Killing zvs boards

[gostavs]

Being an idiot is doing the same thing over and over again expecting different results. I’ve tried and fried the half pint about 20 times now, I guess I’m an idiot!
Can someone spot something I’ve missed or fucked up? I’ll attach a few examples.
They all have been tested in stock configuration before so I know they work, and I’ve wetted the fets to ensure the connections. I believe the diodes are in correct position. The boards are from 3 different orders/resellers.
Shortening the work coil/less turns - immediate explosion of one or two fets when power is turned on.
Lengthening/adding turns on the work coil - adds 10-20 seconds before the mosfet blows.
The fets burn before I press the fire button, it only needs positive and negative connected to smoke/explode.
I have tried with different brand new fets, no change.

 

[gostavs]

I cut the trace just below where the centertap wire is. I cut the trace through the copper and remove it so the white pcb shows.
The caps legs go through the hole and is well connected on the other side.
Fire button is added between positive centertap in and the original positive in hole.
I carefully remove the epoxy on the negative trace between the fet legs and wetted/soldered negative input.
I clean and test every connection before and after all soldering with a multimeter.

 

[gostavs]

I’ve practiced a lot..

 

[badbee]

The expert is @TommyDee , hopefully he'll see this and come comment. If it does this even when you haven't closed the switch you added then that switch must be in the wrong place. Did you follow TommyDee's guide exactly?

 

[TommyDee]

The image host is unavailable atm so I cannot zoom in. I see nothing inherently wrong. Shorts, or maybe even arcs, could be causing an issue but you took care to insulate.

These FETs should be cool-running at idle (w/o cap inserted and firing). That is where the testing begins. The module should not draw over 1.5 amps firing without the Vapcap.

Blown FETs would signify not oscillating. This is where the quality of the center tap comes into play. Source positive and coil positive have to be at the same potential or the FET is turned on into a short circuit. These are DC-AC converters in their simple form, and the output has a 12 gauge wire across it. A current meter (ammeter) will go a long way toward diagnostics.

Do know you can replace those FETs carefully harvested from the MOSFET switch boards. That's once you worked out the glitch.

 

[gostavs]

Thank you guys, you are the masters here.
When testing them in stock configuration they draw 1.3-1.5A at idle. The coil with 10.5 turns draw >~5.1A, the 13.5T draws >~4.5 with the VC inserted.
These links/pictures might work better:
Here you can see the left fet has actually blown to pieces:


The centertap and "subassembly" I have checked again and again. This one came out pretty nice i believe, with wet solders and a good positive connection; Does that count as a high quality centertap @TommyDee?


Heres the other side of that centertap with the positive connection and a fire button.

Blown FETs would signify not oscillating. This is where the quality of the center tap comes into play. Source positive and coil positive have to be at the same potential or the FET is turned on into a short circuit.

I cant figure out where the short is. What do you mean by "Source positive and coil positive have to be at the same potential"? How would i verify this? When everything is soldered, before power is added the values and connections look as they should as far as I can tell.
About the trace cut, does that have potential to damage or short anything? Does it matter if its deep and goes through the copper plane?

Yes, saving the boards for later when things hopefully works out, I have new replacements fets. Ive learned to change them pretty easily by now Im out of new boards right now. I will try to salvage any remaining working fets and make a working board to measure the current on before it blows. Last time i tried to measure the current it blew before I could get a reading.

 

[TommyDee]

I see the switch; that is having the right voltage in the right places.
The trace cut should be through to the fiberglass.
The numbers you are getting from the ammeter are what I would expect. The HalfPint easily does 8 amps or around 100W.
Good wetting of the solder joint on the centertap is critical as all the current is driven through this joint. If it is a cold joint, the resistance can cause issues.
Check to make sure the grid pattern on the board surface (FET side) is still isolated from the solder connections.

Last - are you operating the coil sitting on a metal surface? That can present a massive load.

 

[gostavs]

Grid pattern is isolated, no other metal close by.
I replaced a pair of fets with salvaged ones that had alright numbers on a “fried” board. It doesn’t fire, only lights up the led when fire button is pressed. Doesn’t draw any current at all. Will try another board and pair of fets tomorrow.

 

[badbee]

@gostavs , when I've run into problems where I seem to do be doing something mysteriously wrong, repeatedly, I've solved it by intentionally doing everything differently. Something you are doing, probably something that was never a conscious decision in the first place, is leading to this bad outcome. Maybe try doing it again but try to do every step slightly differently with modified layouts. A new route to the desired outcome might not lead through the mine field you are getting stuck in.

Watch for any lack of symmetry in the mosfet layouts. The oscillation very much depends on the two sides of the circuit being almost identical except for the tiny variation internal to the mosfets which allows one to turn on first, forcing the other momentarily off. If you get excess resistance or capacitance on one side it might not oscillate resulting in the one firing mosfet burning up in a few seconds.

Just a thought...

 

[TommyDee]

Other more sinister problems happened when coils short to themselves or the core. The work coil cannot have a short, typically not a problem; the choke shorted to the core can be a big problem. Haven't had it happened to me but chipped paint on the core is a warning.

The way that the circuits works is to dump energy from each coil to the capacitor. The capacitor dumps its power through the coil when the chokes are at the AC crossover.

No current on the unit with replaced FETs is odd. Check for voltage on the gate while firing.

