Author Topic: Keithley 2001 DVM factory modifications to work with a K1801 nanovolt preamp  (Read 35717 times)

0 Members and 1 Guest are viewing this topic.

Offline chuckb

  • Regular Contributor
  • *
  • Posts: 240
  • Country: us
That's good news. You could probably fit it inside the metal case to minimize noise pickup. How big of a non polarized cap do you have?
 

Offline chuckb

  • Regular Contributor
  • *
  • Posts: 240
  • Country: us
The feedback cap only has the 1 ohms resistor to work with so the cap needs to be huge (0.47ms time constant). A better way would be an RC filter on the output. A 1k ohm and a 1.0ufd plastic cap would give you twice the filtering (1ms TC) that you have now.

Do you know about the trick with some plastic caps? The terminal connected to the outside foil is indicated with a black band or solid bar on the end of the capacitor. You connect that end to your lowest impedance node to minimize noise pickup. I see it on maybe 10% of the plastic caps I work with.

If I had to guess, based on the A10 schematic from LTZ2000 a few months ago, the output integrator cap has broken. I have not traced out my A10 pcb yet but that may be outside the metal can so it could be repairable. But I'm just guessing.

If you need to compare any voltages or waveforms just let me know and will tear down my amp.

I fixed my 30 Mhz osc issue with a few 0.1ufd caps on the LT1021-10 power supply outputs.
 

Online TiN

  • Super Contributor
  • ***
  • Posts: 4064
  • Country: us
  • xDevs.com/live - 24/7 lab feed
    • xDevs.com
That's already with 47uF 63V non-polarized cap inside chassis. It just hanging over precision 1K wirewound resistor.
I don't have much of film caps yet, still looking into digital controlled gain circuitry, so I'll get some for filter as well next year.

This was more like quick try, so I need to put more time and efforts into it, before I ask for comparison or measurements.
Btw, condition are same either battery +/-9V powered or from LT3043/LT3090 on our power card module. One of reasons why I wanted to get A10 going, is to measure noise of power card module, as scopes I have aren't able to go below 2mV/div, on which noise is non-measurable.

P.S. I wanted to be 100% that LDO section works perfect, so you'd have trouble-free unit for testing. Obviously Keithley transformer part I'm not able to test, so it's bit tricky. Module + some bare ones goes on Jan 4, will need only transformers to be populated and tested.
YouTube | Chat room | Live-cam | Have documentation to share? Upload here! No size limit, firmware dumps, photos.
 

Offline alanambrose

  • Frequent Contributor
  • **
  • Posts: 280
  • Country: gb
BTW do you guys have the pin-out for the 32 way connector into the Keithley? I would like to build some custom switching matrices.

TIA, Alan
“A foolish consistency is the hobgoblin of little minds"
 

Online TiN

  • Super Contributor
  • ***
  • Posts: 4064
  • Country: us
  • xDevs.com/live - 24/7 lab feed
    • xDevs.com
Of course we do, what you think :). Related data is in my nV preamp draft article. I'll update some layout screenshots and 3D CAD with keepouts next days. Beware pin numbering for connector :).

Are you going to opensource your design too? I have original TCSCAN-2001 but one can never have enough channels.
« Last Edit: February 04, 2016, 03:52:19 pm by TiN »
YouTube | Chat room | Live-cam | Have documentation to share? Upload here! No size limit, firmware dumps, photos.
 

Offline plesa

  • Frequent Contributor
  • **
  • Posts: 965
  • Country: se
Thats really good idea, PCB should be compatible with few relay types ( low emf, low leakage ). I assume that number of channels will be limited by Keithley cards supported in FW.
 

Offline z01z

  • Regular Contributor
  • *
  • Posts: 140
Did you know that in the older scan manuals you can also find schematics? It's a great starting point.
Apparently there are 2 ID pins on the connector, maybe changing their wiring changes the number of channels. Does anybody know what the different combinations mean?

When I was checking out relays for my home-made scanner board, I've realized that the pinout and pin positions are quite different for various relays. I also had this idea of having a PCB that can hold more that one type but gave up on it in the end. This might be possible for different relay types of the same manufacturer though.
 

Offline alanambrose

  • Frequent Contributor
  • **
  • Posts: 280
  • Country: gb
>>> Are you going to opensource your design too? I have original TCSCAN-2001 but one can never have enough channels.

Ah just got back to this thread. Thanks, and yes, of course.
“A foolish consistency is the hobgoblin of little minds"
 

Offline doktor pyta

  • Frequent Contributor
  • **
  • Posts: 278
  • Country: pl
    • RF Scientific
To chuckb,

Do You have an access to LCR bridge to measure the inductance of windings?
This will give precision info about turns ratio...

