Miller Diversion 165 TIG no arc

[CapLeaker]

Hmm… someone brought me this Miller Diversion 165 TIG welder for repair. Claim is it doesn’t arc. Well it tries doing something if I put the tungsten tip straight to the aluminum plate. All I hear is a little tick… tick… tick…, I have the power LED on, but the other (blue?) LED doesn’t light up. I think it is supposed to when there is an actual arc.

So I opened it up, found a bad looking connector between the main board and the small HF board. Checked the SCR and few other things on it and found nothing. Then tried again. Same thing. I will not for its own life strike an arc.

Took out the board with the big IGBT’s and diodes. 500.000 screws later. Nothing. No shorted diode or IGBT.

I am 100% sure the problem is not on the primary side on the main board The only thing left is that secondary side on the main board (with the 4 large capacitors on it). A new board is $2k.

Oh… ya… before I forget it: Miller does not give out any service manuals or schematics. All the do is change boards. Great price Miller!

BTW. There is no burnt smell, machine is relatively clean and nothing visual that sticks out like a burnt something or other.

I wonder if someone had the same problem before?

[CapLeaker]

Went at it again for a bit. I found this manual that has a little wiring diagram on it, but that's all Miller gives you. Other than that, I poked around again for shorts or opens this time at the smaller components and found zilch! I guess this is going to be painful to figure out and fix. This TIG welder is similar to the Plasma cutter a while back, but way more modern. The HF board doesn't have the to plates where it arcs between them, so there is nothing to adjust. The mainboard has some small chip with the brains and a crystal. Maybe I try out and see if the crystal is running etc. Probably I should test the relay as well. Most of the little chips on board there is no way to read any numbers of it to the tune of 80%.

[stj]

i had a french welder where the igbt board was connected by terminals that had broken off the board - they where huge but smd soldered.
so big i couldnt fix it without a reflow oven!

[CapLeaker]

i had a french welder where the igbt board was connected by terminals that had broken off the board - they where huge but smd soldered.
so big i couldnt fix it without a reflow oven!

The IGBTs and diodes is what holds the board in on this miller. The main board with the primary is on stand offs, same as the HF board. This welder has 2009 stamp on it. But I still think there should be more life in it yet. There is no trigger to start welding either. I got good GND and I got continuity from the machine to the tungsten tip I am going to put a few pictures up. I got me the other day a 700w pre heater. Should arrive soon!

[CapLeaker]

Here is a different Miller Diversion 165 manual that has the pin out for the foot switch and or that wheel. I am splitting the PDF from the pics... otherwise my PDFs disappear.

[CapLeaker]

Here is some Miller Diversion 165 porn.

[petehall347]

long time since i looked at things like this . i seem to remember a spark gap component that stops them working . no idea if that has one or not .

[CapLeaker]

long time since i looked at things like this . i seem to remember a spark gap component that stops them working . no idea if that has one or not .

No. It doesn't have a spark gap thingy. Image 3217 is the HF board in my post #6 you thinking about. You just cant see the post yet. Needs to be approved first. lol

[SMDFlea]

Sorry, the forum spam filter blocked your post. Sorted now.

[momaka]

Have you checked the big caps to see if any have gone open-circuit?
I'm not fond of how these units operate, but if there is any sort of HF current going into those big caps, it might be worthwhile checking them.
Also probably worth checking all of the high voltage PP caps - their capacitance can decrease with time if they have seen many voltage spikes (be it from the design or some external factor.)
I see the boards are coated with protective coating, though. That's never fun to deal with. :\

[CapLeaker]

momaka You are somewhat right with your assumption. There is no full power from the accumulator board to the wand.
Could be causes:
These 4 large caps on the main board
The wand / torch itself
The trigger incoming circuit on the main board
Or the trigger outgoing circuit from the main board to the accumulator board.
I think the accumulator board also has to have a some sort of and enable or on signal from the main board.

Remember the Plasma Cutter LGK60 that I fixed a few years ago? Similar, just with more modern technology.

