Tektronix DM5010 DMM with GPIB

[FlyingHacker]

Hi,

I have a Tek DM5010 -- info here:

http://www.barrytech.com/tektronix/tektm500/tekdm5010.html
http://w140.com/tekwiki/wiki/DM5010

Manual here:

http://www.ko4bb.com/manuals/71.91.230.162/Tek_DM5010_Digital_Multimeter.pdf

The CMOS is backed up by a NiCd battery. This battery has died and leaked a tiny amount of salts. I got that fairly clean (any tricks to that? I just used IPA).

I am wondering if I could replace that battery with a supercapacitor. I have never worked with supercaps before. I was thinking something like this:

http://www.mouser.com/ProductDetail/Bussmann-Eaton/KR-5R5C105-R/?qs=sGAEpiMZZMuDCPMZUZ%252bYlyTGVaxltdWl5cuqcUD%252by%252bU%3d

Here is the section of the schematic where the battery resides (attached image). Do you think the 100Ohm resistor would be OK?

Thoughts? I already have a cordless phone NiCd battery on the way, and this could work, but it is bigger, and will eventually leak again.

[FlyingHacker]

I also wonder how long a CR2032 lithium cell might last if I just disconnected the power input at CR1133 and used only the battery.

[David Hess]

The CMOS is backed up by a NiCd battery. This battery has died and leaked a tiny amount of salts. I got that fairly clean (any tricks to that? I just used IPA).

IPA will work fine.  What matters is the water since the potassium hydroxide and its salts are water soluble.  Scrub the area down with some water with a little bit of dish soap and then rinse in distilled water.

I am wondering if I could replace that battery with a supercapacitor. I have never worked with supercaps before.

Certainly a supercapacitor will work but it needs to be rated at 5 volts.  I would estimate that the backup time would be from 2 to 4 weeks with a 0.22F capacitor.

Here is the section of the schematic where the battery resides (attached image). Do you think the 100Ohm resistor would be OK?

OK for what?  It will dissipate 250mW peak when first charging a dead supercapacitor.

Thoughts? I already have a cordless phone NiCd battery on the way, and this could work, but it is bigger, and will eventually leak again.

This is the simple way to handle it.  Remotely mount the battery and use twisted pair to the circuit board.

I also wonder how long a CR2032 lithium cell might last if I just disconnected the power input at CR1133 and used only the battery.

You would need to change the circuit around a little bit because you do not want to charge a primary lithium cell but assuming U1220 is a micropower SRAM, then it would last 2 to 4 years or so.

http://www.mouser.com/ProductDetail/Tadiran-Batteries/TL2155-P/?qs=sGAEpiMZZMtLoaml0iHoQ%2f%2fIdmmZouk5jrbtVEs2eJA%3d]A battery like this[/url] is more typical and would last 15+ years:

Tektronix may not have used a micropower SRAM in this case but you could change it.

[FlyingHacker]

Thanks for the reply.

I will wash it further with some water as well.

That super capacitor I listed is supposed to be 5.5V. It is a full Farad as well.

I wondered if the 100 Ohm resistor would be adequate current limiting when charging the dead cap.

The phone battery would likely be the simplest, and I have already spent a few bucks on the battery. But NiCds never last very long.

If I went the Cr2032 or similar route then I would remove CR1133 to prevent charging the lithium cell, and perhaps stick a diode in with the cell as well in case there was some feeding from pin 13 of the CMOS through the 22K resistor. Might be cool to have two battery holders in parallel so a second cell could be inserted while changing the first. It is annoying to lose the cal data.

Now I just need to find a source of 200Hz 190V and 700V AC for the calibration procedure. I think I can generator all the other input voltages, and check them with my Kiethley. These "modern" DMMs with the software calibration are a bit of a pain because you can't just use 120VAC, turn a pot, and say that is "good enough" for the 1000V scale.  Instead they want specific voltages on their input.

[FlyingHacker]

Do you think the 3.6V of the long life battery you suggested would be an issue? (vs. the original 2.4V)

Thanks again for the help. Always nice to get help from such a Tek guru. 

[Kleinstein]

Having a higher voltage (e.g. 3-4 V) from the LI cell should not be a problem, as the chip is normally running at 5 V. Supply current might be slightly higher at 3.6 V compared to 2.4 V. Fully charged the 2 NiCd cells can also go up to about 3.4 V.

R1133 would need a diode in series and a second diode to take over the current during normal operation (if CE is set low).

[David Hess]

I wondered if the 100 Ohm resistor would be adequate current limiting when charging the dead cap.

It is probably a 1/4 watt resistor so I think it will be fine.  It had to be designed to take into account a fully discharged or shorted NiCd battery and the supercapacitor will charge quickly.

If I went the Cr2032 or similar route then I would remove CR1133 to prevent charging the lithium cell, and perhaps stick a diode in with the cell as well in case there was some feeding from pin 13 of the CMOS through the 22K resistor. Might be cool to have two battery holders in parallel so a second cell could be inserted while changing the first. It is annoying to lose the cal data.

CR1133 is still needed because the SRAM has to be powered normally when in operation.  The only change is adding a diode in series with the battery.  The diode could replace R1131 but it is better to keep the series resistance as a safety measure.

