miller dynasty what is precharge relay

[coppercone2]

I am not 100% sure but my inverter welder broke again, possibly having something to do with a beetle that got on the circuit.


On of the symptoms is a relay clicking on and off repeatedly. It is called a "precharge relay"

They don't go into detail about what it does.

Does this mean a inrush current limiter circuit that uses a relay and resistor? Is that industry slang for it?


What the hell kind of logic could result in a precharge relay deciding to oscillate, it keeps clicking every 2 seconds or so. Is the machine trying to clean a contact? I.e. I don't know how those circuits work. Does it actuate a relay and check for a voltage drop maybe? Then if there is no voltage drop it keeps trying?

If I made it, it would have a simple timer, thats it. But I guess you can make a more advanced one with fault diagnostics that make sure its acutally open. Is that common at all? Would someone make it to try to maybe like clean a relay contact by repeatedly smacking it? I.e. comparator to measure voltage drop across the limiting resistor. If it does not go away, try again, and again?

[floobydust]

There's no PCB pics or repair info, but which model #?
I would guess it's an inrush limiting relay for the DC bus capacitors, so they get a pre-charge. Some soft-start circuits are dumb - a timer or voltage sense, others "smart" using MCU control.
The pre-charge resistors will cook and burn if they see mains for more than a few seconds. It might use a NTC or PTC (bank) as well.

When it switches in full mains it might reboot if there is a short? It is drawing high current from mains when the relay clicks in?
Worst case you might have a bagged IGBT preventing the bus voltage from coming up. Take it apart and a pic will tell how they did it (pre-charge).

[coppercone2]

Well I bought a LEM module and a replacement relay because if I take that board its not worth saving 35$ on the LEM module to put it back in. because its a big fat hard to remove piece of garbage. I dont like the part# prefix "HY" , I am just gonna replace it. I did not bat an eye gambling on $80 of parts because I really hate that mother board because it means I need to deal with the SKIP MODULE IGBT RECTIFIER and that makes me really angry. I am not taking that board out until I get those 2 replacement parts its like climbing everest. Well I can take it out but I am not gonna put it back in until I change something even if its fine because i want to smash it into the concrete wall like a old carpet   


I did all the diode tests for the IGBT they look OK.

What its doing is...

1) turn on

2a) operate

2b) operate with the inrush relay clicking. The boost board is a transistor that sinks the relay to turn it on. I measure this control pin, it just pulses every second or so.

2c) operate without turning on display

3a) sparks arc HV pilot when I press the foot pedal if display is on
3b) eventually or first time shuts down display when I press the foot pedal

4) shuts down display regardless if I did anything. I got flashed with HELP 8 before.


I put the service manual and this has pictures but does not post schematics

https://community.element14.com/technologies/experts/f/forum/38156/miller-boost-circuit-board-assembly-no-200841


The only thing connected to the machine is the foot pedal. It was doing the same thing without foot pedal. I take a good look at everything I can see and there is no problems. I don't think its related to user input because it will turn off even if I just turn it on and wait a minute or so.

Also, if it turns off, I need to wait a while before I can turn it back on, its almost guaranteed for the screen not to turn on if I turn it off and turn it back on with a wait period of less then like 3 minutes.




The Boost board in that picture is the PCB that is connected to the relay, and I assume the HY-25P module (not 100% sure)


I thought maybe it somehow uses the HY25P module to verify the inrush relay, so if the HY25P is giving erronous signal, maybe the relay go brezerk like that. While relay is going brezerk, the voltage is like ~700V with alot of ripple because it never fully charges or discharges with the relay clicking on and off. If the relay is clicking and I press the foot pedal it turns off.


It stopped working when I left it on and went to go into the house for a minute to go to the bathroom before starting a practice weld on coupons. It was set to ~35 amps in DC mode when it stop working.


I bought a UC3854, n26 optocouplers, IRF9250 transistor. I have all the other active on this PCB in my parts storage. I am just gonna do all the silicon on the booster card because its only my life force and $20. There is some forum post about this happening where it makes noise but its turned off that its the boost card. Maybe the reason why its so loud is because the boost / PFC board was degraded. I think I read some where that pFC can sometimes make inverters sound obnoxious before they break. I think I can do all the ceramic capacitors too because there was a bad ceramic cap on the main board a long time ago and I have several cubic inches of 100nF beads. I did have to replace the other card (which had arc damage) and I did replace all the parts on the motherboard and the IGBT snubber board and whatever else. I figure its natural to assume whatever happened damaged this card too and it just took a dozen feet of high current weld to make it show up. I would just replace it, but since the boost card looks pristine, and I never replaced parts on it, maybe its that. 

no free lunch, I guess  this card did had to get damaged when the other one got totaled.

[coppercone2]

What I am curious about is why the failure mode is so dynamic. Normally it fails 1 way, this one can do a whole bunch of outcomes from a power up.

