Author Topic: Keithley 2001 DVM factory modifications to work with a K1801 nanovolt preamp  (Read 35800 times)

0 Members and 1 Guest are viewing this topic.

Offline chuckb

  • Regular Contributor
  • *
  • Posts: 240
  • Country: us
I tested the On Semi CAT24C01 on a connector. I had to reinitialize it each time power was applied. It stored the 1k and 10k gain constants (or at least looked like it did) but it would not even accept the 100k gain constant.

The XICOR 1991 data sheet does not talk about I^2C data bus. The data sheet refers to Two wire data bus. WIKI says multiple chip addressing was added to the standard in 1992. I guess these meters were designed a long time ago.

I just grabbed an X24C01P Xicor EEPROM from ebay. $9 with shipping. Will be here next Monday.
 
 

Offline TiN

  • Super Contributor
  • ***
  • Posts: 4067
  • Country: us
  • xDevs.com/live - 24/7 lab feed
    • xDevs.com
Won 20pcs Xicor X24C01F on bay..
YouTube | Chat room | Live-cam | Have documentation to share? Upload here! No size limit, firmware dumps, photos.
 

Offline Macbeth

  • Super Contributor
  • ***
  • Posts: 2552
  • Country: gb
Won 20pcs Xicor X24C01F on bay..

You don't do things by halves, TiN. I just did a quick ebay search...

A seller in HK - "0 available /  20 sold"  :clap:

This is a predatory move, and you have cornered the market in X24C01Fs. I reckon you could give Randolph & Mortimer Duke a run for their money  :-DD
 

Offline TiN

  • Super Contributor
  • ***
  • Posts: 4067
  • Country: us
  • xDevs.com/live - 24/7 lab feed
    • xDevs.com
I would never thought I'd need by 24xx EEPROM on ebay, EVER! Never say never..
I did not knew they have 20 total, anyway :)
YouTube | Chat room | Live-cam | Have documentation to share? Upload here! No size limit, firmware dumps, photos.
 

Offline Macbeth

  • Super Contributor
  • ***
  • Posts: 2552
  • Country: gb
Oh just a quickie - do you have the board dimensions and the make and model for the Keithley scan card connector? I fancy making myself a more modern version of the TCSCAN for my 2000.
 

Offline TiN

  • Super Contributor
  • ***
  • Posts: 4067
  • Country: us
  • xDevs.com/live - 24/7 lab feed
    • xDevs.com
Well, my apology to folks who are not so well at soldering and don't have microscope... I used soldering iron only, but in few places it was tricky to do.
To solder DFN and TSSOP center pads I heated those pads from bottom and placed chip on top to get it aligned and soldered at bottom. To ensure best heat transfer I used beefy 4.2mm chisel tip.

Using same wide tip I soldered most of caps.

-





Then switched to 102SDLF0.4mm angled tip and soldered LT3090 and LT3042 DFN's pins and rest of 0603 packages. You need to be patience, as board have no thermal barriers :).

Here's crappy soldering result:

-

-

Checking if it have a chance to work... I don't have proper resistors yet, need order some first.

-

+8.5V LDO works (output 7.07V due to not right resistor to set Vout)
rest don't work :)

6 hours spent.

Macbeth

If you not in hurry, all docs will be posted after I get it all working[/b].
YouTube | Chat room | Live-cam | Have documentation to share? Upload here! No size limit, firmware dumps, photos.
 

Offline Macbeth

  • Super Contributor
  • ***
  • Posts: 2552
  • Country: gb
No hurry, actually I should have read the thread from the start as the connector is on the second page available from Farnell :-[

I can measure up for the board but always good to have someone elses template to work from  :-+
« Last Edit: November 17, 2015, 08:13:41 pm by Macbeth »
 

Offline TiN

  • Super Contributor
  • ***
  • Posts: 4067
  • Country: us
  • xDevs.com/live - 24/7 lab feed
    • xDevs.com
Fixed issues with LDOs, current limiting for negative LDO tied to negative Vin, not GND like on positive ones. Doh.  :palm:

Two 9V Lithium 6F22 accumulators hooked up on power input and output. No load connected.

Positive 8.5V LDO (they all adjustable, I call them this just for clarity, following schematics) = +7.917V output, taking 4mA idle
Negative 8.5 LDO = -7.772V output, taking 1.8mA idle
Positive 5.0V LDO = 5.006V output
Negative 5.0V LDO = -4.910V output

Voltage is set by resistors on pins, so don't worry about voltage levels, they are not important at this moment. I used random jellybean 5% resistors to match up with output voltages. Final will have proper 25ppm 1% ones. Here's are some scope shots with 8.5V voltages connected via direct 1:1 BNC cable:

-

Start-up positive supply (left), negative supply (right)

Noise measurement, scope is AC coupled, minimum voltage scale (1mV/div, TDS5034B), impedance set to 50 Ohm



You see anything? I don't see. Scope shows exactly same with inputs grounded. No surprises, noise well below measurement levels on usual gear :)



1 second scale, same story. Right capture - LDO input power is taken from Agilent E3649A ±10V. PSU causing ~1mV ripple on positive, and ~2mV on negative output. Could be wiring antennae though, did not re-check.



