Ramped SSTC - Power Supply Question + General Improvements

[ZakW]

That entire node (entire copper shape) could use 1-1.5mm clearance, including at the secondary output connector lower pin (both layers), around mounting hole, and gap to other copper shapes.  Probably OK at 0.6mm, but a bit more would be good.  If connector pads themselves are closer together, may not be worth the work of editing the connector part itself in layout tool.  Even 0.6mm will typically handle 500V fine for the limited total use time of hobby equipment.

I was wondering if the additional spacing would make a difference in the end if the connector pins were only 0.6m anyway. To be safe I updated the part to a 1x3 header and just wont use the middle pin. That should give me 1.2mm total spacing without too much effort. I also increased the pad clearance to the neutral plane to 1mm for C19/C20 as well as the output pins.

[davekni]

I was wondering if the additional spacing would make a difference in the end if the connector pins were only 0.6m anyway.  To be safe I updated the part to a 1x3 header and just wont use the middle pin.

Are the pins themselves 0.6mm spaced?  That would imply a very tiny connector.  Or is it the ECB pads of the connector part that are 0.6mm?  If it is the latter and it became a problem, the pad copper could be cut back a bit with a knife.  That's why I was a bit less concerned about that one location.  However, your 3-pin connector is a great solution.  I've seen that used in commercial equipment for the same reason.

[ZakW]

Are the pins themselves 0.6mm spaced?  That would imply a very tiny connector.  Or is it the ECB pads of the connector part that are 0.6mm?  If it is the latter and it became a problem, the pad copper could be cut back a bit with a knife.  That's why I was a bit less concerned about that one location.  However, your 3-pin connector is a great solution.  I've seen that used in commercial equipment for the same reason.

The connector pins are 3.96mm on center with a 0.8mm gap between them. The thin yellow line is the 1mm clearance I added.





Updated front and back copper planes:

[davekni]

The connector pins are 3.96mm on center with a 0.8mm gap between them. The thin yellow line is the 1mm clearance I added.

Thank you for the images.  Looks great!

I was misunderstanding just due to semantics.  To me "pins" refer to the metal pins of the physical connector.  The ECB shape for the connector includes pads to which the pins are soldered.  It is those ECB pads that have 0.8mm space between them.

[ZakW]

Thank you for the images.  Looks great!

Thanks! Excited to get them ordered. While I wait I can start working on the enclosure.

I was misunderstanding just due to semantics.  To me "pins" refer to the metal pins of the physical connector.  The ECB shape for the connector includes pads to which the pins are soldered.  It is those ECB pads that have 0.8mm space between them.

Good point, apologies for the confusion.

[ZakW]

V7 is up and running!!





Aside from KiCads footprint for the SOT-23 version of the 3904 transistor pins being flipped I haven't found any other mistakes.

So far this seems the most stable of all the builds with a very consistent output.

---------------------------------------------------------------

I am still testing and making adjustments. I wound a new smaller coil to get closer to ~450-420kHz range. I am hitting 20in (50.8cm) but I still need to fine tune C19/C20 as well as the primary placement.

The new coil measures in at 1.73in (44mm) using 44awg wire = Output is roughly 11.6x the length of the secondary 





---------------------------------------------------------------

Couple questions regarding the spikes I am seeing on Q1 & Q2 Vgs:

• Using my differential probe set to 50x, are the spikes really there or just an artifact of scoping/long probe length?
• If they are really present, are they harmful to the IGBTs and how might I reduce them?

Here is a close up of Q1 Vgs




I noticed that the spikes develop over time before eventually going away as the coil ramps back down... Here is the beginning, middle, and end.




Here is the Vgs waveform without power to the bridge:


Finally, here is a comparison of the output of the UCC vs Q1 Vgs over time.
Purple = UCC output (after capacitor)
Blue = Q1 Vgs





Again you can see the spikes get worse in the middle of the on period before disappearing again.

---------------------------------------------------------------

Other than that the performance seems great and I am very happy with it. I just want to make sure this issue isn't a design flaw or something else that could hurt the IGBTs.

[davekni]

Is your positive differential probe connected to gate and negative to emitter?  Or reversed (ie. scoping Veg instead of Vge)?  As shown, Vge rise time is shorter than fall time, which would suggest that D6 and D7 are backwards.
Are you scoping low-side IGBT (one with emitter to neutral)?  That reduces affect of differential probe leads.  Twisting leads together from probe up to clips also helps.

[ZakW]

Hello Dave,

Well I feel silly, D6 & D7 were backwards (+1 PCB error). Thank you!. But the scope shots still look really similar, maybe slightly faster turn off.

I confirmed I am probing positive lead to pin 1 of Q1 and negative lead to pin 3.

For the UCC if it matters I am probing output B after the 3.3uf cap, other lead is going to ground.

