Keithley 199 digital board replacement

[diego]

Some time ago, I bought an old Keithley 199. Unfortunately it was broken. The digital board had a shortcut on the "Versatile Interface Adapter" which is only a nice chip which provides digital I/O and some timer functionality to the main processor. Unable to find a replacement (well, I didn't want to find one), I decided to rebuild the whole digital board. I wanted to use a modern microcontroller and provide some extra functionality which the old Keithley didn't originally had, as for example USB interface and Math capability, digital filtering and longer integration times.

I based the new board on a PIC32MX controller which I had laying around. For now the firmware has basic AC/DC voltage and current measurement capability with really good accuracy, which I think is at least as good as the old firmware (I compared the readings with a calibrated Keithley 2000 as shown in attached photograph). You can find the code and the schematics on my github repository:

https://github.com/asanza/keithley_199

[merox]

The digital board had a shortcut on the "Versatile Interface Adapter" which is only a nice chip which provides digital I/O and some timer functionality to the main processor. Unable to find a replacement (well, I didn't want to find one), I decided to rebuild the whole digital board.

I should have quite a bunch of them somewhere in the basement, so if somebody ever needs one (it's a 6522 as far as i know): send me a PM.

[Vgkid]

good job on the repair, but as mentioned above VIA's are sort of common on ebay, but are still produced.
http://www.mouser.com/Western-Design-Center-WDC/Semiconductors/Interface-ICs/I-O-Controller-Interface-IC/_/N-4oneu?P=1yzrr9xZ1yzrr9xZ1yzgd2bZ1z0z63x

[timb]

Wow, cool project, but talk about overkill! At least browsing your code introduced me to libscpi!


Sent from my Smartphone

[diego]

Wow, cool project, but talk about overkill! At least browsing your code introduced me to libscpi!


Sent from my Smartphone

Yea, a little bit overkill. But It was very interesting to discover how they do the analog digital conversion, and I think I just get carried away. A difference of modern multimeters, where the integrator is controlled by precise timing by an FPGA, in the k199 the whole thing is controlled by the microcontroller. It was pretty difficult to do the same with the pic32. The new software is not yet ready (only the ac/dc measurement part is done)... I hope to get some free time in the next months.

good job on the repair, but as mentioned above VIA's are sort of common on ebay, but are still produced.
http://www.mouser.com/Western-Design-Center-WDC/Semiconductors/Interface-ICs/I-O-Controller-Interface-IC/_/N-4oneu?P=1yzrr9xZ1yzrr9xZ1yzgd2bZ1z0z63x

Thanks for the link, but I don't think I am going to need them anymore.  . May be I can get the new digital board to do something more interesting, like increasing the resolution to 6-7 digits. For now, it is possible to set arbitrary integration periods, and I read somewhere in this forum that someone has already changed the k199 voltage reference for something more stable (based on lm399).  https://www.eevblog.com/forum/chat/bench-multimeter-(another)/msg265846/#msg265846

[macboy]

...
Thanks for the link, but I don't think I am going to need them anymore.  . May be I can get the new digital board to do something more interesting, like increasing the resolution to 6-7 digits. For now, it is possible to set arbitrary integration periods, and I read somewhere in this forum that someone has already changed the k199 voltage reference for something more stable (based on lm399).  https://www.eevblog.com/forum/chat/bench-multimeter-(another)/msg265846/#msg265846

Internally (and over GPIB) the resolution is 6.5 digits on the real thing. And that's "actually" 6.5 digits (3300000 count), not 1 million count, and calling it 6.5 digits. One million count is 6 digits, not more, HP/Agilent/Keysight. The last digit is generally not stable unless filtering is turned on. And of course, you need to remember the accuracy (0.009% or 90 ppm) when considering the reading. For fun, I wrote a script to take readings over GPIB (30/second or so), average them, and display the averaged result back onto the meter's display, giving me a 6.5 digit display. Then I bought a Keithley 2001. And another, and...

[timb]

I need to put the 199 on my eBay watch list. I love these 80's vintage Keithley meters so much. My 195A seems to have the best precision out of all my meters, rivaling even my Tek DMM4020!

The only thing I need to do is replace the banana jacks with safety shrouded versions.

Does anyone have a 192? That's 6.5 digit in some ranges, isn't it?

I call this series of meters the "Brown Breadbox Multimeter". There's just something about those multiplexed seven segment displays, the accuracy, fast update speed, low drift...

Though I have to say, my little 197A might be my favorite. I've got an early 90's vintage one with the GPIB and battery options. Such a compact accurate little meter. (Only issue is the LCDs die over time, along with the backlight. Another forum member reverse engineered the LCD protocol and replaced his with a set of 7-segs driven by an Arduino. I found a graphical OLED that will fit perfectly in the display window, so I may replace mine with one.)

Oh yeah, I picked up a 2000 on eBay for $200, should be here tomorrow. I know it powers on, but beyond that... We'll see!

Hi, I'm timb and I'm a Keithley addict.


Sent from my Smartphone

[diego]

Internally (and over GPIB) the resolution is 6.5 digits on the real thing. And that's "actually" 6.5 digits (3300000 count), not 1 million count, and calling it 6.5 digits. One million count is 6 digits, not more, HP/Agilent/Keysight. The last digit is generally not stable unless filtering is turned on. And of course, you need to remember the accuracy (0.009% or 90 ppm) when considering the reading. For fun, I wrote a script to take readings over GPIB (30/second or so), average them, and display the averaged result back onto the meter's display, giving me a 6.5 digit display. Then I bought a Keithley 2001. And another, and...

Thanks for the tip, I didn't know that the internal resolution was 6.5 digits (can't find in the manual). I wonder why the last digit is not stable, may be it has something to do with the jitter in the integration period. I did analyze a working k199 (to find out how they control the integrator) and found that the integration period jitter is about 80uS 3uS.

In the replacement board, I used the output compare capability of the pic32 controller to generate the start-stop integration period, and as it is done in hardware, has almost no jitter (< 1ns, I cannot measure smaller intervals). It should be a lot more stable on the last digit.

[wiss]

Does anyone have a 192? That's 6.5 digit in some ranges, isn't it?

Yes, only in 200 mV it's 5.5.

[cellularmitosis]

Just wanted to dig up this thread to reply and say that this is an insanely cool project.  The thought of engineering an alternate digital board is such a neat idea!