Miller Diversion 165 TIG no arc

[stj]

pic and atmel are interesting, if they are locked they return semi-random data to throw people off.
most other controllers just return FF

[CapLeaker]

pic and atmel are interesting, if they are locked they return semi-random data to throw people off.
most other controllers just return FF

Well... so much to trying to get a dump file from it. That was totally wasted time. Dang! First time I read that Atmega, I figured there is something wrong with the dump. I've seen corrupted flashes on Atmegas before, but none of them did that. Anyway... only because that the Atmega is doing its security protect chit, doesn't mean something all is o,k. with its internal programming. Seen tons of these chips that go corrupt after years of working fine. To get them back in working order was to re-flash them and all is o.k. once more. Not so sure where to go from here now, but the problem is for sure here on the brains side of the board. Would be nice having a working unit to see how they compare.

[sam_sam_sam]

I understand why you would lock down a controller chip but the manufacturer should offer the controller chip for replacement this is nonsense that they are not required to do this so you end up with a piece of equipment that is DOA and nothing else you can really do unless you get lucky and find another one with different issues and repair one of them and have spare parts

Or sometimes you get lucky and they offer a replacement controller board but usually more expensive that way and sometimes it does not make sense to replace it because the cost about half of what equipment cost new really

I have had this happen to many times over the years of doing repairs on my own equipment that somehow you get use to it as the way it is and deal with it

[CapLeaker]

I understand why you would lock down a controller chip but the manufacturer should offer the controller chip for replacement this is nonsense that they are not required to do this so you end up with a piece of equipment that is DOA and nothing else you can really do unless you get lucky and find another one with different issues and repair one of them and have spare parts

Or sometimes you get lucky and they offer a replacement controller board but usually more expensive that way and sometimes it does not make sense to replace it because the cost about half of what equipment cost new really

I have had this happen to many times over the years of doing repairs on my own equipment that somehow you get use to it as the way it is and deal with it

They DO offer a replacement board… for a low price of $2000.00!!! The owner bought the whole machine for $1600 back in 2010 brand new. A replacement machine whatever they have now is $6800.00!!!

[m1ch43lzm]

Found this tool called XPROG which claims to be able to read the code from ATMEGA MCU
https://www.eldb.eu/index.php?route=...ormation_id=18
It's expensive though, about 1300eur for the "base" version, what it does is somehow "glitch" the voltage to the MCU and that way bypass the code protect bits

That tool is mostly used for car ECU programming

I haven't found a DIY way, but it involves voltage glitching the MCU with precise timing

[CapLeaker]

Found this tool called XPROG which claims to be able to read the code from ATMEGA MCU
https://www.eldb.eu/index.php?route=...ormation_id=18
It's expensive though, about 1300eur for the "base" version, what it does is somehow "glitch" the voltage to the MCU and that way bypass the code protect bits

That tool is mostly used for car ECU programming

I haven't found a DIY way, but it involves voltage glitching the MCU with precise timing

Yep, I know about voltage glitching to extract firmware. There are others too besides Xprog. I guess I am screwed with this firmware extraction idea unless I take the chip off and send it somewhere to get it done.

[momaka]

^ And that's with the continued presumption that the issue with this welder is in the firmware IC and not something else.
I do wonder what would happen if you ask Miller to re-program that firmware for you. Well, I know, probably they will just tell you to f-off and buy a new board. But sometimes you can't be 100% sure unless you ask.

[CapLeaker]

^ And that's with the continued presumption that the issue with this welder is in the firmware IC and not something else.
I do wonder what would happen if you ask Miller to re-program that firmware for you. Well, I know, probably they will just tell you to f-off and buy a new board. But sometimes you can't be 100% sure unless you ask.

Tried that and they are only interested on me leaving $2k on the table and replace the board. Even sez so in their service manual which I got my grimy little fingers on.

[CapLeaker]

I went back at it again for a little bit to rule out the optocouplers and all the little MLCC caps. Even chased around resistance readings, no go.

[CapLeaker]

Here are some datasheets for this Miller. Maybe it helps someone in the future. There are still a few missing maybe for later.

[stj]

is it just dead or is the display running and showing info?

[CapLeaker]

is it just dead or is the display running and showing info?

There is no display on this unit, only two LED’s. One LED for power on and the other LED supposed to be on when it welds. It is still doing the same thing. It powers on, but when it supposed to weld, all I get is the HF start (sparks like a single cylinder spark plug) but there is no powerful arc. All the voltage rails are there, all chips get voltage. Some chips get hotter than others like 42C, but they also have positive 15 and negative 15V on them. The trigger input is okay to the opamps. However I think I may found a stuck pin on the Atmega and couldn’t figure out why this is as the circuit board hides a middle layer. But there are some optocouplers that go from way from the secondary where the HF board and the PGM port is all the way back to the primary side to that IXYS ds99037 chip which drives the MOSFETs on the primary with the large heat sinks on them.

[stj]

maybe it's not sensing you touched the surface and withdraw the electrode
some welders autosense that and others need a pedal or button to fire the pulse-stream.

[CapLeaker]

stj
The pedal works and that on signal goes through the quad opamp fine. As soon as I go near metal or ground clamp, the unit starts to spark, but no arc. Also the gas valve doesn’t open and no blue LED either. That means to me the problem is on secondary side of the control board. I’ve tested da chit out of it, but have trouble with the 3 layer PCB. Takes a long time finding something when you can’t see it.

[stj]

electric gas valve?
some people are so spoiled!

does the pcb have a header that could potentially drive a parallel lcd?
just wondering if they used a shared pcb between models?

[CapLeaker]

electric gas valve?
some people are so spoiled!

does the pcb have a header that could potentially drive a parallel lcd?
just wondering if they used a shared pcb between models?

No header for any display on this model. Only extra headers are the programming port and a second hookup for the torch, basically one for the torch and the other for the foot pedal. The biggest problem is the 3rd or middle layer on the PCB. I can see maybe 15% of that middle layer. Very hard to trace, I found one via where I don’t find the other one. Been over that board a few times now, there is zero visible damage. I am sure the problem is looking at me.

[stj]

some via's probably emerge under the microcontroller

[CapLeaker]

some via's probably emerge under the microcontroller

I thought I checked that already, but I’ll do it again.

[CapLeaker]

some via's probably emerge under the microcontroller

I made more progress, but result is still the same. Went though that maze of components and via's and verified the trigger signal working 100% to the ATMEGA16. The foot switch signal and the torch signal are separate, but go to through the LF347 (same IC, just a different input / output pins) and end up at the ATMEGA16 on different pins. Signal is like -0.5V on the Atmega when the trigger is pressed and positive voltage when not.
I can rule out 100% it isn't a trigger problem to the ATMEGA. I think now something should see something, which is not happening, or an on command to something which isn't happening. Without a proper schematic this is a real pain in the @$$.

[stj]

maybe trace back the solenoid drive - that is probably simple.
maybe a buffer is bad
did you post pictures of the cpu area?