UD2.7C thru-hole version populated and bench tested

[profdc9]

I populated and bench tested the UD2.7C thru-hole.  I had to adjust some of the component values on the undervoltage lockout, but it seems to work.

I plugged in a signal generator and tested the output of the MOSFETs, plugged in a signal generator into the overcurrent and tested that, tested undervoltage lock out, and pulsed a laser pointer into the optical receiver to simulate an interrupter pulse.  All of it seems to work.

Here's a picture below:



Dan

[oneKone]

Man that looks slick! How far off are you from testing it in a coil?

[profdc9]

So I just finished winding the secondary.  That was a chore!  1000 turns on a 6" PVC pipe   I insulated all the turns in a thick layer of marine epoxy.

I need to build the primary coil support and the "table" to hold everything.  I hope to put it on casters so I can wheel it around the University for days when we want to show off "science" to visiting students.  My present coil is a little too big for that.  The new one should have everything able to be disassembled, so I can remove the topload and primary coil, throw it in the back seat of the car, and move it around.

I have a circuit board also for the full bridge, and a way I think to build a low inductance power bus by weaving pairs of the positive and negative bus wires together.  So a little ways to go but I'm getting there.

Building out the through-hole board was pretty easy.  There are about 120 parts or so, but nothing that can't be done in 2 or 3 hours at a moderate pace.

Dan

[oneKone]

Would you consider selling blank boards or even the Gerber files?

[profdc9]

I have included the Kicad project below which contains the Gerber files.  There is a bill of materials and schematic in there as well.  I have some extra boards, but these can be ordered for $5 from Seeedstudio or jlcpcb.com, the most expensive part is shipping, and you will get 10 of them, so it will probably be about the same cost to have them made for yourself as to have me send you one.  Just zip up the gerber directory and submit it to them, it should work.

I have a new version of this board I am working on called UD2.9 which will include pulse skip functionality by setting a jumper.  It has only a few changes from UD2.7C.  I have not finished it yet but it should be done soon.  The designs are intended to be somewhat "future proofed" by using parts that are as readily available as possible.

I am going to release all of the PCB's I am working on in the next revision of the "PCB Pack" when I have some revisions to show.

Dan

[oneKone]

wow... thank you for your generosity and releasing these publicly. i'm looking forward to getting the boards made and testing it out

[profdc9]

I am working on getting a coil working with these boards too.  If you have a problem let me know.

Dan

[oneKone]

No worries.

I ended up ordering from seeedstudio, shipping was a bit of a bummer but I think still cheap overall. I can honestly see these boards being big... Especially since not everyone enjoys surface mount components.

[Mads Barnkob]

I am really looking forward to see your coil running with this driver board

Please do keep updating on your original thread with the pcb packs, when I get the time, I will make a sticky post with the best/important/interesting DRSSTCs threads in

[profdc9]

I'll get a new version up of the PCB pack, of course, but there's two experimental boards I haven't tested yet, the UD2.9 and UD3.1 thru hole versions.  The UD2.9 changes the logic so that pulse skip functionality may be added by setting a jumper.  Pulse skip functionality skips a positive and negative pulse, so it's not true freewheeling through the bridge (it doesn't even have two gate drivers) but should add some capability for those who want to try it.  The UD3.1 uses the CY8CKIT-059 PSoc5, which I posted earlier.

I have also been thinking about how to make a simple board 2-layer board for trying to parallel IGBTs without them blowing up because of uneven current sharing.  I figure this is already done in IGBT bricks so it should be possible to do it on a PCB.  As far as I can tell, the following is necessary:  1.  Thermally joining the dies as closely as possible on the same heatsink.  This I do by having them all next to each other spaced exactly 20 mm apart so a heatsink can be precisely drilled to accommodate them, and the common collectors of the IGBTs can all be in contact with the heatsink.  2.  Using a gate resistor and diode on each IGBT, but only a TVS diode on the gate driver output on the PCB, not directly at the gate of the IGBTs.  3.  Minimizing the loop inductance of the gate-emitter drive.  I think this could be done by adding two points on the board where a jumper wire is soldered that can be run in parallel or even wrapped around the gate drive resistor.  Two half bridge boards could be used together to make a full bridge.  The only thing I think needs an extra layer is the parallel run of the gate-emitter signal, so I think this could be done easily just by soldering in a jumper wire and avoid a 4 layer board.

Anyways, I just figured out I was doing some stupid trying to probe the current on the bus, and now I get clean voltage and current signals, so I think I am ready for first light soon...

Dan

[oneKone]

I've just received the boards from seeedstudio. When I assemble and test them did you want me to put my results on this thread?

[profdc9]

It would be great if you posted your results.  If they don't work, I won't be upset, I just want to be able to fix them.

