induction heater troubleshooting help

[aes92000]

Hey all, hope everyone doing fine.
hope this also the right place to post

i have some questions to help me understand what may be happening with my new build.

i had a board working for a while till the pcb burned up on me



i have a 48VDC (~54VDC) @ 75A rectifier


my previous board, i could pull 54VDC @ 73A without problems and use for couple of hours.
i upgraded the original mosfets with IRF4668 (x4) which are rated @ 130A instead of the normal 30A
the fans for cooling are high cfm and kept things at very low temps
it is also water cooled






now that i have redone my board and ALL new components, i cannot get the it to pass 25A before the rectifier shuts down and the board still burned at end. i didnt have water cool on at that point
if i keep under 25A, i can use but the mosfet heatup to about 36-42C.
i added strips of stranded #8 copper to underside of board along cap banks thinking it may help since pcb copper is thin for high amps


questions:
#1) does it matter which way the choke coils run?
#2) do i need to keep working coils close by tank as possibly can?
     closer than the new coils i have made

#3) i have an oscilloscope, how would i connect the leads to heater to measure frequency?
    most of the videos i have watched do not explain how and where to connect leads
#4) what would cause new board to limit itself to 25A?


any help would be much appreciated

[petespaco]

Regarding question 3:
You can get the frequency reading by connecting the scope to DC ground and any Mosfet gate.
IF your scope's input circuit  is COMPLETELY isolated from the mains, you can connect across the work coil. (I have a little battery powered DSO-112 scope that works well for this).

Pete Stanaitis
---------------

[aes92000]

thank you Pete for reply

https://www.amazon.com/gp/product/B07RP9N2BZ/ref=ppx_yo_dt_b_search_asin_title?ie=UTF8&psc=1

this is the scope i have. 
pc based with usb powered. i love the little thing.
i am still a noob using scopes

i hope i understand correctly so i do not blow it up...
i could do either of the two connections with this scope.


being usb power not from AC mains, i can connect at the point where work coil connects to board?  is that correct.

could i use a DMM to get same reading? if can go that high

[davekni]

Your USB scope is most likely grounded through your computer.  If you are using a laptop on battery power only, then you could try connecting across the work coil.  Otherwise, always connect the scope probe ground lead to your circuit ground (presumably negative supply rail, the bottom plane of your ECB).  Then you can scope FET gate or drain (one side of work coil).

Looks like a great board and construction.  I'd guess your initial frying was resistance at the ECB-to-standoff interface.  What material are the standoffs made of?  Your new connections are likely better, but also depends on the material.  Make sure ECB and connection blocks are clean and free of oxide layers.  Light sanding before bolting together would be good.

For your new version, my first guess at issue is the larger gate capacitance of larger FETs making switching too slow.  That's presuming gate pull-up resistance didn't drop proportionately.

Even though the circuit is simple, a sketched schematic with devices labeled and values listed would be helpful.  That way it's easier to talk about scoping a specific circuit node.  Or, at least list values for the gate pull-up resistors, TVS diode (or whatever is across gate-source) and gate-source resistor, and diodes between drains and opposite gates.

The later build looks like it has charring between the two output connections.  Am I seeing that correctly.  Such would imply some contamination or other conductive path bridging the outputs or outputs-to-ground.

[aes92000]

here is the schematic and board design. EagleCad
the top layer (red) is the return side
bottom caries the +VDC, and coil connections




i just copied the board from the 1800W i bought from bangood.com.
https://www.banggood.com/Geekcreit-ZVS-1800W-12V-48V-40A-High-Frequency-Induction-Heating-Module-Without-Tap-p-1235621.html?rmmds=myorder&cur_warehouse=CN

the coil connections that are bolted to the new board are tinned copper with zinc screws in which i soldered the 1/4" copper tubing for brass connectors for easy swaps.

the old board was brass standoffs from the 1800W i bought
i believe that may have burned from changing coils and moving the standoffs enough to get loose enough to start the heating and burning

[klugesmith]

Yay for schematic with reference designators.  I was about to say, be sure to include them.  So we can all talk about "R2", for example, instead of having to identify it by component value and location in the circuit. 

Integers are still inexpensive, and it's OK to skip some (for example when a part goes away in a revised diagram).  Or leave the deleted part in the diagram, with an annotation like "not stuffed", as we see in many mass-produced boards.

[aes92000]

extra info...
i was able to use DMM for frequency

short coil measured ~105kH
long coil measured ~49kH

[davekni]

Coils look fine.  Don't think there's any issue with lead length.  Frequencies look reasonable too.

If you can scope FET gate and drain waveforms and post images (screen capture or whatever is easiest), that would help.  Scope at lower power and again just before shutdown (ie. 24A or whatever the highest you can get).  If you have two probes, simultaneous capture of gate and drain is ideal.  Connect scope ground probes to ECB top side (V- supply).

Is the new ECB charring between the two output connections?  If so, perhaps cleaning that area with sand paper and/or alcohol will remove whatever contaminants are causing conduction there.

At 100kHz, the added gate capacitance may be causing increased FET power dissipation (and temperature), but seem unlikely to be enough to cause oscillation to drop out.  I gather that your supply has some form of over-current shutdown.  Is that what's tripping - current jumps from 25A to something above the supply's capability?  Such current jumps are normal for ZVS oscillators like this if something causes oscillation to stop, such as too low resonant Q or FET forward drop too high to pull the opposite FET gate low enough.

BTW, do you know the current rating for the power-feed inductors?  73A total seems high for the pair just by looking, but I have no specifications.  Even if you keep them cool with high air flow, magnetic saturation of the cores limits useful current capability, as inductance drops too low.