Nixie DMM repair update :OK made some progress I think, grab a cup of coffee.
Started digging seriously on that comparator board, the heart of the DMM.
Started happening last night. At 2AM after posting here, I failed to go to bed, was too excited to start on that board.
So I pushed up to 5AM and woke up at noon
Sometimes it's cool / convenient to be jobless
So last night/early this morning, did what anyone can do on any board even when you have no clue how it works : replace crappy caps !
So I looked at the schematic and found 3 of them, all 5uF 15V. Took a bit to locate them on the board as they look very much like film caps also present on the board...
So replaced them all, no joy. The old ones were cooked though. Measured them with the cheap Chinese tester :
- 7.7uF ESR 9 ohms
- 16,5uF 12 ohms
- 20uF 26 ohms.
Still, had to be done, I can rule them out now. I looked at the good board from the other meter, and two of these caps had already been replaced. There were also many signs of rework on the solder side. Clearly it's been "serviced" before, hence might be why it actually works...
5AM, so disappointed that it didn't fix it, going to bed unhappy.
Woke up at noon, now 9+ PM. Have been working on that board non stop all that time.
Studying the schematic, reading the manual about how it's supposed to work, and trying to figure out the layout of the board to see where is what.
I am glad to say it was not wasted time. This horrible confusing messy schematic that was overwhelming and depressing me... is now very clear.
I took some time with The Gimp to improve the readability of the schematic, and then annotated it to show its various sections.
So basically this board has 7 sections making use of 24 trannies no less. Here goes :
1) and 2) : two comparator sections, identical, one for each signal input (inputs on the front panel are floating, not referenced to ground)
They use 7 trannies. First a differential pair to implement the comparator per se. A third tranny in the tail as a constant current source, as you do. Then another tranny to amplify the output of that comparator. Then goes into a flip-flop made of two trannies. Both outputs Q and Q\ are made available to the rest of the board ("Bus" looking on the schematic, in the center, running horizontally) for whoever wants to use them. For instance, there are two sections that need them : the gate logic circuitry, and the polarity detection circuitry that drive the corresponding symbol on the left most Nixie tube.
3) A 200kHz crystal oscillator that clocks the Nixie boards that implement the counter/digital display.
4) The ramp generator that feeds one of the inputs of each comparator.
5) A trigger circuit (2 trannies) that gives the go ahead for the ramp to happen. It's like a scope trigger. It can either be in "Auto'" or "Manual" mode to trigger the ramp / measurement repeatedly 2/3 times a second, or be in "Manual" mode where you do only one measurement, when the guy presses the push button on the front panel.
6) The Gate logic : implemented in diode & resistor logic style. It "reads" the 4 outputs from the comparators, and generates the Gate pulse, the "Start" and "Stop" signal if you will, to control for how long the clock will be feeding the Nixie counting chain.
7) The polarity detection / display : as said above, it reads the outputs of the comparators to decide if the signal is negative or positive. It's implemented as a simple flip-flop (2 trannies then, as usual).
Now the mega bonus : the manual also explains, in the ramp generator paragraph, all about these special 1,050ppm tempco resistors on the Vref board we discussed earlier ! It's now 100% clear. Zero mystery anymore. I can now rework/fix/redesign that Ref board in total piece of mind, I know what I am doing now !
So here goes : these special resistors have NOTHING to do with compensating for Zener drifts !
I made another annotated schematic below, just for the ramp generator. Hope you like it, worked hard for it !
There are 7 trannies. First one triggers/controls the ramp generation, using the signal in provenance of the trigger circuitry.
The remaining 6 trannies ? It's one big block, they implement an op amp ! Once you realize this, the schematic becomes extremely simple. It's just your usual ompap in "integrator" configuration to generate a ramp. So the non-inverting input it grounded, and you get the big fancy timing cap between the output and inverting input. Then to set the constant current that will charge the cap and produce a nice linear ramp, you get a DC voltage feeding a resistor tied to the inverting input. One opamp, one timing cap and one resistor, could not be simpler.
So what is that DC voltage setting the current coming from ? Yes, that's our Zener voltage on the Vrerf board. And what is the resistor setting the charging current ?
As you can see on the schematic, it's a chain of 4 resistors in series. Starting from the Zener side, we have :
- On the Vref board, the 1,050 tempco resistor, in series with a 1% precision resistor.
- Then on the comparator board, we have a 500 ohms trimmer. It's the "coarse" adjustment for full scale calibration.
- Then on the front panel we have a 200 ohms trimmer, used to fine tune full scale, so the user can compensate for small tempco related variations if needed.
So of course the idea is that we need to control the slope of that ramp in order to calibrate the full scale "deviation" of the meter.
That means, in return, that we must control the current that charges that cap. That means, ohms law, that we need to control The Vref and the resistor chain that follows.
Vref is handled by the Zener. Then the interesting bit is this : manual says that the +1,050 tempco resistor is meant to compensate for the overall/combined tempco of the other 3 resistors in the chain, so that the overall tempco of the complete chain stays below
10-4 /°C in order for the ramp to be stable/accurate enough, hence the meter, accurate enough.
