EEVblog
ฝัง
- เผยแพร่เมื่อ 26 ก.ย. 2024
- Join Dave step-by-step as he attempts to find the problem with a non-booting Aiphone GF NS building intercom system. Will a reset chip make a monkey out of him?
And as always, a trap for young players is explained.
Brochure:
www.zsing.com/d...
Install Manual:
www.aiphone.net...
Setup Manual: www.aiphone.com...
Datasheets:
Mitsubishi M16C: pdf.datasheetca...
24LC128 EEPROM: ww1.microchip.c...
RS232 driver: www.ti.com/lit/...
EEVblog Main Web Site: www.eevblog.com
The 2nd EEVblog Channel: / eevblog2
Support the EEVblog through Patreon!
/ eevblog
EEVblog Amazon Store (Dave gets a cut):
astore.amazon.c...
Donations:
www.eevblog.com...
Projects:
www.eevblog.com...
Electronics Info Wiki:
www.eevblog.com...
I don't see a diode in the RC reset circuit you put in. The data sheet shows that there is one there to bypass the resistor to let the cap discharge quicker. When you turn it off and then on so quickly, could that be causing the latch up since the cap isn't discharged below 0.2v before the power is restored?
That reset chip (5-pin SOT package) must have been monitoring the power rail for stable power and with the bad cap in the power rail it never became stable and so it never brought the controller out of reset. I think if there were maybe a smaller cap and a diode across the resistor to discharge the cap quicker the thing would boot every time with no over-current.
The datasheet for the controller says the reset should be held below 0.8v for at least 20 cycles. At 4MHz that wouldn't be very long at all but no harm in having it low for longer. The cap just needs to get below 0.2v before power is restored after power off. A power on reset chip can hold the reset low for as long as 250ms, so I think this reset circuit is now the trouble here.
i like the repair more then the normal tear down, where tear down is a blast thru and this is way easier to follow and more understandable also for people not really into electronics
Please Dave, don't leave one of your most exciting episodes without a sequel!
thanks Dave, same unit, similar black LCD line, swapped both caps for 5v reg, and lifted reset pin which was grounded, voila it boots, made an rs232 lead for 2.5mm jack and I'm happy it's worked. Thanks for the insight, appreciate it.
Dave DPS
It's like watching an electronic version of House xD
Hi Dave. Thanks for the awesome videos. I had the same experience troubleshooting a treadmill motor board just month ago.I found the faulty components and replaced them right way. The micro-controller on board was just working ok. After a short break and back to troubleshooting i powered the board up and guess what! chip never booted-up. Next step was to check 5v rails and all was just fine. I changed the crystal just to be sure it dum works.but i had no luck. After getting the data sheet for the micro i started with the reset pin and traced on the PCB, the voltage on reset pin was just 1.3v and normally it should be around 5v, down the line it was hooked to the collector ofA337 transistor that has to pull the pin high (5v)with a resistor divider on base and the emitter to 5v. I decided to add a 10ohm resistor on base pin (A733)to 5v, to allow 5v flow from emitter pin to collector to reset, and wow the A733 started heating up really hot due to high current flowing to reset pin. I got the micro out of PCB and did a standalone test on solder-less test board and also added logic components that a micro needs to boot. Lastly i measured how much current the reset pin was consuming and wow 250mah! Really!The chip never booted but was getting kinda warm. Hmm
I got the micro out of PCB and did a standalone test on solder-less test board and also added logic components that a micro needs to boot. Lastly i measured how much current the reset pin was consuming and wow 250mah! Really!The chip never booted but was getting warm. Hmm
When I have power going 'somewhere' due to suspected low resistance or short (like when you had 90mA+ during the direct pull-up through the meter) I just shine the thermal camera at the board and see who's holding the line down (or up).
Has saved me some serious pain :-)
Dave - I enjoy your repair videos, regardless of success. The process for troubleshooting and isolating issues is the fascinating bit.
Jolly good show. I learn no matter what the end result, so keep them coming Dave. I would hate to think this footage would be dumped because the device was 'unrepairable'.
I love this type of investigative troubleshooting even if the end result is not what you want, it still makes a very interesting video.
Most people who commented spotted the same things that I was wondering about but very few people had any useful suggestions especially concerning the high current draw. It seems to me that the current is finding a path that should not be open which points to some other component going bad somewhere else, could be a whole string of faults in there.
That moment when, out of nowhere, EEVblog gives you a close up of a dead and crushed bug.
Even when the task is not successful you can still learn lots from it.
