that the lightning strike somehow cleared the EPROM, so now the CPU doesn't do anything, because there is nothing to do? If so, is there any way I could get a replacement / the binaries for the EPROM?
Any help is appreciated!!!
My source for schematics:
https://download.tek.com/manual/195_901_01C.pdf Site 113 to 121 -
- Have you measured ALL voltages? Page 116.
- 4,8V may indicate there is something wrong, it could be a large current dragging the 5V rail down. Disconnect the load if possible and measure again.
-
Now, I measured every voltage after I replaced the Processor:
5V digital: 5.02V
5V analog: 5.00V
-15V analog: -14.98V
15v analog: 15.01V
-
Do a reverse check with the old processor, see if 5V drops again.
If it does: That's bad news. In that case, the 5V rail was affected by lightning strike and anything powered by that rail may be damaged. -
It drops to 4.96V, better than 4.8V (Which I am no longer sure about, I did measure it with a cheap DMM because I didn't have access to anything better at the time, now using Fluke 83), but not the same as the 5.02V with the new processor. I think that 6mV difference may be because the F6808P that I ordered is not the exact same as the original one (made in Singapore vs original one made in the Philippines and date code is 8319 vs 8035)
-
I can try and read out the EPROM, but I would rather eliminate every other possibility first, since I don't have an EPROM programmer / reader and would have to set it up using an arduino or sth.
-
It drops to 4.96V, better than 4.8V (Which I am no longer sure about, I did measure it with a cheap DMM because I didn't have access to anything better at the time, now using Fluke 83), but not the same as the 5.02V with the new processor. I think that 6mV difference may be because the F6808P that I ordered is not the exact same as the original one (made in Singapore vs original one made in the Philippines and date code is 8319 vs 8035)
Looks like your calculator is faulty too.
-
Leave the old one in, I don't think it's dead, and the new one may cause more trouble than good.
I do recommend getting a proper programmer like Xgecu t48 (around 50 euro on AliX). It allows you to health check logic chips as well.
Do you have access to an oscilloscope?
-
60mV not 6mV, my bad, forgot a 0
-
A lower power consumption is possible for the newer chips. 60 mV difference may also be thing from a different temperature.
I would not expect EPROMs to be especially sensitive to loosing it content from a glitch. AFAIK they are not especially sensitive.
If one has a scope, one could look at the data / adress bus to see if the CPU is running or somewhat stuck. If stuck the clock and reset signal are further points to check.
The signal quality at the bus could also show if there is a bus conflict (usually causes intermediate signal levels). -
I don't have access to a scope at the moment, but I can borrow a picoscope from school tomorrow. I did however find my LA 1010 logic analyzer and hooked it up to the A0-A7, and it looks like the lines are dead. There is a little something when I turn the Keithley 195 on (330ms mark in the data), but nothing more than that. In the zip you will find .csv and .kvdat with the data I recorded on the logic analyzer if you want to take a look. Sample rate is 32MHz.
EDIT: Added png's of the signals
-
The long pauses in the signal look odd. This may indicate a problem with the reset or clock.
-
I got the picoscope and measured the reset and one of the clock lines. Red one is the clock, 4MHz as expected, blue is the reset line, and there is a sawtooth signal. According to the schematic, the IC controlling the reset line should be a ICL8111 (Site 108, U144 in the manual), but the part number on the IC that is actually soldered onto the PCB is ITS6728. I measured the resistors R162 - R164, and got R162 = 15.02k, R163 = 1.000 M, R164 = 15.06k, which is way off, but I am not sure if I can trust the measured values without having to desolder the resistors. Also, R163 looks suspicious, and maybe the last owner replaced it or something (see photo). I also did a diode test on the CR111 and CR112, both have about 0.59V in one direction and 2.85V in the other. R107E reads 2.75k, but, again, not sure if something else in the circuit is affecting the measurement.
Any ideas?
-
Scope the +5 power, threshold and hysteresis pins of that reset chip.
Could be a fault with the power supply unable to supply sufficient current when the processor takes more after coming out of reset. Reset chip sees the voltage fall and resets the processor. Rinse and repeat. -
The clock looks OK, but the reset signal looks suspect.
Another point to check may be supply if there is significan ripple - too much ripple could cause the reset problem. The ripple before the regulator can give a hint on how good the filter caps are still. -
I scoped the 5v power line and the voltage looks clean except for some sub 30mV peak to peak 160kHz ripple. On the screenshot, you can see the threshold (red) and hysteresis (blue) signals.
-
Contained in the zip is my K195a rom dump , .bin .
If you or anyone else needs it.