It makes me think of leakage or contacts, maybe something with capacitors. Why is it being random? This is exactly what I expect if something got corrosion damage, that is, random behavior based on whatever governs the impedance of the corrosion at the time, but this board looks pristine.


I wonder if there is something up with the UC chip or the HY current chip, because those are complicated. Or if one of those CD switches is misbehaving and playing busdriver with me. All that CD stuff is high impedance FET IIRC? it reminds me of mashing buttons

I wonder if its a over heating bond wire that is playing whack-a-mole with me. I see what hot springs do, they behave randomly (contacts), sometimes retracting like a turtle head.

I want to keep smacking it until it finally breaks into a particular failure mode lol (I know you cant and its high voltage but the thought makes me amused)


Maybe I can strip the conformal coat to see (even though it looks great). I need to do it anyway for the parts replacement.

[DisasterNow]

Really a cockroach? If the circuit board was drowned in lacquer like mine, this would not have happened. Have you checked the relay activation signal if it is stable? Mine attacks when it gets 12dc, this happens at about 110ac of the network.

[coppercone2]

not that, its a small round beetle, a rather good looking one, not some nasty bug. Something like a ladybug thats somewhat larger


I can't find where it got burned. And I did coat everything in silicone MG chemicals PCB protectant BTW, everything I can

I did probe the signal, it basically square wave coming off the boost board.


Right now what I am going to do is use conformal coating remover to get rid of all the manufacturers varnish on the boost card, I will give it another try (in case the varnish went bad, sometimes bad varnish can conduct). If it still does not work, I can actually probe it safely without varnish, because probing it through the varnish while its running with 810VDC rails .... is a rather stupid way to have an accident because you need excess pressure




Applying that varnish is not easy, not the ones that are slightly softer. I actually have some peeling on the back of the main board I did last year. I guess it was not clean enough. My work on the control card, and the IGBT snubber board (IGBT controller), looks nice and intact. I like the MG chemicals silicone that comes in a small bottle with a nail polish brush on it.


Now, if I really wanted to protect it, I think I would use red insulating varnish aka transformer varnish... but it makes things look like some bronx boiler room. Its like the "i don't give a fuck" paint they put on cinder blocks in apartment hallways .. nothing shows class like painted brick. I am scared a crack dealer might try to setup operations 



What happens to millers default varnish they supply the unit with
1) PCB is too hot, there is too much corrosive dust
2) varnish turns yellow
3) varnish peels
4) I am suspicious it becomes conductive slightly to 810VDC. I saw evidence of electricity in the varnish near some HV parts, like soot and black cracks 
5) boards look like a burn victim



Basically the history is
1) get machine with 4 damaged cards, had to repair the MCU card diode, the main board flyback, and something on the igbt controller.
2) fix what I can, replace basically all the parts on those 4 boards, fill in holes with 270dp epoxy and fiberglass after cleaning with micro mill.
3) have to replace the inverter control card for $$$$, even though I replaced every part on the old one (every part), and cleaned / milled the soot off, including grooves between some burned traces, it would not work.
4) got it working with new card + my repairs
5) take apart everything and wash it, paint a few rust spots on transformers after cleaning, etc, and siliconed the 3 cards I worked on, *left the new inverter card, the old booster card alone.
6) polish all the connectors, grease everything up, reassemble
7) works
breaks and this thread begins


SO I left the Boost card alone, because it looked nice, the unit worked, and the conformal coating looked nice. Now the symptoms are linked to boost card, meaning it was probobly damaged to begin with but still functional enough to weld.

[DisasterNow]

You know I have some doubts about the conductivity of the old lacquer on my board; not to mention how much it will have hindered the heat dissipation.
Anyway, speaking of insects, a few years ago a small snail I think, blew up the electric board of my gate 

[coppercone2]

The problem I have with the new coating is the big PCB, when i blew out the dust, ALOT of coating came off. ANd I thought I did a good job cleaning it, but its too big to fit into ultrasonic cleaner! The boards that I could get to mostly fit into the ultrasonic cleaner, before lacquering, worked OK.

I guess you need to really scrub the thing down with alot of solvent etc.. for a big board its nasty work.


Snails are extra bad because their wet. IMO the idea that you can protect a board from insects with a coating is BS, they probobly eat it, or destroy it mechanically. It needs a box with hard walls and steel vents. The insects can usually damage wood, and pry things, so I doubt a bit of rubber will stop them. They might think food is hidden under neath by bees or something like that.

[coppercone2]

and for removing conformal coating, I use MG CHEMICALS CONFORMAL COATING REMOVER. Its a good chemical.


You can also use the solvent "THF" , a modern solvent (tetrahydrofuran) if you want a cheaper industrial solution. I found that for some coatings, they are also allergic to spray type flux remover, though this is a expensive choice.