So... back to that Vref board... that means I can just leave the Zener alone, no need to replace it by a modern ref or even add a constant current source to drive it (though that would not hurt I guess, so might still do it). It means I can simply use a normal resistor to replace that special tempco resistor, and leave with the consequences, i.e have to adjust full scale from the front panel maybe more often than normal. Of course it may well be that in this case that pot may not have enough range to compensate any more... so might have to source another +1000ppm resistor for proper/acceptable operation. But since Tggzzz (or Spec ?) said these are STILL made these days... then would probably want to use that. But that will certainly cost money to import, so since I am broke... short term, I can just use a regular resistor. Now that I understand how it all works, I guess I could do like this to determine the value of that resistor : center both coarse and fine trim pots. Replace resistor with a pot and adjust it to get full scale deviation. Measure the value of that pot. Replace with closed fixed resistor. Won't be spot on of course, so then I would adjust the coarse pot. And then I would use the fine pot on the front panel to compensate for deviations as needed during use (and witness just how much deviation it can exhibit, and if that's within the range of that front panel pot).
OK ! Now that we understand pretty much how this beast works, we can go do some more involved troubleshooting !!!
So to recap the main symptom I am having : the display / counter looks like it's "free running". All digits are showing all digits all the time, and they look like they are never stopping : it's not stopping briefly 2/3 times a second like it should, to let you see the result of the measurement. It's like really, counter is free running, overflowing constantly, confirmed by the "Overload" Neon indicator on the front panel, that's lit solid.
So that means that at least the counter part is working, and that the clock is working. Also means that the Gate logic is not doing its job of starting and stopping the counting. It lets the counter run forever. So why could that be ?
1) Gate logic section is defective
2) Or the comparators are faulty because they feed the Gate logic
3) Or the ramp is faulty because it feeds the two comparators
4) Or the Trigger logic is faulty because it controls the ramp.
So, I power up the meter, counter is free running again. I start scoping the output of some sections because the board provides a few test points nicely designed for you to shove a probe tip inside, very thoughtful of Rochar
So I looked at the output of the 200kHz crystal oscillator.. was fine, running a sine looking signal between ground and the -6V rail, spot on 200kHz. Well we already knew it was working, but still good to see it for yourself and check frequency...
Then after a few minutes, go figure, the meter suddenly stopped misbehaving ! Was reading and operating properly !
What ?! An intermittent problem, that's my luck !!!
So then I probed the ramp generator. Looked fine. Nice clean ramp with appropriate voltage levels (swinging between the positive and negative 6V rails) and duration, about 12ms which is what you need to get full scale deviation with a 200kHz clock on a 2,500count meter...
OK so I thought, maybe a parasitic capacitive effect that takes a few minutes to clear up ? So I unplugged the meter and let it sit for a minute, the time for caps to discharge, if that was indeed the problem. Then power up again... nope, meter is working fine now, damn it...
So maybe a mechanical problem/bad connection somewhere ? I tapped the board everywhere, no luck, can't get it to misbehave, now it WANTS to work !
Not capacitive, not mechanical...so maybe a thermal effect then ?
So I unplugged the meter again, and this time let it sit for an hour or two, the time for it to cool down real well.
Power it back up.... AH !!! display free running again, yes !!!
So thermal related it might be !
So while it was free-running/exhibiting the symptoms, and knowing it might not last, I rushed to scope the board in search for clues.
I checked the clock, thinking maybe when cold it misbehaves and generates super high frequencies (harmonics, instabilities/ringing, I don't know...) whereby maybe it could be that the meter would want to display say just few counts (zero offset count), and these very short gate time would be enough to get the counter to overflow. Just an idea.. so checked the clock.. but it was fine, still at 200kHz, normal.
Then I scoped the ramp. No signal, flat line... fiddle with the scope in case it's having problem triggering on that signal, though it was an easy signal really. Got kinda convinced that maybe indeed the ramp was not there... then played with the trigger circuit : operated the auto/manual switch, and pressed the manual trigger push button, I thiiiink.... and that's when the meter started to work properly !
So... was the ramp generator faulty, or was it good but simply not being asked to start, because the trigger section failed to trigger it, and me playing with the trigger controls shook things up and got the trigger section working again ?
Or was it just a coincidence, a correlation and no more ? Hence a red herring...
But at least that gives me a lead now : let's see. Knowing how that board works, could a non working ramp, at power up, cause the counter to free run ? In short, an "illegal" state at power up, causing this "bug".
At first I would say no, because having no ramp at power up would be the equivalent of powering up the meter in "Manual" mode, I think !
So... it's not yet obvious what, deep down, is the root cause of the problem.
However it's clear that NOT having ramps being produced in Auto mode from cold, is NOT normal.
So I will first investigate this, and go from there.
So as I said, first looking into the trigger circuit to see if it is it not triggering the ramp, or if it is, if it's rather the ramp circuitry that's a bit deaf when cold...
Anyway, I have a plan now, so let's follow it !
First, I need to go get a few spray cans of CO2 so I can cool down the board locally, to see if I can get to misbehave again once it's hot and working.
This way I could confirm or deny that it's a thermal related issue, and what part of the board it's coming from.
That's all for now folks, you can now resume your normal radar detection activities !
-group-therapy-thread/?action=dlattach;attach=1458601;image)
Home
Help
Search
Login
Register
Topic: Test Equipment Anonymous (TEA) group therapy thread (Read 18878363 times)