Fluke 87V displaying dashes only

[Manul]

This does sound like something else. Maybe something turns on after 1/4 of second and causes overcurrent condition. TPS77050 is only 50mA rated and has overcurrent protection.

[giosif]

Ok, so the meter is showing some signs of life. That's good.
For the 5V rail, it could be overcurrent but, at the same time, unless I am reading the schematics wrong, the Enable pin on U10 is active Low.
And, since you mention the voltage on the En pin is always 3.3V, that would normally mean the output of the regulator should be off.
Still, probably worth testing U10 out of circuit, maybe there is a trend here (i.e. multiple supply rails having failed).

[Multiple Cheese Slices]

Actually, your response made me realise I was wrong just now.
The 3V3 DISABLE EN isn't actually on all the time, just when the output voltage of U10 drops low. My bad - I assumed the delay of it turning high was simply the DMM booting up.

Somehow I have missed the little line above the EN in the datasheet and therefore the part that the EN pin is negated and supposed to turn the regulator off, and without that assumption it all makes sense now - the regulator is working as it should - it turns on, which in turn turns on the 2.5V reference, which i assume allows the controller to turn on properly, but for whatever reason, it decides that the pin 69 (V5*, leading to the (not)EN on the regulator) should be high - at 3v3.

Since I don't have a scope at the moment, I have used a working multimeter to confirm that the voltage values match up with my theory and a logic analyzer to record the waveforms.

So, If i'm not incorrect yet agian, for some reason the μC wants the 5V regulator to be off. Question is - Why?

Here's what I have found in the service manual about "5V*":
U10 is a 5.0-V regulator that generates the +2.5-V (-2.5 V + 5.0 V) power supply. U10 is enabled and disabled via control line V5* from microprocessor U2. C15 bypasses the output of U10.
When U10 is enabled, 5 V is supplied to the voltage reference U8 input. U8 is a 2.5-V reference with a buffered output and accurately holds the COMMON input 2.5 V above the -2.5 V power supply or at 0 V. The triangular ground symbol represents this reference level throughout the Meter schematic. C16, C57, C50 and R109 bypass the output of U8 keeping the voltage between COMMON and the -2.5 V power supply noise free and stable. C50 and R109 bypass the
output of U8 keeping the voltage between COMMON and the -2.5 V power supply noise free and stable. C50 and R109 form a low Q bypass that is directed at electromagnetic interference (EMI).

Not really sure what implications this would have (is this *supposed to happen?* If yes, that would mean there is yet another issue somewhere.), but posting it here in case it helps.

Oh, and I think it's worth adding that the issue isn't the battery itself - it measures sightly over 9V and is brand new.

[giosif]

Like in most cases with broken devices, it rarely is just one issue, so this doesn't come as a surprise here.
I am guessing the reason the microprocessor U2 is disabling the 5V rail is because it thinks the battery is too low and it shuts off the meter (meters can become quite inaccurate when their power supply - i.e. the battery, in this case - is running low and that could pose a safety risk - e.g. the meter might display a voltage measurement of 20V while, in fact, it might be more like 30V and the meter operator might decide to work on that circuit thinking it is safe).

Can you measure voltage at TP16 and then at TP22 and let us know the readings?
Also, with the battery disconnected from the meter, can you confirm continuity between TP22 and pin 6 of U2?

[Manul]

I think it is quite clear that MCU ADC reference is wrong. Reference comes from LT1790 basically and should be 2.5V. My guess is that +2.5V (5V regulator output) is excessively sinking into circuit ground (LT1790 output). So all MCU ADC readings are wrong, inluding 9V battery voltage (reads lower, then it actually is).

[Multiple Cheese Slices]

Regarding Giosif's reply;
There is continuity between TP22 and pin 6 of U2.

The voltage at TP22 when the DMM is on is 1.87V, and TP16 is 9V. I'm using the battery (-) as a ground point.
The output of TP22 matches up with my calculations of the voltage divider consisting of R15 and R16.

Funnily enough, I think this is a bad sign, since the 1.87V does reach U2 (i have even probed the pin directly), so it's not a matter of a shorted resistor or cracked trace/via. Do correct me if I'm making a wrong assumption here tho.


Regarding Manul's reply;
Would there be any way of checking that / making sure that is the case? It would in fact explain this behavior.

