EDC 521 Voltage Standard - replacement precision 100:1 voltage divider?

[Ash]

Hi,

I've been a lurker on this forum for a while.. I recently acquired an EDC 521 DC calibrator which a known fault on the 100mV range. All the other ranges are stable and look to be very close against my best ability to test on a Keithley 6.5 digit DMM.

I've diagnosed the fault as a failure on the 100:1 voltage divider (a KRL branded 4-wire module about 2" x 3/4" x 3/4"). It is designated A-3600 on the schematic. I have contacted KRL Bantry to see if they can help, but not holding my breath on that one..

The 100mV range configures the output to a 10V range and places this across this divider, the output of which is presented to the output terminals. There is a small trim network around the divider, but it is not able to bring the voltage back into cal. In this mode, the input voltage to the divider is feed back to the output amplifier so after the divider, the output is "open loop" and specified as 20ohm output impedance, and EMF into 1Mohm compliance.

I was measuring about 88mV out for 10V in, about 12% down. After removing the resistor and verifying out of the unit, I can see the output leg is measuring ~17 ohms, not 20. Guessing it has a few shorted turns or some other damage from a "transient event" 

So, what can I use to replace this? Does anyone have a parts unit from this range?

Alternatively, does anyone have a recommendation on where to go to get a set of matched resistors?

I assume I need TC matched resistor pair, 1980 + 20 ohms. I don't know the tolerance, but given the cal pot allows me to adjust the voltage approx +/- 0.3% (+/-0.3mV for 10V in) I'm guessing an initial tolerance of 0.1% or better.

Thanks,
Ash.

[Ash]

Ok, so a positive update on this!

I heard back from KRL Bantry, very helpful - they identified the part (now well and truly obsolete) but they still had the last 5 units IN STOCK!!!!   

Because they were so old, their engineering team checked them all and they are still all in spec!

They were (understandably) pushing hard to sell me all remaining 5, but I couldn't justify that. I bought 2, so I'll have a spare and leave some on the shelf for anyone else that needs one.

But if anyone is looking for this part in the future, it is worth checking in with them as I know they still have 3 as of May 2017. The price wasn't too extravagant either.

http://www.krlbantry.com/

I would have been happy to use the other voltage / current ranges that were ok, but now I'll be able to fully restore this and hopefully one day get it properly calibrated.

Anyone have a friendly cal lab in Brisbane that is happy to do "hobby level" calibration (don't need certificates etc)? 

Cheers,
Ash

[Dr. Frank]

Would you mind shooting a few pictures of this divider, maybe one from the interior of the defect one?
Maybe you can also re-engineer the inner schematic of this part.

This divider is not fully described in the manual / schematics of the EDC 521, therefore it would be  very interesting to get more detailed information.

THX - Frank

[Edwin G. Pettis]

Interesting, I built one of these dividers for a customer not too long ago.  Nothing really special, just two resistors stuck inside a case and sealed.

[Ash]

Would you mind shooting a few pictures of this divider, maybe one from the interior of the defect one?
Maybe you can also re-engineer the inner schematic of this part.

This divider is not fully described in the manual / schematics of the EDC 521, therefore it would be  very interesting to get more detailed information.

Hi Frank,

I can't get into it because it is fully potted with epoxy, however I asked for the datasheet / specs from KRL. Hopefully they can find something for me. If/When I get the details I'll add the details here.

But basically it is just a 2 resistor voltage divider, with very stable and accurately matched temperature coefficient resistors. The first between the input and output is 1980 ohms, the other between the output and both the commons (grounds) is 20 ohms. I'm sure there is some careful placement and concern for kelvin connections... Something like the attached schematic maybe? 

FYI: I also have an EPROM image from this device if anyone needs it, but I have no idea what firmware version it is (the sticker had long since fallen off) Date codes on most of the parts are late 90's.

Cheers,
Ash.

