Hi all, and thanks for the replies.
I am bit late in replying, because well, I spent the entire week-end working on the Metrix ! Friday night I actually did not sleep at all, I was up all night and morning at the bench !
So, I tested it as well and as thoroughly as I could, given my limited/modest equipment. But with the help of a little creativity, I managed to achieve more than I first thought was doable... My other/working multimeter, the only one I have, is a trusty 20 year old Fluke 11, which I bought new back in the day. Horrendous 0.9% accuracy (but much better in practice/reality, probably because despite it's advertised crappy accuracy, it DOES have a hybrid ceramic network !) and cannot measure currents. But at least it can go down to a one mV resolution, and measures ohms and capacitance. Then using a breadboard and lab power supply, I was able to wire up some simple test circuits to check the mV range and mA/uA range. I used the current limiting adjustment of the power supply, to test the main 10A current range. It doesn't have a digital display, only galvanometer, so not very accurate, but still useful as we will see further down. main DV Volts I cold test up to 60Volts by putting the two outputs of the power supply in series.
So, what works in the end ? Most of it.
- Diode test
- Continuity test, though scratchy but well if it's not latched what can you do.
- Frequency measurement. Quite sensitive : even with the probes in free air, it picked up the mains zand displayed a rock solid "50.00Hz".
- Ohms
- Vdc
- mA/uA : works.
- mVdc works
- Capacitance
So, what are the remaining problems ?
1) Capacitance range : when you first select it, it shows, as expected, a steady 00.00 nF, with the occasionnal "00.01" LSD poping up erratically, I guess this is OK. What is not OK in my view however, is that if I measure a capacitor (or just short the leads, making it believe it's a massive/out of range capacitor), AFTER I have measured it, and the leads in free air then, not connected to anything, then from this point on, the display is NOT a solid 00.00nF as it was before and is it should be. Instead, the last 2 digits (00.XX nF) are floating and dancing quite a bit, like if there was some capacitive effect at play inside the meter, following the measurement, that would not go away.
2) the main 10A current range : I took a few data points from 0?7A up to about 2.5A, which is what I can achieve with the current limiting know of my power supply. Data points show that the met basically reads about 40% higher than it should ! Yes, quite a lot indeed...
For reference, the data points where : 1,4A read instead of 1A, then 2.1A instead of 1.5, then 2.8A instead of 2, then 3.5A instead 2.5 and lastly 3.9A for about 2.7A on the galvanometer, which was the maximum the power supply deliver.
So it seems like a (huge) calibration error ? Seems crazy to be THAT off, even without calibration ? But well... what else. The Cal as usual help in an external serial EEPROM, and these are usually rated for 10 years right ? Meter is around 20 years old... so why not. Thanks to "Macboy" in his comments further up, it seems the calibration software is luckily freely available, and that the serial cable to connect to the IR ports on the meter's main board, is not that big a deal to construct, and that even the protocol is documented ! What more could I dream for ? So I guess I will have to take this path.
3) The mV range and the dreaded alarm... spent quite some time fiddling with the thing, trying to understand what was going on, and chasing a few red herrings in the process ! The alarm does not trigger when you short the probes, or if you actually measure a voltage. But it will trigger invariable if I touch a metallic part with one of the probes (the other probe remaining in free air). Work every time if I touched the ground/earth/chassis connection of the power supply, or the ground/outside part of the BNC sockets on my scopes or function gen. Works also if I touch the tip of my soldering iron (be it plugged in or not, makes little difference). At some point, I didn't even have to touch anything ! I mean... I noticed that once, the simple fact of APPROACHING one of the test leads to the tip of the iron, was enough to make the Metrix go nuts ! I would see the mV readings on the display, go up and up as I was progressively approaching the probe to the iron. The Metrix saw 500mV, and hence the alarm would trigger... and this happened when the probe was about 4 Inches / 10cm away from the iron ! Crazy. So yes, as I said, the alarm is not the "high voltage/danger" as I though initially, but instead it's the overload/out of range alarm, indicating that the meter sees more than the allowed 500mV on its inputs.
I noticed that that if I select the "AC + DC" feature on the mV range, it becomes even more sensitive ! in mV dc, when disconnected, the meter would read something reasonable, a few millvots, but if you set to AC+ DC, then it reads HUNDREDS of mV right from eh start ! So it's already very near the 500mV trigger point of the alarm, meaning it doesn't take much at atl to trigger it.
So there is something wrong going on here for sure, but what ?..... Why does it read hundreds of millivotts when unconnected ? The fluke, which can go display down to one mV, does not do this ! It displays 0.000 Volts when not connected !
So, the mV is clearly way too sensitive. I should note that I desoldered the large RF shield at the back of the main board, in order to get access to the components et see where the traces where going etc... could this have such a huge impact ?
