160mm DRSSTC II project | Questions

[Late]

Hello everyone,

After finishing my DRSSTC and playing around with it for a while I noticed some bad designs practices that I had used when making it so I decided to make a second bigger DRSSTC but this time with a way better design which should allow it to run more reliably.


Recently I started planning the DRSSTC II project and the current specs that I'm going for are as follows:

Secondary coil: 160mm x 730mm with ~2000 turns of 0.355mm enameled copper wire

Top load: 170mm x 810mm toroid constructed from aluminum air duct

Primary coil: 200mm x 500mm flat spiral coil with 8 turns of 8mm copper pipe with possibility to water-cool the primary in the future

Coupling: ~0.15 with possibility to adjust it

Primary capacitor: MMC made from 6 220nF 6000VDC dawncaps in 2s3p configuration

Bus: 3 phase input for ~565VDC on the bus

Bus capacitor: 2s2p from some nice 6800uF 350V Itelcond electrolytics with 15kΩ 15W resistor across each parallel bank of caps to act as both discharging and loading resistor.

Inverter: Full-bridge from 2 CM300DU-24NFH half-bridge bricks with a 10kΩ loading resistor across the output and with 2 3uF 1kV snubber capacitors across the IGBT terminals

Driver: My own modified version of the UD2.9 pulse skip drive

Here's the inverter schematic and JavaTC inputs and outputs:




The frame for the whole construction I'm planning to make out of some 30x30 aluminum profile. Whole frame will be around 520mm x 520mm x 300mm. With the primary coil platform being 150mm higher than the whole construction to avoid the heating of the frame. Here's a crude drawing of the frame:



For the build I have already acquired the IGBT's and the BUS capacitors which I got a pretty good deal on. For the case the plan is to cover the openings in the 3030 alu frame profile with some clear acrylic or polycarbonate and to maybe add some LEDs inside to illuminate the whole construction. For the bus itself I plan to go with a laminated bus made from some 3mm alu sheet with fiberglass and resin as the insulating layer to keep the bus stray inductance as low as possible.

Id love to hear any input about my coils plans and to see if there's anything I should improve or fix when making it. And also is there anything I should pay more attention to when assembling/building the coil?

[davekni]

With the primary coil platform being 150mm higher than the whole construction to avoid the heating of the frame.

Perhaps a bit close for a 500mm diameter primary to metal frame.  Hope you get other opinions too.  I've tried to keep one primary radius gap to other sizable metal objects, but I don't know if that's actually needed.  A FEMM simulation could predict effects.  Perhaps there are simpler tools to estimate coupling between primary and metal frame.

Here's the inverter schematic

Presume the 5R6 gate pull-down resistor values are place-holders.

Looks like a good design in general.

[Benjamin Lockhart]

I would reduce your primary turn to turn spacing to something like 6mm.
Making the coil smaller is always nice, and that should actually increase your coupling slightly as well.
If you play around with the primary outside diameter in javatc you will find there is a sweet spot with the highest coupling.

[Late]

Presume the 5R6 gate pull-down resistor values are place-holders.

Yes the 5R6 resistors are just the placeholders. When ill have everything assembled ill be adjusting both of them for each gate to get the fastest rise and fall times without having big spikes.

[Late]

I would reduce your primary turn to turn spacing to something like 6mm.
Making the coil smaller is always nice, and that should actually increase your coupling slightly as well.
If you play around with the primary outside diameter in javatc you will find there is a sweet spot with the highest coupling.

I put the turn to turn spacing to roughly 10mm to be able to easily connect the primary tap clamp without risking any shorts to adjacent turns. And the 0.15 coupling seemed like a good value but yea sure ill be playing around a bit with javaTC to find the optimal spacing.

[Benjamin Lockhart]

The turn spacing on my big coil is 4.8mm with 6mm tubing and I'm using fuseholders for the tap connection and there's lots of space.
6mm tube was too small for 1000A though and demands water cooling.

[Mads Barnkob]

99% of the time, a Tesla coil is in storage, so think a lot about storage, modularity, transportation etc. The coil you are building here, will take up about 1½-2 cubic meters of space, Tesla coils always seem smaller on paper

Your design does look pretty sound and follows best-practises. Maybe ditch the connection between DC bus capacitors and give them individual resistors instead, to balance them better.

[Late]

The turn spacing on my big coil is 4.8mm with 6mm tubing and I'm using fuseholders for the tap connection and there's lots of space.
6mm tube was too small for 1000A though and demands water cooling.