 

[gostavs]

@gostavs , when I've run into problems where I seem to do be doing something mysteriously wrong, repeatedly, I've solved it by intentionally doing everything differently. Something you are doing, probably something that was never a conscious decision in the first place, is leading to this bad outcome. Maybe try doing it again but try to do every step slightly differently with modified layouts. A new route to the desired outcome might not lead through the mine field you are getting stuck in.

Thank you, that was exactly my thinking the first 15 times. I thought Id maybe just did a little mistake here and there, so i tried to change the trace cut, move a wire/button/positive/negative, change coil placement/thickness length/turns etc. Still trying to do things different and be mindful of every step I take. I take extra time to get the fets symmetric and not too much solder on the legs.

Other more sinister problems happened when coils short to themselves or the core. The work coil cannot have a short, typically not a problem; the choke shorted to the core can be a big problem. Haven't had it happened to me but chipped paint on the core is a warning.

The way that the circuits works is to dump energy from each coil to the capacitor. The capacitor dumps its power through the coil when the chokes are at the AC crossover.

No current on the unit with replaced FETs is odd. Check for voltage on the gate while firing.

I checked the diodes on the board i replaced the fets on yesterday, they were damaged as well.

Today was good and bad as you will se later on..
I found another board with diodes that looked alright on the multimeter, removed the subassembly, wires and fets.

The vias(?) came off but that didnt matter since I was planning on using the "Out" holes instead this time.


Built a new subassembly. This time I used the legs from the choke to attach through to the other side of the board to get som extra strength and contact. The third leg is soldered on the middle "out" hole and is soldered onto the choke on the outside instead of inside. I found that was closer to the middle.


Another different thing this time was I soldered the work coil before the fets to avoid unnecessary heat.

Then the fets, negative, positive and fire switch.



This one looked pretty good!

 

[TommyDee]

Looks promising...

 

[gostavs]

And it was! This was the first test, I’m scared to push the button since it has blown so many times https://streamable.com/l77ej3

Amazing! I was stunned that it finally worked!

Then I moved it, changed nothing else. Flipped the main switch, not the fire button, and: https://streamable.com/8vo1xt

 

[gostavs]

569w and 51A later the board is toast to say the least.

I checked the surfaces and table with a metal detector to rule out nearby metal! Nada.
Everything looks exactly like the first test. No components touching. Measured across several points to look for shorts, nothing. I barely moved/touched ut before the second test. This has me so confused.

 

[TommyDee]

51 amps! - that's a heck of a battery pack. I don't know what to tell you. Your setup is almost exactly what I am using to check gain ratio on caps.

 

[gostavs]

Yes 50c rate lipo can delinger som power!
Rescued another board today, new fets and subassembly again.
I installed a 10A fuse before the board to see if that could save it if it shorted, it did not. The fet didn’t explode to pieces like before, but still fried.
But at least now I can maybe see a pattern. Three times now have a board worked on the first try, ie: I turn on power, then press the fireswitch and it fires as I should with correct numbers. I shut power off and the next time power turns on it shorts before the fireswitch is pressed. What can make it short after one run before the fire button is pressed?

 

[TommyDee]

I'm leaning toward a cold solder joint on a high current leg. Show me that center tap solder connection at the choke.

 

[gostavs]

Here’s todays build:

 

[TommyDee]

That all looks great. Okay, I'm out of guesses.

 

[gostavs]

Okay thanks for your time and effort! I’ll order another batch of zvs boards and hope for the best with those. It’ll take a month for them to arrive so I’m getting a break from the IH projects for now. Maybe I’ll get a chance to complete either either a lathe, a forge, tools, or other ideas all spawning this weird multiplying hobby

 

[DrJynx86]

What Mosfets are you using? are you sure they aren't knockoffs? Are they being used to spec?

 

[gostavs]

What Mosfets are you using? are you sure they aren't knockoffs? Are they being used to spec?

Im buying the boards from china, no idea how to check if they are knockoffs. They are cheap so they might be?
My replacement fets is: NTD5865 N, 60V 46A
They are bought locally and have no indication of being fake. I noticed the last ones I replaced I had cold soldered, the solder iron is starting to give me problems. It’s shutting off temperature and is unstable. Have ordered new mosfets and are looking for a new iron to try and rebuild some boards with.

 

[DrJynx86]

Are you sure those mosfets are safe to use with your voltages? I did read the datasheet and cannot find anything about using them with 5-12v but I don't know as much electronics as you people, that's why I'm asking

 

[gostavs]

i don’t know for sure but the stock ones blows as well and they are rated for those voltages

 

[gostavs]

I may have got this sorted. The solderingiron was too weak I think, and generally of bad quality. The new station is digital and have 60W, and it was night and day in difference! Getting wet solders is a breeze now, it is so much faster.
I measured all the diodes on the fried modules, almost all of them were toast. Found one with good values, rebuilt it to stock but with new fets and it worked! It may actually have been a bad soldering iron the whole time.
With all of the earlier trauma with the half pint tries I still feared modding one of my only two working boards, so I tried some other well known mods instead. With a 6+5.5T coil, still very hot. Removed a cap, a lot better but still quite hot. Tops out at ~50w and 4A. Will try a 7+5.5T tomorrow. The quest to find what configuration fits you the best with the most effective voltages is fun!