If this is not it, then I guess there must be some voltage inside the meter that's somehow "leaking" into the mV range input path ?
Inspection of the meter and trouble-shooting :
I disassembled and re-assembled the meter several time.. and eventually one of the pins of the LCD header broke off... but I was able to temporarily fix it by soldering a thin/wire wrap wire between the two boards. Works just fine. I will order a replacement header shortly. Found a gold plater on Farnell' site, so even better than the original.
I checked the PTC, reads just fine a 1100 ohms, which is spot on it's rated value, according to the data sheet I found for it online.
The 4 MOVs look fine as weel : like new on the outside, and all 4 are very high impedance. don't know how much because the Fluke couldn't even measure it, so I guess that means at least 20Mohms, which I guess is good enough. The readings on the Vdc range where all fine, so I guess it must be OK ?
I found that the LM385 1.25V voltage reference (small TO92 package) was bent in order to be able to close the meter enclosure, and that as a result, pins 2 and 3 looks like they were touching ! So I re-arranged them a bit, but that didn't change anything to my test results.
I also found a group of 4 few dipoles on the back of the board, in the fuse area. I don't know what kind of device it is... but they work in pairs : one pair is connected to one of the fuses, the other two, to the other fuse.. Not knowing the nature of the component, I ventured testing them with the ohm meter. They read in the order of the Mega ohm (both ways). Ranging from 0.7M up to 3M. I guess they this high impedanc is normal, because I also found teh exact same component, lonely this time, on the top side of the board. It is stuffed between 2 jacks, and is electrically connected straight across the 9V battery terminals ! So indeed it batter by high impedance... now what are there purpose and technical nature.. no idea. They just dont look like resistors, and although they have color coded rings, they carry only 2 color rings, a bit too few to be a resistor...
Two of them looked like they had not so good solder joints, so I reflowed them all... but didn't change or improve a thing. Oh well, at least it couldn't hurt I guess !
I also found a little bit of corrosion on the PCB, at the base of the negative battery terminal, due to the old battery leaking I guess. So I desoldered the terminal so I could give it a good clean, then soldered it back. Again, this didn't change a thing to my problems...
Lastly, when examining the main PCB to see how it compared to the pics posted by Robomeds, I noticed that it's all similar except that the 4 MOVs are laid out in a row instead of a 2x2 square, and that the PTC and the yellow 200ohms resistor (R1 and R2 then), were relocated further down, in between the Volts and COM jacks. Not that it matters of course. Anyway, nex tto the MOVs, I notice R25, a 10M 1% resistor. I supposed it's purpose was to achieve the typical 10Mohm input impedance that's pretty much standard for digital meters. So I could not help but measure it, in circuit first. At first, as soon as I applied the probes to its leads, it would read 10M... great, how unexpected... but a few seconds later, the value started DROPPING ! Slowly but surely, after a couple minutes measuring the thing, it would eventually settle at around 8,5M, so much for the 1 % tolerance ! What an odd behavior for simple resistor, I thought. Maybe some capacitive effect in the surrounding circuitry then. So, I lifted one of the resistors terminals, so I could get a valid measurement. guess what ? It behaved exactly the same ! So this resistor is clearly acting very funny ! Actually I am not so fussed about the actual value of it, as I don't think it would influence the slightest bit the validity of the measurements, after all its only purpose is to provide a high impedance... but that's not required in my test procedures, since I feed the meter with low impedance source voltages, my lab power supply ! No, what really worries, puzzles etn troubles me, is the fact that is unstable ! This means there is something definitely wrong in this resistor, and since it is by essence, right in the signal path, god knows what effects that might cause... Sadly I didn't have any 10M or very high value resistor in stock to replace it with for test purposes.
So... where ma I now ?
My best guess :
- for the 10A range reading too high, try and do what it takes to recalibrate the thing and take it from there.
- For the mV range being overly sensitive when the probes are disconnected, two things : 1) the faulty 10M input resistor my be partially "open", on its way out, which means the meter is kinda measuring straight from the DMM5 chip input, which must be extremely high of course, explaining why it is so sensitive. That, and/or the absence of the RF shield.
- About the capacitance range bug, it may as well be related to the above problem, we shall see.
So, to conclude : I will get a brand new 10M 1% resistor, then solder the shield back on, then replace the LCD header on the main board, then build a cable a recalibrate the 10A range. Ah, and contact Chauvin-Arnoux/ Metrix to see if they would be kind enough to send me a service manual for this meter, or at least the schematics, given that the meter is 20 year old and not marketed any more... so hoping they are not going to make too big a deal about sharing data on this particular product !
Will report back as I progress...