CPU card - where to start ?!

[Cozzmo]

Hi all.

Hoping to get some pointers from you experienced people. I have this CPU daughterboard that plugs onto a main board which is used for MDB coin mechanism testing used in vending machines.

Most of the time a dead RTC 2032 battery is the main issue. I have a couple of these boards however which draw an insane amount of current from the 2032 and do not operate as designed.

Normal board draws about 5.5 µA and the bad boards are drawing 1.5-1.8 mA !! It’s a complex board and likely to be 4 layers as it’s a real bugger to solder the battery in/out.

I imagine the RTC and the SRAM would be the only parts getting supplied with backup power ? Or would the main micro also be getting power usually too ? My thoughts have been to remove components and see if the current draw suddenly drops. But I thought this might also cause other problems which may mask the real problem. I have one that will power up but fails to ‘clear’ (I assume it does a clear of the SRAM??) as it thinks the EPROM (removed in the picture) has been upgraded and it needs to clear memory or something. I am not sure if a faulty RTC, logic,SRAM or main micro would be the likely cause for the high current draw. The other board seems more dead as it gives no display, the unit just beeps.

Thanks for getting this far. I went to fire up my scope to start probing around on a ‘good’ board but the smoke escaped from inside and I haven’t had a chance (or the space) to get it on the bench for dissection.

[NivagSwerdna]

imagine the RTC and the SRAM would be the only parts getting supplied with backup power ?

I don't see a dedicated RTC chip so assume the uP is doing the time keeping so a regular short somewhere?  If the damage was due to external abuse then an obvious approach would be to heat off the 74HC245 and see what happens.

[wraper]

Sometimes it's as simple as leaky diode. Often it's a dual diode in sot-23.

[wraper]

imagine the RTC and the SRAM would be the only parts getting supplied with backup power ?

I don't see a dedicated RTC chip so assume the uP is doing the time keeping so a regular short somewhere?  If the damage was due to external abuse then an obvious approach would be to heat off the 74HC245 and see what happens.

I don't think that 74HC245 has any relation to the circuit relevant to the problem. I'd guess this PCB has double sided load, what's on the other side?

[james_s]

Can you tell precisely what the battery powers? It will be as little as possible, usually an RTC chip, possibly a SRAM. I would definitely look for a diode as mentioned above, I encountered that fault on a PC motherboard years ago. That poor little battery might be trying to power up other parts of the circuit.

[SeanB]

Going to guess the RTC is IC4, as Epson made a whole range of RTC chips with built in oscillator, and the RAM is IC10. Most likely bet is the ram is faulty, as that is a common failure of them, plus the 32 pin Ram has some rather obvious heat damage where the die is in the package. Simple test is to apply power and feel the chip, which will be a lot hotter than the rest if it is faulty, though with a failing RAM chip it might just have some stuck blocks and draw only a small amount of extra power.

[Cozzmo]

imagine the RTC and the SRAM would be the only parts getting supplied with backup power ?

I don't see a dedicated RTC chip so assume the uP is doing the time keeping so a regular short somewhere?  If the damage was due to external abuse then an obvious approach would be to heat off the 74HC245 and see what happens.

I don't think that 74HC245 has any relation to the circuit relevant to the problem. I'd guess this PCB has double sided load, what's on the other side?

The board is single sided only, nothing on the bottom except traces.

[Cozzmo]

Can you tell precisely what the battery powers? It will be as little as possible, usually an RTC chip, possibly a SRAM. I would definitely look for a diode as mentioned above, I encountered that fault on a PC motherboard years ago. That poor little battery might be trying to power up other parts of the circuit.

I haven't been able to tell as the thing is so densely populated, but the battery goes to the opposite end of the board, through the jumper in the middle, through the through hole resistor, powering the Max 691A which then goes to the EPSON RTC but i haven't traced it much further.

I bought a couple of replacement SRAM chips as i thought seeing as it couldn't perform its 'clear' on startup that it was possibly the logic driving the clear or the SRAM itself.

[Cozzmo]

Going to guess the RTC is IC4, as Epson made a whole range of RTC chips with built in oscillator, and the RAM is IC10. Most likely bet is the ram is faulty, as that is a common failure of them, plus the 32 pin Ram has some rather obvious heat damage where the die is in the package. Simple test is to apply power and feel the chip, which will be a lot hotter than the rest if it is faulty, though with a failing RAM chip it might just have some stuck blocks and draw only a small amount of extra power.

What looks like heat damage is i think sticker residue from the "QC PASSED" sticker that was on it. I removed the sticker to find out what the chip actually was. Your assumptions are correct as to the chip functions. I'll try apply powering for a bit longer and see if i can feel anything get warm. I don't have any kind of thermal imaging to easily see subtle temperature changes.