Now need get those pesky EEPROMs, order rest of parts and we can unbox preamp module and try some real stuff.  8)
YouTube | Chat room | Live-cam | Have documentation to share? Upload here! No size limit, firmware dumps, photos.
 

Offline ManateeMafia

  • Frequent Contributor
  • **
  • Posts: 719
  • Country: us
I see you are running the 3458A without top covers. Does this have something to do with your drift issue or is that blue cable monitoring something?
 

Offline TiN

  • Super Contributor
  • ***
  • Posts: 4067
  • Country: us
  • xDevs.com/live - 24/7 lab feed
    • xDevs.com
Yes and no :) Blue coax was A9 output monitoring. All the drift results were with covers on.
YouTube | Chat room | Live-cam | Have documentation to share? Upload here! No size limit, firmware dumps, photos.
 

Offline chuckb

  • Regular Contributor
  • *
  • Posts: 240
  • Country: us
I tested the On Semi CAT24C01 on a connector. I had to reinitialize it each time power was applied. It stored the 1k and 10k gain constants (or at least looked like it did) but it would not even accept the 100k gain constant.

I just grabbed an X24C01P Xicor EEPROM from ebay. $9 with shipping. Will be here next Monday.

The XICOR X24C01P arrived early today. I just installed it in a socket on a bare connector. NOW the meter remembers that it has a preamp between power cycles! However when I manually store the gain constants they return to 1.000 between power cycles. I still can't manually set and store the 100k Gain value. Maybe it needs a normal calibration first.

 

Offline TiN

  • Super Contributor
  • ***
  • Posts: 4067
  • Country: us
  • xDevs.com/live - 24/7 lab feed
    • xDevs.com
Based on my LA captures, meter does not write ROM if you just change gain values, it only does so after calibration or GPIB command ...SAVE from 1801 manual.
YouTube | Chat room | Live-cam | Have documentation to share? Upload here! No size limit, firmware dumps, photos.
 

Offline chuckb

  • Regular Contributor
  • *
  • Posts: 240
  • Country: us
I also have an old (1970s) N2a nanovoltmeter from EM Electronics. It needs a little work but I don't have any info on circuit details. I should probably start by changing the PS caps. Get those nasty electrolytics out of there.

I believe these are all modern versions of the K148 nanovoltmeter. Reference the Keithley web site for the schematic. The K148 uses a mechanical chopper to turn the DC signal into an AC signal to feed to the step up transformer (1:100 ratio?). Then you can use normal electronics to amplify the signal and demodulate it. The transformer lets the input noise be 100 times less than the first stage amplifier noise. Nothing is free so the trade off is 100 time higher input current. But in most low noise applications that is not an issue. The newer versions probably use low resistance MOSFET transistors to do the switching.

I saw a very good design article for a chopper transformer nv preamp in The Review of Scientific Instruments back in the 90s. I'll try and find it. I had to go to the local college to look it up.

Just an update on this comment.

I found the old article. It was in the March 1992 (Vol 63, No 3) issue of Review of Scientific Instruments, page 2089. "Low-noise preamplifier with input and output transformers for low source resistance sensors" by J. Lepaisant, et al.

They work through a very good analysis for a 100 pV / rt hz preamp with transformer coupling. This is the Johnson noise level for a 0.5 ohm resistor! We would need to add a chopper to the front end for DC operation.

Always on the lookout for new ideas...
 

Online Zucca

  • Supporter
  • ****
  • Posts: 2184
  • Country: it
  • EE meid in Itali
Can't know what you don't love. St. Augustine
Can't love what you don't know. Zucca
 

Offline TiN

  • Super Contributor
  • ***
  • Posts: 4067
  • Country: us
  • xDevs.com/live - 24/7 lab feed
    • xDevs.com
Revive, revive!



Some goodies are here:



Will it crack? :)
YouTube | Chat room | Live-cam | Have documentation to share? Upload here! No size limit, firmware dumps, photos.
 

