- Thread starter JOEPRO
- Start date
That AI reply is amazing....!
The explanations about your meter charging the capacitors, and the capacitors probably being in series to get the required high voltage rating, etc., are exactly the words I was about to write. I can see that I will soon be out of a job.
The explanations about your meter charging the capacitors, and the capacitors probably being in series to get the required high voltage rating, etc., are exactly the words I was about to write. I can see that I will soon be out of a job.
JOEPRO
Member
- Messages
- 5,290
- Location
- UK
It is hard to trust though!
But its the first time ive been able to use AI sucessfully feeding it the ciruit diagrams etc
I am not sure yet, I need to re-check the troubleshooting results. I am not as fast as Mr./Ms. AI
Here's a thread about another Miller welder, from 6 years ago:
Apparently I should have remembered all about Miller using 3-phase 6-diode bridge rectifiers with a built-in precharge thyristor!
That thread confirms that the two 2800uF 450 volt capacitors C3 and C4 are in series, to reach a high enough voltage rating.
Apparently I should have remembered all about Miller using 3-phase 6-diode bridge rectifiers with a built-in precharge thyristor!
That thread confirms that the two 2800uF 450 volt capacitors C3 and C4 are in series, to reach a high enough voltage rating.
Last edited:
I disagree with the AI result in Post #40.
It has misread the resistance reading on your meter from the rectifier "constant positive" DC output to the +ve terminal screw of capacitor C3. It thinks the value is 20.16 Ohms, but I see it as 201.6 Ohms. ( Does your meter read to a hundredth of an Ohm anyway? )
It is correct in assessing 20 Ohms as a reasonable value for a pre-charge resistor, but the actual 200 Ohms is too high.
However, if AI thinks that the pre-charge resistor is 20 Ohms, then the bus voltage should rise as the capacitors charge, so the first point of the AI section "This explains everything", i.e. "No bus voltage rise", is illogical. The second point "No relay click" is wrong, since the role of a soft-start relay is fulfilled by the thyristor here. In the Miller XMT 304 thread that I linked above, their use of rectifiers with a built-in pre-charge thyristor was confirmed by @Jan-DIY , there is no relay.
It is also illogical for AI to blame the "No bus voltage rise" on "The internal thyristor section of SR1 is failed open or not being triggered", because that thyristor gets triggered to short the pre-charge resistor and pass full load current when welding, and all this would happen after the soft-start delay, when the bus voltage would already have risen.
Having said that, AI may still be correct that SR1 is faulty, but that is not proven yet....
Please confirm that the resistance of the pre-charge path from the +ve of the rectifier ( terminal R2 ) to the +ve ( right-hand ) terminal screw of capacitor C3 is 200 Ohms not 20. Then identify the resistive component - resistor or thermistor - and measure resistance directly across it. Look for bad or burnt connections or lifted PCB tracks along that circuit path.
Then by opening the "-", SCR, Gate, and "+" screws, the small board which sits on the output side of the SR1 3-phase rectifier can be disconnected and moved aside. The welder can then be powered on and you can check for a DC voltage of about 240v across + ( terminal R2 ) and - at the rectifier.
I suspect that AI may be correct in thinking that the internal thyristor in the rectifier has failed or is not being triggered, but that fault has caused the pre-charge path to be burned when attempting to supply full welding current by that route instead of via the thyristor.
It has misread the resistance reading on your meter from the rectifier "constant positive" DC output to the +ve terminal screw of capacitor C3. It thinks the value is 20.16 Ohms, but I see it as 201.6 Ohms. ( Does your meter read to a hundredth of an Ohm anyway? )
It is correct in assessing 20 Ohms as a reasonable value for a pre-charge resistor, but the actual 200 Ohms is too high.
However, if AI thinks that the pre-charge resistor is 20 Ohms, then the bus voltage should rise as the capacitors charge, so the first point of the AI section "This explains everything", i.e. "No bus voltage rise", is illogical. The second point "No relay click" is wrong, since the role of a soft-start relay is fulfilled by the thyristor here. In the Miller XMT 304 thread that I linked above, their use of rectifiers with a built-in pre-charge thyristor was confirmed by @Jan-DIY , there is no relay.
It is also illogical for AI to blame the "No bus voltage rise" on "The internal thyristor section of SR1 is failed open or not being triggered", because that thyristor gets triggered to short the pre-charge resistor and pass full load current when welding, and all this would happen after the soft-start delay, when the bus voltage would already have risen.
Having said that, AI may still be correct that SR1 is faulty, but that is not proven yet....
Please confirm that the resistance of the pre-charge path from the +ve of the rectifier ( terminal R2 ) to the +ve ( right-hand ) terminal screw of capacitor C3 is 200 Ohms not 20. Then identify the resistive component - resistor or thermistor - and measure resistance directly across it. Look for bad or burnt connections or lifted PCB tracks along that circuit path.
Then by opening the "-", SCR, Gate, and "+" screws, the small board which sits on the output side of the SR1 3-phase rectifier can be disconnected and moved aside. The welder can then be powered on and you can check for a DC voltage of about 240v across + ( terminal R2 ) and - at the rectifier.
