Ok, I soldered in my ERZ-V10D151 varistors from Drain to Source on each MOSFET
and retested with V+ at 12V and again at 24V, and it is seems to be working acceptably.
With the 10uF cap disconnected from V+ now, I get no heating on the UF4007 diodes even at V+ = 24V.
The MOSFETS stay completely cool to the touch as well at 24V.

The only components that are getting hot at 24 V are the 220 Ohm 2 Watt resistors I am using
in the series RC snubbers, so I will eventually change those out with 5 Watt ceramic block resistors
and I should be able to run steady at 24V without the resistors overheating.

The non-dissipative snubbers are working OK using the UF4007 diodes with no electrolytic
connected to the yoke center tap, so I think I will stick with this for now unless I run
into new problems later on when I am doing my experimenting with different setups.
So there you go. The component values I am using may not be optimal, but they are at
least keeping the switching spikes below 200V, so not so bad I guess.

Itsu, I notice the current limiting light on my bench power supply is flickering on when I
use it to power the PWM driver board, so that probably accounts to some extent for the drooping
waveforms you mentioned. Something odd is that when I put even just a 100uH choke in the
power supply V+ line, it chokes out the rectangular waves on the toroid primary. I just get
big switching spikes but the rectangular waveforms are all degraded. What choke value
are you using in your power supply V+ line, if you are using one?

All the best...

How I known and understand, shottky diodes have wery small input capasistance, so they not works like capasitors, UF diodes or other, works like capasitors partialy. Expeshaly low voltage shottky diodes are wery fast, much faster (maybe 10 or 100 times), than UF and SF diodes and other silicon diodes and almost not have capassitance. So maybe this peaks in Void oscilograms is from slow UF diodes, whose works like capassitors? And to reduce peak voltage on snubbers, maybe need use bigger capasistance capasitor in not dissipation snuber?

Void,
thanks for this info, i wonder why this 10uF cap has such an impact, it should only more filter the 12/24V supply voltage.
I have build a small pcb with a non dissipative snubber consisting of 1mH inductors, 0.1uF caps and UF4007 diodes.
I will be testing tonight.   
I have some BAT46 schottky diodes  (100V), but i doubt they can handle the abuse like you mention. 
Else some D12G60C schottky diodes (12A / 600V) are available too.

The last weeks i was using batteries for both 12V and 24V supply voltages, so without any filter chokes or caps.
Since the last video i use a 24V (and 12V) cheap led PS with added common mode choke (7mH) and filter caps (900uF elco, 230nF poly and 0.1uF ceramic) on its output.
For protection i have a 6A fast fuse in the 24V line to the yoke.
Itsu

Hi Itsu. It will be interesting to see how your non-dissipative snubber compares
to my setup. I also still don't understand why my adding the 10uF cap at the toroid primary
center tap causes some of the UF4007 diodes to get hot. I should test with that further as I was
getting better switching spike reduction with the 10uF connected.

I find it very odd that when I place a 100uH choke in my power supply V+ line
that it degrades the rectangular waveforms quite a lot. I need to investigate this further
as well to see if I can figure out what the problem is there.

How I known and understand, shottky diodes have wery small input capasistance, so they not works like capasitors, UF diodes or other, works like capasitors partialy. Expeshaly low voltage shottky diodes are wery fast, much faster (maybe 10 or 100 times), than UF and SF diodes and other silicon diodes and almost not have capassitance. So maybe this peaks in Void oscilograms is from slow UF diodes, whose works like capassitors? And to reduce peak voltage on snubbers, maybe need use bigger capasistance capasitor in not dissipation snuber?

Hi Menof. You may well be right that using Schottky diodes could give a lot
better results. It is worth trying to compare the results. I would have to order some,
as the Schottky diodes I have on hand only have a reverse voltage rating of 100V and 60V.
Yes, changing the C value might give better results as well.

All the best...

Hi Itsu. That's looking pretty good. I am not clear if you are using two
non-dissipative snubbers, one for each MOSFET?

I will reconnect my 10uF cap at V+ and retest to see if any of my UF4007 diodes
still get hot now that I have the varistors connected in.

All the best...

Hi Itsu. Ok, thanks for doing that test. It is helpful to compare results.

My UF4007 diodes still get too hot when I connect the 10uF back to V+,
expecially the D3 and D4 diodes. I get a lot cleaner waveform and better
spike reduction with the 10uF cap connected however. I think my varistors
do not have a low enough clamping voltage however. The max clamping voltage
for my varistors is listed as 250V, so I think I will have to get some varistors with
a lower max clamping voltage.

Also, I wonder how the non-dissipative snubbers would work if the value of C is reduced somewhat.
I may try that, as it may be a way to reduce the current in the diodes and keep them cooler.


All the best...

Hi ,

The readings on scope shots are interesting with the non-dissipative snubber circuit as shown . I wonder what It would do to the device.
gives something like a spike at some point.

Currently modding the Akula Kacher, it looks like the output signals were not perfected as how it should be, my dead time to form nice spike
has to be adjusted ( Collapse time ) , might be that this is limiting me to get more out at output.
Have tried many points of tuning( many harmonics of the same spot) but nothing increased, but last video was best result till now,
it's all about the Tesla coil operation spikes that make the differences .

cheerz~

Hi Itsu. Good video! 

My coils in the non-dissipative snubbers seem to just get warm but not hot at 24V, 
by my coils are maybe using thicker gauge magnet wire on them. I think the wire gauge
is either #20 or #22 AWG. The coils are rated for 1.9A, and specified as 0.28 Ohms.
A coil with a higher resistance would likely show more heating.

Since the UF4007 diodes get hot when using 24V, I think that if anyone wants to
try this type of snubber it would be better to go with Schottky diodes that have a reverse
voltage rating of at least 500V and a current rating of at least say 8A or 10A.
Then the diodes should stay a fair bit cooler, I would guess.

So, with coils made with heavier gauge magnet wire to keep the coil resistance low, and
suitable Schottky diodes it seems non-dissipative snubbers may be workable for use at 24V
without excessive heating.

I am still testing at 24V. I removed the UF4007 diodes and put in 12R06D diodes instead,
as I had some available. They look like they should be OK for use at 24V, but they may still get
on the hot side if running for a long time, but at least small heat sinks can be mounted on them if needed
for running for long periods of time.

The Allega snubber is maybe simpler to implement, and probably workable at 24V for long   
run times if a 10W or higher wattage total resistance is used.


GeoFusion: Thanks for the update.

How are you doing my friends? Nice to see the team again ready for new discoveries.

I just wonder, why don't you build Verpies lossless design to end this never ending push pull story? It is a rhetorical question. I know.  it is because you love to learn.. But for this specific job, i find Verpies design just what it needs. Lower consumption, no heating problems, clean drains. What else can someone may want?

All the best