DANA 5500 (part II) chasing a gremlin in mV function

[valley001]

Hello all,

Thanks once again with the help repairing the ohms section of this meter.  It is functioning quite well.

I have now moved on to working out a problem with the mV function.  When the mV function is selected, in the 1000mV range it will hold at 0 with the input leads shorted.  In the 100mV range the counter will start at 0 and then begin counting upward and will not stop.  If I take a reading in the 100mV range it will display the correct reading and then begin counting upward continually.  In the 1000mV range it will take an accurate reading and remain stable.

In the DC and AC mode the display will hold at 0 with the leads shorted in all ranges, so I know the problem is limited to the mV function.

The manual for the 5500 is very similar to the 5524 which seems to be available on line.

The sections on "count up" and "count down" are beyond the scope of my current understanding. I believe 4.13-4.19 describe the relevant circuits and functions. 

My current plan is to test diodes and transistors on these boards (range control) hoping to find something.  A more refined approach would be of course be welcome. 

[valley001]

Just bumping this to see if anyone might have an idea.

I do have a parts unit and swapping boards does not improve the situation with the 100x mV range.  This weekend I have some time so I will be testing diodes and transistors in the range control board and possibly other boards.

[valley001]

Update, well it seems like an issue with a relay.  Any advice cleaning these types of relays?  I have Deoxit D5, would this be advisable?

[floobydust]

Update, well it seems like an issue with a relay.  Any advice cleaning these types of relays?  I have Deoxit D5, would this be advisable?

A chemical cleaner doesn't work so great on heavy oxide buildups.

I'd take the relay apart (cover off) and put a piece of paper in the contacts and squeeze the armature down to close the contacts. Then I drag the piece of paper through the contacts, like it's a sheet of soft sandpaper. You can see the black oxide come off, on the paper. Then, applying a dab of DeoxIT will add a film to prevent oxidization.
This is a trick an old telephone tech taught me, from the days of mechanical rotary relays in phone exchanges.
Careful you have the perfect tool to pry off the relay cover. The plastic gets hard with age and wants to crack. A screwdriver or blade with a thin edge works for me, and not over prying.
Make sure the pins (socket) are clean too, I use a chunk of pencil eraser or maybe a Q-tip with DeoxIT to rub off the oxide on the pins.

As far as the A/D (DAC counter) not stopping on mV, this is a difficult multimeter to understand.
At first I thought the Null Detector trimpot (R24 on 5524) might need to be adjusted. I think the Null Detector tells the counter to stop (latch/hold) the reading which is not happening. But this does not explain why other ranges are working.
Leads me to... I think the mV amplifier (Reference and Buffer Amp) has a problem or large offset on mV range, causing the Null Detector Output to malfunction or stay stuck. This whole multimeter is full of balance trimpots, so I would check it on Buffer amp with the ranges that are causing trouble.


This is quite the piece of art for 1967: 19" rack-mount sealed enclosure, 100,000 count with Nixies, fan cooling, runs to -55°C with the internal 400W heater, explosion-proof, 0.005% DCV accuracy...  It must have cost a fortune in its day. It's serious MIL-spec. 

[valley001]

Well..mine is the humble civilian version, no internal heating etc.      

The interesting thing is the counting in 100x MV range has somewhat resolved after replacing the associated relay on the control and attenuator assembly.  If I rock the relay in its seating the issue becomes intermittent...leading me to believe the entire problem is a relay/contact issue rather than an issue in pots/solid state components etc. 

Another issue (I figured was endemic to this meter) was that various ranges on DC volts would also count unless the leads were shorted.  This has also somewhat resolved after changing the relay. 

I will follow your advice on cleaning and see how things react. 

Update, well it seems like an issue with a relay.  Any advice cleaning these types of relays?  I have Deoxit D5, would this be advisable?

A chemical cleaner doesn't work so great on heavy oxide buildups.