[Cozzmo]

Sometimes it's as simple as leaky diode. Often it's a dual diode in sot-23.

The only SOT-23 is near the MAX691A but its labelled K6C which googles up as a BC817 transistor.

[Cozzmo]

Alrighty ! update time, first off you guys are awesome, great suggestions to give me a direction.

I sucked the SRAM chip off one board and immediately the current draw dropped to ~2.5-3.0µA ! So I figure that's definitely the culprit ! I need to get better with the hot air though, even with foil around it i still melted the jumper just above the SRAM.

I suspected this might be the case only based on the one board that did still work but failed to do the "clear" on initialisation so i bought 2 replacement SRAM chips a few weeks back which arrived on Thursday.

Bad news is NEITHER of the replacement chips work at all. The board is basically 'dead' with them installed. Re-install the "semi-working" SRAM back onto the board and its alive again but with the same fault. The other good news is the second board that seemed totally dead, when i swapped the semi-working SRAM onto that board, it comes to life but also fails to "clear" so i know that the fault is that SRAM and now i guess i should find a better source than eBay for replacement chips.

Once i get them i'll update whether that totally fixed it or not.

[wraper]

so i know that the fault is that SRAM and now i guess i should find a better source than eBay for replacement chips.

Did you get exactly the same part? If yes, likely they are fake. Do the markings look suspicious, rough or just different surface on top of the IC? What's the model BTW, it's not clear from the pic.

[Cozzmo]

Yeh exact same part but the writing on the top is a smaller size than the original and the pin 1 mark is also different. I thought it’s different batch but I imagine they are fake now as you suggest. I am a tight arse like most and I wasn’t prepared to take a $50 gamble on a pair from UK. With some digging I have found suitable replacement parts RS Components and ordered a pair for a little over $30 posted be here in a week or so.
Original part from one board, semi working, BSI BS6LV4006SIP55. Other board that was dead had a R1LP0408DSP-5SI which RS has. There was also a cross reference with a Cypress part too but I ordered two Reneseas chips.

[Cozzmo]

two original on top and two new that dont work on the bottom

[james_s]

Fake parts are a real problem, there's a good chance it's a mask ROM or whatever other useless IC someone found a batch of that have been re-marked as SRAM.

[wraper]

IMO they look like blacktopped. Try to rub them with acetone, nail polish remover or alcohol. Logo has wrong proportions and font as well.

[Cozzmo]

Very frustrating when I have no real way to test//verify it. But I think once the ‘genuine’ parts arrive, if they work as I expect I will press the original seller for a refund. I’ll have a go at cleaning the tops off later tonight.

[wraper]

BTW, if you used hot air on them, in my experience blacktopping will hold much stronger than on non heated ICs.

[Cozzmo]

Rubbed it pretty vigorously with both alcohol and PCB cleaner, no change in the visibility of the markings. If they do turn out to be fakes and the 'genuine' ones do the job, at least they didn't damage the board. Will find out next week.

[wraper]

Rubbed it pretty vigorously with both alcohol and PCB cleaner, no change in the visibility of the markings. If they do turn out to be fakes and the 'genuine' ones do the job, at least they didn't damage the board. Will find out next week.

There is no question, they certainly are fakes. Package does not match original IC and datasheet, logo has substantial differences. If there is no blacktopping, then they were originally marked with ink. Then it was removed with solvent and new laser marking applied.

[NivagSwerdna]

The S on the originals has a very distinctive hook on the lower left hand side.... seems to be missing in the replacements.  Evil.

[Cozzmo]

UPDATE!

So, after making a stupid decision in order to receive them faster, have them couriered from RS to home with Couriers Please, that ended up delaying me 4 days... what a stupid system they have in place for 'missed delivery' ... anyways ...

Chips i received from RS Components were identical to the original R1LP0408DSP-5SI.... and what do you know... BOOM perfect. Worked first time. No fuss, no muss !! So i'll be emailing the seller of the original BSI chips and be asking for a refund as they're clearly defective.

Thanks everyone for your help, more hardware saved from the dumpster and very useful hardware at that !

[james_s]

Good to hear. For curiosity sake you might try reading one of the defective chips in an eprom programmer, I suspect they're some sort of mask or OTP ROM that has been re-labeled as SRAM. Could be any number of other things though.

[Cozzmo]

I Do have an EPROM reader but that chip is not specifically supported. I don’t really know how to make it work (I’m sure custom chip/profile can be done?) and for the time I’m not really fussed to mess around with it.

[james_s]

Fair enough. Since you don't know what chip it is anyway it doesn't really matter though, I read unknown chips occasionally and just use one of the largest sizes in the same pin count, for reading it doesn't really matter. The specifics only come into play for writing.