[SOLVED] Voltmeter over-range on lowest voltages.

[d123]

Hi.

Having a trouble-shooting problem with a home-made voltmeter which reads the output voltage of a power supply.

The power supply transformer has two secondary windings, so there are two DC supplies that cover +5/6/7/9/12/15V outputs.

Both supplies, and the soldered voltmeters and ammeters worked fine when I was bread-boarding and testing the last stage which was the specific resistor values to get the output voltages approximately right. Since adding the resistors and rotary switches, one of the voltmeters can measure 12 and 15V fine, but now says "1" (over-range) for 5/6/7/9V outputs - it didn't do that a few days ago, it functioned correctly.

I checked the banana plugs with a DMM, and the output voltages are correct, it's one of the home-made voltmeters that has the anomalous behaviour.

Not sure why it would matter, it didn't during the testing stage I mentioned: supply/voltmeter/ammeter #1 run off the same line as the fan, supply/voltmeter/ammeter #2 have no fan on their line, and #2 voltmeter works fine.

As far as I can see there are no wiring mistakes (both are the same in every respect, unless I'm missing something).

Has anyone had a similar experience? I could understand the higher voltages causing an over-range if something were wrong, but not those below 9V, doesn't make sense to me.

Attached is a schematic with the power supply, and a schematic of each 'meter.

Before anyone rightfully rubs their hands to point out a lot of things unrelated to the question , it's not a very good power supply, I'm trying to disown it in a crowd, and I wouldn't make one this way again if I had to.

Does anyone have any guesstimate ideas why would one have over-range for lower voltages, but the other be fine, when in principle they are identical, and have been through the same process?

I can upload a photo of the back of the front panel where all the relevant wires mingle, if that would help.

View attachment Power supply - Schematic.pdf

Thanks.

 

[KlausST]

Hi,

i assume it is an issue with the common mode input range of the voltmeters. They use ICL7107.

1) Datasheet says: • COMMON MODE INPUT VOLTAGE: (V- + 1V) < VlN < (V+ - 0.5V)

2) Usually In_Lo is connected to COM. COM voltage range is given in datasheet: 2.4...3.2V.

I don´t know if it helps, and I don´t know if it hurts the voltmeter:
* There is an 1M resistor at the voltmeter input: maybe use a 500k resistor at each input. (or just try to connect the 1M to the other input)
* or you may lift both V_in+ and V_in- to valid (common mode) voltage levels.

Hope this helps

Klaus

 

[d123]

Hi, that does help a lot to bear in mind try to understand this concept properly as I'm still not sure I fully understand it, no matter how many times I've read that datasheet..., and it's a good place to start to work on the problem from, thanks.

What I can't understand is why one voltmeter does this but the other one doesn't.

If you don't mind, and if you could take a look at this pdf, I'd appreciate it, it's the readings on outputs and inputs, and a couple of other measurements. I imagine that even if one has an oscillator frequency of 24kHz (the voltmeter that works correctly), and the other which does the over-range has one of 19kHz this probably has nothing to do with the problem (?).

View attachment PSU voltmeter readings.pdf

Oh, right... I'm not sure which way round it was, so ignore this as being factual, I have to look at the underside of each board/circuit to refresh my memory, but I do remember I tried that when bread-boarding (what seems like a long time ago), for this circuit when I connected Analog Common to Vin- it doubled the reading, so I left Vin- connected to 0V but not to Analog common on the voltmeter. But the ammeter did need that connection to give a correct reading (which puzzled me at the time, as they seem to be the same circuit - both go to the main circuit common/0V, except for one input coming from the current shunt monitor).

 

[KlausST]

Hi,

could you pleas measure the voltage of
* COM
* IN Lo
* IN Hi
referenced to V- (all directely at the 7107 pins)?

This gives the most reliable information.

Klaus

 

[d123]

Hi,

I did as you requested, the pdf below has the measurements taken on the pins you said (referenced to V-). I didn't know about checking what was happening on those pins, that's interesting to learn, yesterday I only measured from In_hi to In_lo on the device.

View attachment PSU voltmeter readings 2.pdf


I checked yesterday, and on the ammeter (TC) 7107 In_lo (/Vin-) is connected to Com (/Analog Common); it isn't on the voltmeter 7107; as I said before, when bread-boarding the voltmeter, connecting Com to In_lo doubled the output reading and sent the device to over-range when it shouldn't have done, that's why I didn't make that wired connection on the voltmeter.

I'm surprised at the change in correct behaviour on one voltmeter but not on the other, as both have worked correctly as a breadboard version and as the final soldered versions (tested on a manufactured power supply's voltages) and also correctly with the soldered power supply with the voltage regulators, etc., and with breadboard version of resistors for output voltage included as well, until I soldered the resistors and rotary switches, and finished the power supply and connected everything together, when #1 started doing this.

 

[d123]

Hi again, don't worry, thanks for your help, for example it has helped me to focus on things I've read several times but the information never sank in as to what it actually meant.
I tried to do the sums with the last readings, and if I haven't got them wrong (which I might have), it looks like the Com voltage range is within the acceptable parameters for the TC7107 (2.7 - 3.35V), I think, so when I have a little free time to actually concentrate on the circuit I'll have another go. It had nothing to do with the fan being on the same supply line (I was clutching at straws there!).
The best I can think of for now is changing the oscillator capacitor by trial and error based on the value back-calculating from the frequency of each voltmeter to get the same frequency as the one that works okay.
Bit of a mystery.

Once again thanks.