How rare are duff chips?

R

retepsnikrep

Senior Member
I just built a fairly complicated circuit using a 28X1 and it won't accept the program download. I have tied all the floating inputs to ground via resistors.

I have tried without a resonator and with an 8mhz res fitted. The same board also contains an 08M which is working and downloads fine via the serial cable. I'm using the latest editor. The download just times out with the standard "it's not connected or powered message". The chip has the correct voltages on the power pins. The SV2000 is working and outputs it's welcome screen before then starting to print garbage which is related to it's floating input and noise on pin 3 and the fact trhe 28X1 is not working. (I'll add a pull down resistor for this in the next version)

Have I been unlucky and got a duff one? Anything glaringly obvious on the Schematic?

The schematic is here :eek:

www.solarvan.co.uk/bms/Master120808.jpg

A picture of the board is here

www.solarvan.co.uk/bms/MasterPcbPop.jpg

Jumpers JP1- JP3 are closed to pull down the unused ADC Inputs.

Jumper JP4 is open during this testing phase, so the Watchdog 08M U7 has no effect on the 28X1

I wish I had used an IC socket now!!

How best to unsolder a 28 pin IC from the board?

How critical are the decoupling capacitors? I used 100n poly is that a big mistake? I can soon solder some ceramic ones onto the board, suggestions?

The SV2000 chip works at 20mhz producing an RCA Video signal output so It could be generating some noise.

This is not a commercial project, it's my battery management system, for which the schematics and source are freely available.
 
Last edited:
A

Andrew Cowan

Senior Member
Best way to unsolder a chip - solder wick and solder desucker to remoe all solder. Then pull. It is very hard to do.

It is possible you overheated the chip while soldering.

Andrew
 
D

Dippy

Moderator
I've NEVER had a duff PICAXE or PIC. I've sometimes made them DUFF after careless use.
Remember, a PICAXE is loaded with a 'bootloader' of sorts and therefore has to have been working in the factory programmer. Any 'dufferisation' concerning programming can only be introduced afterwardds by clumsy distribution or by you.

Well, Andrew could be right about soldering causing damage. It is an obvious possibility.

No ic socket... oh dear. Lesson learnt I hope.

If noise is not too much then your decoupling cap should be OK. You could always solder a ceramic in parallel with it and try.

FIVE questions:
1. Have you checked the resonator is OK with a 'scope?
2. Are you sure your programming cable is connected to your terminal correctly? Have you scoped the programming Serial In and Out pins?
3. Have you done the long boring bit where you check continuity of every track?
4. Have you checked the discrete components? I check the value of every discrete prior to soldering into first board. Tedious yes. Failure rate 0%. Smugness factor 100%.
5. Can you check the PC-PICAXE programming link with another PICAXE?

And one delicate question:
Didn't you do a home-brew d/s prototype before the expense of a 'production' board?

Can you remove the SV2000 and try again? Oh dear, soldered directly again?
Never do direct soldering at early stages esp on double-sided board (unless SMD of course, but with SMD you make a pth prototype first).

I can't see anything obv wrong but I've only looked for a couple of mins and a piccy of a D/S can't help me much.
And only you can check the quality of your soldering with a multimeter.

Best way to get it out? hard work. I think you will almost certainly destroy it getting it out of a double sided board as the solder 'rivets' the component in place.
You may be lucky with wick or solder-plunger-sucker. The best way is a solder station which can heat and suck like mad simultaneously. Speed is the key to get it out quickly to mimise component heating and pad damage.
 
C

ciseco

Senior Member
Hi retepsnikrep,

Some good work there fella! fancy doing me a leccy landy? got a 1960 one, would love the 600+ bhp and 4x4 out of 4 motors like that blue mini :D

Your circuit looks fine to me at first glance (3-4mins), would need to get a scope on it to answer about the caps, doubt it though, if you are anywhere near nottingham I'd consider lending you mine for a day or two.

You could very carefully slice through pin 8 (and probably 6) on the SV2000 (so it could be resoldered after) to isolate that if you really suspect he's the culprit, I'd do that as a final desperate act, though.

Not had a duff one yet and been through quite a few.

I'd try desoldering the resonator first because it's easy and is the only thing on your circuit that if faulty might stop the 28, everything else looks pretty standard. A 28x will run on the internal just setfeq m8 instead of e8 from memory.

Think I'd have pulled JP1 to 3 through 3 x 10K not one, but thats for no good reason than thats what I'd do and wont be affecting you here.


Got a spare 28 and 40 if thats any use, I'm in all weekend, you are welcome to have it and I'm sure I've got a DIL socket too.

Miles
________
buy extreme q
 
Last edited:
M

moxhamj

New Member
Yes, I just learnt the socket lesson the hard way with a board with 15 ICs on it. Zap one ( a 74HCT32 as it turned out), and the whole board doesn't work. But then, how do you troubleshoot it. Start desoldering and invariably damage the board? With sockets you can swap and change quickly.

I used to worry about the cost of sockets till I found Futurlec in the US. 4c for an 8 pin. 12c for a 28 pin.

I think zapped chips from wave soldering is rare, but I think zapped chips from my *grin* soldering may be not so rare! Chips are rated at 300C for 10 secs, and I think I may take more than 10 secs to solder a chip.

So it is sockets for everything now!
 
B

BCJKiwi

Senior Member
Presume you have tried starting the download then powering the chip?

Find on 28X1 (at 16MHz with external Res) that more often than not, it won't download without powering the chip after the download starts.
 
D

Dippy

Moderator
If you think it's noise then why haven't you already tried //ing a ceramic across your PICAXE Vdd/Vss?
Don't mean to sound cheeky, but it's the obvious thing to try if you suspect it or have detected it using your 'Scope. (You've got one of course).