[Manul]

Well, if you would connect TP22 to the LT1790 output you would likely bypass low battery alarm and have it running, so you can check voltages. BUT the problem is that if 5V is really sinking into 2.5V (and I'm guessing that this is what damaged the old LT1790) then you could expect more failures if you keep it running like that. Because now it just disables the 5V regulator quickly after starting up, but if you bypass low battery, likely it will not do that.

Or there could be another scenario, that LT1790 failed, its ouput got low and something which was powered from +2.5V (from 5V regulator output) got over voltage and is now failed. But this I would say, unlikely.

Thermal camera would help here so much... How is ouput of LT1790 when you start it? Does it look, that it exeeds 2.5V for a moment?

[Multiple Cheese Slices]

Unfortunately I do not have a thermal camera laying around, however I've sprayed the whole board with isopropyl alcohol (which should evaporate if anything gets hot, indicating where a short may be), but unfortunately nothing seems to heat up enough to cause the IPA to evaporate.

Checking the output of LT1790 yet again with my other DMM shows stable 2.5V. I do not have an oscilloscope to record the waveform, but what I can see from the DMM, it's unchanging even in the milivolt range (does not exceed 2.5V). Of course it drops to 0 after the 5V is cut off.

[Manul]

Then you may try to bypass low battery alarm, but I suggest to monitor current consuption from battery and if it is excesive be ready to turn it off. Maybe try to understand what is getting hot. Normal consumption should probably be less the 10mA. If everything looks good, recheck all supply volatges, ADC reference and basic functionality of meter.

[Multiple Cheese Slices]

I've done as you have suggested. Now there is a direct connection between the output of LT1790 (C57) and TP22 (R16/C28) as shown in the picture attached.

The multimeter does in fact start up, however it still seems to be freaking out. Here's a vid of what's happening: https://streamable.com/vb4yb2

The current consumption from the battery is around 5mA, I have measured the consumption before making the modification to the board and it was around 5mA aswell. Of course it was 5mA for a short period of time before the multimeter went into low battery mode, in which it was around 0.1mA. No surprise nothing is getting hot at those currents. (as you can obviously see, I'm not exactly using a top-of-the-line ammeter, but I've calibrated it to be at least somewhat accurate ).

Not idea what is causing the 87V to behave in such a way.    Any ideas?

P.S. I feel like this may be important information - Out of all the modes, one actually seems to work properly - the diode testing. The 87V successfully measures a voltage drop of a rectifying diode, and isn't that far off from my other (two) DMMs. That's a start! 

[Manul]

So 5V, 3.3V, 2.5V are all spot on? Check VEREF-, VEREF+, also look at ADIN and FASTADIN at different meter modes.

[Multiple Cheese Slices]

Spot on? Well, not quite 
Everything (including the input current( starts with 7mA, drops to 4mA) drops over time to not-good levels.

2V5(U8) 2.8V, at start, then gets lower and stabilizes at 1.5V. not good...
3V3(U6) is 2.4V and also is slowly lowering.
5V(U10) is 3.8V at the beginning, but slowly drops to around 3V over time.

I assume the new connection somehow caused everything to misbehave, as the voltages on those were good and stable before.

VREF-   is between 140mV and 250mV. Seems to change as well.
VREF+   2.7V (but then drops down along with current consumed from the battery)

FASTADIN (pin 96) starts off at around 250mV and follows the trend of other measurements and drops to 96mV.

and ADIN? It may be just me, but I can't find anything about ADIN in the schematic itself, just some text in a section about "secondary analog circuits" in the service manual. May you point me to the exact spot I'm supposed to measure? Thanks!


Those voltages are pretty abnormal. This definitely wasn't an issue before. All measurements were taken relative to battery (-), just like before. 

[Manul]

Something is hugely wrong here. With 7mA total current consumption from battery none of these regulators can be overloaded. If they are not overloaded, but output is too low, that means only one thing - input voltage (battery) is low. Can you take a stable 8-9V source like bench power supply or something?

However, looking at 2.5V output (2.8V) it seems that it sinks qiute a lot of current, so much, that it can not keep regulation. So maybe your 7mA is hugely incorrect? You need to sort some things here, because it is not logical at all.

