Did you notice its doing it in a pattern? It goes: 0 5 0 5 0 5 0 5 pause, repeat. It's extremely consistent. Something in the logic is definitely doing it. Looks like it's timed with the refresh of the LCD too....
My guess would be it has something to do with the autoranging "leaking over", maybe even internally accidentally connecting the reference wrong. Its probably on the asic.
Brian I saw that too. It's static on the 5, but flashes the real reading four times quickly and then pauses on the 5. The period for one complete cycle is just slightly over 2 seconds (close to 2.2 seconds, maybe). Only the designers would know what on the processor might be cycling at this speed and with that pattern. Some auxiliary function is bleeding over and affecting the display output, but is otherwise leaving the critical functions unaffected. Another clue is that it is causing a display error on V and A settings, but is only causing some barely noticeable flickering on resistance and diode settings - again that might only mean something to the ASIC designers.
I like that you keep investing failures of "your" meters. Makes interesting videos and improves probably not this but multiple lines of devices. Keep on going Dave!
@@hotgluegunguy asside from the different chemistries of the batteries, one potentially charging/overcharging the other, and different material types mixing, and the potential to cause a fire, the actual current they are capable of is different also, resistances, allsorts, so who knows exactly what issues it could cause...
@@Ghozer I still don't see what damage it would cause here, as the batteries are not charged by the device. High series resistance could cause unstable operation, but I can't imagine it damaging the meter. I'm not saying you're wrong, I'm just curious as to why you arrived at that assumption.
Looks like it's generating some noise somewhere and it is trying to autoscale to the fast fluctuations in the noise. EDIT: So, it shows the correct voltage and toggles range. Then it can't be noise. I'm probably wrong. Definitely looks like a processing issue now.
I have one and I regret buying it. The first firmwares were slow and the recent ones tend to drift. The display hasn't got a lot of contrast (even at max setting) and there's no way to read the SD without taking it out of the meter. I recommend watching ua-cam.com/play/PLZSS2ajxhiQDDs_mWPLavaveGe0RGEw1M.html before ordering one.
@@henninghoefer Agree with you, I really don't like the 121GW. It messed up some measurements because of a low battery, and at that point you've just fucked up my confidence completely.
I had both types of EEVBlog meters (121GW and BM235). The BM lost all function of its buttons, and the 121GW I barely trust to measure DCV under 10 volts. The AC measurements are downright dangerous to trust when checking if an AC circuit has 110V/230V live voltage on it and trying to measure ripple on something like +24V DC is just impossible. The AC+DC measurements ranges are not specified in the datasheet. The fanboys at the EEVBlog forum thinks this is fine, as they know the inside of the meter by heart...that is NOT what I expect to need when I buy a DMM. I want a datasheet. I bought it from the KS campaign so naturally there was no datasheet at the time. I have learned my lesson, I will never again any instrument carrying the EEVBlog on it...But each to their own!
Since Dave has access to the company he should definitely get a replacement CPU and put it in. What if that does not fix it? Reasons to fix it: 1) You got to be very curious why the meters you sell died. 2) It will probably sell another 50 meters. 3) You can call it a tutorial on replacing SMD components.
I wonder why Dave doesn't put much time into his channel anymore. He said that this is his main job. So what else eats up so much time that he cannot do fundamentals or anything with some work behind anymore?
Yeah that's a glorious example of terrible marketing by Energizer! First thing I thought when I first saw Energizer "Eco" branded batteries was that they must be rechargeable cells. So I'm not surprised they've been mixed up with an actual Ni-Mh cell. But no, turns out they're just made of a whopping FOUR percent recycled materials. I'm not sure you could get much more disingenuous.
My theory is the LCD driver is damaged. The segments change each time the processor refreshes the display, more often when measuring resistance than when measuring voltage. It looks like some bit of some internal counter (frequency divider or memory address counter) has a short to memory area of first digit's segments.
I bought a 235 directly from Brymen and 1 month after the warranty expired it developed a calibration error that prevent it from measuring any voltages. Shipped it back to Brymen and they re-calibrated it and sent it back - it did take about 90 days to get it back but at least they repaired it.
@@Yrouel86 Yeah, I think so too, I dont see a skid mark on the pcb near that wire. So probably flux.. My bet is an internal fault of the prosessor and that flashing 5 four times, then pauses and repeats consistently. Is some sort of fault indicator. Like maybe the eprom has corrupted somehow. It's weird for sure
Swap the asic (processor) with a good meter and see if it fixes it. Might be an internal fault in the asic, perhaps even reflashing it, if possible, with new firmware would fix it.
It's an interesting and very unusual fault, the fact that it is a repeating pattern makes it fascinating ! I would like to see a scope reading to measure the frequency of that pattern. Most of these guesses here show that most people really don't understand what they are seeing and either didn't watch the whole video or are assuming that Dave knows nothing about electronics and that he can't see what we can see in the video !
You must have another scrap unit around that you could swap the processor with just to see if that cures it. it doesn't matter about calibration as you are unlikely to put the unit back in to service. It would just be good to see it through to the end.
Hi, I've had a *very similar problem* with APPA305. Reason was, it somehow managed to erase its' own calibration EEPROM. Because I got it free with this error and it had some decent amount of counts, I decided to dig into this. It took me several hours to reverse engineer eeprom data mapping, and in the end, I've mapped all calibration data (I think except of temperature) and recalibrated it.