Here are some more captures after correcting D6/7.

Q1 Vge without powering the bridge


Q1 Vge Beginning vs Middle



Q2 vs UCC output



I overlaid Q1 and Q2 and made one transparent. If I understand correctly, it looks like Q1 (brighter blue) is turning off before Q2 is turning on. Which is what I am after right?



Here I overlaid Q1 before and after fixing D6/7. Green trace is with the diodes fixed showing the IGBTs turning off more quickly, right?

[davekni]

But the scope shots still look really similar, maybe slightly faster turn off.

Yes, looks to me like slower turn-on too, as it should be.

I overlaid Q1 and Q2 and made one transparent. If I understand correctly, it looks like Q1 (brighter blue) is turning off before Q2 is turning on. Which is what I am after right?

Yes.  Gate signals by themselves don't guarantee turn-off is before turn-on.  Most IGBTs have less turn-on delay than turn-off delay.  I'd guess you have sufficient dead-time (enough extra gate time to accommodate difference in IGBT delays).

The spikes in Vge are most likely caused by Cdg (Miller charge).  They likely occur simultaneously with half-bridge output transition.  That is one reason they are worse at the middle where half-bridge voltage transitions are largest.  I doubt the spikes are causing any problems.

[ZakW]

Yes, looks to me like slower turn-on too, as it should be.

Thanks for confirming that. I am glad it looks correct now.

The spikes in Vge are most likely caused by Cdg (Miller charge).  They likely occur simultaneously with half-bridge output transition.  That is one reason they are worse at the middle where half-bridge voltage transitions are largest.  I doubt the spikes are causing any problems.

I appreciate the context! I tried googling the Miller charge and waveform examples to see if I could find something similar before asking but I was unsuccessful. Glad to know they are not causing any issues.

[davekni]

I tried googling the Miller charge and waveform examples to see if I could find something similar before asking but I was unsuccessful.

Yes, classic Miller charge effect is a flat spot (plateau) in Vge.  However, in cases where Vge has finished transitioning before IGBT output switches, the Miller charge becomes a glitch in Vge.

[ZakW]

Question on grounding my coil while I am working on the enclosure for it. 

For all of my testing I have been using the mains ground via a cheap line filter I picked up on Amazon - https://www.amazon.com/Comimark-Filter-CW1D-10A-T-Suppressor-Power/dp/B07X3C345J/ref=sr_1_3_pp?keywords=ac+line+filter&qid=1694802673&sr=8-3 wiring my Secondary directly to it. I have tested a quick one sheet tinfoil counterpoise but I noticed the arcs were louder (snapping and popping). I assume the overall size was not providing sufficient capacitance to sink currents from the coil?

Ideally I would like to run this coil in several environments, home and work (standard commercial office) mainly. But I am concerned with EMI/RFI in the office area especially using the mains ground (I understand this is not suggested). What are my options? I was going to try testing a larger counterpoise, spray gluing tinfoil to a 18x18in piece of thin plywood to see if the arc appearance and sound matches what I see when I am connected to my mains ground.

Is my basic line filter doing much at these frequencies (~400kHz)?

For running it at the office it would be in a large open conference room not near computers or other office equipment... I also run the coil at very low BPS. I did see a nicer line filter with 2 inductor stages as opposed to the single stage mine has, might that be a better choice?
 

[davekni]

I have tested a quick one sheet tinfoil counterpoise but I noticed the arcs were louder (snapping and popping). I assume the overall size was not providing sufficient capacitance to sink currents from the coil?

Perhaps.  If too small, maybe snapping and popping is due to some small arcs between bottom of secondary and primary, if secondary bottom isn't close to ground potential?  I don't have personal experience here.  I always ground everything to both a large counterpoise and to line ground.

Is my basic line filter doing much at these frequencies (~400kHz)?

I'd guess so.  400kHz is above many power supply PFC input stage switching frequencies.  Filtering applies to line and neutral.  Ground is not filtered.

[ZakW]

Perhaps.  If too small, maybe snapping and popping is due to some small arcs between bottom of secondary and primary, if secondary bottom isn't close to ground potential?  I don't have personal experience here.  I always ground everything to both a large counterpoise and to line ground.

I haven't noticed any arcing from secondary to primary when using the counterpoise. The counterpoise was about 1.5in below the coil base. The popping sounded more like when I add too much smoothing capacitance and I flatten the mains half cycle ramp.

I'd guess so.  400kHz is above many power supply PFC input stage switching frequencies.  Filtering applies to line and neutral.  Ground is not filtered.

I was thinking about making a custom filter for my frequencies but I am not sure at this point. My coil is physically small and relatively low powered compared to other builds I have seen. Will grounding it to mains really be an issue?

Also, is using a counterpoise inconjunction with grounding to mains better than not using a couterpoise? Grouding is still a confusing topic to me.