I'd take the relay apart (cover off) and put a piece of paper in the contacts and squeeze the armature down to close the contacts. Then I drag the piece of paper through the contacts, like it's a sheet of soft sandpaper. You can see the black oxide come off, on the paper. Then, applying a dab of DeoxIT will add a film to prevent oxidization.
This is a trick an old telephone tech taught me, from the days of mechanical rotary relays in phone exchanges.
Careful you have the perfect tool to pry off the relay cover. The plastic gets hard with age and wants to crack. A screwdriver or blade with a thin edge works for me, and not over prying.
Make sure the pins (socket) are clean too, I use a chunk of pencil eraser or maybe a Q-tip with DeoxIT to rub off the oxide on the pins.

As far as the A/D (DAC counter) not stopping on mV, this is a difficult multimeter to understand.
At first I thought the Null Detector trimpot (R24 on 5524) might need to be adjusted. I think the Null Detector tells the counter to stop (latch/hold) the reading which is not happening. But this does not explain why other ranges are working.
Leads me to... I think the mV amplifier (Reference and Buffer Amp) has a problem or large offset on mV range, causing the Null Detector Output to malfunction or stay stuck. This whole multimeter is full of balance trimpots, so I would check it on Buffer amp.


This is quite the piece of art for 1967: 19" rack-mount sealed enclosure, 100,000 count with Nixies, fan cooling, runs to -55°C with the internal 400W heater, explosion-proof, 0.005% DCV accuracy... It must have cost a fortune in its day. It's serious MIL-spec. 

[valley001]

Well I thought I had this sorted but this morning discovered the problem has returned.  I cleaned some relay contacts with no effect.  I then discovered if I rock K7 out until it disengages the 100x range mV will settle to zero (K7 is normally engaged in the mV function setting).  I am not sure why this is happening but at least it seems like I am getting closer to discovering the problem.   

The really confusing thing to me is with K7 disengaged it was taking mV readings on both 100x and 1000x ranges and they seemed fairly accurate. 

I have attached the schematic of the reverent section in hopes someone might see something that I should check or that could cause these problems.

[magic]

In the 100MV range the counter will start at 0 and then begin counting upward and will not stop.

I doubt you actually waited until it hit overrange? Or is it really rising so fast?

The really confusing thing to me is with K7 disengaged it was taking MV readings on both 100x and 1000x ranges and they seemed fairly accurate.

Weren't they wrong by one or two decimal places? Your schematics suggest that K7 enables 10x or 100x gain in that ISO amplifier which appears to be the one buffering the input signal to the ADC, is that correct?

[valley001]

I did not notice this before but with the device set to 1000x range in mV connected to a batt/divider setup outputting 78.5mV I get a reading of 0078.52 mV.  If I switch to 100x scale it will display 007.852 mV and start to count upward immediately (4x/sec).

I missed this before, the reading not shifting with the decimal in 100x, and yes with K7 disengaged the readings are off by (I think) 2-3 decimals. 

I have not waited to see if it counts to over range, but I have left it on for a while and it does not appear to ever stabilize/slow down/stop counting in the 100x range. 

In the 100MV range the counter will start at 0 and then begin counting upward and will not stop.

I doubt you actually waited until it hit overrange? Or is it really rising so fast?

The really confusing thing to me is with K7 disengaged it was taking MV readings on both 100x and 1000x ranges and they seemed fairly accurate.

Weren't they wrong by one or two decimal places? Your schematics suggest that K7 enables 10x or 100x gain in that ISO amplifier which appears to be the one buffering the input signal to the ADC, is that correct?

[magic]

Nitpicking, but 1MV is 1 megavolt

I did not notice this before but with the device set to 1000x range in MV connected to a batt/divider setup outputting 78.5MV I get a reading of 0078.52 MV.  If I switch to 100x scale it will display 007.852 MV and start to count upward immediately (4x/sec).

Check that "attenuation assy 40231" on the right. All voltages relative to negative input:
E9 should be 0.000000V
E10/E13 should be 10x the input in 1V range and 100x the input in 100mV range.
E11 should be 1/10th of E10
E12 should be 1/100th of E10.

Check continuity from those points to "MECCA" (WTF - it's just signal ground), "ISO OUTPUT" and "- ISO INPUT" according to the schematic. The latter goes through K5 so it will depend on whether you are in 1V or 100mV.