I find it very interesting that it toggles 0, 5, 0, 5, 0, 5, 0, 5, always in the same sequence with the same time period, then waits a fixed period of time and repeats that toggling. It's not at all random, which I would have expected. It certainly looks like the processor is displaying an error message (like when your car shows the diagnostic codes on the malfunction indicator light). Maybe contacting the manufacturer will yield some results. Very interesting indeed.
@@steverobbins4872 I was looking for the reference voltage, U2 looked like the candidate for being one. A scope probe on that, and a few on the power pins for the micro should sort out where the issue is.
@@station240 I couldn't agree more. Scope the shit out of that - Hook up a good meter right next to it, and just compare what is going on at various voltage-rails. (where you find deviations that rhythm of full-scale readings). My opinion is: having a non-broken copy of the device might be even better than having a schematic ;-)
1.check OP amps input stage. 2. check Ref. Voltage input stage (ref diide). 3. clean the contakts of the rotary switch (oxyde). 4. Glitch in the memory or cpu register?. 5. re-solder pins specially Opamp and mpu.
Maybe ESD damage to the micro. Might be worth scoping the meter supply and comparing its current draw against a known good one. If the chip is damaged, its supply current may be higher.
Maybe a stupid suggestion, but could it be the most-significant bit (MSB) of their custom ASIC going weird? IDK how they make their ASICs or what each of the pins does, but maybe a bond wire somewhere that's related to the MSB is faulty. Not that I've ever heard of them failing, but it's all I can think of. The randomness makes it look like a floating pin or something.
Its not random. On voltage anyway. Its a regular sequence of 5 pulses. its four pulses showing the correct reading then the 5th pulse just shows the max range. edit: the other ranges have a pattern too but different from voltage.
the way the 5 flashes from 5 to 0 is almost like some sort of error code.. like it flashes 4 times then pauses then flashes 4 times and so on. almost like the check engine light on a car
What happened to: "though shalt always check voltages"? :P seriously though, isn't 5v a standard digital voltage value? Something being polled at regular intervals and it's somehow leaking onto the main sensing line/one of the grounds being weak and unable to sink enough stray voltage, would be interesting to see whether the relevant pin on the micro was actually seeing any stray voltages? One interesting thing to note: the volts reading is dc, and the millivolts is AC, but they both show something that looks like 5 to some exponent, that's important, because the only waveforms that have the same peak to peak and RMS values...are rectangular waves, i.e. some kind of logic and it's quite a regular pattern too
@@muppetpaster 5v seems equally random otherwise it's not too of the range and it's not a power of two, plus it's a multimeter-i'd imagine it has to have precision voltage sources floating around to even work properly, and it's not quite bang on 5v every time anyway- sometime it shows a .01/.02 if I recall correctly Edit: hard to say it's switching to fast to say whether that .01 is tied to the 5v or to the 0v reading
When I saw this, I wondered if the design has a auto-cal, or sanity check circuit, that’s executed on boot up. I can imagine a built-in circuit, like BIT, that switches the input between ground, 5V, etc. If the control signals for such a circuit went berserk, it might cause these symptoms.
Both fuses good? Voltage spike perhaps trashed the asic driving the display? I had a Uni-T that someone blew the fuses which caused a voltage skip that got past the input protection and took out the main IC. Spot on except it had all the wrong ranges. Replaced the IC and it's been perfect.
@Dave if you can, try giving it a re-flash. What's the betting of a random bit flip in the firmware that causes a jump to the wrong part of code or some crap?
Can you measure the input to the logic side to see if something is feeding 5v into it? Although this kind of failure doesn't explain the A/mA weirdness to me. I'm brand new to electronics, green as grass, so I'd love to learn more about this, and if this develops into anything. Will scoping any of the logic test points (if there are any?) help at all? I'm probably talking complete nonsense, but I'd love to see more diagnosis of this.
Just guessing. It is showing the correct voltage intermittently, so the downstream of the ADC is ok. It is weird how the display is messing up which may show an electrical issue, like voltage leaking on a line, or dropping. It could be some kind of ADC communication breakup which is going to the max value.
Definitely look like exceeded range measurement due to different ASIC chip properties in this particular multimeter or it can be software bug too. I'm curious are this glitch is only BM235 specific or it affect all BM2xx line? I ordered BM257s at yesterday and hope it will not show these funny 5-s.
Internal bondwire fracture - maybe on something related to the voltage reference. Making just enough contact to supply a reference, then draws current and the failed junction goes non-linear, add a tiny amount of capacitance and it oscillates?
Could you test this with some percussive maintenance while the thing is powered and see if anything changes, or is everything kind of set in stone with the packaging? Would be curious to know if this is a viable diagnostic method.
Dave, check the EEPROM. Read and check its contents several times. I have a 25q series EEPROM with similar weird behaviour, it bit flips at MSB every exactly 16 bytes sent. It was used as a font lookup table memory and a particular character on the display changes, as your multimeter does. I exchanged emails with the manufacturer several times and they refused to believe that their EEPROM was delivered as faulty, they carefully test them, etc. The EEPROM was delivered with factory programmed fonts and I never used it. Just plugged it in my flash reader to check if their fonts are fine and bootloaded the fonts in my code. I had instantly saw the weird glitch, with exact repeatability. I copied its content in a blank EEPROM and it works flawlessly. It's not my circuit design fault, as the manufacturer claimed. Maybe you have similar issue.
I was going to buy one of these, but I got no reply to my question on the EEVBlog Store, or on the eBay listing. So, I figured that support communication would probably also be nonexistent. So, didn’t end up getting one.