It can't hurt to re-do the coatings, yellow is suspicious


I actually found if you can bend a stainless rod to the shape of a vent hole, its easy to spot weld stainless steel mesh on it with a K-weld.

[DisasterNow]

I then discovered that it was quite common for slugs to cause problems with gate board.
As I wrote in my topic, I used acetone, but the lacquer does not come off. But I will leave it alone, the only thorough cleaning I will have to do is around all the soldering pads of the two chips, to facilitate soldering. I think I will use the Dremer with a very fine tip, and I will go at low speeds and very lightly, I usually get good results.

[coppercone2]

you want this for PCB and dremel

https://www.amazon.com/dremel-rubber-polishing-bits/s?k=dremel+rubber+polishing+bits

[coppercone2]

well I got the coating off the boost board, unpleasant work.

After it sun dries for a while I will bake it in PCB oven and then see if it works.. its always possible the coating was bad or there was a tiny tin whisker or something under the coating on that card, scrubbing and ultrasonic might have improved something 

or perhaps it failed more to where its obvious what the problem is

[DisasterNow]

I already have those inserts for the dremel, I should be able to clean the pads well before soldering the two chips.

For your electronic board, let me know if it will work well 

[coppercone2]

well cleaning it off aint do anything really

the only thing it works at is giving me a killer migraine

[coppercone2]

I think how it works is when the 15V rail is stabilized on the card, it turns on the transistor that lets the relay drain. Because the transistors connected to the relay, the base of one of them seems to go to VDD, thought a resistor?

So I wonder if something is dragging down the 15V rail. It seems to be the output of the LDO on the card. And I think that LDO is connected to the output, and the input is the BUS.



the 15V LDO that is connected to one of the transistors for the relay driver is dead stable at 15V. As far as I know, the 30V powers the relay. That one was stable too, I think. I need to figure out exactly what is providing power to the relay, and see if thats stable when its clicking. I don't get what it could be. The circuit is mad annoying to trace out. But I think the relay is connected to a rail, through a resistor, and the other side of the relay is connected to the boost board.. so if I probe ahead of that resitor, I should be able to see if the rail that powers it is sagging, or if a switch on the low side is opening and closing. or if the card is providing current,




because from the looks of things, out put and input is confused on the data sheet. The 15V LDO output, is connected to a pin labeled "input" on the manual!!!!!!!!!!!!!!!!!!!!!

I think they copy paste it from the fucking main board and forgot to adjust polarity... let me see... thats right... both PC2 mainboard and PC10 have RC1 listed as "output 810VDC". Of course the little inverter board is not generating 810VDC ~!~!@@@@ wow that ws driving me up a fucking wall . high quality unverified documentation 

[coppercone2]

I replaced all the DIP chips on that board. It seems to work... better?

It does not always turn on, but it seems to stay on. I still wanna replace that controller chip on there.


I do recall finding a busted CD series gate on the other board (which was replaced instead of repaired). Maybe those rails took damage,.


I also notice, when it fails to start up, the BUS voltage is around 310V, and the +30V and +15V is OK, but -12 (confirmed on another forum that its not -15, but -12) is like -8 V. The relay "pin" is 4V when its on, and 30V when its off. So its like there is a device that is... effecting the -12V rail.. sometimes.


Now I don't see any negative -12V regulator on there, but there is a coil. I wonder how the hell its getting negative voltage


I found this drawn by john76
https://forum.millerwelds.com/filedata/fetch?photoid=597488



ANy idea what could cause that regulator to act up intermittantly? They are using the NAND gate to make some kind of oscillator to power the transistor? The 30V is there on mine, but I get like 8V from the negative thing. Can someone explain to me WTF is the point of that crazy circuit. They really need all that to get a minus voltage?!?!!

[coppercone2]

well interesting it turned on 10/10 times  after it was left alone for a while.

I guess the solder needed a bit of time to set (people have a misconception that it works as soon as its solid) I have no other explanations. I am gonna give it a try some time in the future when it has a chance to acclimate to its new parts  the electrons got used to it ?


the only thing I did different when it started to work, is I flicked the mains switch rather hard, instead of letting the spring do the work, I kinda slammed it. I guess it pushed the electron into correct spot

[DisasterNow]

But what output signal do you have on this welder? Just a dc voltage? And what value should it be?

[coppercone2]

The output was getting near the usual 80V before it shut down, but I did not try it on load with an arc, because I thought it might make damage worse, until I prevent it from shutting down randomly and killing the display



However I am happy to report that after changing the opamps, comparator, logic gates on the boost board, and cleaning the small amount of dust on the inverter control board (even though its insulated with manufacturers varnish) it does seem to spark an arc and weld.