Is your Ground/0V path good? Weedy/convoluted tracks can sometimes cause problems, prob not in this case.

I use a slightly different Reset/MCLR ciruit nowadays. Recommended by Microchip. Its in most PIC Data Sheets.
It ensures MCLR is always 'catching up' during power up, this makes sure of a clean power-up reset. And provides transient protection to MCLR too.

Anyway, you've got a lot of things to check starting at basics... so good luck.
INCLUDING your download connection naturally.
And as you were bursting to show us a picture of your pcb I'd better say "nice board", though it would be nicer if it worked :)

I think spending 40 years soldering that IC has probably overheated it.

PS. I've been doing electronics for almost 20 years and I STILL put a wrong value resistor in a breadboard prototype yesterday.
So age/experience only REDUCES the probability of messing up .... until the point of senility is reached of course.
 
H

hax

New Member
If all else fails, you asked how to desolder the IC without damaging the board.

If you are pretty sure it is the chip that is faulty, you can use side cutters and cut each leg individually.

Then use a soldering iron and solder wick to desolder each leg.
 
R

retepsnikrep

Senior Member
In answer to some of the questions.

I tried without a resonator before I put one it, In fact I only allowed provision for the resonator on the board so it could be overclocked later if reqd. I intended it to use only the internal one.

Stupidly I have run out of cermaic decoupling caps so can't add any at the moment which is damn annoying.

The programming cable link works with the the 08M also on the board.

I do have a scope but haven't resorted to that yet. time to break it out.

I think cutting the legs and then removing it and replacing with a socket is my only chance if all else fails.

I have tried various variations of the start the download and then power up the board scenario.

I'll keep you posted and thanks for the ideas so far.
 
G

gbrusseau

Senior Member
You didn't say you double checked the resistor values for the download circuit and the reset circuit. That 10K resistor could be a 100 ohm one.
 
D

Dippy

Moderator
If you've got another pcb then inspect it and continuity test the power tracks.
Then insert 28pin ic socket and 2 res download circuit. Feed in 5V and try the most basic circuit. (Pre-test the PICAXE).

Assemble a bit at a time until it conks. If it doesn't conk and works OK then it means you fried the first PICAXE.
I don't know what your timescale/budget is but I'd get on with a slow build-a-stage-at-a-time.

I assume you've done all the component and continuity testing?
As there seems nothing obvious then you are going to have to put your Detective Hat on and get probing. We can all say check this and test that but I'm sure you've checked everything(?) inlcuding the board? Is it a pukka pth board? Has a via been missed for pth? Has a via been damaged or drilled? (I've done that).
 
H

hax

New Member
Also, did you swap the TX and RX lines of the download plugs between the 08M and 28X?

That might explain why it works on the 08 but not the 28?
 
R

retepsnikrep

Senior Member
It's Alive :D

I built up a second Master board at 04:00hrs this morning :shock: after laying awake fidgeting in bed with it all going round in my mind, it worked straight away, indicating a duff main pic on the first one which is damn annoying. Might be able to unsolder that and replace it. Must have fried or zapped it somehow even after soldering for 40 years :rolleyes:

Spent the next few hours getting it to talk to the single slave built so far and just got succesful comms. Failure to talk straight away was due to my bad choice of resistor value, I went for a 100k pull down on my opto coupler output and it was too feeble, changed to a more normal ttl 4k7 and works fine :eek: As usual, spent a lot of time arseing about with the scope looking at serial signals and scratching head :roll: Think that's sorted now. Must remember to tweak the schematics to reflect new values.

Current consumption of digital mode Slave running at 8mhz with no speed/current reduction implemented is 800ua. Should be able to slash that when proper software back up and running. That's not bad though and could run slave for 1000hrs for 1Ah drain.

I shall build up another 4 slaves, wire them up and start working on building up functioning subpacks. All the little glitches in the software are comming out now of course :roll: Display alignment is all over the place. Hope to have a pic/vid of 5 slaves and master working together by tuesday :wink:


Thanks to everyone on the forum for there help. ;)
 
D

Dippy

Moderator
"I don't think I have overheated it, pretty quick and been soldering for 40 years "
- we'll remember that ;)
 
B

BeanieBots

Moderator
Glad you sorted it. Duff chips are so rare it's not worth considering.
Desoldering tips:-
Get a GOOD QUALITY solder sucker.
Turn your soldering iron up by 10C.
Apply tip to pad and component leg simultaneously.
Place sucker tip over leg at slight angle.
Keep iron in place for two seconds after solder melts.
Then, very quickly, remove iron, straighten and push down sucker and release plunger.

Waggle the component let with moderate force. If it does not come loose, then RESOLDER again. Do NOT try to use the sucker again without re-soldering. This may be required several times before success. Allow the chip to cool down between desoldering each pin.

It is also possible to buy desoldering tips which contact all pins at the same time. I've never manged to get on with them. I've used the technique described above to salvage many 40pin chips from boards. With a little practice it can be just as quick as soldering in.
 
D

Dippy

Moderator
Good points BB. But it can be a bitch.

Can you work out a costing of desoldering vs buying an IC socket? (Cheeky!)
 
B

BeanieBots

Moderator
Dippy said:
Can you work out a costing of desoldering vs buying an IC socket?
Click to expand...
At my hourly rate it would be FAR cheaper to throw away the board and start again. However, if I had to sit around waiting for another chip to arrive, then might as well go for the damage limitation option.;)

Always, always, always use IC sockets unless it's high volume fully automated production line with ATE fault finding. If you do the risk/bucks calculation, you'd save more by not bothering to renew your houshold insurance and I can't say that I'd recommend that option either.
 
You must log in or register to reply here.
Top