My other theory is a short circuit on the zebra stripes connecting the display to the board. My meter alternately displays 0.000 and -0.000. Maybe in this Brymen there is a short circuit between the driver lines of the minus symbol and the G segment of the first digit.
Possibly get a new asic and solder that in. As everyone said it is flickering at regular interval. Atleast putting a new one would eliminate some faulty silicon
Could the 5.000 be the reading from the internal reference voltage (after scaling)? It may be a faulty internal multiplexer is feeding it into the A/D converter intermittently? PS. You're more frugal with the flux than a certain Mr Rossman. 😉
It could be a problem with the voltage reference. Or it could be an op-amp with a cracked solder joint or feedback resistor somewhere. Those are the only 2 things that could shoot the reading to the roof and back I would guess.
I had a fault in a scope, where a rubber key was stuck inside and made a permanent contact. Couldn´t be seen from outside. So the reaction on any input was weird. Check all rubber keys for correct operation.
Do the eeprom chips carry any write data. I had an incident where a character byte was accidentally written to the wrong address on an eeprom and it caused the display to output the wrong character when called. Just something to check.
I have a 25q series EEPROM with similar weird behaviour, it bit flips at MSB every exactly 16 bytes sent. It was used as a font lookup table memory and a particular character on the display changes, as Dave's multimeter does. I exchanged emails with the manufacturer several times and they refused to believe that their EEPROM was delivered as faulty, they carefully test them, etc. The EEPROM was delivered with factory programmed fonts and I never used it. Just plugged it in my flash reader and bootloaded the fonts in my code. I had instantly saw the weird glitch, with exact repeatability. I copied its content in a blank EEPROM and it works flawlessly. It's not my circuit design fault, as the manufacturer claimed.
This is a dissappointing video. Essentially "I got a faulty meter back with a weird fault. I tried a couple of things but coluldn't really be arsed. Thanks for watching".
If it should measure the reference, the result would obviously be full scale. But I don't know why it would measure the reference unless it alternates between signal and reference using a switch.
It could be a both programmatical and hardware error. Because the digit 5 blinks with a constant pattern. (It blinks 4 times then stops for a while) Why don't you replace the MCU to see if it helps?
I'm not sure if' it's affecting the other digits, since you didn't connect up anything to measure, they show as zeroes (with the least significant digit 'bobbling'). Measure, say, 1.234 volts and see if it alternately displays "1.234" and "5.000"
My first guess on seeing its behavior, and the fact it happens in any mode, is that it has dual ADCs that it alternates between to get measurements and one of them is not working anymore for some reason.
@eevblog2 I would probably take another meter out of the scrap bin and swap the processor. Not too many pins so shouldn’t take too long I don’t think. Would be interesting to see if it persisted, might be the silicon.
Why is this video posted on the secondary channel? I usually don't watch EEVblog2, but this one seems perfectly suitable for the main channel that I'm subscribed to.
The etchant trap thing still matters if you use the cheapest, worst process available on the market for high volume production, where you still can't afford any failure.
Not that it matters, but what firmware is in that meter? I noticed the RM1, RM2, RM3, and RM4 positions on the PCB. They look like option jumpers. I wonder what would happen if you played with them. I have two of the BM235 meters and they have different firmware revisions. That is one thing about the 121GW over the 235. Easy update of the built-in firmware.
Is it possible its switching in a reference voltage for self calibration and the switching signal is not functioning? Or is there a pin on the ASIC that's pulsing at that frequency? Is there any ghost signal coming out of the input connections?
My $0.02 would be to thoroughly clean the rotary switch and contacts... (My guess is that the CPU is being asked to be in multiple 'modes' at the same time due to leakage between switch contacts)
Shot in the dark here - Is the reference crystal on spec? Can they be fractured internally yet still 'function' ? Will replacing the crystal still require re-calibration?
Press the selector down and it starts to be ok...The usual fault on Uni-t ...only some contact lub in the selector, turn a few times and, problem solved.
4 роки тому
Always very interesting videos. Thanks Dave for the time you spent for us. I just have a quick question : I saw on your PCB some resistors that is labeled RM in place of R. What is the meaning?
Probably a stupid solution, but have you tried connecting it in parallel with another multimeter? Maybe it's working fine but leaking power to the measuring side somehow?
I would be nice if you would try to swap chips, just to confirm whether the fault follows the main CPU or the PCB, calibration wouldn't be important for that.
You don't need to recal the meter just to prove the processor is the fault. Just swap it out with one from another busted meter and see if it stops blinking 5.
Dave please don’t leave it there. Please get them to send you a new processor, swap it over and see if it makes it work. If not please please please look at repairing it
I'm looking to buy either the BM235 or the 121GW, just not sure which to get. I mostly do some Auto/Home DIY and minor electronics repair, nothing too deep.
The BM235 is an excellent meter. I have 3 (one at home, 2 at the office) and they're the first meter I tend to reach for. If you need more advanced functions then the 121GW is good, but I generally prefer the BM869S or a bench meter in those cases. The 121GW is pretty impressive for feature set vs size.
