Restoring (TRYING TO FIX) and upgrading ROM of a Keithley 2000

[ElektroQuark]

I have:

At TF245,

- Pin 1 = 13.986V
- Pin 3 = -13.986V

At R315,

- One side = 6.9929V
- Other side = 15.888V

My 15V lines are a little of the maximum 5%, and 13.986V are pass your stated limit for the CPU timeout.

So I have to low those voltages or try another firmware in hope the timeout is longer, and I will lost calibration.

[EMUD]

Hi,
there are some other things wich may affect the comunication. I will make some additional tests with my 2000 and let you know, where the tolerances precisely are.
Do you have a scope with memory? I will shot some measurements on tuesday.

best regards

Emud

[ElektroQuark]

Yep, I have the scope.

[EMUD]

Hi,
your English is much better than my Spanisch ...

OK, the 15V are ok, the regulators are set high by a silicon diode each, so it should be about +15.6V and -15.6V . (The voltages in the schematic are for information only, not a must be!) The 20V after the rectifier i.e. are about 24V depending on the mains voltage.

Your reference is ok, I have seen references from 6.91V to 7.01V, your voltage matches.
The +14V and -14V have to be the exact double of your reference, this matches, too.

But there are some other parts wich are influencing the A/D converter, i.e. the resistors in the TF-245 network.

My K2000 has an issue with this TF-245, one resistor has changed its value. So there is a 101.2 error.

Good luck

EMUD

[EMUD]

Hi,
I have had a lot of cummunication with K2000 users about their problems. It seemed, that many problems were caused by a bad resistor network. So it would be good to extract and measure the resistors in the network. It is possible, that many of your trouble is caused by it.

Good luck

EMUD

[ElektroQuark]

Do you mean desoldering TF-245 and reading it?

[Galaxyrise]

Do you mind desoldering TF-245 and reading it?

https://www.eevblog.com/forum/repair/keithley-2000-gain-repair/msg242713/#msg242713

[EMUD]

Hi,
yes, the easiest way is to desolder it and check it. I will do this if the new one arrived. Thanks for the hint with the divider, but this seemed not to be the issue in this case. In the diagram there are some resistors for the integrator, these have to be very precise. They may cause such errors like 100.1 and 100.2 .

Thanks

EMUD

[ElektroQuark]

Thanks EMUD.
I will desolder and check the resistor array this afternoon.

[ElektroQuark]

Here they are the TF-245 values:

PIN        OHMS           PIN

1           999.58          16
2           8997.6          16
2           9998.4          15
3           19K994         15
3           14K995         14
4           30K171         14
4           29K698         13
5           30K163         11
6           999.65          9
7           8997.5          9
7           89K974         8
10         27K412         12

Two much difference between both 30K resistors? The rest seems to be OK.

[EMUD]

Hi EdoNork,

this may be the problem. I have desoldered and measured my network, too. It has much more tolerances.
The problem is, we only can speculate about the allowed tolerances. I will, when I get the new TF-245s
(I will get 4 of them) check all the 4 and post the results. This will give us enough confidence in the correct
values.

EMUD

[Galaxyrise]

Many of those are ratio tolerance, not absolute tolerance, and even then it's more about long-term stability and TCR tracking than absolute ratio.

[eas]

It helped me to learn that I could get the measured value of failed tests by running the tests with the option to pause after faults. Hold down the shift key and it will display the raw value.

Test 100.1 was failing on my unit, along with most of the others. Once I figured out how to get the raw value, it was obvious that there was communication with the ADC, but the value was really odd. While orienting myself to the signals on the pins, I noticed a hair, or some sort of fiber. Cleaning it out fixed everything. I was lucky.