Keithley 2010 KiCad Schematic Project

[KungFuJosh]

This started over in the Repair section, but we decided it would be better to have a thread dedicated to the K2010 schematics rather than getting lost in long repair threads.

The original schematics are from @picburner. I've done a bunch of updates and corrections along with guidance from @Kleinstein. Many thanks to both of them for their help and making this possible.

Anyway, there's only so much I can spend time on developing this further, so it would be better as a group project for anybody with a K2010 that wants to play along.

For anybody interested in helping, this is my suggestion to contribute:
- Preferably, pick a specific page to work on.
- Reply to the thread and let us know what page you're working on.
- Share screenshots for review.
- After review, share the updated KiCad page to be included in the main archive again (attach the single page as a zip, or add a .txt extension to it).
- Be sure to keep everything dated and numbered, and don't overwrite dates or titles for other pages.

Of course, it would be much cooler if Keithley would release the schematics for this and save us from doing the work. Oh well.

Thanks,
Josh

[Cyclotron]

Thanks KungFuJosh!
I'm going to grab this and add some updates that I've found during my K2010 repair journey.
 

[Cyclotron]

@KungFuJosh,

I was hoping to use Global Labels for connections that cross sheets. Such as +ICB, -ICB, FSRES, and so on.
Is it ok if I use those?   That will allow the netlist to generate connecting the sheets where they are supposed to be connected.

I'm a KiCad newbie but that's the way I understand Global Labels.

[KungFuJosh]

@KungFuJosh,

I was hoping to use Global Labels for connections that cross sheets. Such as +ICB, -ICB, FSRES, and so on.
Is it ok if I use those?   That will allow the netlist to generate connecting the sheets where they are supposed to be connected.

I'm a KiCad newbie but that's the way I understand Global Labels.

Absolutely! What's on there now is left over from the import. If you use Global Labels, set them as Passive unless you're sure it should be something different.

Thanks,
Josh

[Cyclotron]

Here's a first pass at the ADC. I need to validate some more items.
The labels have been put in as Global and I've adjusted the snap grid because it was off.
CR110 has the proper identifier and pins are showing.

I considered adding grid coordinates from the service manual layout image to components but I'm not sure how valuable that might be.
I can also note it in a table and attach.  A road map.

[Cyclotron]

I have trouble with bus diagram so I broke out the GPIB into a separate page to make it easier for me to trace lines.

[Doctorandus_P]

How do you approach a project like this? Apparently it started with a schematic made by picburner. What's his background?


My own approach would be with a handful of good pictures of the PCB. Then do some perspective and other corrections (Gimp, or similar) and then load the pictures directly into KiCad. (They can be scaled in KiCad itself).

Then go to the schematic, Create the symbols for a (sub) section of the PCB, attach footprints, and put them on the PCB again.
After that you can build up the schematic wiring, sync with the PCB and also draw the tracks over the drawing.

KiCad also has some more reverse-engineering functions. You can build up the netlist in the PCB Editor, and then push that back to the schematic. In the schematic the netlist will then appear as labels directly connected to symbols. I don't have much experience with this though.

[Cyclotron]

How do you approach a project like this? Apparently it started with a schematic made by picburner. What's his background?


My own approach would be with a handful of good pictures of the PCB. Then do some perspective and other corrections (Gimp, or similar) and then load the pictures directly into KiCad. (They can be scaled in KiCad itself).

Then go to the schematic, Create the symbols for a (sub) section of the PCB, attach footprints, and put them on the PCB again.
After that you can build up the schematic wiring, sync with the PCB and also draw the tracks over the drawing.

KiCad also has some more reverse-engineering functions. You can build up the netlist in the PCB Editor, and then push that back to the schematic. In the schematic the netlist will then appear as labels directly connected to symbols. I don't have much experience with this though.

I started adding to and adjusting the schematics due to working on a K2010 to repair it. I do not have a method because this is all new to me. I'm just trying to help and give back as there are several sources of data about the K2010 schematics but there remains gaps and some minor errors.

For me, I would love to be able to document in a "DOT" language model and have a tool that did the schematics for me.  But I've not looked into a way to do that because the KiCad project already existed and has substantial data.

There's likely significant differences in how to best do something like this depending on it being you alone or a community/team effort. I wasn't aware of the features you mentioned in KiCad. I'll take a look.

[KungFuJosh]

This is not my first time reverse engineering something. Theoretical tools are nice, but unless you have all components removed from the board (with a maximum of 2 layers), photos aren't that helpful. Too many traces pass beneath components with no clear way to see where they go.

The most realistic method, without clearing the board, is to trace things out as we have been. This will give a generally good idea of what connects where, but there will likely have some extra connections missing anyway. For example, if you trace one item that goes to multiple points, and don't catch all the common points. Some of those missing connections will get corrected anyway as more components are added.

It's always better to have somebody like @Kleinstein review the work, since (at least in my case) he understands what's going on a lot better than I do.

If you have a light panel, you can take photos like I will attach, and that can certainly help...especially if some components are removed.

[KungFuJosh]

A few more

[KungFuJosh]

..and a few more.

[Kleinstein]

There are 2 more helping points: the ADC part looks very similar to the Keithley 2182. There is schematics for this, but some more clues.
The front end part, before the main amplifier looks very similar to the Keithley 2000 - even quite some of the part lables are the same.
So one has rather good guesses to start with.

[KungFuJosh]

Q193, Q194 part numbers are wrong. They're shown as SMT SST4393. I'm guessing those were copypasta'd from Q140, Q188.

Does anybody know what those correct part numbers are? I don't want to cut the heat shrink.

[Kleinstein]

Q193/Q194 should be a pair of matched, low threshold FETs. The PN4393 (or similar) would make sense. If even lower noise is wanted a SK170 would be possible. At least the feedback resistors are rather low - maybe to get low noise. It would still likely fail from the odd AZ loop however, that leaves it sensitive the 1/f noise even in AZ mode.
The choice of availabel JFETs in a TO92 case is anyway limited.

The noise specs for the 100 mV range may give a hint, bit are a bit unclear if they are for AZ or non AZ mode. The odd AZ mode also makes it a bit hard to tell which frequencies are relevant.

[Cyclotron]

Q193, Q194 part numbers are wrong. They're shown as SMT SST4393. I'm guessing those were copypasta'd from Q140, Q188.

Does anybody know what those correct part numbers are? I don't want to cut the heat shrink.

Hmm,

I was looking at U185 and Q184 next door to the Q193/Q194 pair and found similar discrepancies in the service manual and the actual parts used to some extent.
At least as far as Q184 goes.  U185 is a TL431CLP and the part on mine agrees.

Q184 is "TRANS, N CHANNEL FET, 2N4392" per the manual, but that's not what I see. 
If I've decoded the part that's installed (see photo), it is a 2SC2611.  This makes no sense to me.

[Uunoctium]

With a bit of imagination, you can recognise an ‘i’ within a filled ‘S’.
If it’s a J-FET, then Siliconix could be the right choice 

[Cyclotron]

With a bit of imagination, you can recognise an ‘i’ within a filled ‘S’.
If it’s a J-FET, then Siliconix could be the right choice 

Brilliant!  Thanks. I did a search, and you're absolutely correct.  This is a Siliconix J2611, and a 2N4392 would be a similar part.
The J2611 seems to be a "binned" part per my searches.

The 2N4392 appears to be more commonly available in a TO-18 package.    
PN4392 TO-92.

Assuming the service manual is right, later boards would have the TO-8 2N4392, where mine has the J2611.

Thanks again.