If the measurement is spot on, could it be the communication between the processor and the lcd? if it's an spi or i2c, could the processor be reading the values correctly but experiencing noise when sending that over to the lcd controller? 5 is 0101 in binary btw
So here I was, thinking it was a range overflow issue (+1, 0, then -1 manifesting as 5000), but noooo... you had to go and rain on my parade, saying it's a 6000+ count meter, at the end of the video!! :D (totally j/k)
Seems like a internal silicon issue to me. Not sure how the main processor and LCD driver communicate and the format of display data, but it could be just a display data corruption considering it only affects the most significant digit and in a repeatable way.
Not for economic viability obviously but rather curiosity / makeshift QA department, I'd have swapped the processor with another one. Especially should another dodgy meter turn up some day. Note batch / date code of affected processors, it could be a bad lot. But it would exclude software issues.
I imagine this sort of crap happens all the time, for example with cheap routers. Customers think "oh it must be me" or "the ISP are throttling me" or "the damn kids are on the WiFi again", when in actual fact one of those poor processors got tired on running at 1GHz with zero cooling, full of dust and knocked off the table once every couple of months. These kind of "half" failures are probably happening everywhere. One major problem is that software makes the assumption that the hardware is always working correctly, for example that a bit flip is impossible and an ADC will only ever read between a value range, when in actual fact hardware failures are actually common place.
Depends on the application and to what cost the equipment is being manufactured. A home broadband router needs to be built as cheap as possible as a first priority, hardware and software wise. A piece of life support equipment will be built to a higher specification and cost may be only a secondary factor.
Lower trace resistance due to the extra solder metal, I think Dave made a video about that practice. P.S. found it ua-cam.com/video/L9q5vwCESEQ/v-deo.html
@@sysghost ah, I see which one you meant. Hmmm, I know only two other reasons, but the traces don't look like that would apply here... (1. to improve heat conductivity to air, 2. acting as guard traces/rings, actively driven to the same potential as they are guarding, to get very high insulation/ultra low leakage from other parts of pbc, e.g. ultra low current/charge measurements => leakage flows over the PCB into these traces)
Just a random thought. But could there be a 5v rail connected to a mux? I assume it's in package package on die so there's not an external chip to replace but might that be doing it?
14:40....If you get a proper reading when an input is connected and an erratic one when nothing is connected....suspect an input ground fault or an input fault of another kind.
Did you notice its doing it in a pattern? It goes: 0 5 0 5 0 5 0 5 pause, repeat. It's extremely consistent. Something in the logic is definitely doing it. Looks like it's timed with the refresh of the LCD too....
My guess would be it has something to do with the autoranging "leaking over", maybe even internally accidentally connecting the reference wrong. Its probably on the asic.
Brian I saw that too. It's static on the 5, but flashes the real reading four times quickly and then pauses on the 5. The period for one complete cycle is just slightly over 2 seconds (close to 2.2 seconds, maybe). Only the designers would know what on the processor might be cycling at this speed and with that pattern. Some auxiliary function is bleeding over and affecting the display output, but is otherwise leaving the critical functions unaffected. Another clue is that it is causing a display error on V and A settings, but is only causing some barely noticeable flickering on resistance and diode settings - again that might only mean something to the ASIC designers.
...... It's Morse, I decoded it,
"Seven castaways marooned on deserted isle"
@@robbieaussievic WILSON!!!!!!!!!
It says "Help, I'm stuck in a Chinese multimeter factory"
I like that you keep investing failures of "your" meters. Makes interesting videos and improves probably not this but multiple lines of devices.
Keep on going Dave!
Hey Dave, 3:30, you can see why they discharged 'unevently'. Top is a NiMH while bottom is alkaline!
quite a careless usage of the owner I might say :(
I spotted this too, potentially the cause of the failure....
@@Ghozer it's definitely not a good idea, but how would it cause the failure?
@@hotgluegunguy asside from the different chemistries of the batteries, one potentially charging/overcharging the other, and different material types mixing, and the potential to cause a fire, the actual current they are capable of is different also, resistances, allsorts, so who knows exactly what issues it could cause...
@@Ghozer I still don't see what damage it would cause here, as the batteries are not charged by the device. High series resistance could cause unstable operation, but I can't imagine it damaging the meter. I'm not saying you're wrong, I'm just curious as to why you arrived at that assumption.
there is an electrician imprisoned in a workplace somewhere and he reprogrammed it for SOS but didnt get it done right...
Aye, that would be 5000 5000 5000 2001 (sooo sooo sooo cool)
Brilliant guess 😁
5:31 "Light at the right angle" - no no no Dave, it's all about getting the tongue at the right angle 😁
The first time I read that, it sounded dirty. Lol
😂😂😂😂
try the range button. if it is a 5 V issue then the "5" will jump a decimal, if it is a leftmost digit issue then the "5" will stay on the left
Looks like it's generating some noise somewhere and it is trying to autoscale to the fast fluctuations in the noise.
EDIT: So, it shows the correct voltage and toggles range. Then it can't be noise. I'm probably wrong. Definitely looks like a processing issue now.
And this is why Dave is able to sell meters, absolutely no BS .. I have two already; thinking about a 121gw :-)
I have one and I regret buying it. The first firmwares were slow and the recent ones tend to drift. The display hasn't got a lot of contrast (even at max setting) and there's no way to read the SD without taking it out of the meter. I recommend watching ua-cam.com/play/PLZSS2ajxhiQDDs_mWPLavaveGe0RGEw1M.html before ordering one.
@@henninghoefer Agree with you, I really don't like the 121GW. It messed up some measurements because of a low battery, and at that point you've just fucked up my confidence completely.