I suspect that AI may be correct in thinking that the internal thyristor in the rectifier has failed or is not being triggered, but that fault has caused the pre-charge path to be burned when attempting to supply full welding current by that route instead of via the thyristor.
Last edited:
JOEPRO
Member
- Messages
- 5,290
- Location
- UK
Turns out I was measuring the DC wrong.
I tried the above test this morning before work and measured the pins below going by the + & - symbols on the board itself.
But this was WRONG and I should be measuring from the top left - terminal and + R2. I will re check this when I get home from work later.
I tried the above test this morning before work and measured the pins below going by the + & - symbols on the board itself.
But this was WRONG and I should be measuring from the top left - terminal and + R2. I will re check this when I get home from work later.
JOEPRO
Member
- Messages
- 5,290
- Location
- UK
The results and AI exhange just now -
Nice catch—that 221 V DC reading is exactly what we needed.
Good, that 200 Ω reading is now a solid, repeatable fact—not a fluke.
Here’s what that tells us and where to go next.
Nice—those numbers are really telling a story now.
You’ve got:
In other words:
Grab that Gate–SCR(−) reading and we’ll know which side of the line the fault lives on.
That 0.00 V on Gate–SCR−, even right at power‑up, is the key result.
Here’s the clean picture now:
In other words:
See my thinking
Nice catch—that 221 V DC reading is exactly what we needed.
1. What 221 V DC tells us
- SR1’s rectifier section is working.
- With 240 V AC fed in single‑phase, ~220–240 V DC is a believable result under light/no load.
- So the fault is not “no DC from SR1” anymore—it’s what happens after R2.
2. Next tests to run
A. Do the bus caps actually see that DC?
- Measure DC across C3 (and C4 if accessible)during power‑up.
- Black probe on cap negative, red on cap positive.
- If they stay near 0 V while R2–(−) is at 221 V → the path from SR1 to the caps is compromised.
- Check continuity from R2 to C3+(power off, caps discharged).
- You should see a low resistance path through the pre‑charge element.
- Right now you’ve seen ~200 Ω end‑to‑end—that needs to be pinned down.
B. Find and measure the actual pre‑charge component
- Trace from R2 to C3+ and locate the resistor/NTC in that path.
- Measure directly across its leads:
- If it’s ~200 Ω and looks cooked → likely damaged, needs replacement.
- If it’s meant to be ~20 Ω but reads much higher → same story.
- Solder joints
- PCB tracks
- Any connectors in that path
3. Check if the SCR is being told to fire
Once you’ve confirmed the caps aren’t charging properly:- Meter on DC volts.
- Black probe on SCR(−), red on Gate.
- Power on and watch for a brief gate signal (often a few volts).
- Gate signal present, but no conduction / no bus rise: → Internal SCR in SR1 is suspect.
- No gate signal at all: → Control board / gate driver path issue.
4. Where we stand now
- Rectifier half of SR1: OK.
- Fault is now between R2 and the DC bus (pre‑charge path, SCR section, or control/gate drive).
- Voltage at C3 during power‑up
- Direct resistance of the pre‑charge component
- Whether you see a gate signal
See my thinking
Good, that 200 Ω reading is now a solid, repeatable fact—not a fluke.
Here’s what that tells us and where to go next.
What 200 Ω from R2 to C3+ really means
Between R2 (SR1 constant +DC) and C3+ you should have:- A deliberate pre‑charge element (resistor or NTC), typically on the order of tens of ohms, not hundreds
- Solid copper / tracks / joints around it
- The pre‑charge component itself has drifted high or been cooked
- There’s extra resistancefrom:
- Cracked solder
- Burnt or lifted PCB track
- Bad connector / terminal
Next concrete steps
- Physically trace the R2 → C3+ path
- Follow the copper / wiring from R2 to C3+.
- You’re looking for:
- A power resistor (ceramic, wirewound)
- Or an NTC thermistor (often disc‑shaped)
- Measure the component directly
- Power off, caps discharged.
- Meter across the two leads of that part only.
- Compare:
- Does it read ~200 Ω like your end‑to‑end measurement?
- Does it look heat‑stressed, cracked, or discoloured?
- Inspect the joints and tracks around it
- Reflow any dull/cracked solder.
- Look for:
- Darkened PCB
- Hairline breaks in tracks
- Loose screws / terminals
After that
Once you’ve:- Identified and measured the actual pre‑charge part, and
- Either confirmed it’s bad or the path around it is damaged,
- Replace that component / repair the track, power back up, and recheck:
- Bus voltage on C3
- Whether HELP‑6 clears
- Or, if the pre‑charge path checks out but the bus still doesn’t rise, we move on to:
- Gate signal measurement (Gate to SCR−)
- Deciding if the SCR section of SR1 or the control board is the remaining culprit
See my thinking
Nice—those numbers are really telling a story now.
You’ve got:
- R2 to −: ~221 V DC
- C3/C4: ~161–162 V DC, stable
- The rectifier is fine (we already knew that from the 221 V).
