Overvolted multimeter

[Wojta]

Hey all,
Ive probably ovevolted my multimeter recently, but Im not willing to throw it to the trash can. At least not yet. The thing is, it looks like it still works, just in a weird manner. When it comes to meassuring resistance, the meter shows pretty constant value of 16.7 on any of the ranges. The value doesnt change when i change the ranges (on ohm range it shows 16.7 an on Kohm 16.7, same on the Mohm). When i apply resistance greater than 16.7 (i think Ive tried it at ohm range so 16.7 ohms) the value doesnt change but when i apply resistance lower than that, the value lowers. When it comes to meassuring voltages, the meter works, but shows only 1/30 of the real voltage. For example on 4.5V battery it shows 0.15V. When switched to diode test, it shows constant number of 0.08.
What could be possibly wrong? What parts could be dead?

 If you have any idea please let me know.

Ive attached a image of the board. There is just a handfull of components. The 2 recognizable ICs are Op amp at the bottom (probably for shunt voltage amplyfication) and NOR or OR gate at the top left.

Image: (was too big for attachments) https://1drv.ms/i/s!AnCjEyxUhqCPifxVH9Hg4dNlgCyesG

Thanks for any suggestions 

[Rerouter]

Use imgur or something similar, i dont want to have to signup for onedrive to see it

As for possible issues, You have the input clamping diodes, the measuring reference resistor, the range analog switch,
|
If you can figure out the input to the main blob from the leads in, measure what voltage is there when running to see if its a fault at the blob, or in the repairable components.

[retiredcaps]

Just resize your pictures and attach them onto the eevblog server.  Like the other person, I'm not willing to create an account just to view your pictures.

A clear focused picture of 1600x1200 under 2MB will be sufficient for us to help you.

[Wojta]

Sorry for the onedrive.
Here is the picture.

[jeroen79]

One thing you could check is if the input arrives intact at the chip.
Apply a voltage to the input and look for it on the chip's pads.
It would be reduced depending on the range selected.
When you have found the input pad apply 100mV to it and check the readout.
This will tell you if it is the chip or the circuit before it that is at fault.

I'm guessing this is some icl7106 or derivative that takes up to 200mV on the input.

[capt bullshot]

I can spot some diodes near the device labeled PTC1. These may be input protection diodes. Check 'em for shorts. Check also the diodes on the lower right of the board.

[Wojta]

Update:

Ive borrowed the exact same multimeter from a friend of mine for checking the components. The diode all seem good. Nothing broken. The PTCs also are the same as on th working one. But there is a problem arround (or in) the chip. First four pins of the bottom row are either 100mV or 0V on the working one. On mine, the first one is something like 6.3V, second one is pulsing, something like cap being charged and discharged repetively. The other two are only connected together with a cap and they are both also 6.3V. Ive also noticed that there is too much voltage on the only transistor on the board. It is also connected through some resistors to the IC. Im gonna try to dig in that part of the board more. Hopefully I can find out whats going on.

[twice11]

I recently troubleshooted a similar multimeter that was connected to live in ohms range. That meter had one transistor, which is used as clamp on the A/D converter input. This transistor was shorted out, and killed the ohms range (constantly showing some low value). Just removing that transistor revived the ohms range, and installing a new transistor with low leakage current fixed that problem. There was some collateral damage, though. The buzzer driver chip and the capacitance measurement chip (both custom hybrids) were blown, too. The buzzer driver can be replaced by a transistor when the passive buzzer is replaced by one with an integrated oscillator, but I guess the capacitance is beyond economical repair.

For your meter: If the transistor only has two connected pins, it might lightly be the culprit, too. Just unsoldering should fix the problem, replacement of it will restore protection. (Without that transistor, overvolting would most likely have killed the chip-on-board).

[Wojta]

Twice11: Thats exactly the case. Only two pins of it are connected. I was wondering what is it for... Thanks for your advice. Ill try it as soon as I can. Could you please recomend a transistor I could use? Olny transistors I have on hand are BC807 and 817 but Im not sure of their leakage current...