Dan

[profdc9]

Well, I have first light   Take a look.  The UD2.7C through-hole driver works, now I just have to figure out how to make the spark longer.  I'm still new to this.  I put the interrupter through its paces in the video.

/>
Dan

[Mads Barnkob]

That seems quite good at driving high BPS, long on-time at a "high" frequency.

It sounds very high pitched, what BPS are you running it at here? Was it running in burst-mode all the time?

[oneKone]

congrats on getting the coil running! you'd be amazed on how one turn of the primary tap can throw the tune off. would you be able to share your secondary length and diameter of wire used?
   

[profdc9]

So the specs of the coil currently are:

J A V A T C version 13.2 - CONSOLIDATED OUTPUT
4/17/2018, 10:25:34 AM

Units = Inches
Ambient Temp = 68ºF

----------------------------------------------------
Surrounding Inputs:
----------------------------------------------------
0 = Ground Plane Radius
0 = Wall Radius
0 = Ceiling Height

----------------------------------------------------
Secondary Coil Inputs:
----------------------------------------------------
Current Profile = G.PROFILE_LOADED
3.25 = Radius 1
3.25 = Radius 2
20 = Height 1
38 = Height 2
1100 = Turns
26 = Wire Awg

----------------------------------------------------
Primary Coil Inputs:
----------------------------------------------------
Round Primary Conductor
4.5 = Radius 1
7.314 = Radius 2
19 = Height 1
19 = Height 2
5.6283 = Turns
0.25 = Wire Diameter
0 = Ribbon Width
0 = Ribbon Thickness
0.08 = Primary Cap (uF)
0 = Total Lead Length
0 = Lead Diameter

----------------------------------------------------
Top Load Inputs:
----------------------------------------------------
Toroid #1: minor=4, major=14, height=42, topload

----------------------------------------------------
Secondary Outputs:
----------------------------------------------------
151.4 kHz = Secondary Resonant Frequency
90 deg ° = Angle of Secondary
18 inch = Length of Winding
61.1 inch = Turns Per Unit
0.00042 inch = Space Between Turns (edge to edge)
1871.9 ft = Length of Wire
2.77:1 = H/D Aspect Ratio
75.7758 Ohms = DC Resistance
57753 Ohms = Reactance at Resonance
1.44 lbs = Weight of Wire
60.711 mH = Les-Effective Series Inductance
63.23 mH = Lee-Equivalent Energy Inductance
61.599 mH = Ldc-Low Frequency Inductance
18.202 pF = Ces-Effective Shunt Capacitance
17.477 pF = Cee-Equivalent Energy Capacitance
31.016 pF = Cdc-Low Frequency Capacitance
7.2 mils = Skin Depth
12.352 pF = Topload Effective Capacitance
190.0093 Ohms = Effective AC Resistance
304 = Q

----------------------------------------------------
Primary Outputs:
----------------------------------------------------
151.4 kHz = Primary Resonant Frequency
0 % = Percent Detuned
0 deg ° = Angle of Primary
17.41 ft = Length of Wire
2.89 mOhms = DC Resistance
0.25 inch = Average spacing between turns (edge to edge)
1.468 inch = Proximity between coils
0 inch = Recommended minimum proximity between coils
13.869 µH = Ldc-Low Frequency Inductance
0.07996 µF = Cap size needed with Primary L (reference)
0 µH = Lead Length Inductance
138.48 µH = Lm-Mutual Inductance
0.15 k = Coupling Coefficient
0.137 k = Recommended Coupling Coefficient
6.67  = Number of half cycles for energy transfer at K
21.71 µs = Time for total energy transfer (ideal quench time)

I ran the coil for a few seconds at a time not in burst mode but most of the time I ran it in burst mode. The BPS had to be high to get decent streamers.  I do not know what the BPS is but I had it turned up to the maximum BPS Steve Ward's interrupter will do.  I can plug it into a frequency counter and find out what it is.  The energy transfer time is very short, so I wonder if I should add some capacitance and use IGBT bricks or parallel IGBTs.  The primary capacitor is only 80 nF @ 10 kV currently, because that was left over from my other coil.

I have the overcurrent set to what I think should be 300 A, but perhaps I have that wrong and the current is not allowed to rise high enough.  When I measured the current before putting on the secondary coil, I know the overcurrent detection did not kick in until at least 200 A, because I measured the voltage on the burden resistor and calculated the current.

The other thing I'd like to do is make the pulse skip version and try that because it could prolong the interrupter time while keeping the current below a threshold.

Do you have any suggestions?  Yes I know the BPS is very high, and thinking of options.  According to JAVATC, the energy transfer time is very brief, which means the peak currents in the IGBTs are also very brief, which is probably why I can use such a high BPS.

Dan