Repair : EDC MV106 DC voltage standard

[TiN]

Hmm, unplanned repair log

My EDC MV106 DC voltage standard stopped working suddenly without visible reason. Was fiddling with freshly received Keithley Model 182, sourcing various voltages from EDC while suddenly overload lamp lit and now unit stuck to it. Disconnected everything, toggled AC power, reseated sense shorts, no any improvement.

Output is stuck to 13.18V on 10V range, or -13.18 in inverse polarity switch, or 0.125V/-0.125 in 100mV or 13.18mV/-13.18mV in 10mV range. Changing knobs does not affect voltage output, and lamp overload stays on.

Measured zener voltage, it's same voltage as I had it measured in February 2014, 6.1698V, so it's likely output amp went dodgy?

Bummer 

[TiN]

Fixed, was loose wire between decades rotary switch.



So beware, loose wobbly mechanical switches can cause issues like this and trap young players with failures.

[PTR_1275]

I had my MV105G stop the other day, it blew the fuse fairly majorly. All sorts of thoughts were running through my head, turns out it was the neon in the switch causing the fuse to blow. I temporarily removed the neon and it was back in working order.

I'll keep the loose wires on the back of the switches in mind if I have this happen.

[mij59]

The solder joint of the rotary switch to the left in the picture is not the best, the solder did not flow very well, the other wire (leaving the picture to the right) may also be affected by the motion of the switch.

[texaspyro]

I recently scored a free MV-100H from a recycle pile.   It is basically the same as the MV-106.  The previous owner was apparently attempting a repair or was stoned out of his mind and disconnected the coax from the first decade(s).  The far end is currently flapping in the breeze, other end is soldered to the board.

Also there is no longer a connection between the first and second decade switches.  (First decade refers to the highest value one with the 6 trim pots on it...)

Does anybody have any nice photos  showing where these should be connected (or a description).  The photos on TIN's site are not clear enough to determine how this thing should go back together properly.

[Vgkid]

Give me a few (2-3)hours, and I will get you the wiring pics from my mv-106.

[dacman]

The decade with the 2 kOhm resistors should be 10 kOhm when set to 10 and zero when set to 0.  The decade with the 20 kOhm (or 19.99 kOhm resistors and 20 Ohm trimmers) should be 100 kOhm when set to 10 and zero when set to 0.  The unit is a big op-amp and the decade resistors are the feedback.

[texaspyro]

The decade with the 2 kOhm resistors should be 10 kOhm when set to 10 and zero when set to 0.  The decade with the 20 kOhm (or 19.99 kOhm resistors and 20 Ohm trimmers) should be 100 kOhm when set to 10 and zero when set to 0.  The unit is a big op-amp and the decade resistors are the feedback.

Yes, I've cleaned all the switches and they measure correctly.  The question now is how / where those two decades interconnect and how / where the coax connects.  I can make an educated guess, but would prefer some "ground truth" on the proper way to wire it up.

Overall, the unit is in excellent condition and the Vref seems to be good, so should work once the decades are connected.   There's not a whole bunch to go wrong in these units.  The potted amp module might be a source of grief, but I think the original owner was cleaning the switches before he got sidetracked, stuck it on a shelf, and later it wound up being trashed.  The hardware for the first decade switch was missing, but I fixed that.

[dacman]

The potted amplifier (which was used on several EDC units, including the CR103) was replaced with physical parts on the CR103.  It consists of an ICL7652 and NE5534 and associated circuitry.  In the first post of the thread below, there is a link to a schematic of the CR103 showing the ICL7652 and NE5534:
"https://www.eevblog.com/forum/testgear/edckrohn-hite-cr103-currentvoltage-standard-teardown-(large-images)/msg386101/#msg386101"
EDC also made the "potted" amplifier without potting it.  There were several resistors and capacitors on it, along with the op-amps.

[texaspyro]

I took my educated guesses and hooked the two decades together and hooked the center of the coax to the first decade.  Voila!  it works.   I need to fire up one of my 3458A's and check it out and do a cal.  But first, I need to verify my educated guesses... particularly where/if to connect the shield of the coax.  I also need to lube the switches... the Deoxit did a good job or removing who knows how many years of scuzz and probably all the old lube.

Also, there are two ground wires coming off the main board.  One connects to the rear panel.  The other is flopping in the breeze.   Anybody know the official place to hook that one?

[dacman]

This is Dave's teardown of an MV106J.  It has the unpotted chopper amp.

[texaspyro]

My module (potted) is labeled:

801-A-225
Series "D"
700723

[texaspyro]

I whipped out my diddle stick and went through the calibration procedure (with the upper two decades connected by clip leads).  I used a 3458A as the volt meter.   Everything seemed to go well.  The even numbered voltage settings on the upper decade are within 10 uV.  The odd numbered voltage settings on the upper decade are off (but still within spec):

3V -> +58 uV
5V -> +118 uV
7V -> +176 uV
9V -> +219 uV

The upper decade switch has trim pots for 1,2,4,6,8, and 10V.  The calibration procedure has the rather cryptic comment that after adjusting an even voltage value to look at the next lower (odd) voltage and to perform a "minor tweak" to the even setting.  This sounds like they want you to split the error between the even and odd voltages.  I'll wait until I have it back together properly before trying that.

I think the previous owner removed the upper decade assembly to try and "fix" the even/odd voltage errors (even though they are within spec).  It would be nice to figure out a way to get the odd voltages closer to true values without sacrificing accuracy of the even voltages.

[texaspyro]

I watched TIN's cal video and it had a clear shot of the upper decade switch connections and confirmed my educated guess how they should be wired.  What was throwing me off if the previous owner had exposed and tined the shield braid of the coax at the decade switch end.  I was assuming it needed to be connected to something, but apparently not.

[dacman]

The upper decade switch has trim pots for 1,2,4,6,8, and 10V.  The calibration procedure has the rather cryptic comment that after adjusting an even voltage value to look at the next lower (odd) voltage and to perform a "minor tweak" to the even setting.  This sounds like they want you to split the error between the even and odd voltages.  I'll wait until I have it back together properly before trying that.

I think the previous owner removed the upper decade assembly to try and "fix" the even/odd voltage errors (even though they are within spec).  It would be nice to figure out a way to get the odd voltages closer to true values without sacrificing accuracy of the even voltages.

The MSD uses 20 kOhm resistance legs yet has 10 kOhm steps.  The first resistance (19.99 kOhm + 20 Ohm trimmer) is used on all odd steps by placing it in parallel with the second resistance.  The first resistance is never used by itself.  The second resistance is used by itself when the dial is set to 2.  Because of this, the second resistance needs to be calibrated first, then the dial is set to one and the first resistance can be calibrated (which is in parallel with the second).  Once the even steps are all calibrated, the odd steps will be calibrated also.