I had both types of EEVBlog meters (121GW and BM235). The BM lost all function of its buttons, and the 121GW I barely trust to measure DCV under 10 volts. The AC measurements are downright dangerous to trust when checking if an AC circuit has 110V/230V live voltage on it and trying to measure ripple on something like +24V DC is just impossible. The AC+DC measurements ranges are not specified in the datasheet. The fanboys at the EEVBlog forum thinks this is fine, as they know the inside of the meter by heart...that is NOT what I expect to need when I buy a DMM. I want a datasheet. I bought it from the KS campaign so naturally there was no datasheet at the time. I have learned my lesson, I will never again any instrument carrying the EEVBlog on it...But each to their own!
Dave don't see enough sellers doing this, levels of honesty in new nevermind older products that you show.
Please spend some time on finding the fault on the BM235, this subject is interesting.
Since Dave has access to the company he should definitely get a replacement CPU and put it in.
What if that does not fix it?
Reasons to fix it:
1) You got to be very curious why the meters you sell died.
2) It will probably sell another 50 meters.
3) You can call it a tutorial on replacing SMD components.
I wonder why Dave doesn't put much time into his channel anymore. He said that this is his main job. So what else eats up so much time that he cannot do fundamentals or anything with some work behind anymore?
nice job of mixing batteries!... noticed one is rechargable (the low voltage one)
That's why I stick to Eneloops (and not just because I bought them from DSE like they were toilet paper)
@@leonkernan hahahaha
went to check the comments before i mentioned the very same. good eye mate
I seen that too right off the bat. 1.3 V is about as high as a rechargable one gets from my experience
Yeah that's a glorious example of terrible marketing by Energizer! First thing I thought when I first saw Energizer "Eco" branded batteries was that they must be rechargeable cells. So I'm not surprised they've been mixed up with an actual Ni-Mh cell. But no, turns out they're just made of a whopping FOUR percent recycled materials. I'm not sure you could get much more disingenuous.
My theory is the LCD driver is damaged. The segments change each time the processor refreshes the display, more often when measuring resistance than when measuring voltage. It looks like some bit of some internal counter (frequency divider or memory address counter) has a short to memory area of first digit's segments.
I bought a 235 directly from Brymen and 1 month after the warranty expired it developed a calibration error that prevent it from measuring any voltages. Shipped it back to Brymen and they re-calibrated it and sent it back - it did take about 90 days to get it back but at least they repaired it.
The white lead on the input connector board looks a little toasted (visible between 9:11 and 9:20)
It looks to me like a hardened drop of rosin flux not a burn mark
@@Yrouel86 Yeah, I think so too, I dont see a skid mark on the pcb near that wire. So probably flux.. My bet is an internal fault of the prosessor and that flashing 5 four times, then pauses and repeats consistently. Is some sort of fault indicator. Like maybe the eprom has corrupted somehow. It's weird for sure
I noticed the same thing!
Just some stray flux or conformal coating
@@Kyle-Veilleux Except that the folks that built the thing said there is no internal fault code.
Swap the asic (processor) with a good meter and see if it fixes it. Might be an internal fault in the asic, perhaps even reflashing it, if possible, with new firmware would fix it.
5:59 J1 or U5 looks some thing burn out
@@neosandi6 J1 is a jumper
it look 's you are right , but it still look burned at place J1
replace the chipset. take one from another broken bm235.
would love to see you repair it and give it away to someone in need!
18:10
It's an interesting and very unusual fault, the fact that it is a repeating pattern makes it fascinating ! I would like to see a scope reading to measure the frequency of that pattern.
Most of these guesses here show that most people really don't understand what they are seeing and either didn't watch the whole video or are assuming that Dave knows nothing about electronics and that he can't see what we can see in the video !
It looks like it blinks every 500 milliseconds for 4 times than stops for a second
At 2:30 My guess is there is a broken capacitor attached to the voltage reference
It is very regular. The pattern repeats. I guess some cross talk from some counter, on a high bit causing a saturated value.
You must have another scrap unit around that you could swap the processor with just to see if that cures it. it doesn't matter about calibration as you are unlikely to put the unit back in to service. It would just be good to see it through to the end.
Hi, I've had a *very similar problem* with APPA305. Reason was, it somehow managed to erase its' own calibration EEPROM. Because I got it free with this error and it had some decent amount of counts, I decided to dig into this. It took me several hours to reverse engineer eeprom data mapping, and in the end, I've mapped all calibration data (I think except of temperature) and recalibrated it.
Notice the 5 cycles on regular timed intervals.
I find it very interesting that it toggles 0, 5, 0, 5, 0, 5, 0, 5, always in the same sequence with the same time period, then waits a fixed period of time and repeats that toggling. It's not at all random, which I would have expected. It certainly looks like the processor is displaying an error message (like when your car shows the diagnostic codes on the malfunction indicator light). Maybe contacting the manufacturer will yield some results. Very interesting indeed.
Is there a voltage reference chip in there somewhere? If Vref is dropping to zero the display might read full scale.
It looks like U2 is a LM385-1.2 voltage reference diode from Texas Instruments. Looks like it's being fed bias current via R27.
@@steverobbins4872 I was looking for the reference voltage, U2 looked like the candidate for being one.
A scope probe on that, and a few on the power pins for the micro should sort out where the issue is.