- The bus caps are charging, but only up to about two‑thirds of the rectifier voltage.
- They’re not being pulled up to full bus voltage.
In other words:
- The path through that ~200 Ω is working enough to bring the caps up to ~160 V.
- But the SCR inside SR1 is never being turned fully on, or it can’t conduct even when asked.
- So the caps never see the full 221 V that exists at R2.
- Control side issue:
- The control board is not sending a gate signal to SR1.
- Power device issue:
- The control board is sending a gate signal, but the SCR inside SR1 is dead/open.
- Measure gate drive during power‑up
- Meter on DC volts.
- Black probe on SCR(−).
- Red probe on Gate.
- Power on and watch what happens in the first few seconds.
- If you see no gate voltage at all: → Control board / gate driver problem.
- If you see a clear gate signal (a few volts) but the bus stays at ~160 V: → Internal SCR in SR1 is not responding → SR1 is bad.
- (Optional but useful) Measure DC between “+” and − on SR1
- That’s the output after the SCR.
- If R2–(−) is 221 V but +–(−) sits near the same 160 V as the caps, it confirms the SCR never fully connects R2 to the bus.
- Pre‑charge path: working, though high.
- Caps: connected and charging.
- Rectifier: good.
Grab that Gate–SCR(−) reading and we’ll know which side of the line the fault lives on.
See my thinking
That 0.00 V on Gate–SCR−, even right at power‑up, is the key result.
Here’s the clean picture now:
- SR1 rectifier: OK → you’ve got ~221 V DC at R2–(−).
- Pre‑charge path: High (~200 Ω) but working → caps sit at ~161–162 V.
- Caps: Connected and charging.
- Gate drive: Absent → 0 V between Gate and SCR− at all times.
In other words:
- The machine is happy to let the bus sit at a “limbo” pre‑charge level.
- It never takes the final step of fully energising the DC bus.
- That’s exactly the behaviour you’d expect if the control board:
- Has a failed gate‑drive circuit (opto, driver transistor, series resistor, track, connector), or
- Is deliberately inhibiting enable because some other condition isn’t met (fault sense, missing supply, interlock, etc.).
Where I’d go next
- Trace the gate drive path physically
- From the control board connector that feeds SR1’s Gate and SCR−
- Through any series resistors, optos, or driver transistors
- Look for burnt parts, cracked joints, or broken tracks.
- Check continuity
- From the control board gate‑drive output pin to the Gate terminal.
- From the control board reference/return to SCR−.
- Any open here = no gate signal at SR1 even if the logic is trying.
- Confirm low‑voltage supplies on the control board
- If a driver IC or opto isn’t powered, it won’t ever produce a gate pulse.
- Check the usual suspects: 5 V, 12 V, 15 V rails if they’re accessible and labelled.
- Keep SR1 replacement in perspective
- The rectifier half is good.
- The SCR might still be fine—right now it’s just never being asked to work.
- I’d fix/understand the missing gate drive first before condemning the SCR section.
Apart from being annoyingly chatty, I think that AI is doing quite well now!
However, there's two points that I would like to question:
- is 200 Ohms a valid design value for a precharge resistor? It seems too high for me, I would expect less than 50 Ohms.
- can a DVM "see" a thyristor gate trigger pulse? I think it cannot, you need a 'scope or a logic pulse catcher probe.
Instead of dismantling everything to get to the components and circuit path, it might be possible to get a better clue about the integrity of that "200 Ohm" precharge value by testing it with more current than just a meter set to Ohms. I'm thinking of something along the lines of a 12v car battery and a 60W H4 bulb...? ( Obviously with mains power off.) That might tell you if it is really a 200 Ohm resistor, or a burnt 20 Ohm resistor and/or a lifted track.
From the EV world I know that a Tesla uses a 23 Ohm precharge resistor, a Nissan Leaf uses 30 Ohms, and 230v filament lightbulbs are often used for tests. You could try powering up with a 60W mains bulb jumpered from R2 to capacitor C3, which would bypass the suspect 200 Ohm precharge resistor. The bulb is about 40 Ohms cold.
However, there's two points that I would like to question:
- is 200 Ohms a valid design value for a precharge resistor? It seems too high for me, I would expect less than 50 Ohms.
- can a DVM "see" a thyristor gate trigger pulse? I think it cannot, you need a 'scope or a logic pulse catcher probe.
Instead of dismantling everything to get to the components and circuit path, it might be possible to get a better clue about the integrity of that "200 Ohm" precharge value by testing it with more current than just a meter set to Ohms. I'm thinking of something along the lines of a 12v car battery and a 60W H4 bulb...? ( Obviously with mains power off.) That might tell you if it is really a 200 Ohm resistor, or a burnt 20 Ohm resistor and/or a lifted track.
From the EV world I know that a Tesla uses a 23 Ohm precharge resistor, a Nissan Leaf uses 30 Ohms, and 230v filament lightbulbs are often used for tests. You could try powering up with a 60W mains bulb jumpered from R2 to capacitor C3, which would bypass the suspect 200 Ohm precharge resistor. The bulb is about 40 Ohms cold.
