Microcontroller replaceable?

[partyanimallighting]

Hi wizards, I'm trying to resolve some problems I'm having with some olddd Martin MX-4 scanners. I'm having some problems with 2 of the 4 units that I use. I removed all the ULN2803 drivers from the pcb's of all 4 units and installed IC sockets on all the pcb's so that I can swap around the 2803's between pcb's to find the faulty drivers, my simple troubleshooting method. I can now replace any defective drivers easily from the pcb's whereas before they were all soldered in, and by swapping around, I've discovered that 5 of the 12 drivers on the 4 pcb's were defective and have to be replaced. However, on 2 of the pcb's, I'm still getting an issue with color on one and shutter on the other so I'm assuming that the AT90S8515 microcontroller is faulty. When another pcb is installed the units in question function properly so it's definitely not a motor or motor harness problem.I've found the schematic for the pcb and it's attached. Here's my question....I've searched online and found the Atmel AT90S8515 microcontroller and ULN2803 drivers. Can I simply purchase these parts and replace them in the defective pcb's or do these drivers and microcontrollers have to be programmed to suit the pcb's?

 

[(*steve*)]

Simple answer, the microcontroller is almost certainly replaceable.

But the bad news, if it doesn't have the firmware loaded, it won't do anything.

 

[partyanimallighting]

Figured so. I'll have to see where I can source these preloaded then. What about the drivers? Are they replaceable without a problem?

 

[(*steve*)]

The other parts should be easily obtainable, look on digikey or similar.

The microcontroller would be a real pita to replace without the right equipment.

 

[partyanimallighting]

Well I do have a rework station on order and I've done a little poking around with SMD parts and I've had some success with replacing parts with the help of a heat gun, a pair of tweezers and a huge magnifying glass so there's hope. So I can just order the 2803 drivers, replace them and they'll work fine, right? It doesn't make sense replacing the microcontroller if I can't get it preloaded with the correct firmware anyway and it's over 300 bucks for a new (used) pcb. I can pick up a complete used unit on eBay for around 100 bucks but that doesn't guarantee that the pcb would be fully functional.

 

[kellys_eye]

I'm still getting an issue with color on one and shutter on the other so I'm assuming that the AT90S8515 microcontroller is faulty.

What tests did you make to come to that conclusion?

Processors are usually 'good-or-dead' though this isn't always true!

The 2803's generally offer protection between input and output (any fault has to feed back through a darlington pair) so any fault is unlikely to feed back to the processor side.

With the 2803's removed from circuit you can fit pull-up resistors and check the logic levels at the processor outputs to see if they do as they should.

 

[dorke]

Well I do have a rework station on order and I've done a little poking around with SMD parts and I've had some success with replacing parts with the help of a heat gun, a pair of tweezers and a huge magnifying glass so there's hope. So I can just order the 2803 drivers, replace them and they'll work fine, right? It doesn't make sense replacing the microcontroller if I can't get it preloaded with the correct firmware anyway and it's over 300 bucks for a new (used) pcb. I can pick up a complete used unit on eBay for around 100 bucks but that doesn't guarantee that the pcb would be fully functional.

Replacing the μC at theses prices is probably not economical.
If you buy large amounts of products("become a distributor") the manufacture of the equipment may be willing to provide you a master device or file that can be copied/programed for servicing.

If you do decide to replace the μC ,
note that PLCC packages should be removed with a heat gun!

The ULN2803 may be replaced easily(don't buy them on Ebay! go for reputable sellers such as digikey/mouser etc.)
I think you should investigate what exactly failed and why ,as well as is the failure at the same places etc.

 

[dorke]

What tests did you make to come to that conclusion?

Processors are usually 'good-or-dead' though this isn't always true!

The 2803's generally offer protection between input and output (any fault has to feed back through a darlington pair) so any fault is unlikely to feed back to the processor side.

With the 2803's removed from circuit you can fit pull-up resistors and check the logic levels at the processor outputs to see if they do as they should.

For processors,that is mostly true("every bit counts"),but for μC, not so ("every I/O pin for itself").

Actually a darlington doesn't provide more protection than a single BJT,
if there is a breakdown in the CB junction it will reach and damage the driver the same way.

I think it would be better to probe the circuit in it's faulty state before removing any parts.

 

[kellys_eye]

if there is a breakdown in the CB junction it will reach and damage the driver the same way.

whilst I agree, the current in the 'top' transistor is an order of magnitude (or two) less than that of the 'bottom' one and, imho, less potential for failure.

 

[(*steve*)]

In theory, with a Darlington you will also have a fairly large base resistor which will help protect the microcontroller, and the path to ground through the two base emitter junctions should limit the voltage at the base anyway.

 

[partyanimallighting]

Hi guys, I got the faulty PCB's working again!! kellys_eye post about good-or-dead got me thinking so assuming that the microcontroller was indeed good, and the drivers were tested working, I figured that the issue would be between the driver and the MC. So I checked from driver pin to MC pin and there was the problem. One broken solder trace. One of the pins on the microcontroller didn't have a proper solder connection to the pcb. So I soldered that up, fired up the unit and voila!! Fully functional. As I continuously state here, I am not a technician but, with the excellent advice I get here, I am able to solve a lot of issues over and over again. With this learned, I now know to check solder traces for continuity before posting any problems here, and that will save both myself and you guys a lot of time. Thanks guys!!!