I hope this will solve it. 

[twice11]

Twice11: Thats exactly the case. Only two pins of it are connected. I was wondering what is it for... Thanks for your advice. Ill try it as soon as I can. Could you please recomend a transistor I could use? Olny transistors I have on hand are BC807 and 817 but Im not sure of their leakage current...

I hope this will solve it. 

I don't know what the original transistor is. There should be a two- or three-letter code printed on it which you can look up in SMD code databases. This will tell you about the kind (NPN or PNP, it's bipolar for sure) and the pinout. I guess they are using the base-emitter diode which works as zener and thus can clamp voltages in both polarities and the traces I make out in your photo match B-E for the BC816 layout. Diodes Inc specifies the leakage current of that diode in their BC816 as "Emitter-Base Cutoff Current" which is max 100nA at 25°C. They specify the same value for the BC846-BC848, so the exact model likely doesn't matter. I don't expect them to have chosen a less jelly-bean transistor if its just the garden variety 3.5 digit multimeter.

You might also replace the transistor by what it really resembles: a low-leakage protective Zener (for example the ESD9R3). They specify 1nA leakage, but the comparison is unfair, as this it at 1V reverse voltage thile the BC816 specification ist at 4V reverse, a further data point for the ESD9R is 1000nA (1mA) at 4.8V, so the device ist not advertise to perform significantly better than the BC816. On the other hand, in normal operation the protective diode in your meter is at +/-200mV, so the 4V leakage is irrelevant after all. As a circuit designer I would feel more confident with the ESD9R3, as it specifies the low-voltage leakage current at all, but as a user I assume jelly-bean transistors to have a similar drop of leakage like that diode, so anything with the right polarity is fine. (and you can just reverse-solder a transistor of the wrong polarity).

[Wojta]

Im actually not sure whats going on with the transistor. It seems to be an NPN one. I tried desoldering it and that didnt help. When I desoldered it, I realized that there was a hidden trace under it. That means that the transistors base and collector are connected together and emitter goes to the blob. Im now thinking more about the caps surrounding the blob. If I meassure a voltage across them, its not the same as on the working meter. Also the charging of the big cap seems to be messed up. On the broken one it charges to 60mV and starts again, on the working one the cap voltag slowly rises to 7V and than is slowly dropped. There is 224J100 written on the cap. I suppose thats 224J and 100V, am I right? How do I find replacement for it, Ive never seen cap markings in Joules... Neither my local shops seem to have caps in joules.

Thanks

Edit: Ive attached a picture of the caps in closeup. I probably wont help but at least you will know what Im talking about.
BTW: Dont even look at the solder joints  I had no flux on hand...

[twice11]

I tried desoldering it and that didnt help. When I desoldered it, I realized that there was a hidden trace under it. That means that the transistors base and collector are connected together and emitter goes to the blob.

OK, the wiring means they are in fact using the B-E junction as diode, but connected the collector too to allow a greater current to flow. If desoldering didn't help, there must be another fault.

Im now thinking more about the caps surrounding the blob. If I meassure a voltage across them, its not the same as on the working meter.

Comparing voltages is a good idea to find out possible locations of a fault, but oftentimes wrong voltages on a cap are not caused by the cap itself, but by the surrounding elements. The interesting parameters of a cap are its capacitance (which might be difficult to measure in-circuit) and, especially for electrolytics, its ESR (which usually measures fine even in-circuit).

Also the charging of the big cap seems to be messed up. On the broken one it charges to 60mV and starts again, on the working one the cap voltag slowly rises to 7V and than is slowly dropped.