I always thought the DM5010 memory backup time was too short for reliable operation.  If I had one, I would probably change it just to fix that aspect of its operation.

As far as dual battery circuits, it would be almost as good to just be able to change the battery while the instrument is in operation.

Now I just need to find a source of 200Hz 190V and 700V AC for the calibration procedure. I think I can generator all the other input voltages, and check them with my Kiethley. These "modern" DMMs with the software calibration are a bit of a pain because you can't just use 120VAC, turn a pot, and say that is "good enough" for the 1000V scale.  Instead they want specific voltages on their input.

Finding a calibrated AC or high voltage source is a perennial problem.  Some instruments allow entering the actual applied calibration voltage as long as it is within a specified range.

Do you think the 3.6V of the long life battery you suggested would be an issue? (vs. the original 2.4V)

No, it will not be a problem.  The SRAM drops into low power mode depending on its control signals and is quite happy anywhere between about 2 and 5 volts.  A primary lithium cell would require a series diode anyway to prevent charging when normal power is applied so the 3.6 volts would be about 3.0 volts anyway.  Your 3.0 volt lithium 2032 cell would become 2.4 volts.

[FlyingHacker]

Thanks for the info, guys.

I will probably see how long the cordless phone battery lasts, and then decide if I want to take an alternate approach.

Some instruments allow entering the actual applied calibration voltage as long as it is within a specified range.

Sadly, this one just wants you to apply the prescribed voltage and press enter. I was wondering about a signal gen into multiple 200Hz transformers with a high enough breakdown voltage. Perhaps I could get the voltage up high enough that way. I would use a fly back diode to protect the sig gen from the inductive hit when powered off. I have been trying to find home brew solutions for this. Oddly, I can't find anything online. I should think you could tweak it to match a known calibrated meter, and the. Use that as the source.

[David Hess]

Sadly, this one just wants you to apply the prescribed voltage and press enter. I was wondering about a signal gen into multiple 200Hz transformers with a high enough breakdown voltage. Perhaps I could get the voltage up high enough that way. I would use a fly back diode to protect the sig gen from the inductive hit when powered off. I have been trying to find home brew solutions for this. Oddly, I can't find anything online. I should think you could tweak it to match a known calibrated meter, and the. Use that as the source.

A kit form of AC and DC high voltage calibration source would be nice but I do not think the market is there to support such a thing.

I would use a clean sine wave source, step up transformer, and then a compensated divider with high accuracy average responding rectification to close the loop and produce an accurate output.

As an alternative to average responding rectification, sampling of the divided high voltage output followed by an instrumentation style slow but high precision ADC to directly measure the RMS output would work *if* the sampling can be done accurately.  I think a closed loop track-and-hold could meet that requirement since it is only dealing with low frequency AC.  I bet you could get significantly better than 0.1% this way limited only by the high voltage divider which itself can be calibrated with DC and a transient response test.

[FlyingHacker]

A kit form of AC and DC high voltage calibration source would be nice but I do not think the market is there to support such a thing.

Maybe not. Though I would think there would be a lot of people interested in some kind of meter calibrator in the $100-300 range that would provide up to 1000 or even 1500V AC/DC with reasonable stability.

I would use a clean sine wave source, step up transformer, and then a compensated divider with high accuracy average responding rectification to close the loop and produce an accurate output.

You are saying the rectification would be for DC, right? Would a sine wave into a step up transformer be adequate as an AC source? You would fine tune this until it read 1000V (or whatever) on a known calibrated meter. Then use that same output on the meter to be calibrated.

[David Hess]

... Though I would think there would be a lot of people interested in some kind of meter calibrator in the $100-300 range that would provide up to 1000 or even 1500V AC/DC with reasonable stability.

It gets more complicated if it is a full featured design and supports any output voltage but maybe that would be necessary.  It would be nice if the calibrator itself could be calibrated against a single accessible standard like that old HP multimeter.

You are saying the rectification would be for DC, right?

I was not thinking of high voltage DC but that is sure one way to provide it.  My 7CT1N curve tracer works that way.

Would a sine wave into a step up transformer be adequate as an AC source?

I think it would work but the alternative of generating a high DC voltage and then using a high voltage low current linear amplifier to produce the AC or DC output is feasible as well and probably better if AC and DC calibration is desired since the high voltage DC output would require something like that anyway unless the transformer runs at a high frequency.

You would fine tune this until it read 1000V (or whatever) on a known calibrated meter. Then use that same output on the meter to be calibrated.

It would be something like that but to be useful, the calibrator itself needs to have a minimum number of calibrations; otherwise you might as well use the calibrator of the calibrator.

Design the high voltage output divider for low temperature coefficient, low voltage coefficient of resistance, and low drift; this is where the expensive parts go.  Calibrate it at a lower voltage versus a known standard.  Then use it to produce the calibrated high voltage based on its performance.  For AC, its capacitance also needs to be calibrated but that can be done with a low voltage reference flat pulse generator and oscilloscope.  I wonder how it could be done without an oscilloscope; I know of a couple of frequency counters that manage it for adjusting x10 probe compensation so maybe I need to study that.