 

[partyanimallighting]

Oh! And based on your recommendations, I'll order the 2803's from someone reputable like Digikey, not eBay. I'm assuming these drivers are standard, just purchase and pop into the pcb, right?

 

[(*steve*)]

Good job finding the poor solder joint and broken trace.

The driver ICs should have the same pinout and can be swapped. Just make sure you order the right part

 

[partyanimallighting]

Great!!! Found it on DigiKey.

https://www.digikey.com/product-detail/en/stmicroelectronics/ULN2803A/497-2356-5-ND/599591

 

[kellys_eye]

Congrats on finding the problem.

But it leads me to suggest that you need to follow the basics of fault finding more closely in that 'sight and smell' are (perhaps) the first two techniques used.

I always give any faulty equipment the #1 eyeball (with attendant 'sniffing') to spot areas of stress, failure etc before making static and dynamic tests.

Although the track issue you eventually found was potentially miss'able under general observation you sometimes get lucky!

 

[partyanimallighting]

That fault was definitely not really visible. The break was where the pin on the microcontroller went into the pcb so I didn't see it at all. I'd definitely put it down to lack of experience too. But yet another lesson learned. I'll be more observant next time and do some tracing before I seek the wizards' advice.

 

[kellys_eye]

If the break was that small then the 'casual' inspection of a first-go look-see of broken kit is bound to miss it - but if it was a manufacturing fault then so-be-it, however if the break was as a result of (possible) corrosive effects then the other symptoms would be fairly visible and alert you to the possibility. i.e. residues, other tracks looking 'worn'/thin etc.

Given you mentioned 2 of the 4 units you have are faulty, does the dodgy pin connection on the μC happen on any of the other sets? Can you relate it to manufacture or environmental effects?

Given the potential for high humidity where you are perhaps you might consider tossing a sack of silica gel into 'sensitive' equipment if corrosion is atmospheric????

 

[partyanimallighting]

I'm putting it down to old age, corrosion over the years and lack of use as these have been out of service for years, sitting down in a storeroom waiting for me to get some time to eventually take a look at them to resolve the problems. If you look at the picture of the pcb you'll see a manufacture date of 1999!! That's old but these are really great "old school" units that I'm glad to have in my possession. The other defective unit had a barely visible broken solder trace that I did not pick up initially but I discovered it afterwards after confronting the pin problem on the microcontroller. So now, it's just to order the drivers and put them in and I should be good to go. Then on to the next issue, that fog machine pcb that I never got a follow up on. I'll do my best to check everything on this pcb as best as I can beforehand with my limited knowledge. Hopefully I'll find the issue and not have to re-post.

 

[dorke]

That fault was definitely not really visible. The break was where the pin on the microcontroller went into the pcb so I didn't see it at all. I'd definitely put it down to lack of experience too. But yet another lesson learned. I'll be more observant next time and do some tracing before I seek the wizards' advice.

Well,
My advice is to test the board in it's faulty state,in my experience it is the best way to find the fault.
Using vision to locate faults is at best being lucky,since many, if not most faults are invisible.

Avoid swapping parts as much as possible,PCB's and devices may be damaged that way quit easily!

Here is how you should have found the problem systematically in this case:

1. Identify the non-functioning property of the equipment.
2. Locate the "path" that operates it on the board.
3. Verify along the path what happens in the on-off state (in your case).
4. That would have led you to the non-changing state of the ULN2803 input.
5. Check the driving pin of the corresponding μC pin,which would be fine,in your case.
6. Conclusion ,the connection between 4 and 5 is suspect,ohm it to verify.
re-solder joints(even without visual inspection...
Problem solved in a systematic process.

Note:
If there is one cockroach, there are probably more,
hence re solder all the pins of that μC!!!

 

[dorke]

I'm putting it down to old age, corrosion over the years and lack of use as these have been out of service for years, sitting down in a storeroom waiting for me to get some time to eventually take a look at them to resolve the problems. If you look at the picture of the pcb you'll see a manufacture date of 1999!! That's old but these are really great "old school" units that I'm glad to have in my possession. The other defective unit had a barely visible broken solder trace that I did not pick up initially but I discovered it afterwards after confronting the pin problem on the microcontroller. So now, it's just to order the drivers and put them in and I should be good to go. Then on to the next issue, that fog machine pcb that I never got a follow up on. I'll do my best to check everything on this pcb as best as I can beforehand with my limited knowledge. Hopefully I'll find the issue and not have to re-post.

1999 isn't really old,
the problem is poor solder when the equipment left the factory(Chinese?).
Corrosion though, is a big problem in wet areas,equipment should be stored wrapped( and sealed if possible)
with absorbing bags.

When ordering these parts from Digikey,
order a bunch, they may be useful in the future,and cause delivery is relatively. expansive.