Offline TiN

  • Super Contributor
  • ***
  • Posts: 4067
  • Country: us
  • xDevs.com/live - 24/7 lab feed
    • xDevs.com
Assembly push-pull supply based on LTC3439 - Check
Assembly Xicor X24C01 - Check
Correct TVS directions - Check
Add connectors - Check
Test with 5V power from SMU - Check
Plug into meter - Check
Init preamp EEPROM - Check
Do dummy calibration - Check
Confirm preamp card is detected properly in 2001M B17 - check
Check preamp card in K2002 - Check (data is garbled though, perhaps 2002 use different format?)



Tests with real preamp head hopefully this week!
Need make few dividers to get microvolt outputs for calibration first!
« Last Edit: December 01, 2015, 04:37:17 am by TiN »
YouTube | Chat room | Live-cam | Have documentation to share? Upload here! No size limit, firmware dumps, photos.
 


Offline TiN

  • Super Contributor
  • ***
  • Posts: 4067
  • Country: us
  • xDevs.com/live - 24/7 lab feed
    • xDevs.com
I was trying preamp EM A10 today with 2002, but had very noisy readings.
Connected output of preamp to scope instead, got this:



287 Hz ripple. Turned off all lights and gear around, no difference. Preamp powered from 9V batteries (+/-8.5V measured). Tried 10Meg and 50 ohm at output, zero difference.
Cleaning board did not help either. :\
YouTube | Chat room | Live-cam | Have documentation to share? Upload here! No size limit, firmware dumps, photos.
 

Offline chuckb

  • Regular Contributor
  • *
  • Posts: 240
  • Country: us
I just checked my A10 and even with a gain of 10,000,000 it clean down to 1mV at those frequencies. I was expecting some chopping noise. The K1801 uses 488Hz. I don't know about the A10. I did notice some 30Mhz noise at 10mvpp. I will have to look into that.
 
I'm using +-12v lead acid batteries feeding LT1021-10 regulators.

What gain are you running? 20k?
 

Offline TiN

  • Super Contributor
  • ***
  • Posts: 4067
  • Country: us
  • xDevs.com/live - 24/7 lab feed
    • xDevs.com
I tried with gain 10k and 100k (feedback resistors 1k and 10k), the noise was exactly same.
YouTube | Chat room | Live-cam | Have documentation to share? Upload here! No size limit, firmware dumps, photos.
 

Offline chuckb

  • Regular Contributor
  • *
  • Posts: 240
  • Country: us
If I am remembering correctly, the internal feedback resistor is 1 ohm so the 1k resistor should provide a gain of 1000.
 

Offline TiN

  • Super Contributor
  • ***
  • Posts: 4067
  • Country: us
  • xDevs.com/live - 24/7 lab feed
    • xDevs.com
Mm, yes, sorry. So it's even lower gain then. I'll play with feedback cap (stock should be 10pF) to see if oscillations so away.
YouTube | Chat room | Live-cam | Have documentation to share? Upload here! No size limit, firmware dumps, photos.
 

Offline chuckb

  • Regular Contributor
  • *
  • Posts: 240
  • Country: us
Are you running the signals through the power supply pcb?

With battery power, the A10 can be a stand alone amplifier. I have operated the amplifier with a short across the input, battery power and connected the output directly to a K2000. Just set the y=mx+b scaling to 0.0001 if you use a 10k feedback resistor.

I believe the A10 has internal protection if both supplies are not energized at the same time, to prevent lockup conditions. Still, sometimes it takes a minute for the A10 output to come off the rail voltage (4v) and settle at 0 volts.
Good Luck!
 

Offline TiN

  • Super Contributor
  • ***
  • Posts: 4067
  • Country: us
  • xDevs.com/live - 24/7 lab feed
    • xDevs.com
Exactly like you saying, two 9V batteries powering A10 via 10cm wires, inputs shorted with copper wire, output connected to scope via coax (tried with active probe, same result). Voltage settles around zero RMS with this oscillation in place after a minute warmup. It does not change over 10 minutes time.

If I supply 10uV, 100uV, 1mV from DC voltage standard, signal is properly amplified, but oscillation stays same (in both amplitude and frequency).
« Last Edit: December 28, 2015, 05:04:49 am by TiN »
YouTube | Chat room | Live-cam | Have documentation to share? Upload here! No size limit, firmware dumps, photos.
 

Offline TiN

  • Super Contributor
  • ***
  • Posts: 4067
  • Country: us
  • xDevs.com/live - 24/7 lab feed
    • xDevs.com
If I add 10-47u cap over feedback resistor, this ripple reduce tremendously.

Shorted amp input:



EDC MV106 output connected to amp input (10V range, output 0V):



Gain : ~1000 (1K resistor).


YouTube | Chat room | Live-cam | Have documentation to share? Upload here! No size limit, firmware dumps, photos.
 


Share me

Digg  Facebook  SlashDot  Delicious  Technorati  Twitter  Google  Yahoo
Smf