Around 7V is the typical operating voltage of ICL7106 based meter designs. The reference section in that chip is often co-used as linear regulator, which will result in 7V output voltage. You likely find similarities around the blob to schematics of the garden variety 7106 meters, but it might also be an entirely different design. If it is a bypass cap for the operating voltage, everything is that meter is going haywire if it doesn't come up above 6 volts. Check the cap for shorts (might indeed the fault of the cap), but if it is not shorted out, check the resistors around it for being way too high. Measuring resistors in-circuit usually results in too low garbage values, but nearly never results in too high values. If some resistor that *should* charge the cap is burnt, it will be orders of magnitude too high. This can oftentimes be detected in-circuit.

There is 224J100 written on the cap. I suppose thats 224J and 100V, am I right? How do I find replacement for it, Ive never seen cap markings in Joules... Neither my local shops seem to have caps in joules.

That's not how to read that marking. 224 is the capacitance, in picofarads, encoded like in SMD resistors, so the third digit is the number of zeroes. This makes it a 220000pF cap (aka 220nF or 0.22µF). J encodes the tolerance and indicates +/- 5%, and you are correct about the 100 being the maximum voltage of that cap.

[Wojta]

Sooo.
How is the transistor wired up? The pin that is not connected to collector shuold be wired up to ADC? Or its the other way? Okay, Ive discovered that the two common pins of the transistor are connected straight to "input" ground. Not the multimeters battery ground.
Ive meassured resistance between the meters "input" ground (I mean the common terminal where your test leads come from) and pin of the transistor (of the pin I suppose is connected to ADC) and there is a big difference in resistance between these two. On the working one it is 2,22Mohm or 40Mohms vs 215kOhm or 0,7Mohm on the bad one. There are two values for each one because they depend on how I meassure. COM to meassured point or +lead to the meassured point.
So Im guessing some resistance that stood in the way was blown by the voltage I put in it?

thanks for the cap explanation twice

[twice11]

Okay, Ive discovered that the two common pins of the transistor are connected straight to "input" ground. Not the multimeters battery ground.

Again, this is typical for 7106 based meters. Check http://www.hobby-hour.com/electronics/icl7106.php for a couple of schematics using that chip.

Ive meassured resistance between the meters "input" ground (I mean the common terminal where your test leads come from) and pin of the transistor (of the pin I suppose is connected to ADC) and there is a big difference in resistance between these two. On the working one it is 2,22Mohm or 40Mohms vs 215kOhm or 0,7Mohm on the bad one. There are two values for each one because they depend on how I meassure. COM to meassured point or +lead to the meassured point.
So Im guessing some resistance that stood in the way was blown by the voltage I put in it?

Spot on! You could cut the trace going to the +IN pin (the pin you suppose is connected to the ADC) and compare the "chip side" and the "meter side" of that trace. If you find a difference in the chip, it's gone. Forget about that meter. If the difference is caused by something outside of the chip, chances of repair are not that bad. You "just" have to find the faulty component. You can reconnect the cut trace using a bit of solder afterwards.

[Wojta]

Spot on! You could cut the trace going to the +IN pin (the pin you suppose is connected to the ADC) and compare the "chip side" and the "meter side" of that trace. If you find a difference in the chip, it's gone. Forget about that meter. If the difference is caused by something outside of the chip, chances of repair are not that bad. You "just" have to find the faulty component. You can reconnect the cut trace using a bit of solder afterwards.

Meassure in respect to multimeters groung lead I guess.

[twice11]

Spot on! You could cut the trace going to the +IN pin (the pin you suppose is connected to the ADC) and compare the "chip side" and the "meter side" of that trace. If you find a difference in the chip, it's gone. Forget about that meter. If the difference is caused by something outside of the chip, chances of repair are not that bad. You "just" have to find the faulty component. You can reconnect the cut trace using a bit of solder afterwards.

Meassure in respect to multimeters groung lead I guess.

Yeah, that's what I intended. Measure resistance in respect to multimeter ground, which is also connected to IN-, if that chip-on-board-blob is indeed a 7106 clone. The A/D in that chip has differential input.