[Ash]

Interesting, I built one of these dividers for a customer not too long ago.  Nothing really special, just two resistors stuck inside a case and sealed.

Hi Edwin,

Cool, good to know - I had remembered that you made precision resistors and if all else failed I was going to try to reach out to you for a suitable replacement. 

Any ideas on the TC and tolerance spec on these?

Would you mind shooting a few pictures of this divider

Sorry Frank, I forgot the photo of the divider. I only have one of the unit when I took it out however I'll add some more photos of the insides of the calibrator when I open it up again for repairs.

[Edwin G. Pettis]

Hello Ash,

I checked my records, the 1.98K was ±0.02%, an 807, and the 20 ohm was ±0.1%, an 802, the tracking TCR was under 2 PPM/°C and copper tape was wrapped around them.  The customer reported that it worked very well.  Despite the four pins, it is simply a series divider, apparently EDC asked for the rather unusual configuration for proprietary purposes (likely) but there is nothing proprietary about the network.  The epoxy casing does little to enhance thermal conduction either, copper tape will do at least as good or better. As to cost, about $16.00.  By the way the original networks were made by RCL who were bought up by KRL Bantry some years ago (one of the few takeovers that wasn't Vishay).

[Ash]

I checked my records, the 1.98K was ±0.02%, an 807, and the 20 ohm was ±0.1%, an 802, the tracking TCR was under 2 PPM/°C and copper tape was wrapped around them.  The customer reported that it worked very well.  Despite the four pins, it is simply a series divider, apparently EDC asked for the rather unusual configuration for proprietary purposes (likely) but there is nothing proprietary about the network.  The epoxy casing does little to enhance thermal conduction either, copper tape will do at least as good or better. As to cost, about $16.00.  By the way the original networks were made by RCL who were bought up by KRL Bantry some years ago (one of the few takeovers that wasn't Vishay).

Hi Edwin,

Awesome, thanks for taking the time to find the details. I should have my KRL replacement in the next day or so. Should have contacted you first the KRL parts weren't too expensive, but not $16 

Where can I find out more about your resistors? Do I just send a PM if I need something in the future?

Also, I'm interested in the technology behind these precision resistors, is there any books or similar you could recommend to give me some background on it?

Thanks,
Ash.

[Edwin G. Pettis]

Hello Ash,

Easiest way is to contact me at pettiseng@q.com, everything I build is 'custom'.

Outside of bits and pieces, I don't know of any books on the subject that are anything more than general discussions of the various resistor technologies.  Depending on the source it may be anything from accurate to little more than marketing BS.  Wire wound is the oldest of resistor technology in the very early days with various flavors of carbon composition resistors coming close behind.  Some aspects of the various resistor technologies are deemed proprietary (whether they actually are or not).  PWW and film/foil are the two main competing resistor technologies today and that depends to a fair degree on what the application is, they each have their advantages and drawbacks.  Both technologies, in their best forms are rather complex in nature, more so than many people think, particularly when getting into tight tolerances and low TCRs, things get very tricky to control.

It would take a small book to generally explain how a given resistor technology is manufactured even without getting into so-called proprietary details.

[Ash]

Hi Everyone,

The replacement parts arrived today  along with some measurements. See the attached image.

It looks to me like the original specs were 0.02% tolerance with a +/- 3ppm/K Temperature Coefficient.

I'll probably get some time to fix the calibrator in the next day or two and I'll add photos.

Just thinking through now how I'll "calibrate" the 100mV range given that I don't have anything to cal against at the moment. I guess I'll lock my Keithley 2015 (last cal in 2010) into the 10V range, and measure the voltage on the input and output of the divider and tweek the number so they match volts to millivolts. That will get it into the ballpark.. Then I have to decide how far down the volt-nut / test gear addiction rabbit hole I want to go. 

  Yeah, right.. we all know how that is going to turn out...