@@station240 I couldn't agree more. Scope the shit out of that - Hook up a good meter right next to it, and just compare what is going on at various voltage-rails. (where you find deviations that rhythm of full-scale readings). My opinion is: having a non-broken copy of the device might be even better than having a schematic ;-)
1.check OP amps input stage. 2. check Ref. Voltage input stage (ref diide). 3. clean the contakts of the rotary switch (oxyde). 4. Glitch in the memory or cpu register?. 5. re-solder pins specially Opamp and mpu.
Maybe ESD damage to the micro. Might be worth scoping the meter supply and comparing its current draw against a known good one. If the chip is damaged, its supply current may be higher.
"Let's switch this puppy on", bless, not heard that for too long, brilliant. Thank you for that happy nostalgia thought.
Maybe a stupid suggestion, but could it be the most-significant bit (MSB) of their custom ASIC going weird? IDK how they make their ASICs or what each of the pins does, but maybe a bond wire somewhere that's related to the MSB is faulty. Not that I've ever heard of them failing, but it's all I can think of. The randomness makes it look like a floating pin or something.
Time to get the nitric acid out and decap the ASIC.
Its not random. On voltage anyway. Its a regular sequence of 5 pulses. its four pulses showing the correct reading then the 5th pulse just shows the max range. edit: the other ranges have a pattern too but different from voltage.
5:59 J1 or U5 looks some thing burn out
the way the 5 flashes from 5 to 0 is almost like some sort of error code.. like it flashes 4 times then pauses then flashes 4 times and so on. almost like the check engine light on a car
What happened to: "though shalt always check voltages"? :P seriously though, isn't 5v a standard digital voltage value? Something being polled at regular intervals and it's somehow leaking onto the main sensing line/one of the grounds being weak and unable to sink enough stray voltage, would be interesting to see whether the relevant pin on the micro was actually seeing any stray voltages?
One interesting thing to note: the volts reading is dc, and the millivolts is AC, but they both show something that looks like 5 to some exponent, that's important, because the only waveforms that have the same peak to peak and RMS values...are rectangular waves, i.e. some kind of logic and it's quite a regular pattern too
Bit unlikely that it is 5 volts bang in everytime.....
@@muppetpaster 5v seems equally random otherwise it's not too of the range and it's not a power of two, plus it's a multimeter-i'd imagine it has to have precision voltage sources floating around to even work properly, and it's not quite bang on 5v every time anyway- sometime it shows a .01/.02 if I recall correctly
Edit: hard to say it's switching to fast to say whether that .01 is tied to the 5v or to the 0v reading
When I saw this, I wondered if the design has a auto-cal, or sanity check circuit, that’s executed on boot up. I can imagine a built-in circuit, like BIT, that switches the input between ground, 5V, etc. If the control signals for such a circuit went berserk, it might cause these symptoms.
Both fuses good? Voltage spike perhaps trashed the asic driving the display? I had a Uni-T that someone blew the fuses which caused a voltage skip that got past the input protection and took out the main IC. Spot on except it had all the wrong ranges. Replaced the IC and it's been perfect.
Batteries are mixed, rechargeable and ordinary, thats why voltages are different.
@Dave if you can, try giving it a re-flash. What's the betting of a random bit flip in the firmware that causes a jump to the wrong part of code or some crap?
Dave didn't design or build this meter, Brymen did. He isn't going to have the code or hardware to flash it with.
@@stargazer7644 They might have given him a sealed black box flashing tool like you do for untrusted field techs.
Can you measure the input to the logic side to see if something is feeding 5v into it? Although this kind of failure doesn't explain the A/mA weirdness to me. I'm brand new to electronics, green as grass, so I'd love to learn more about this, and if this develops into anything. Will scoping any of the logic test points (if there are any?) help at all? I'm probably talking complete nonsense, but I'd love to see more diagnosis of this.
Swap the asic with another meter, my bet's on a damaged asic... It toggles at a consistent frequency
Just guessing. It is showing the correct voltage intermittently, so the downstream of the ADC is ok. It is weird how the display is messing up which may show an electrical issue, like voltage leaking on a line, or dropping. It could be some kind of ADC communication breakup which is going to the max value.
Definitely look like exceeded range measurement due to different ASIC chip properties in this particular multimeter or it can be software bug too. I'm curious are this glitch is only BM235 specific or it affect all BM2xx line? I ordered BM257s at yesterday and hope it will not show these funny 5-s.
Ive got a similar lower model to this one, a 210A I think but in the Greenlee brand, and so far its been perfect, love it.
Internal bondwire fracture - maybe on something related to the voltage reference. Making just enough contact to supply a reference, then draws current and the failed junction goes non-linear, add a tiny amount of capacitance and it oscillates?
Could you test this with some percussive maintenance while the thing is powered and see if anything changes, or is everything kind of set in stone with the packaging? Would be curious to know if this is a viable diagnostic method.
Dave, check the EEPROM. Read and check its contents several times. I have a 25q series EEPROM with similar weird behaviour, it bit flips at MSB every exactly 16 bytes sent. It was used as a font lookup table memory and a particular character on the display changes, as your multimeter does. I exchanged emails with the manufacturer several times and they refused to believe that their EEPROM was delivered as faulty, they carefully test them, etc. The EEPROM was delivered with factory programmed fonts and I never used it. Just plugged it in my flash reader to check if their fonts are fine and bootloaded the fonts in my code. I had instantly saw the weird glitch, with exact repeatability. I copied its content in a blank EEPROM and it works flawlessly. It's not my circuit design fault, as the manufacturer claimed. Maybe you have similar issue.