If you're struggling to read part numbers that are waterproof coated, like in the video, try using light of just a single wavelength. You can pick up red, green, blue laser pointers, and so IR bulbs and UV lights can be used. Just trial and error, but it's handy to have at least one of each in your workshop. They are really cheap and can be extremely handy in reading difficult things. You'll be amazed how some things look like they have no writing on but under the correct sort of light appear crystal clear.
You’re still a rockstar Dave! Never would’ve gotten to the resistor pull-up/pull down whatever it was.
Check the processor data sheet for the spec of the reset signal, it might have a maximum rise time which is why the simple R-C circuit won't cut it. I've had fun and games with them not resetting properly before now.
Alan Cordwell Yes, but that's the least of my worries, there is a serious intermittent over-current problem.
EEVblog Well, you never replaced the proper reset generator with a new working one. I talked to a friend the other day and he said, those act as Brownout detectors and are designed to deliver a well-defined RESET signal when the power is cut shorty. The reset generator in question looks like a SOT-23-5 package and might be a MAX823/824/825, for example.
EEVblog If you want to save the building sup. some money, you can buy a new LCD module off ebay with the same model number for around 200 USD.
EEVblog
A slow rise-time on the reset line may very well put the processor in a really bad state where it draws alot of power.
I agree with Alan Cordwell. You should restore a proper reset edge before troubleshooting the power issue.
The original reset chip was probably fine, just put it back in there, but don't connect it to the reset line on the board, since the board seems to be partially shorted. Connect it directly to the CPU using your jumper wire.
It would be very interesting to know if the reset line continues on beneath the processor. Perhaps it goes to other devices as well. Such as the LCD?
EEVblog If the processor isn't reset properly, it may go to an invalid state and drive some pins high/low, that are not intended to go high/low. This might explain the high current draw of the system. I'd suggest refitting the reset IC and see if it works then.
Win! The diagnostic vids are how the rest of us learn :)
Probably the voltage regulator was causing the reset chip to stay low. Put the reset chip back in and recheck.
Very nice troubleshooting Dave, we some times cant fix everything perfect, but least you tried. Great work.
Great video, and I couldn't agree more that the outcome is not as important as the troubleshooting steps and in spite of the fact that the device did not function in a useful way it was fun to watch and I did learn a lot. thanks Dave and please keep these repair videos coming.
I enjoyed that demonstration of tear down security immensely ;)
Hello EEVBLOG.
I think I had found a similarity in a research which I had done as student. I had study the aging of the sequentially digital circuits. So I had three set of two 8bit FF register, the one part of the set is made with discrete logic MSI chips and the second one was made with a PLD. I had study the starting condition in both of this device. By had the reset off, and switch on the power for some time, store the output value of this two devices on a logger, and repeat the process. I did that for about 1 month.
The result was the following.
Initially both of this is start with random values, and the study of this values is give a rectangular distribution in both of that devices. But after of some days the initial values start to take more some values than others, the distribution of the random initial values goes to be Gaussian. First, that happened to PLD, and later to MSI chip (remaining, both CMOS technology chips, temperature of the both device was constant 25°C and also isolated from the outer world, actually I had it on a oven, the distribution analyses had be done with than sampler of one day). So, as a conclusion every sequential circuit had a unique tuning destination (I think you understand what I mean). And this one is actually so unique where is very difficult to make any approximation on that. Even, device with close parts numbers had totally different result.
So that one my Story about my study on the aging silicon chips.
For your problem with the door-telephone, you can try to apply a sequence of pulses on the reset pin of the μC. Probably it's will make it to wake up from the latching statement of the internal Software reset circuit.
And some words for this BLOG.
It's the first time where write to your blog, but it's one of the most beautiful blogs on the concept of electronics. Always you try to make the things Easy, and this is good, very good. One of the thing where I learn on school of the life is: "The most brilliant people on the world, was brilliant not because they know many complex things, but because they make this complex things so easy where even a 10 years old child can understand it". This I can say it with one word as "Μεταδοτικότης" or "transmissibility".
As you say "Thumbs up".
I'll call this a win. Always enjoy your repair videos, Dave.
Its seems to be more trouble than its worth, good video
The M16C Datasheet clearly states (page 14) that "When the supply voltage is in the range where operation is guaranteed, a reset is effected by holding the reset pin level “L” (0.2VCC max.) for at least 20 cycles. When the reset pin level is then returned to the “H” level while main clock is stable, the reset status is cancelled and program execution..."