Outside of bits and pieces, I don't know of any books on the subject that are anything more than general discussions of the various resistor technologies.  Depending on the source it may be anything from accurate to little more than marketing BS.  Wire wound is the oldest of resistor technology in the very early days with various flavors of carbon composition resistors coming close behind.  Some aspects of the various resistor technologies are deemed proprietary (whether they actually are or not).  PWW and film/foil are the two main competing resistor technologies today and that depends to a fair degree on what the application is, they each have their advantages and drawbacks.  Both technologies, in their best forms are rather complex in nature, more so than many people think, particularly when getting into tight tolerances and low TCRs, things get very tricky to control.

It would take a small book to generally explain how a given resistor technology is manufactured even without getting into so-called proprietary details.

Thanks for the contact details Edwin.

I had hoped there was something to read, but I guess the secret sauce and exact tongue-angle to moon-phase charts are how precision manufactures keep in business   

Ash.

[Ash]

Hi Everyone,

I've replaced the 100:1 divider with a new one and the unit is now functioning well. There are still some periodic "glitches" where the relays all reset and I suspect this is because of a protection event occurring. This happens a bit if I mess with the knobs, the sometimes show an "F" or other symbol if positioned on the "edge" and I think the CPU is detecting this invalid state and tripping out the protection circuits. I guess I need to now take the front panel off and clean up all the switches.

Here are a bunch of photos of the inside of the calibrator. Note that it is a bit dusty, I will certainly carefully clean it up now that I know it is working. I will also replace all the electrolytic caps, they are all getting on a bit (date codes in the mid '90s). I want to be able to run this thing for many years without worrying about it, so I'll do it now before it a proper cal.

I went through the full cal process to make sure everything checked out (yes, I know I don't have anything to cal against properly, but I intend to get it working happily first, then can properly). As a bonus I now have perfect agreement to the last digit on most of the ranges on my K2015 and the standard  (I know that's not worth anything) 

Enjoy.. If anyone wants more details or other photos let me know. Sorry I had to shrink them to upload them.

Ash.

[nidlaX]

I've replaced the 100:1 divider with a new one and the unit is now functioning well. There are still some periodic "glitches" where the relays all reset and I suspect this is because of a protection event occurring. This happens a bit if I mess with the knobs, the sometimes show an "F" or other symbol if positioned on the "edge" and I think the CPU is detecting this invalid state and tripping out the protection circuits. I guess I need to now take the front panel off and clean up all the switches.

Nice repair! Before you clean the front panel switches, I suggest you check ALL the solder joints on vital components along the signal path, especially active components such as transistors and op-amps. From experience, these units can develop small cracks in the joints leading to intermittent conductivity and lockup / overload events. Drove me nuts when I was trying to debug my 522. Good luck!

[Ash]

Nice repair! Before you clean the front panel switches, I suggest you check ALL the solder joints on vital components along the signal path, especially active components such as transistors and op-amps. From experience, these units can develop small cracks in the joints leading to intermittent conductivity and lockup / overload events. Drove me nuts when I was trying to debug my 522. Good luck!

Hi nidlaX,

Thanks for the recommendation - I will do that. I did give it a good look over before, but I didn't pull it down enough to get to the front panel. I did fine a few suspect looking joints which I re-soldered.

I'm also going to give the boards a good clean and probably re-seat all the ICs as they are in sockets and who knows lurks there.

The glitches seem more common as the unit is warning up, but they don't stop. I've left it running for a few hours connected to my meter and even if I don't touch the control, I still glitches periodically, so you may be onto something there with an intermittent joint.

Now that I know it seems to be fully functional, I'll put some more time into it and get it ready to put back into constant service.

Interestingly there are 2 diode "calibration" stickers, the original one crossed out and a new "KH" labelled one (Krohn-Hite?). It seems that this unit may have been refurbished at some point after KH bought EDC and the diode re-tested. The KH label shows a much more precise voltage (and different) from the original EDC one, but the reference number is the same.

EDC: 6.1774V     at 7.0mA
KH:   6.177615V at 7.0mA


Ash.