Yea on my other DR I'm also using 6mm pipe and at 1kA it gets pretty toasty fast. Hoping that on this coil using 8mm pipe should prevent it getting too toasty. Might need to even use 10mm pipe though not sure. For the tap connection I'm planning to make a custom connector from some copper like I did for my current DR.




For this DR ill make something very similar to that but with 2 16mm2 cables going to it. And will reduce the turn spacing to around 6-8mm as suggested. Though with that small of a distance wont that cause issues with the outer turns induction heating the inner turns?

[Late]

99% of the time, a Tesla coil is in storage, so think a lot about storage, modularity, transportation etc. The coil you are building here, will take up about 1½-2 cubic meters of space, Tesla coils always seem smaller on paper

For easy transport i plan to add some wheels on the bottom of it so it can easily be wheeled in and out of my workshop. And I'm planning to make each part of the DRSSTC easily swappable or upgradable if in future ill want to do anything like that. Size wise i made it smaller than the largest size i could easily fit in the bottom shelf of a commercial shelving unit i have against the wall in my workshop. I will try to make a 3d model of everything in fusion or solidworks to see how much free space is left over and based on that ill see by how much i can decrease the whole size as smaller ofc means lighter so easier for me to move, store and set it up alone.

Your design does look pretty sound and follows best-practises. Maybe ditch the connection between DC bus capacitors and give them individual resistors instead, to balance them better.

Glad to hear that. I spend quite a lot of time researching and looking at other DRSSTC builds to make my coil the best design i can.

[Late]

I have acquired both the bus capacitors and the IGBTs. The bus capacitors are bit bigger than i thought but they will work wonderfully for this build and the red color of them will stand out.




I'm thinking of arranging them something like this in the build with the capacitors being held together by a plastic plate and on the other side connector to the laminated bus.


I have ran into one issue though, I cant seem to find suitable heatsink for cheap anywhere locally. Does anyone know where i should look for heatsinks or where i could get one for not too expensive? I'm located in Europe, Latvia and I'm looking for a heatsink that's around 280-500mmx150-300mmx50-120mm

[thedoc298]

Really nice build and you gave me an idea for the variable primary tap, I like yours and will be making some for my other coils. Be nice to see it fire.

[Late]

Recently i have had some free time and have continued working on the coil. Also have been continuing to gather parts.
Main things that have happened since last post are:


1. I ordered, received, soldered and tested my UD2.9 SMD driver boards which are based on the ud2.9 driver. After thoroughly testing it i ca say that the V2.0 of my boards work very well even at higher frequencies and these us2.9 boards have the same size and form factor as the very popular UD2.7 driver board but with some of the things improved and reduced BOM cost.







2. I started making up the MMC bank by machining some brass rods and threading them for use in connecting up the MMC bank.








3. Last but not least i acquired a nice alu heatsink. Big thanks to Mads from kaizerpowerelectronics for offering to sell one of his big heatsinks.






Now with all these parts acquired ill start planning out the final layout and size of everything.

Also id like some feedback on the MMC capacitor connections. Mainly what would be the optimal way to connect the 3 series strings in parallel and connecting main leads to them.
And what would be the best way to cool such a heatsink? I was thinking of getting some 80mm fans and blowing air trough the end of it.

[Mads Barnkob]

Also id like some feedback on the MMC capacitor connections. Mainly what would be the optimal way to connect the 3 series strings in parallel and connecting main leads to them.
And what would be the best way to cool such a heatsink? I was thinking of getting some 80mm fans and blowing air trough the end of it.

Be careful with tightening the interconnect screw between two capacitors, by twisting them hard, this risks ripping the bonding of the foil from the terminal. It is actually prohibited all virtually all threaded terminal capacitor datasheets. Usually I recommend mounting them on a plate of busbar (which doubles as heat sink to dissipate capacitor losses), but with only two in series, you have enough cooling from the end busbars. Which leads to next question.

Next question, which is about leads. Connect the three string in parallel with two even length busbar, something like 5-10mm thick and maybe 20-30 mm wide would make a nice heat sink and solid current path. Connect leads in opposite sides, so that current path from lead, through any string of 2 capacitors and to other lead, its exactly the same length.

Even the slightest moving air does a huge difference, form no moving air, when it comes to heat sinks. So you do not need to ram air through the heat sink, just make sure its adequate flowing without giving you too much noise. But then again, fan noise is not exactly a priority to limit, in a DRSSTC

[Late]

Be careful with tightening the interconnect screw between two capacitors, by twisting them hard, this risks ripping the bonding of the foil from the terminal. It is actually prohibited all virtually all threaded terminal capacitor datasheets. Usually I recommend mounting them on a plate of busbar (which doubles as heat sink to dissipate capacitor losses), but with only two in series, you have enough cooling from the end busbars. Which leads to next question.