So you maybe really replace that 5223 thing? It also says "Free Delay Time Setting [...] Delay Time Controlled by external Capacitor" in the datasheet. So your RC might just not meet the specs for a proper reset here.
Maybe you could trace where these LEDs are going? The M16C seems to have some 8 Channel DACs and all the rated I/O current is only 5 mA so there probably is a driving transistor for the LEDs or something?
Dave, you had the reset dip switch on all the time! I guess that's what is pulling the damn reset line low.
***** No, the DIP switch goes to an I/O pin.
Great job Dave. I love the troubleshooting videos you make. Great learning experience for me.
that was good one, gave me some ideas what to look for in my current repair :)
I loved watching this video as it followed my technique but maybe in a different order . I always go for supply then reset then data to display .thanks for a great video
Ive seen these processors used in electronic oven controls, and some washer/dryers, and i can say on occasion that I have seen these processor spontaneously combust.
A wild guess: First it was the ridiculous amount of ripple on the 5V rail keeping the supervisor, and later the processor, in reset. Maybe connection through DMM (in low impedance mode?) smoothed out the supply voltage enough for the processor to start up, direct jump with resistor did not work. Then when the 5V rail was fixed, the reset timing was wrong. I guess the supervisor chip wasn't there just for "good engineering" after all..?
As someone who doesn't know nearly enough about this stuff to provide practical advice, (though your vides are amazing and I do learn from them some.) as the 2 LED's seem to be sporatic in behavior, I'd check out the circuits they are involved in, a component attached to them may be somewhat sickly, if not dead. Least that's what I can come up with.
In a situation like this, since you cannot get a replacement, I would whack in a PIC programmed to wait a few mSec on power up and perform a reset on the main processor. You could keep it old school with a 747541 and an RC time constant to do the same thing. There's a million ways to fake a reset on a micro.
The troubleshooting was definitely interesting.
Maybe the oscillator? The processor may be picking up the frequency from inside the multimeter?
Looks like brown out, I'd replace the IC you removed with a voltage monitor. Had a very similar experience with an entire product line one, check Microchip AN686. A quick fix would be a 555 timer to hold in reset until the power supply settles.
Some thoughts. 1) get rid of the damn *^&! conformal coating first, then desolder the R/C network and the uP. 2) Clean the board thoroughly. 3) resolder the R/C network, then power on the board without the uP. Poke gently but forcefully, around the resistor; you might have a break in an internal board layer! 4) with the uP isolated, measure from pin 12 to Gnd and Vcc in both directions with a DMM that has a "low voltage" mode and see if the protection diodes on the reset pin have failed shorted. 5) If, not, resolder, and try again. If it still fails, well... parts donor time!!
Best video ever! Thank you so much for going through the troubleshoot steps.
Dave, really enjoyed the repair video for the methodology and agree the unit was both beyond economic repair and potentially dangerous. This time it was the journey, next time it might only be the destination. Keep up the repair videos and we might eventual get a video with both. The reason I originally got into electronics and became an EE was through repairing a non-working Apple IIe at the age of 15 yrs. It cost me $12.25 in components to fix it and was a really empowering experience. I'm sure your videos are doing the same.
This means that the LCD Display is initialized. Power up the LCD the first Line is not visible but after the Command Power ON this will displayed. The Cursor is not at position 1 - left upper first line. Measure the EEPROM first that looks like a death counter. If you measure a Communication on the EEPROM first after Reset and then show nothing on the LCD you could be sure it depend on the information in the EE. If you have a younger System like this you can copy the information from the other 2nd System to this EEPROM and you'll see that is working. I think so I have a problem like this in a other System the System was from a Time Stamp Company.
Well, I called a 9/10 dead cap in the power circuit on twitter, glad I was at least partly right. The overcurrent issue does rather point to an unhappy chip. Either it's expressing serious unhappiness with the timing of the reset signal from the bodge, or it's in some more serious physical pain.
really enjoyed that video Dave,
I feel your pain when you got conflicting information that led you astray.
I felt I was with you all the way with your thinking and procedures so that for me is reasuring.
Electronics isn't an exact science but it should be.
Maybe the microcontroller's reset pin has some internal diodes for protection that were wrecked somehow and are now partially shorted out? Possibly the lower protection diode.
I realize this is a very old video, and I don't know what the outcome of the fix was, but the number 1 dip switch which Dave mentioned as a "reset" is turned on again at 35:45.
I'm patting my back for calling the brown out reset when you thought the uC was dead. Seems it is afterall but at least that got it to boot
I like it how Dave uses a different multimeter every time. (:
You left the DIP switch on! -- Nah, just kidding. IO pin and all that.