[Multiple Cheese Slices]

I have made sure that the ammeter is reasonably close to displaying the real current (example: measuring 9.5mA when calculated 10mA should be flowing in a test setup), so it's likely not the issue.
I have also connected the 87V directly to a 9V bench PSU (13A max, no issue with limited current here), and the voltage/current measurement functions were still broken. I have decided to re-take the measurements of voltage on regulators/references output, and now the 2V5 and 3V3 seem to be proper and stable, however 5V is still dropping to 4-3V. Not sure why that is, since nothing changed from the last time I have measured them. (I have tested this with both battery and PSU as the power inputs, makes no difference).

Checking the "notEN" pin on the 5V regulator yet again, connecting TP22 to 2V5 does indeed make it stay low all the time, and without the connection it still rises to 3V3 after a fraction of a second, like it did before. So nothing changed here.

It would be great to find a proper way of fixing the 5V reg turning off because of "low battery", I assume hooking up 2V5 to TP22 is bound to cause problems in the future. No idea why everything but the diode measurement is behaving strangely as well... Could it be an issue with a broken ADC?

[Multiple Cheese Slices]

I feel like this may be helpful information; I have probed the digital pins on the external ADC with my logic analyzer yet again, and there does in fact seem to be communication going on, so that's one thing replacing the 2V5 reference helped. (also, rules out one possible point of failure?)

[Manul]

It would be great to find a proper way of fixing the 5V reg turning off because of "low battery", I assume hooking up 2V5 to TP22 is bound to cause problems in the future. No idea why everything but the diode measurement is behaving strangely as well... Could it be an issue with a broken ADC?

I suggest to forget about low battery for a moment, because it is just a side effect of other things not working properly. Just keep it bypassed for debugging purposes until you will fix the real problem.

When measuring current, don't forget that at 20mA range your multimeter likely has around 10 Ohms shunt resistor plus some inductance because of long leads, which drops voltage and also might cause instability with some circuits which rely on having low impedance source (battery) and do not include enough bulk capacitance. So if in doubt, add bulk electrolytic capacitor on the board where battery is supposed to be connected. It will suppress possible oscillations and will lower the impedance.

Now about 5V regulator with droping voltage. There is really not much options:

1. There is not enough voltage at input
2. Current limitting (>50mA ?)
3. Cycling of enable signal
4. It failed and needs replacement

So you need to setup current measurement, voltage at regulator input and output, also to be sure that it is enabled all the time. It seems to be working and enabled, so either it does not have enough input voltage (check exactly at its input) or it is overloaded.

[Multiple Cheese Slices]

I have added a 330μF (is it enough?) electrolytic cap parallel to the battery as you have suggested.

Though it appears that the issue causing U10 to have low output was the 9V battery which suddenly started to massively drop voltage with just a few mA of load. (drops to 4V with just (what should be) 9mA (1k resistor across the terminals) of test load), no wonder the voltage regulator didn't produce enough output voltage. This is a recently developed issue though, I have previously checked the input voltage on all the regulators and it was fine. Not sure what made the battery suddenly "die" too, it was new and unused, as far as I know at least. But hey, accidents happen, I may have confused it with some other (used) battery.

With a new battery, the output of U10 is 5V as it should be, the 3V3 and 2V5 work fine as well. However the issue seen on the display (constant overload condition) (issues with ADC?) persists, so it appears that the battery/power supply section was not the key culprit. 

[Manul]

I think that diode mode is working, because it is the only mode which is not using autoranging. Because diode mode does not need autoranging, probably it is not using FASTADIN and uses just the dedicated ADC feed from ADIN.

I suggest such test: put meter into milliamp mode (almost guaranteed to produce zero, voltage modes may have noise) and measure ADIN and FASTADIN. Maybe you will find that OPA2347 is broken. But be sure to use a GOOD battery and that all power supply voltages are good

[Multiple Cheese Slices]

I have set the 87V to mA mode and made sure yet again that all power rails have proper voltages.

I have found that:
FASTADIN is at 5mV
ADIN         is at 2.5V

furthermore, I have probed all the pins on the (OPA2347) dual Op-Amp in case that will help.

pin1= 1mV      |   pin8= 5V
pin2= 1.666V  |   pin7= 2.505V
pin3= 2.512V  |   pin6= 2.505V
pin4= 0mV      |   pin5= 2.441V

(I have attached a pic marking all my measurements, hopefully it is intelligible  )
all measurements are relative to -2.5V (battery (-))

[Manul]

And again you measured relative to -2.5V

Anyway, looks bad. Either R34 is somehow not connected (solder/trace cracked) or OPA2347 is bad. You can buzz pin2 to R34 and then pin1 to R34. Make sure it is really connected. If ok, then try to replace op amp. That could be the end of problems if you are lucky. Because now, op amp output (FASTADIN) seems to be stuck at max scale, so no wonder that software endlessly tries to autorange. It is like constant overload.

