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

[ElektroQuark]

Hello,

Just arrived a non working Keithley 2000.
It powers on. Self test gives tens of fails, pointing to a power supply failure.

It has been reworked, there are lots of flux and solder blobs in +15 and +5 regulators and in other points. Capacitors have been changed.

But, the 79M15 regulator is untouched. And the -15 rail is dead. I have not a 7915 at hand now but this is very promissing: if the negative rail doesn't work the analog section can't work correctly neither, so there is that bunch of fails on seft test.

On the other hand, it has 07 version ROM.

In the Keithley forums there is the las version (20):

HERE

It looks like this:

Code: [Select]

S00600004844521B
S214000000008100000000045000000540000005408C
S21400001000000540000005400000054000000540C7
S21400002000000540000005400000054000000540B7
S21400003000000540000005400000054000000540A7
S2140000400000054000000540000005400000054097
S2140000500000054000000540000005400000054087
S21400006000000540000318B20000054000000546E9
...
Motorola style, I think.

It must go in two ROMs, one with even data and other with the odd data.

I don't know how to do that. I have a Minipro TL866 as burner. Any help?

I will be updating it only if I can make it work properly. It was last calibrated in 2010 and is in very good shape internally: clean, no burns, etc.

Thank you.
[/code]

[jfuller000]

That does look like a Motorola S-REC file - As far as I know, the TL866 software doesn't support these, so you'll need to convert it to a binary (or intel hex) file  - srec2bin from http://www.s-record.com/ would probably work on windows (For linux, there's a similar utility called SRecord).

When you load this binary file into the TL866 software:
Select the ROM type first (because it clears the buffer if you change the selected device).
When you click open, after you select the file name you'll get a "File load options" screen.
For the even ROM, change the load mode to "Load a low byte of a WORD", then program the ROM. Repeat the file open process but this time with "Load a high byte of a WORD" selected to program the odd ROM.

(I am not 100% sure I've got odd/even the right way round - although a quick test with a file looked correct)

[ElektroQuark]

It seems simple now
Thank you.

[andy_silicon]

Be aware - you WILL LOSE YOUR CALIBRATION if you change the firmware.

[ElektroQuark]

Noted, thank you Andy.

-15v rail repaired. There was and open diode on 7915 GND pin also. But the 7815 runs somewhat hot.

Same errors, none has been fixed.

Main custon IC pinout anyone? I can't find schematics of this beast.

[andy_silicon]

My third K2000 ( a restoration project ) had lots of supply issue problems - in fact only +5v was good.
Most of the faults ( not all fixed yet ) were due to capacitor leakage.

Underneath the large caps ( sorry no open unit in front of me ) leakage had occured - and totally eaten lots of traces - mostly hidden under psu components.
Some removal of regulators/capacitors , lots of cleaning , usage of pcb varnish to protect , replacement of an 18v zener , the killing of the lm399 :-( , and replacement of a couple of op-amps.
Now I have a machine which fails the low side AC volts ( I forget which related tests these are ) but from the description I suspect either another dead opamp or open trace in the RMS->DC converter section.

I think the 7815 does run very hot ( in later revisions of the board - there are larger heatsinks and the regulators are moved away from the large electrolytic caps ). However - see above as your issues sound like I recall - opamps near Lm399 ( or however the white ceramic heated zener is marked ) also the tiny little 18v zener. Also check the lm399 is producing 6.9v and the opamp buffer which it feeds is okay - these appeared to have died in my k2000. ( sorry if some of this makes no sense - jetlagged )

[ElektroQuark]

Capacitors have been changed. No leakages nor bad tracks. PCB is like new.
The reference is working OK, 6.99v. Zener too.
Will check OpAmps.

[ElektroQuark]

Sorry, I forgot about this.
The meter has been in a corner and today I have put my hands on it again.

With the schematics now I have been able to test the signal path of the 100 self test. All is OK, 0v found at the test point, and it is feed correctly to the A/D converter. This is sending signals too to the MCU, and they are too in the MCU receiver pin.
But self test complains about not been able to communicate with A/D.

Another symptom: the meter responds if I play with configuration settings, but if I change, for example, from DCV to ACV, or to Ohms, etc., the meter hangs, it becomes irresponsible.

Any idea?

[ElektroQuark]

Little update.
When powered and opened, the meter makes a very low creeping sound.
I make a tube of paper and applying it to my ear I found the sound origin: the processor is making it 
Can it be partially fried?
Its an MC68306FC16A.

[ElektroQuark]

The MC68306FC16A is not available anymore, so looking for a donant now.
Another revision, MC68306FC16B is a drop in replacement.

[Kjelt]

While youre at it there is some custom communication ic that gets fried a lot, put a small heatsink on it if it is not dead already.
I wanted to change the firmware but the unit will be worthless than since all calibration data will be lost. How were you going to handle that?

[ElektroQuark]

Someone told me that too, so if I can fix it and if it isn't out of calibration I will not upgrade ROM.
What is that custom IC?

[Kjelt]

I tried to find my emails about that ic but can not find them anymore 
All my friend remembers is that it were "Orbit" marked ic in the back of the pcb. If it still works, just stick a small coolingplate on it so it becomes less hot 

[TiN]

There are no components on the back of 2000's board.