That board is becoming a bit of a mess with all your mods running around! You're a trooper for taking it this far, good show Dave.
Hey, a non-fix is just as informative as a fix. For example, I had no idea there were chips specifically designed to handle the reset state of a CPU.
Comms between controlling station and keypads may be using DTMF A,B,C,D codes not normally on consumer 0-9 + * + # keys.
I'm sure you probably did off camera, but did you check the supply at the processor? Possibly being powered through the reset pin because no supply at the vcc pin?
Dave, an idea: Well first, great job troubleshooting, a very logic approach and you found the problem area very quickly. Looking at the data sheet, I suspect other pins in the vicinity of the reset pin may be affected. Check the voltage level on the CNVss pin and the BYTE pin, both two pins over from RESET. They need to be at a particular logic level for proper chip operation. The CNVss trace runs suspiciously close to the RESET pin trace. I don't know why CNVss and BYTE are connected through a 10k, but that means there must be a trace hidden somewhere that is connecting those to Vcc or Vss. Perhaps a dodgy trace under the uC or a defect in an inner layer is affecting that entire board area. Good luck! You better get some free rent out of the bastards if you fix their keypad. :)
I hope you pointed to this video when you invoiced the strata manager.
As AIPHONE specialist GF-NS is old model GH-NS newer model and the last virsion is GT-NS but unfortunately it's totally not compatible but I think if you ask AIPHONE agent request PCB XC-884A it may be available at factory
1 of the only channels where i have watched every ep
Great video as explains well learn skills needed from it so even though couldn't fix it I have learned a lot from it, so a bit like Apollo 13 in the fact I would call it a successful failure, as failed to work but a greater good has come out of it educating people.
Dave, did you try to toggle the reset DIP-switch again after the 'repair'? I see you left it on, while the manual stated that it should be off...
Robbe Derks Yes, makes no difference, the DIP switch goes to an I/O pin. Some issues either way.
EEVblog
Have you checked if that dip is really making/breaking connection? In my past ive seen sticky or failed dip's.
if the switch is not full of that mask it might be corroded to hell ?
Dave... If you don't get to put it back into operation I don't think you should be able to call it a repair success even though you got it working.
I had to go google SCR latchup. I had no idea that was possible, but would certainly explain the overcurrent situation and the bright LEDs. The question is then, what's causing the latchup, and why/how is it doing it consistently?
Not a bad job for the little information you had to go off of.
leds on full as out put ports on micro turned on when processor locks up.
try making reset line from psu input voltage.
Sounds like old silicon latching... transistors start acting like SCRs when they get that old...
It's dead jim!
Love the repair vids.
I've seen these exact units in heaps of apartments, would imagine that there would be some spare parts for them out there.
operation is a success the patient is dead :-)
smbrob Good way to put it!
Moose *is* fine! Finally dead!
I would have cleaned up that power rail first. I reckon a new cap would have got that power-on reset chip happy, and she would have started first kick.
There looked like cracked solder joint on back of that board early on in the video.
Just wanted to say about the reset dip switch but found out that a tone of people allready done that
26:41 Or it could be a brick chip. A chip that will brick your device after cartain expiration date.
Sadly it's verry common these days, our washing machine was bricked this way and i found this tiny PIC on power suply circuit.
Funny the "bad circuit desig" was a dead giveaway, cause there was a backup battery right next to a real time chip on power suply pcb. :P
Dave, you said in the video that the reset trace doesn't go anywhere, but maybe it goes to a via under the chip. There could be another component pulling reset low. For example, what are the optocouplers for? If I were designing this thing, I might use an opto as a means for resetting the unit remotely.
these episodes should be done more
I bet its something with the regulator. I would have checked the 5V rail while the unit was drawing 200mA
Awesome video, I wonder, for expensive electronics where space is not too much of a concern, why can't they just socket as many components as possible in order to make repair easier?
By the way thats a Aiphone GF is over 10 years old. You could of found a used GH system in Ebay and replace the unit with it. The new system is Aiphone GT but that model is not compatible at all.
Try resoldering the reset chip back down, not using the suspect trace
Dude, couldn't the "random" power-on behaviour be related to the missing reset chip? After all, it's supposed to take the CPU outta reset only when it's supply is stable.
Jaroslav Malec Sure, but in the end that's not the point. There is a serious intermittent over-current issue.