I was going to buy one of these, but I got no reply to my question on the EEVBlog Store, or on the eBay listing. So, I figured that support communication would probably also be nonexistent. So, didn’t end up getting one.
My other theory is a short circuit on the zebra stripes connecting the display to the board. My meter alternately displays 0.000 and -0.000. Maybe in this Brymen there is a short circuit between the driver lines of the minus symbol and the G segment of the first digit.
😂 😂 😂 😂 😂 That's Good
Possibly get a new asic and solder that in. As everyone said it is flickering at regular interval. Atleast putting a new one would eliminate some faulty silicon
Could the 5.000 be the reading from the internal reference voltage (after scaling)? It may be a faulty internal multiplexer is feeding it into the A/D converter intermittently? PS. You're more frugal with the flux than a certain Mr Rossman. 😉
I'm guessing there's a 5v rail going into a mux, reference is more likely 1v-3v
That wasn't even one miliPaul of flux...
But its powered by 2 AAA batteries so unless there is a boost circuit to generate 5V, there won't be 5V on the board. It likely runs on 3.3V or less.
It could be a problem with the voltage reference. Or it could be an op-amp with a cracked solder joint or feedback resistor somewhere. Those are the only 2 things that could shoot the reading to the roof and back I would guess.
That's a weird fault, I'd be fascinated to see more in depth troubleshooting...
Seems the best solution is to replace main chip with another good DMM ic's and do some comparison ...
I had a fault in a scope, where a rubber key was stuck inside and made a permanent contact. Couldn´t be seen from outside. So the reaction on any input was weird. Check all rubber keys for correct operation.
Did the broken inductor issue you made a previous video about turn out to be a common issue or was it a one off?
Do the eeprom chips carry any write data. I had an incident where a character byte was accidentally written to the wrong address on an eeprom and it caused the display to output the wrong character when called. Just something to check.
I have a 25q series EEPROM with similar weird behaviour, it bit flips at MSB every exactly 16 bytes sent. It was used as a font lookup table memory and a particular character on the display changes, as Dave's multimeter does. I exchanged emails with the manufacturer several times and they refused to believe that their EEPROM was delivered as faulty, they carefully test them, etc. The EEPROM was delivered with factory programmed fonts and I never used it. Just plugged it in my flash reader and bootloaded the fonts in my code. I had instantly saw the weird glitch, with exact repeatability. I copied its content in a blank EEPROM and it works flawlessly. It's not my circuit design fault, as the manufacturer claimed.
This is a dissappointing video. Essentially "I got a faulty meter back with a weird fault. I tried a couple of things but coluldn't really be arsed. Thanks for watching".
If it should measure the reference, the result would obviously be full scale. But I don't know why it would measure the reference unless it alternates between signal and reference using a switch.
Hello, did you try freezing the processor a little bit to to see if there is any change?
What happens when you switch ranges? Let's say, 50V instead of 5? Will it flash second digit or first?
It is an auto-ranging multimeter. But maybe applying a higher voltage might help.
@@devrim-oguz There's a manual ranging button in the row under the display.
I've used a Mantis for years and I agree they are the 'Ducks Guts'....RM3 looks almost unsoldered on the left side (11:39) got to be a problem
It could be a both programmatical and hardware error. Because the digit 5 blinks with a constant pattern. (It blinks 4 times then stops for a while) Why don't you replace the MCU to see if it helps?
I'm not sure if' it's affecting the other digits, since you didn't connect up anything to measure, they show as zeroes (with the least significant digit 'bobbling'). Measure, say, 1.234 volts and see if it alternately displays "1.234" and "5.000"
It does measure fine, he said that then it alternates between measured voltage and 5.
Very nice that you sent him a new one
My first guess on seeing its behavior, and the fact it happens in any mode, is that it has dual ADCs that it alternates between to get measurements and one of them is not working anymore for some reason.
@eevblog2 I would probably take another meter out of the scrap bin and swap the processor. Not too many pins so shouldn’t take too long I don’t think. Would be interesting to see if it persisted, might be the silicon.
I would change the processor even without calibration just to rule out or confirm the processor failure theory.
The fault produces a repeating fixed pattern not random, internal cpu crosstalk fault??
Was about to comment the same. Doesn't look random, there's an actual pattern for some reason.
#12 Freeze Spray the ICs
#13 Send it off to the hydraulic press channel and get lunch.
I have had meters that a voltage reference ic takes a dump after the meter gets hit with high voltage...dose similer things
might be an issue with one of the registers getting an improper clock and/or ref voltage.
It's a consistent pattern, that's for sure.
Why is this video posted on the secondary channel? I usually don't watch EEVblog2, but this one seems perfectly suitable for the main channel that I'm subscribed to.
I think he said it's because he isn't able to fully diagnose or fix the problem.
8:10 - The surface of the black component in the lower screen just right of center.
R13?
The etchant trap thing still matters if you use the cheapest, worst process available on the market for high volume production, where you still can't afford any failure.
Not that it matters, but what firmware is in that meter? I noticed the RM1, RM2, RM3, and RM4 positions on the PCB. They look like option jumpers. I wonder what would happen if you played with them. I have two of the BM235 meters and they have different firmware revisions. That is one thing about the 121GW over the 235. Easy update of the built-in firmware.
Is it possible its switching in a reference voltage for self calibration and the switching signal is not functioning?
Or is there a pin on the ASIC that's pulsing at that frequency? Is there any ghost signal coming out of the input connections?