Thanks for warning me about this. Im not quite sure tho that i get what u mean by mounting them to a busbar. Do you mean having them offset from one and other by screwing them to a small piece of busbar? Could you provide a drawing showing what you mean?

Also would just using a big plate at each end be better or worse for the current sharing?

I also have  a fewguestionS about GDTs:
1) If i use 2 GDTs is it better to use one for each of the bricks or one for top IGBTs and other for bottom IGBTs?
2) If i use signle wires instead of a cat5 cable is it better to have as many primary wires as secondary wires on the GDT? And having them twisted together
3) How much for coil this size does GDT wire crosssection matter and is there optimal crossection for a GDT.
4) Am i correct to assume that its better to keep the GDT wires as short as possible and use shielded wire for the primary? (GDT would be way closer to igbts than the driver circuit)

[davekni]

1) If i use 2 GDTs is it better to use one for each of the bricks or one for top IGBTs and other for bottom IGBTs?

One GDT per brick (one per half-bridge) is better.  Uniform load on both edges for each GDT, one gate rising while other gate is falling.  Keeps dead-time uniform.  And some pulse-skip drivers require 2 GDTs connected this way (one per half-bridge).

2) If i use signle wires instead of a cat5 cable is it better to have as many primary wires as secondary wires on the GDT? And having them twisted together

Yes.  Twist each pair (one primary and one secondary) together within coil winding.  Twist each primary winding lead pair separately.  Parallel the four (or two) primary windings at driver.

3) How much for coil this size does GDT wire crosssection matter and is there optimal crossection for a GDT.

Don't think wire gauge is very critical, but get other opinions too.  Ratio of wire diameter to insulation diameter does matter some.  Thicker insulation is best combined with larger wire diameter too.  That keeps parasitic (leakage) inductance from rising.  (Leakage inductance for twisted pair windings is proportional to log(wire spacing / wire diameter).

4) Am i correct to assume that its better to keep the GDT wires as short as possible and use shielded wire for the primary? (GDT would be way closer to igbts than the driver circuit)

If primary leads are twisted, shielding shouldn't be needed.

[Late]

One GDT per brick (one per half-bridge) is better.  Uniform load on both edges for each GDT, one gate rising while other gate is falling.  Keeps dead-time uniform.  And some pulse-skip drivers require 2 GDTs connected this way (one per half-bridge).

Yea that was my first thought too but wanted to make sure. Currently i plan to use a driver that doesn't require two GDTs but in the future i want the possibility of easily upgrading to a pulse skip or freewheeling capable driver without making too many modifications.

Yes.  Twist each pair (one primary and one secondary) together within coil winding.  Twist each primary winding lead pair separately.  Parallel the four (or two) primary windings at driver.

Alright, will likely add one primary turn per each of the secondary turns then

[Late]

I have decided to make some small changes to my coil design. I've decided to use 10mm copper pipe for the secondary with 8mm spacing between turns. This should allow the primary to keep cooler during longer runs without the need of water-cooling the primary. Recalculating with java TC i get a coupling coefficient from 0.14 to 0.17 depending on how high the primary is. Which brings me to my first question. What's the optimal coupling coefficient for DRSSTCs of this size and power? And how high coupling coefficient is too high for a DRSSTC? Looking at other builds on this forum ive seen coils with a range of coupling coefficient from around 0.1 to even i think 0.2. From what I've read the higher the coupling coefficient is the more power can be transferred from primary to secondary but at the risk of having racing sparks. That's why I'm wondering about the optimal coupling to get the max spark length without increasing the risk of racing sparks too much.

Second question i have is about my secondary coil. Originally i planned to use 0.355mm wire and have the winding length of around 730mm with 2000 turns, Calculating this gives me spacing between turns of 0.01mm. How realistic is it to wind it that tight? And would it be better to get 0.315mm copper wire instead of 0.355mm in order to achieve the 2000 turns and having close to 730mm of winding?

Last but not least ive started to plan out the laminated bus design for my coil. I plan to make it from 2mm alu sheet with 2mm of fiberglass material as the insulating layer.
This is a quick sketch of what ive imagined it to look like:

This is the layout i have planned out based on how i wanted to fit the components and to match with as many of the holes in the heatsink as possible so i don't have to drill and tap too many new ones. The clearance between positive and negative will be around 2-3mm and ill be filling the empty space with some fiberglass material for better isolation. and will use spacers to match the height for the terminals.