[Multiple Cheese Slices]

Sorry about the -2.5V again, fortunately it's only offset.

There is continuity between R34 and pin1/pin2 (to different sides of the resistor, of course, as shown in the schematic).

While desoldering the op-amp for testing, I have accidentally knocked off R34 and R36 into my carpet... considering they're 0603, it's safe to say that I won't ever see them again.

I have already replaced R36 with an equivalent value resistor, and will be doing same to R34 as soon as I find a proper one or decide to order it if I can't find it. Although that leaves me with a question - are those resistors important for calibration by any chance? the 200K will be getting replaced with 200K and 100K with 100K of course, but there always will be minor differences. Should I be worried about those at all or is this non-essential for good measurement accuracy?



I'll be testing the op-amp one more time, outside of the PCB, and will order a new one if it's definitely fried. The delivery shouldn't take long; I will post an update as soon as I solder the new one in or have any new info. 

[Manul]

I often find some SMD resistors and capacitors to be slightly ferromagnetic, so you may try a strong neodymium magnet on the carpet. These resistors should not be critical, any 1% part shoud be ok. But speaking about calibration, the 2.5V reference is quite critical. So there is no guarantee that meter will be exactly as it was. Not sure, you will see I guess.

[Multiple Cheese Slices]

Yes, I know about them being ferromagnetic. Thanks for the tip tho! I have actually tried looking for them for more time than I'm willing to admit. Sometimes stuff seems to teleport into an another dimension when you drop it, you know what I mean?

Anyway, It's good that they're not critical. Worst case scenario - I'll send the multimeter to get calibrated once it's up and running properly. 

[Multiple Cheese Slices]

Hello! Both good and bad (but mostly good) news. 

Replacing the op-amp has fixed all modes except Vac and mA/A(ac)/ (which I will talk about in one sec). The measurements aren't that far off from my other DMM, however the 87V should probably get calibrated anyway to be able to show its potential; It' supposed to be accurate after all.

That is in fact great! Since, well, the DMM is working almost perfectly fine. Almost being the key word - unfortunately there's something weird going on with Vac and A/mA(ac) measurement functions.

Firstly - Vac - it seems to be okay when in the 600mV range (the auto default when probes aren't touching anything, tho unfortunately I do not have a way to test the measurements in this range) however when taking a measurement of a voltage in any other range, the overload condition occurs. Example - 7V, 32V and 230V AC all return overload. This happens regardless of the range being selected manually or by the auto-range function.
One anomaly is when measuring 18Vac - for some reason it returns 5Vac. I'm 100% sure I am in fact probing a 18Vac source, and that it doesn't drop to 5V when being measured. I have double-checked it with my other DMM.


Secondly - also related to AC - the current measurements.
the DC current measurements seem to be working fine and the measured values are reasonably close to the theoretical (calculated) ones. AC, however, always returns something close to 2-3mA, despite definitely way more (should be ~86mA, as calculated prior to testing) flowing through (judging by the resistor heating up alone).

Not sure if this is one issue or two separate ones which coincidentally both have something to do with AC. Nothing seems wrong with the input sections at the first look, but I'll be taking a closer look just in case. Not sure what to check besides that, what are your thoughts?

EDIT: Oh! And I forgot to mention: replacing the op-amp somehow fixed the battery issue. The wire is no longer required.

[Manul]

This is quite weird, LT1790 2.5V reference, then OPA2347 op amp, which is not even exposed much to environment (electrically speaking), and now something more again. It is suspicious how there can be so many seemingly unrelated failures. Maybe this meter got struck by lightning or something.

You need to study schematic a bit. Hope this AC failure is not main chip failure (U1). Maybe it is RMS converter.

I would probably put it to VAC, short the probes (to be sure input is zero), then measure ADIN at different ranges selected manualy. If some ranges gives weird ADIN voltages, then I would try to follow the circuit backwards and see where is the failure.