 I am waiting for a vial of silicone lacquer from MG chemicals to re coat the PCB that I worked on, then I will reassemble it. I will leave the LEM current sensor and relay in parts storage if their ever needed. I had to scrape off all the old varnish with mg chemicals varnish remover so I can work on the PCB


Based on that picture schematic someone drew of the board, I started to think that the NAND gate is acting like a oscillator for the -12V rail, and that it was damaged.... maybe thats why the -12V rail would drop to like -7 and dance around. I don't know HOW that switching power supply works, I don't think I want to know, I think I would use a control chip instead of whatever the hell they tried to do there.


Also, I have no idea why that 20V IGBT driver power supply diode fried, its not related to anything. But I did replace it with a 5W diode, when it was a 1W diode before. The other two diodes I modified for 5W, but I left one of the smaller ones in place (3x 20V isolated rails). I feel like that diode ends up breaking if anything goes wrong on the two cards. I think one of em controls a switch that makes the input voltage for the flyback transformer, and when one of the rails get messed up, it makes too high a voltage and blows the diode (maybe goes to very high duty cycle).  Or that its just not specced properly, and needs a very robust diode to survive on 1W, because other resistors on the PCB for this welder are known to brown.... again I would have put them on standoffs away from the PCB and used a higher value resistor!

[DisasterNow]

It is better to cover with lacquer, after all these are machines that are used in environments where there are many iron splinters. When I opened mine, there was a lot of metal dust on the board, without lacquer I think it would break down quite often. A test would be to try at least a couple of electrodes continuously at high amperage. I would avoid 3.25mm electrodes, in my opinion on these machines they only serve to stress the power part too much.

[SteveThackery]

I guess the solder needed a bit of time to set (people have a misconception that it works as soon as its solid) I have no other explanations.

A joke?

[coppercone2]

You tell me. It needed to rest with the new components for a while before it works. All people can do is blame the card contacts but they look great to me. I cleaned them very well and inspected all of it with a 14x magnifier


Honestly what it reminds me is de-gassing a radiator for a house or something, after you do plumbing you got the air bubbles that take about that amount of time to really come out. Maybe the circuits get loaded up with bad electrons or something. They  get stale in storage and need to be drained into the circuit. The machine was plugged into earth ground during this rest period.   

Professionally at work of course you would just blame some kinda nebulous mechanical contact related effects or tin whiskers or what ever there is in the "old thing dont work" box of tricks, if you had to write a report to sate someone about a transient issue. Personally I think there might be more to it

[coppercone2]

I got the fucking flaky conformal coating off the main board (wow that took an hour and like 1/2 liter of solvent), plus 100 task wipes, so I can replace the sensor and inrush relay, just incase, because they were connected to the dodgy PCB with the bad CD and op amps.

The machine was working before but I figure I should not leave it in the shop with a conformal coating that looks like it has the morphogenic changling virus

What a god damn mess. It turned into like... pizza crust blisters on the bottom. Wow i hate conformal coating. 


I don't know why. Maybe I turned it on too soon the first time I fixed it and it caused residual solvent to evaporate too soon before it was cured, but I dunno. I guess I should wait like a week 

[coppercone2]

now some fucking resistors blew up 


I have a feeling this whole thing is under rated, they should have used 1/2 watt metal oxide parts instead of this film bullshit on the high voltage part.

I wonder how well it would drag behind my car

I almost want to make it my lifes work to remake this entire thing with heavy copper PCB (double thick fiberglass), with isolation traces,  300% bigger parts, and see if its still a piece of shit. it baffles me


I want to know if its actually a bad circuit that is unreliable or just shitty finite element engineering where they think they can predict everything to keep shaving the BOM to nothing. The only reason I keep working on it is because the parts are so cheap in comparison to the machine.

I wonder if it has the electrical equivalent to 10-32 screws holding on a truck tire.


I have a feeling every single component on those little PCB cards is basically electrically over stressed for any reasonable reliability at that voltage. All the parts near the card inputs near HV on the boost and inverter control card probobly need a change to a more durable resistor.  This machine has "lol 1/4 watt" disease



"mains is never gonna be more then 124V AC with the 270V mov right?"

"the power is 249.95 miliwutts in the simulator!"

"don't worry, I put a nichicon capacitor!"

"the monte carlo is a casino right?? you want me to do a team building exercise there??"

- miller expert project engineers


Maybe I can scan the boards and we can have a betting pool on which one will blow up next

finding the deficiencies in their design is more fun then actually welding I suppose.



You need to be like Stalin to fix this thing. Kill everything that might even look at you the wrong way in the future. I just feel paranoid and I keep wanting to replace more and more parts. its not a good design that makes you feel like this after only ~10 years.   


Fortunately there is a river between the two cities, There is a few suspicious grain silos that need to be cleansed on the other side.



I found brown residue under one of the 1000 ton grain silos.