P.S. In famous "Low-Noise Electronic Design: C. D. Motchenbacher, F. C. Fitchen" book I read that:
'Do not check the dc continuity of any of the transformer windings
unless they can be adequately demagnetized!'
and
'
To measure the inductance of a low-level transformer accurately, use a
low-frequency impedance bridge with very small drive signals. Measure the
inductance at the lowest frequency used.
'

However Your transformers are ferrite core so they should be more resistant to that.

BTW. Here is the book:
http://www.pearl-hifi.com/06_Lit_Archive/14_Books_Tech_Papers/Motchenbacher_Connelly/Low-noise_Electronic_Design.pdf

Best Regards

Transformer update.
The K1801 power supply card has two transformers. I ordered them from Tektronix and they arrived in under a week. They are not cheap.
These are both Ferrite POT style transformers in 18 mm x 11 mm packages. The bobbin foot print is given in the attached pdf.

TR-285B ($193), core material 3E2A.
Primary, wire dia 0.0315" with insulation, this may be 20 ga. DC resistance = 0.0033 ohms. This is 3.9 inches of 20 ga wire.
Secondary, wire diameter 0.004" over insulation (difficult to measure), this may be 38 ga wire. The secondary winding is center tapped so there are two identical windings. DC resistance for one of the two matched secondary windings is 10 ohms. This is 15.4 feet of 38 ga wire.

When 0.1V RMS at 4.8 kHz is applied to the primary, half of the secondary generates 11.2V RMS.


TR-286B ($393), core material 3E27.
Primary wire dia 0.004" over the insulation (difficult to measure), this may be 38 ga wire. The primary is center tapped. The DC resistance of each half of the primary is 1.55 ohms.
This would be 28.7 inches of 38 ga wire.

Secondary #1 winding. This wire is smaller than the primary. I measured 0.003", again, this is very difficult to measure. I broke one of the wires trying to measure it. It may be 40 ga. This output is center tapped. The DC resistance of half of the winding is 12.6 ohms. This would be 11.7 ft of 40 ga wire.

Secondary #2 winding. This stranded wire is 0.028" in diameter (22 ga?). The wire is insulated as it goes into the transformer. The insulated wire is 0.047" in diameter. It has a DC resistance of 0.0061 ohms. This would be 4.5 inches of 22 ga wire.

With 2 V RMS applied to half of the primary, half of Secondary #1 developed 3.4 V RMS. Secondary #2 developed 0.0715 V RMS.

The TR-286 Secondary #2 feeds the primary of the TR-285 transformer. This looks like a technique to reduce leakage currents to the preamp. Insulated wire is used to reduce capacitive coupling inside the transformer and a low voltage excitation AC of less than 200mv is used to reduce capacitive coupling some more.

« Last Edit: May 05, 2016, 01:44:59 pm by doktor pyta »
 
The following users thanked this post: DiligentMinds.com

Offline chuckb

  • Regular Contributor
  • *
  • Posts: 240
  • Country: us
doktor pyta
Thanks for the ideas! No, I don't have access to an LCR meter. That's on my list to finish out my lab.

TiN did a great job developing an alternate power supply for the 1802 preamp interface card. That's probably the easiest way to get a working system.

I will be posting the results of testing the 1802 preamp and TiN's interface card soon.

chuckb
 

Offline chuckb

  • Regular Contributor
  • *
  • Posts: 240
  • Country: us
Here is some noise data I collected from a battery powered A10 preamp over the last month. The A10 preamp is very similar to the 1801 preamp produced by EM Electronics. I am collecting 1801 noise data right now and will have results later.

My A10 preamp is powered from two 10 AH 12v lead acid batteries. The batteries power two LT1021-10 preregulators to stabilize the + and - 10v power to the A10. The input of the A10 was shorted and the gain was set to 1,000,000 with a 1% metal film resistor. The output signal was digitized with a K2002 DVM set to average 100 samples. The preamp had a 2nv offset so the K2002 was working with a large 2mv signal. Also the K2002 was set on 8.5 digits and 100 NPLC so it did not contribute any noise to the measurement. A GPIB interface cable sent the data to a PC that runs a small HP VEE program to collect data every 30 seconds.

Because temperature change and temperature rate of change is a issue with nv measurements the preamp is inside two styrofoam containers. The inner container holds two preamps, 2 lead acid batteries, several lbs of metal for thermal mass and the temperature sensor for the HP2804A Quartz temperature sensor (0.0001 C resolution). The temperature sensor measure the temperature of the small metal enclosure around the A10 preamp. That container is wrapped with bubble wrap to minimize air currents. All of that goes inside a square container built with 2" thick styrofoam sheets (R-10). That seems to eliminate short term temperature changes even with the airconditioning cycling on and off. I still see temperature changes on a daily time scale.