[valley001]

I took some measurements per your suggestion.  I assumed these should be taken with an input voltage applied, so I applied my 78.5mV source, below are the results using my handheld DMM:

In 1000mV range:

E9- 78.5mV
E10/13- .788mV
E11- 78.7mV
E12- 7.8mV

In 100mv range:

E9- 78.5mV and stays stable
E10/13- starts at .78mV and slowly rises
E11- starts at 78.6mV and slowly rises
E12- starts at 7.8mV and slowly rises

I have not had a chance to check continuity on the other points but I will later. 

[magic]

Bitten by autoranging, eh? Because sure as hell that 0.78mV is actually 0.78V.

So far, so good.
78x.xx mV going into the ADC ought to produce a readout of 0078xx and that's what you see. Nothing wrong with the ADC at least.

Definitely do the continuity checks from E11,E12 through K5,K7 to -ISO IN. E11 should be connected in 1000mV, E12 in 100mV. Something funky is going on in the second case.

And whatever you measured as E9 isn't E9. Just check continuity from the junction of those 1.111k/100.03 resistors to "mecca" and the negative input terminal, that will do for peace of mind. I'm sure the problem is elsewhere.

[valley001]

Oops 

You are correct. I am not taking my time with some of this. I mistook the junction of C12 and R27/33 as E9.  Revisions below, hopefully without more mistakes.

Ill start looking into the continuity checks.

In 1000mV range:

E9- 0 Volts
E10/13- .788 V
E11- 78.7mV
E12- 7.8mV

In 100mv range:

E9- 0 volts
E10/13- starts at .78 V and slowly rises
E11- starts at 78.6mV and slowly rises
E12- starts at 7.8mV and slowly rises

[magic]

Okay, I know what's going on.

Power it up, switch it to 100mV and only then apply some voltage to the inputs. You will see it completely oblivious to your new voltage and still counting up from the old value
Looks like you only ever switched it to 100mV from 1000mV with the voltage already applied.

In case it still isn't obvious: K5 doesn't contact at pin 10.

edit
Err, wait, sorry, you will probably see it immediately hit overrange.

[valley001]

Interesting. You are correct I only ever switched to 100mV while there was a test voltage at the inputs. 

I removed the 78.5mV source. 

Powered unit on.
Switch to 100mV scale
Applied 78.5mV source.
Display shows 023.150ish and continually count up. 

Unfortunately it is not obvious to me.  I am new to this but learning as much as I can.  So my next step is to check continuity between K5 pin 10 and E12 correct? 

[magic]

Yes. That's feedback connection to the isolation amplifier, without it the amp is going to react in hardly predictable ways to changes in the input and slowly drift up or down if the input is constant. Exactly what you see.
It also may be possible that the contact isn't completely broken but "only" has too high resistance to fully work.
At any rate, I think the problem is connection between R5 and E12.

R5 itself appears to be working because without it the 100V range wouldn't work.
BTW, check the 100V range. Maybe it also is off by a decimal place or two

[valley001]

K5 pin 10 to E12 is approx 920K.

[magic]

What? No!
I thought it's K5 itself which is broken and has no contact inside at pin 10.
At any rate, K5 to E12 is a straight wire. So... if it measures 920k... that should be obvious, damnit 

[valley001]

Right.  I actually have several spare relays and the problem persists with all of them, so I dont think its the relay.  I will disassemble and check that connection.

What? No!
I thought it's K5 itself which is broken and has no contact inside at pin 10.
At any rate, K5 to E12 is a straight wire. So... if it measures 920k... that should be obvious, damnit 

[valley001]

   I had pin 10 wrong, the schematic refers to the relay itself not the socket so I had them reversed.  I have continuity between K5 pin 10 and E12, less than 1 ohm. 

With the relay out I inspected the socket more closely and noticed pin 10 spread a bit more than others in the socket. I bent pin 10 on the relay inward and now mV is functioning in the 100mV range. 

Thank you for helping this noob sort this problem out.