[Wojta]

Okay, Ive cut exatly 3 traces. The first one was the one I thought is ADC. That cut trace disabled the multimeters voltage reading except for the 1000V range which was still reading something. Then I cut other 2, one of which was also connected to the transistor (that seemed to disable the beeping and the ohm range so the multimeter showed 1 on every selection) and one that was going to the selector wheel (donk ask me why that one...) that trac disabled the entire meter and caused the 1 to change to -1. I guess thats good. I mean if it shows a certain value adn doesnt count up/down or act weired, the chip could still be allright. I also meassure the resistance. The pin side shows about 10Mohm to 15Mohm, the part side shows nothing.
Thank you twice for help.

[twice11]

On the working one it is 2,22Mohm or 40Mohms vs 215kOhm or 0,7Mohm on the bad one. There are two values for each one because they depend on how I meassure. COM to meassured point or +lead to the meassured point.

The pin side shows about 10Mohm to 15Mohm, the part side shows nothing.
Thank you twice for help.

These numbers don't match up yet (and after cutting 3 traces, you should have 4 measuresments, not just 2 of them). You need to find the source of the low resistance (215k/700k).

[Wojta]

Okay, Ive had enough of it now. I spent most of my time looking for the errors but everything is as it should be, the voltage that goes to the chips ADC is okay and is dividing as I turn the wheel, all the voltages on other pins look okay. Ive checked out the 7106 datasheet and the chip on board seems to be exactly that. The voltages match up to the datasheet. Basically I cant find an error. The IC is probably dead... I traced the ADC pins and they were NOT clamped in any way maybe only by the PTC but certainly not by the transistor. So Im just giving up right now...

I wanted to ask a few questions on what multimeter to buy before I get a new one but I suppose Im better of starting a new thread. Maybe?

Anyways Thank you very much for the support 

[Gyro]

Well if it is the 7106 then you always have the option of carefully demolishng the COB die and soldering down a packaged replacement on the available footprint. Only you can say whether it is worth it though.

[Wojta]

Well if it is the 7106 then you always have the option of carefully demolishng the COB die and soldering down a packaged replacement on the available footprint. Only you can say whether it is worth it though.

Not even a bit it would cost me even more than the meter itself  Our local shops are awfully expensive and I dont really want to wait a month to get one from aliexpress...

[twice11]

Okay, Ive had enough of it now. I spent most of my time looking for the errors but everything is as it should be, the voltage that goes to the chips ADC is okay and is dividing as I turn the wheel, all the voltages on other pins look okay. Ive checked out the 7106 datasheet and the chip on board seems to be exactly that. The voltages match up to the datasheet. Basically I cant find an error. The IC is probably dead... I traced the ADC pins and they were NOT clamped in any way maybe only by the PTC but certainly not by the transistor. So Im just giving up right now...

How are you so certain that the transistor did not (try to) clamp the voltage? The transistor is (mis-)used as Zener diode between IN+ and IN-, and should work as clamp (+5V/-1V). It's just not beefy enough to handle mains voltage in Ohms range, especially in 230V-like areas. The PTC might limit current, but it needs enough time to get hot that the 7106 might have been blown before. As you reported low impedance (below 1M) between IN+ and IN- in the broken meter, you should easily be able to measure which part of the IN+ network has that low impedance after splitting it. It might very well be the 7106 itself, and in that case, I agree that the meter is beyond economical repair.

[Wojta]

How are you so certain that the transistor did not (try to) clamp the voltage? The transistor is (mis-)used as Zener diode between IN+ and IN-, and should work as clamp (+5V/-1V). It's just not beefy enough to handle mains voltage in Ohms range, especially in 230V-like areas.

I think it did not have a chance. The transistor is connected between GND and Volgage Hi reference on the chip so the inputs dont seem to be secured in any way.

As you reported low impedance (below 1M) between IN+ and IN- in the broken meter, you should easily be able to measure which part of the IN+ network has that low impedance after splitting it.

I am pretty certain that its not the ic. Ive measured every single component on the board and compared it to the values written on it. I didnt find any faulty one. If it wasnt the ic I would have to go through meassuring all the traces and I dont really think thats worth it.