Keithley 6485/6487 teardown

[plesa]

6487 does not have guard at all.
Input stage of 6485 and 485 it different, 6485 use the opamps and 485 use discrete FET.

Razberik, did you tried LMC replacement? I do not think there is any matching nor preselection of them for this entry level meter.

[razberik]

Hello there.

A big credit goes to SKE. After a few talks he came with a theory.
Lets analyze the most simplified loop for nanoAmpere range. I omit capacities and use most simple opamp model.
Firstly we apply Vsource+resistance to give about 2nA.
U26 creates about 1V across 500Meg resistor. This is the way it should be. When I place 1TOhm at the input to GND, it simulates open input, the U26 output settles aprox. at zero.
We can now believe that U26 and U50 are OK. So the wrong part can be propably U25 or U27.
If we assume that the instrument died after crazy relay clicking and U27 works probably as an input protection (I was told that the only input FETs are used, it is easier to buy this cheap opamp than low-leakage diode itself).
So If I bring some leakage into input, I receive the exact behavior like on real instrument.

Shortly ... this was true. When U27 LMC662 was desoldered, instrument started to work. I attach pictures to prove it.
One thing interesting. I soldered new LMC662 and washed the soldering area using IPA in few cycles, but there is still some residue remaining at the pins (tiny flux ring around pin). This doesnt affect the input current.

[plesa]

Congratz to successful repair. In your box is lot of space for input protection so I will consider to add it.

[TiN]

Very nice job documenting and getting down to a rabbit hole, thank you!

Can I use your info to add in my 6485 article, with link to this thread posts?

[Smith]

The 6485 is a nice device, just like the older 485. I use both, but the 485 has two main advantages: the connector is positioned on the front (just like me) and the autorange doesn't tend to keep switching between two ranges in particular cases. The latter can be solved on the 6485 by lowering the reading speed.

For everyday use I mostly use the 485, it's just so easy to use and fast and stable. I think it was build somewhere in the 80's, but it's still spot on.

[razberik]

Can I use your info to add in my 6485 article, with link to this thread posts?

Yes, of course. No problem.
I expect to do reversing K2400 SMU in future.

[plesa]

Can I use your info to add in my 6485 article, with link to this thread posts?

Yes, of course. No problem.
I expect to do reversing K2400 SMU in future.

If you have faulty 2400 you can check 23x service manual, there is schematic.
Two units I needs to repair has blown power mosfets only  (MTP2P50 and 3N100 or in more recent meter 7N100).

[razberik]

I am still not sure if I understand used relays correctly. They use EE2-3SNUH which arent special, they dont have excellent resistance.
But these relays are latching. This means that when relays not operated, coil is connected to GND. Does it mean that the coil acts as shielding/guarding ?

[pelud]

Hi All,

I have some questions about how the analog circuitry of the meter works.

1. What is the purpose of the 2N3904 and 2n3906 transistors (Q4, Q19 and Q20 on kei_6487_reverse_preamp_sch2.pdf) and surrounding circuitry?  It looks like some sort of compensation network, but I'm not sure what for or how it works.

2. What is the null adjustment on the AD829 for?  I would have thought that the LMC6081 is the "boss", and would do whatever it can to make it's inputs the same, including compensating for any null offset on the next amp.

3. What are R185, C140, R50 and C141 (kei_6487_reverse_preamp_sch2.pdf) going into the AD829 for?

Also, some comments:

I noticed that some people mention that the triax is not necessary.   However, it very much is, as you would logically expect that the exposed metal shell of the connector is at earth ground potential, while the center conductor can be at whatever voltage source you are using.  Therefore, the center conductor as well as the internal circuitry can "float", and the internal feedback of the device will drive the second "guard" conductor of the cable to same as the center in order to "intercept" any current that would otherwise leak to ground.

[TiN]

You can get away w/o triax if use floating BNC with external "shell" connected to GUARD. However you would still need/want low triboelectric noise cables with good insulator material. Using coax with driven shield at guard potential is very often used inside the instruments, that don't need worry about user's touching live wires (that might be at lethal potentials!). For example Keithley 7xxx series low-current switch matrix cards use coax to route signals from triax input to PTFE-floated reed COTO relays.

Caveat with this approach is obvious = your outer BNC metal shield is now under live voltage, controlled by guard driver, so if you trying to measure something with dangerous voltage levels - you need exercise great care. 
Also this solution may provide more noise in samples, as you don't have as good shielding as with triax or quadrax cables.

[Minikisscool]

Hi,

I have a 6485 which is not working.
There is no display of value.
Error reading Read command in GPIB (timeout)
I read all topics here, used preamp schematic v2 to check input
preamp is working, I have proportional voltage on analog output


I look to check ADC, it's OKI component (2000-802A04)
In my board, it's a bit different, it's a new version 2X00AD component.
Does anyone have documentation on this component or old version ?


Thanks in advance