My $0.02 would be to thoroughly clean the rotary switch and contacts... (My guess is that the CPU is being asked to be in multiple 'modes' at the same time due to leakage between switch contacts)
Shot in the dark here - Is the reference crystal on spec? Can they be fractured internally yet still 'function' ? Will replacing the crystal still require re-calibration?
Apply high heat to processor. And see if problem gets better or worse.
Press the selector down and it starts to be ok...The usual fault on Uni-t ...only some contact lub in the selector, turn a few times and, problem solved.
Always very interesting videos. Thanks Dave for the time you spent for us. I just have a quick question : I saw on your PCB some resistors that is labeled RM in place of R. What is the meaning?
Probably a stupid solution, but have you tried connecting it in parallel with another multimeter? Maybe it's working fine but leaking power to the measuring side somehow?
The meters that get sent back for repairs, do you then sell them on at a reduced price? If so ide be keen on one.
I would be nice if you would try to swap chips, just to confirm whether the fault follows the main CPU or the PCB, calibration wouldn't be important for that.
The alternating between 0 and 5 does seem to have a bit of a rhythm to it...no idea if that's relevant.
Not just a bit, it is completely regular.
You don't need to recal the meter just to prove the processor is the fault. Just swap it out with one from another busted meter and see if it stops blinking 5.
Dave please don’t leave it there. Please get them to send you a new processor, swap it over and see if it makes it work. If not please please please look at repairing it
I'm looking to buy either the BM235 or the 121GW, just not sure which to get. I mostly do some Auto/Home DIY and minor electronics repair, nothing too deep.
The BM235 is an excellent meter. I have 3 (one at home, 2 at the office) and they're the first meter I tend to reach for. If you need more advanced functions then the 121GW is good, but I generally prefer the BM869S or a bench meter in those cases. The 121GW is pretty impressive for feature set vs size.
The BM235 in that case. No need for anything fancier.
@@EEVblog2 thank you Dave, cheers
Ordered the BM235
@@EEVblog2 Got the BM235 in today, excellent bit of kit. Had fun testing it out on a few things. Thanks Dave!!
Lenny from Boston USA (Soon London UK)
If the measurement is spot on, could it be the communication between the processor and the lcd?
if it's an spi or i2c, could the processor be reading the values correctly but experiencing noise when sending that over to the lcd controller?
5 is 0101 in binary btw
Someone inside is sending an SOS before letting the smoke out.....
Man, that microscope is fine, amazing quality.
What happened to "always check power supplies"? If the processor is given an out-of-spec supply voltage then it may misbehave.
So here I was, thinking it was a range overflow issue (+1, 0, then -1 manifesting as 5000), but noooo... you had to go and rain on my parade, saying it's a 6000+ count meter, at the end of the video!! :D (totally j/k)
Seems like a internal silicon issue to me. Not sure how the main processor and LCD driver communicate and the format of display data, but it could be just a display data corruption considering it only affects the most significant digit and in a repeatable way.
Not for economic viability obviously but rather curiosity / makeshift QA department, I'd have swapped the processor with another one. Especially should another dodgy meter turn up some day. Note batch / date code of affected processors, it could be a bad lot. But it would exclude software issues.
At round about minute 10 you can see a hand above the input jack, look like making high 5. I guess this is what the meter want to say. 🖐
I imagine this sort of crap happens all the time, for example with cheap routers. Customers think "oh it must be me" or "the ISP are throttling me" or "the damn kids are on the WiFi again", when in actual fact one of those poor processors got tired on running at 1GHz with zero cooling, full of dust and knocked off the table once every couple of months. These kind of "half" failures are probably happening everywhere. One major problem is that software makes the assumption that the hardware is always working correctly, for example that a bit flip is impossible and an ADC will only ever read between a value range, when in actual fact hardware failures are actually common place.
Depends on the application and to what cost the equipment is being manufactured.
A home broadband router needs to be built as cheap as possible as a first priority, hardware and software wise.
A piece of life support equipment will be built to a higher specification and cost may be only a secondary factor.
Question: I see that some traces on the PCB are without soldermask the whole length. Why is that? What benefit does it provide?
Lower trace resistance due to the extra solder metal, I think Dave made a video about that practice.
P.S. found it
ua-cam.com/video/L9q5vwCESEQ/v-deo.html
@@icestormfr Neat, but here they have no solder on em. Just bare traces.
@@sysghost ahh, ok, I thought you meant the traces seen at 8:37
@@sysghost ah, I see which one you meant. Hmmm, I know only two other reasons, but the traces don't look like that would apply here...
(1. to improve heat conductivity to air, 2. acting as guard traces/rings, actively driven to the same potential as they are guarding, to get very high insulation/ultra low leakage from other parts of pbc, e.g. ultra low current/charge measurements => leakage flows over the PCB into these traces)
Less capacitance perhaps? I've seen that kind of thing in RF voodoo situations before.
Just a random thought. But could there be a 5v rail connected to a mux? I assume it's in package package on die so there's not an external chip to replace but might that be doing it?
Very interesting issue. Following with interest.
Maybe give it some crazy vibration and see if anything changes?
What about an unstable crystal o/p? Worth a quick probe ??
14:40....If you get a proper reading when an input is connected and an erratic one when nothing is connected....suspect an input ground fault or an input fault of another kind.
p.s. could this be made into a how to use your oscillisucope lesson to trace down the fault?
Is there a bit from the ADC that is flickering?
I think it is because I've seen many faulty ADCs doing the exact same thing