Heres just the top layer of the bus


and here's just the bottom layer

Going off of a recommendation from a user here (i think it was here on the forum at least) i will be connecting the capacitors in series in a cross. Though i still don't know if ill short together the both middle points or will keep them isolated. I will however have individual resistors for balancing and discharging purpose


How does the current layout look in terms of good design practices and keeping the inductance low? Is there anything i should improve in the layout or the overall DRSSTC design?

[Mads Barnkob]

I have decided to make some small changes to my coil design. I've decided to use 10mm copper pipe for the secondary with 8mm spacing between turns. This should allow the primary to keep cooler during longer runs without the need of water-cooling the primary. Recalculating with java TC i get a coupling coefficient from 0.14 to 0.17 depending on how high the primary is. Which brings me to my first question. What's the optimal coupling coefficient for DRSSTCs of this size and power? And how high coupling coefficient is too high for a DRSSTC? Looking at other builds on this forum ive seen coils with a range of coupling coefficient from around 0.1 to even i think 0.2. From what I've read the higher the coupling coefficient is the more power can be transferred from primary to secondary but at the risk of having racing sparks. That's why I'm wondering about the optimal coupling to get the max spark length without increasing the risk of racing sparks too much.

Second question i have is about my secondary coil. Originally i planned to use 0.355mm wire and have the winding length of around 730mm with 2000 turns, Calculating this gives me spacing between turns of 0.01mm. How realistic is it to wind it that tight? And would it be better to get 0.315mm copper wire instead of 0.355mm in order to achieve the 2000 turns and having close to 730mm of winding?

Last but not least ive started to plan out the laminated bus design for my coil. I plan to make it from 2mm alu sheet with 2mm of fiberglass material as the insulating layer.

How does the current layout look in terms of good design practices and keeping the inductance low? Is there anything i should improve in the layout or the overall DRSSTC design?

1) 0.15 to 0.2 is normal for coupling of a DRSSTC, is does not have the same high voltage primary as a SGTC, which JavaTC is designed for, so suggested coupling should be a bit higher for DRSSTC.

2) Enamel thickness on a thin wire is roughly adding around 10% to diameter, so what I input to my helical coil calculator ( https://kaizerpowerelectronics.dk/calculators/helical-coil-calculator/ ) to get winding length for JavaTC with a wire size of 0.355 mm would be 0.0355 mm between turns (1 layer from each wire). You could also check out the manufacturers specification of enamel thickness, I got a table of thickness depending on the wire classes here: https://kaizerpowerelectronics.dk/tesla-coils/drsstc-design-guide/secondary-coil/

3) Your laminated bus is about as low inductance as you can get. Only thing would be to move rectifier to the side, so have shorter path from +/- to the capacitors, instead of going over one of the IGBTs, placing it in the middle would make it even for everybody. Properly a totally neglectable detail.

[flyingperson23]

Adjustable coupling secondaries are very useful. I usually build a geometry that gives high coupling, then raise the second secondary with various height spacers to lower the coupling. Maximum coupling without racing sparks is dependent on coil parameters. Make sure the inner turn of your primary isn't too close to your secondary and you should be fine.

Your bus design is very low inductance, but may be prone to arcing over, depending on how you design it. Running at 565Vbus, you have more headroom for high inductance bus design than at 850V. If I were you, I'd keep at least 5-6mm between + and - so it won't arc over even if you get big voltage spikes or conductive dust blown around.

[Late]

2) Enamel thickness on a thin wire is roughly adding around 10% to diameter, so what I input to my helical coil calculator ( https://kaizerpowerelectronics.dk/calculators/helical-coil-calculator/ ) to get winding length for JavaTC with a wire size of 0.355 mm would be 0.0355 mm between turns (1 layer from each wire). You could also check out the manufacturers specification of enamel thickness, I got a table of thickness depending on the wire classes here: https://kaizerpowerelectronics.dk/tesla-coils/drsstc-design-guide/secondary-coil/

Thanks, will use that calculator to recalculate. How tightly can i expect the winding to be irl? Id assume that its hard to get 0mm between the turns in a secondary coil.

3) Your laminated bus is about as low inductance as you can get. Only thing would be to move rectifier to the side, so have shorter path from +/- to the capacitors, instead of going over one of the IGBTs, placing it in the middle would make it even for everybody. Properly a totally neglectable detail.

Thanks for the advice. ill move the components around a bit. This is what i have planned out atm:

The image ofc doesnt have the bus caps and bus and some other small things but the main parts should be there. and on the space on the right i plan to put the MMC. One question tho. Is the orentation of the bus caps optimal or is there something to improve?