EdoNork

As of repair, I would start off checking all power rails and convertors.

If processor is doing it's job and meter is working - it's likely to be OK.
Check it's power pin resistance to ground. I can measure it on my broken 2000 to compare, if u have reference measurement.

Before changing firmware, read your current ROM contents and save it. Better post , so we can have dump too
I suspect Calibration data to be stored in I2C EEPROM U136, near CPU. At least cal data in 2001's are in I2C EEPROM.
Read it and publish, and then you have no need worry about calibration, as can revert it back anytime.

I have extra 2001's or 7001 board with MC68302FE16, but not sure if you can use it instead of 306, need check DS.
But I doubt that CPU or firmware is faulty, if meter boots.

Photos of affected/repaired areas around on your board may help too, to understand what was done, and what parts were debugged.

Here's my thread, since I got 2000 too, but it's kinda collecting dust since then, more focused on 2001's lately.

[Kjelt]

Damn english  in the back i do not mean on the other side of the board but i mean the opposite of in the front so in the back in the rear

[ElektroQuark]

I have checked all power rails, they are OK.
Power pins on MCU have proper voltages too.
I have checked step by step signals of the 100 test, and they look OK. ADC sends pulses to MCU, an intermediary IC inverts them, and they arrive to the MCU. But K2000 complains about no data received from ADC. That's why I suspect about MCU. Maybe a bad pin? Or the ADC is sending bad data.
I will try to find some free time to capture the ADC sent data with the oscilloscope and will post here (its a short train of pulses that is repeated periodically). I will post the firmware too.

[TiN]

Did you check optocouplers between digital and analog?

[ElektroQuark]

Yes, and they work.
Controller receives the data from ADC. Can the ADC be sending the wrong data?
I have check data sent by the ADC with the scope at ADC pin. I have done the same thing in the receiving pin of the micro. The data received is identical to the sent data but inverted.

[TiN]

Which exact errors you having?
I can check signal path on my 2000 if you share more details.

[ElektroQuark]

Errors? All of them...

• 100.1
• 100.2
• 101.1
• 101.2
• 101.3
• 200.0
• 200.1
• 200.2
• 201.0
• 201.1
• 201.2
• 300.0
• 300.1
• 301.0
• 301.1
• 301.2
• 302.1
• 302.1
• 302.2
• 303.0
• 303.1
• 303.2
• 304.0
• 304.1
• 400.0
• 400.2
• 401.0
• 401.1
• 402.0
• 402.2
• 403.0
• 403.0
• ...

If the micro can't connect to ADC all of them appears.
I'm not near the equipment now, but I followed signal path for Test 100.1 all is OK.
I will capture ADC data to micro and post here.

[TiN]

100.1 says that If ADC not getting correct measurement of 153.6M +/-1.2M counts, then test fail.
So measure 0V at A/D in first. If it's correct, check operation of ADC circuitry , digital side:
U154 74HC175, 12MHz clock from Y102, +5VR rail. Check U146 74HCT14.
Is your ADC digital part custom Keithley ASIC or ALTERA MAX CPLD?

Then proceed with analog side of ADC - (U149, U148, Q127,Q131, stuff around U138, U137, U142, U153, Q132,Q133, U145).
You should have +14, +15, -14, -15, +20, -20V rails correct.

[ElektroQuark]

I checked the 0V at ADC input pin. Correct.
12MHz clock correct.
Don't remember about U154 and U146...
Rails were OK.
The ADC is a Keithley ASIC.

I will check around the rest.

Thank you.

[ElektroQuark]

Here again.

I have rechecked voltages:

+5VD ---- 4.914
+37V --- 37.035
+15V --- 15.909
-15V ---  -15.901
+5VRL -- 5.040

+-15V are out by a 6%, out of specs, it they must by within 5%. Can be there the issue?

Now TEST 100.1 passes, so the ADC is working correctly.
TEST 100.2 fails. Fault message: SIGNAL NOISE. Can the opamps be acting funny for that overvoltage?

UPDATE:

During TEST 100.2 U138 output drifts slowly from -14.985 to -14.975 when I put the DMM probe.
U137 from -14.767 to -14.757
U142 something similar.


UPDATE 2:

Now TEST 100.1 failed again.

UPDATE 3:

A now it passes.....

[ElektroQuark]

UPDATE

Checcked registers U106, U109, U130, U134 and U121 bit patters and U109, U134 and U121 differs from those stated in the repair manual.
I changed a couple of then and the errors remains.
The ADC is sending the wrong data or there is a problem in the data lines.
Need to check. (Grrrr, no free time lately for this).

[EMUD]

Hi EdoNork,

does the test 100.1 work without problems ? Is the voltage on pin 1 of the TF-245 Network precise double the hight of the ref voltage?

To have this test working you need a data connection between the micro and the adc. The next is you need the two voltages of +14 and - 14V. (They must be double the hight than the Vref)  Connect TP105 to ground.
Now test 100.1 and 100.2 should work.
If there is an issue in this tests refer to the simplified diagram in my post.

Good luck
EMUD

I have just discovered some interesting:

If the 14V are too high (in my 2000 about +13,88V) test 100.1 brings "no communication with adc". Seemed that the cpu has a timeout for this data and displayed, if the data is too late, this message.