Over the 25 days of the testing the preamp noise was constant at 0.5nVpp. The zero reading drifted inside a 2nv band as the preamp temperature changed 2 deg C. Not to bad.

What good is this kind of performance? Well, lower noise is always better...
If this was measuring the difference between two 10V JJ arrays the difference could be determined to better than 0.0001 ppm or 1e-10. With a little noise averaging and some simple temperature control of the preamp that performance could be 10 times better.


I was able to take a few days off from work so now the Keithley 1801 preamp card that TiN designed and manufactured is working and working very well. Thanks TiN! I will post that data later.

 


« Last Edit: May 22, 2018, 05:53:49 am by chuckb »
 
The following users thanked this post: EmmanuelFaure

Offline chuckb

  • Regular Contributor
  • *
  • Posts: 240
  • Country: us
To prepare the K1801 for it’s new life the 22 year old electrolytic capacitors were replaced. The new capacitors are higher voltage, higher temperature and have a very long rated life.

The original Rubicon 1000ufd 16v caps were replaced with 35v capacitors. These capacitors are in a 9v circuit so the 35v caps have a good margin. There were two smaller 100ufd, 16v caps. These were also replaced with 35v units. I chose large caps for long life but they were a little difficult to fit inside the housing.

It’s now ready to go for another 30 years or more.

A preamp interface cable was built using a 14 ft length of CAT 6 7 network cable. This has the advantage that all 4 wire pairs are foil shielded and there is an overall outer foil shield.
The 4 command lines were placed in two of the shielded pairs. The commands are 1k gain, 10k gain, 5ufd filter, 50ufd filter. These are simple +-9v dc signals. The +- 9v power lines were in another shielded pair. The preamp output and the ground reference were the last two signals to be placed in a pair. The overall shield was tied to preamp case on one end and to the isolated ground at the interface card.

TiN did a great job designing and building the interface card. He built one version using the original very expensive transformers. Then he designed another version to work with much less expensive transformers. I should have just waited for the second version! I did not use the bias current cancellation circuitry and so far it has not affected the noise readings or stability. The card was working in my K2001, the cable was built, now I just needed to plug it into the rare K1801 preamp. After triple checking voltages and my connector pinout I plugged it in and it worked fine.

Adding the preamp to the K2001 or K2002 adds several new ranges to the DVM. For example now there is a 20uV, 200uV, and a 2mV full scale range for DC volts. The new 2m ohm, 4 wire ohms range should read down to u ohms now.

I let the equipment warm up a bit and followed the factory calibration procedure. I used an EDC 521 calibrator monitored with a K2002 DVM to provide voltage to K262 Low Thermal Voltage Divider. Luckily I had a Keithley low thermal cable to connect the K262 to the K1801 preamp. The K2001 built in calibration procedure has you use a K262 divider so that worked out well. After the 20uV full scale calibration on the new 20uV range. The readings were spot on at 20.000 uV, even the negative full scale reading was right on. The calibration limit is about 1 nV because of the noise floor.

The 200uV range was just as accurate. The 2mV range would not calibrate. It looks like my K1801 has an issue above 0.5mv (0.5ma output stage current) on the x1000 gain range. Oh well, I'm just using it for the more sensitive ranges anyway.

More test data to follow but the K1801 noise data looks as good as the A10 noise data.

Thanks to TiN and others, it looks like this project was everything I hoped it would be.

 
« Last Edit: June 08, 2016, 02:47:01 am by chuckb »
 
The following users thanked this post: TiN

Offline chuckb

  • Regular Contributor
  • *
  • Posts: 240
  • Country: us
The battery powered A10 preamp and the K1801 preamp were both installed in the insulated containers. Data was collected for several hours. The K1801 has a little more noise than the A10 preamp. Both preamps were operating at a gain of 100,000 with the inputs shorted and floating. I plotted the raw data (30 second samples) and added a line that averaged 15 minutes of the data. I shifted the line 0.5 nV up so the two lines could be clearly seen.

The temperature smoothly changed -0.3 deg C inside the enclosure during this data run.

I'm a happy Voltnut. Two nV preamps that can discern a 0.1nV change!

« Last Edit: June 08, 2016, 02:46:23 am by chuckb »
 

Offline doktor pyta

  • Frequent Contributor
  • **
  • Posts: 278
  • Country: pl
    • RF Scientific
chuckb, thanks for sharing the measurement results !
P.S. When I wrote about measuring inductance of the transformers I was totally convinced that You have the signal transformer that are inside the "front-end magic can". Now I see these transformers are used in the power supply section.
Sorry for the confusion.