[CapLeaker]

One thing is for sure: Miller doesn't have a schematic diagram of the boards. Best thing you get out of the SM is kind of a block diagram that may or may not be your board and you get to know some troubleshooting tips. Basically it doesn't help much, as they just say: replace board x.
I verified the 28V rail and the 15V regulator and that is working. However I don't get any control signal going to the arc board and not the gas valve either. More check-a-roos to do. If all these little power supplies are indeed working, then I may have to look at that micro controller.

[CapLeaker]

O.k. I figured out how all these power supplies on that main board work. There are 6 different voltages to check and they are all present. So I decided to feed some power with my bench power supply and see with a thermal camera what non PSU parts are getting hot. Well… there I found some all right. The MCU seems to be all right, some chip is good, others are at least twice as hot and are in the same row. What these are, I don't know yet, ran out of time for today.

[CapLeaker]

Board number I am working on is 230058B. The processor is an ATMEGA16 -16AU TQFP style with a 16Mhz crystal. Very tempted to read that thing out with my old TL866II. I have to look for the adapters and may not have the right one.
Off course this is a 3 layer board and hard to work on. Some other chips are LF347M x3, LM324AM x1 and an AD597 x1. A lot of the SOT-23 diodes are marked M5C = MMBD7000.
Verified today the trigger input to one of the inverting op-amp IC (LF347M) working fine. These LF347M are powered with +15V and -15V and that is a reason for them to run hotter than the other chips (I get like 42C). Then I tried to trace it to the Atmega and the trace started to split off in branches but was able to find a pin where it goes to. Problem there is that this pin on the Atmega is stuck high?
The HF start works making a small spark, the rest does not (no gas valve and no power from the other PCB with the large IGBTs and diodes). There are tons of SMD resistors with a high ohms value that have power on one side but none on the other. Then I poked around the test points on the board with my oscilloscope. The one closest to the Atmega is doing something like data but slow and the square wave voltage is inconsistent. Most of the time it is 0 to 5V but sometimes for a brief period it goes like 0v to 5V to 10V? Weird. Crystal works fine at 16Mhz. Another test point is doing nothing and the rest are stuck high. That's all for tonight.

[stj]

the atmel was probably programmed on the board, look for a programming header to link to the programmer

[CapLeaker]

the atmel was probably programmed on the board, look for a programming header to link to the programmer

Yes, there is a PGM port, but I don’t know if I can hook the TL866 up to it? The port is very similar to these old serial ribbon cables, just this one is 10 pins, 2 row.

[stj]

meter it to see which pins go to the programming pins on the atmel, the tl866-II can do ICP through the connector at the end.
if you select the device and ICP it will show you the wiring needed.

[CapLeaker]

Hmm… I might do just that. Sound easy enough for a Sunday morning while cooking steak for lunch.

[CapLeaker]

meter it to see which pins go to the programming pins on the atmel, the tl866-II can do ICP through the connector at the end.
if you select the device and ICP it will show you the wiring needed.

Well, I made a dump file from the Atmega16 with my old TL866II using its ICSP port and it is not what I expected. Not sure what to make out of this. One thing... I had to select the Atmega16 on a 40 pin. I could not select the TQFP package, wasn't in the list. But shouldn't matter since the Atmega did properly ID itself to the programmer software. I am sure this isn't right. It's almost like it lost its programming or defaulted somehow.

[m1ch43lzm]

There's some "fuse bits" on the Atmega MCU to protect the code from being read, check here
https://electronics.stackexchange.co...ng-through-isp

https://www.avrfreaks.net/s/topic/a5...ZBeEAO/t149052

[CapLeaker]

There's some "fuse bits" on the Atmega MCU to protect the code from being read, check here
https://electronics.stackexchange.co...ng-through-isp

https://www.avrfreaks.net/s/topic/a5...ZBeEAO/t149052

After I read the Atmega, all the Lock Bit Bytes are checked, even so I didn't check them before reading. Seems like Miller locked that Atmega down, so it cannot be read. Dayum! It also perfectly explains my strange dump.