EEVblog that reset pin can have some short circuit internally, but you prove that it can be override it, , but it will anyway require a well defined reset signal, not just a RC circuit, that is the reason why it sometimes boot up and sometimes don't, and sometimes you gets a dirty boot up, the leds for sure are PWM controlled so, it is normal f sometimes it hungs up in HIGH state, the leds gets brighter, that will overdraw more current from the I/O pins that control those leds, so be careful, why it seems to boot up more times when you do it manually, because you are inserting noise, and that noise with random luck will meets the requirements of the chip for reset it.
flyguille You're right - that's what I pointed out in the comment, but Dave also has a point. The device stopped working because of a short circuit failure inside the IC. This not-so-clean reset thing may be causing the excessive draw due to improper startup, but that's not the point of failure.
Jaroslav Malec I once did a repair like that, and the chip worked during years, if there is an internal shortcut, bit the pin is stil sensing it, you can try to cockup the short cut, sometimes it is just matter of overheating that exactly point, lets it broke, and test again if it is still sensing, if the rest of the circuitry works, it will work for years
Ohhhhh I'm so disappointed that he didn't get it working!
I was thinking maybe the VCC of the CPU was high impedance and delivered too little power. When you shorted Reset to VCC, the CPU probably drew it's power from the Reset pin.
Usually reset pins / circuits inside micros include MOSFETs in their innards, maybe they got FUBAR in this exercise?
this was very interesting for me, i love fault finding, and love all the steps you took but as you say, if its not going to be reliable then its still no good! fair play for all the steps you did take though and keep it coming :) loving the channel!
Hey Dave, why didn't you use a thermal camera? If the device takes more power than its suppose to, the defective component gets hot. If its the processor, you're right.
The installers should really have supplied a spare panel(s) and certain auxiliaries at the time of ordering or made it an "futures" option...but then again why would they.
Love this video. Big thumbs up.
Google "CAF PCB failure". Conductive Anodic Filaments are an increasingly common failure mode of FR4 PCBs. Copper filaments grow along moisture trapped inside the pcb during processing.
Maybe related to this wierd problem.
Maybe you should have replaced the cap. first and kept the reset chip in there. The ripple on the 5V rail at 16:37 might actually be the issue.
(not so) Modern processor need a clean rising edge on the reset signal, else it could be interpreted as brown out reset ... If you wanna demonstrate, why don't use a circuit based on RC + 74HCT14 ... or simply use a pulse generator/another brand reset chip
Thou check voltages!!
what you are seeing is the processor being powered through the reset pin, because it's 5v supply is missing.
Well. This Video is old. I love to play around with microcontrollers und simple electronics, but here i don't understand much. and still its sooo interesting!
you can remove coating with right chemicals easily
EEVblog btw. I thin the marking is bv3223 which curiously leads in google search ("bv3223 datasheet") to an as1917 supervisor ic ...
"excuse the crudity of my finger" I died laughing :D
Old Processors are very susceptible to starting before the voltage was acquired this is why they used to use the supervisors a lot. This is why probably 1 of 5 reboots worked. So a pull up in reset alone will not work. If you add a push button to do a pull up and a Resistor to GND you can test this. And in this case you will require a supervisor.
What interests me more than anything here is why those LEDs are so bright. If they have current limit resistors then that should not happen ever. Does the processor have optionally current limited I/O pins that are driving the LEDs directly? Are they driven from switching current regulators? What could overdrive them like that?
Probably PWM'd off the chip.
Dave McAnulty And the extra current is from the hyper-bright LEDs stuck full on when the CPU resets.
EEVblog
Hey Dave, couldn't you follow the current to see where most of it goes when it draws the 200mA? You do have those special probes for following current through traces. Maybe you can spot another component that is failing.
What do you think ? Worth a re-try ?
Also, you don't use those special current probes often and the board doesn't look too complex, might be interesting !
That company had an iPhone three years before Apple!
Intermittent faults, the most favourite fault of all time, cos they're hard to find and a pain in the arse to fix... :P
Very informative video Dave, you gave a good attempt at repairing it!
Just wondering... did you put DIP switch #1 back into the correct position after you tried the reset? Also, it might have been worth seeing if the processor was temperature-sensitive, which would rule out other parts of the cct maybe.
Thanks for the upload. i learned a lot today
You could rig up a 555 monostable with a resistor and transistor to kick the reset, possible?
It didnt go mad untill you applied that electrolitic across the supply.
very cool video, thanks for sharing. Too bad it's getting more and more strange along the way.
Great trouble shooting video, bravo more.
I see a lot of Aiphone stuff here in Phoenix.
Has the capacitor on the reset circuit failed short, maybe?