Online TiN

  • Super Contributor
  • ***
  • Posts: 4064
  • Country: us
  • xDevs.com/live - 24/7 lab feed
    • xDevs.com
Long overdue update from my side, finally got the setup ready last night during impulse livestream session.



Modified EM A10 with three gain range switch resistors (100K, 10K, 1K) and pair of MOSFETs to add digital gain control to let it use with Keithley 2001/2002.
Next step is to move thing to K2002 and calibrate.

Now more difficult task - find a store to buy M3 copper nuts  ???.
« Last Edit: February 09, 2017, 05:03:15 am by TiN »
YouTube | Chat room | Live-cam | Have documentation to share? Upload here! No size limit, firmware dumps, photos.
 

Online TheSteve

  • Supporter
  • ****
  • Posts: 3033
  • Country: ca
  • GHz
Now more difficult task - find a store to buy M3 copper nuts  ???.

copper stock, drill, tap&die set - make your own perhaps.
VE7FM
 

Online Vgkid

  • Super Contributor
  • ***
  • Posts: 2517
  • Country: us
Closest version of copper is brass(You could try it, but it would not be ideal). Lets see what a nanovolt source uses.
We will go with the Keithley 560. I recently lost a bid on the 181 that had popped up, forgot to up my bid. :-[  :phew:

I recognize those binding posts.
CSI Enhance go.


In the Keithley manual, it uses copper nuts, but it doesn't say how they are used.
nut-spade-nut
or
screw flange-spade-nut
------
Furthermore it says that if copper nuts are not available, then twist the wires around the binding posts.
Anyone own a Keithley 560, that they want to open up the shielded nanovolt divider portion. >:D
---Sorry for typo's I can't fall asleep.
« Last Edit: February 09, 2017, 07:13:27 am by Vgkid »
If you own any North Hills Electronics gear, message me. L&N Fan
 

Offline ManateeMafia

  • Frequent Contributor
  • **
  • Posts: 719
  • Country: us
Isn't the Keithley 7168 using generic screws to tighten the copper lugs?
You might try nylon washers with brass nuts and get good results.
 

Offline chuckb

  • Regular Contributor
  • *
  • Posts: 240
  • Country: us
TiN, congratulations on the mod!

Your 1801 Power supply / Interface card has been connected to a K1801 and running 24/7 since I started it up last year. It's working great!
 

Online Vgkid

  • Super Contributor
  • ***
  • Posts: 2517
  • Country: us
Looking at other keithley microvolt meters. They sandwich the copper lugs between 2 brass nuts.
If you own any North Hills Electronics gear, message me. L&N Fan
 

Offline USMC_Spike

  • Regular Contributor
  • *
  • Posts: 73
  • Country: us
This is all really good stuff y'all.  I want to play too.

But Alas...My 2001 got hit on a deadly corner...and UPS
refuses to pony up for a costly new display.

So I'm in limbo until I can find a good working affordable display.
Hoping that is the only problem I'll find on a former working unit.

If you have any ideas....I'm all ears.

Spike
 

Offline ap

  • Regular Contributor
  • *
  • Posts: 248
  • Country: de
    • ab-precision
The display of a 7001 scanner should fit. These often come cheap.
Metrology and test gear and other stuff: www.ab-precision.com
 

Online TiN

  • Super Contributor
  • ***
  • Posts: 4064
  • Country: us
  • xDevs.com/live - 24/7 lab feed
    • xDevs.com
I'd listed to ap and buy 7001 with confirmed working VFD to replace the glass (FP PCBs not compatible directly).

Alternative - contact plesa member here, he got few new VFDs and was selling them at 100$ cost some time ago. Not sure if he have any left.
YouTube | Chat room | Live-cam | Have documentation to share? Upload here! No size limit, firmware dumps, photos.
 

Offline mimmus78

  • Supporter
  • ****
  • Posts: 673
  • Country: it
There is a guy in UK selling 50pcs 7001 on eBay. If you contact him I think you can get one very cheap.

Inviato dal mio Nexus 6P utilizzando Tapatalk

 

Offline rastro

  • Frequent Contributor
  • **
  • Posts: 368
  • Country: 00
...
So I'm in limbo until I can find a good working affordable display.
Hoping that is the only problem I'll find on a former working unit.

If you have any ideas....I'm all ears.

Spike

USMC_Spike;
Have you tried running the unit remotely over the serial port?  You could then at least verify operation.  adapt-improvise-and-overcome :-+

Also I've heard UPS is notorious for flatly rejecting initial clams.  It not too much time has elapsed you may want to challenge their response.
« Last Edit: February 21, 2017, 02:34:05 pm by rastro »
 


Share me

Digg  Facebook  SlashDot  Delicious  Technorati  Twitter  Google  Yahoo
Smf