Tardief's DRSSTC

[Hydron]

It is basically 3 parallel single phase voltage doublers. Instead of having 400VAC fullwave rectified to 564VDC, you have the midpoint/neutral to get 2x 325VDC ~ 650VDC across the capacitors.

I am unsure about how this balances in regard to load on each phase or neutral.

I imagine the load sharing would also depend on interactions between the coil BPS and line frequency. The DRSSTC is not a nice constant load; it will take big chunks of energy out of the bus capacitors at a low frequency. At lower BPS, especially close to multiples of line frequency, there may be significant mismatch between phase currents just because the bus caps get emptied just before the peak of a particular phase. I have not done any simulations however, though they would not be difficult (I suggest LTSPICE as a more powerful tool than online simulators - it's easy to model most of the coil in it).

[Tardief]

Hi,
The topload is under construction.  I will make some photos when its done.

We still have some wuestions about the primary coil.
We will make it from 12mm copper tube.
What is the suggested distance between the turns? We thought about 1cm.
And what about the distance between the primary and the secondary?
The last question is about the coupling. It has to be an exact value or it's enough that is about 0.15?

[futurist]

There is very little benefit from 10 mm to 20 mm copper tubing, for coil of your size I wouldn't go anything higher than 10 mm if you didn't already bought the tube
Also 12 mm tubing will be more difficult to bend in shape than 10 mm

I suggest coupling k ~ 0.16 so you can lower it by raising secondary, for example with wood disc

[Tardief]

We wanted to use 12mm because we can pump through more air/water.

How did you choose the size of the heatsink?

[Mads Barnkob]

Hi,
The topload is under construction.  I will make some photos when its done.

We still have some wuestions about the primary coil.
We will make it from 12mm copper tube.
What is the suggested distance between the turns? We thought about 1cm.
And what about the distance between the primary and the secondary?
The last question is about the coupling. It has to be an exact value or it's enough that is about 0.15?

12 mm copper tube is fine for a primary, 10 mm is a very common size in Denmark, so that is what I always use.
I have 20 mm between turns on my large DRSSTC, less would make it harder to mount a heavy tap contraption on, simply from risk of short-circuiting two turns.
I think its about 20-30 mm from secondary to most inner turn of the primary coil
Coupling is a very interesting parameter, the higher the better, but you have to take measures against flash-over, tuning can be more tricky etc. Somewhere between 0.15 to 0.20, as futurist mentions, its much easier to lower the coupling with a few disks that making it higher later (like in, good luck sawing a hole in the middle and lowered the coil down through it)

[Tardief]

Hi,
We are still alive.

What ferrit core do you suggest for the GDT and feedback?
I found that the AL should be big enough, but nothing about the physical size.

[Mads Barnkob]

Hi,
We are still alive.

What ferrit core do you suggest for the GDT and feedback?
I found that the AL should be big enough, but nothing about the physical size.

For the GDT I calculated it for my DRSSTC1, but just doubled the physical size for the DRSSTC3   ( http://kaizerpowerelectronics.dk/tesla-coils/kaizer-drsstc-i/ scrol down to 25th September 2009 )

For feedback, just use the same core type as for the GDT, but here the physical size is more dependant on where you want it to sit in the circuit, it simply has to be large enough to make the primary cable/busbar go through, have enough distance for isolation from the inverter voltage and not be a source for a possible short circuit from inverter into control circuit. I made a nylon bobbin for them to be mounted on, with a piece of brass rod going through for the DRSSTC3.

[Tardief]

Hi,
We disappeared a little. Unfortunately the university ruined our scheduel.
We had some problem with the GDT.


Here is what we got.
It should be a perfect square wave, doesnt it?
The blue one is the input. The yellow is the output.
The function generator makes a perfect square wave until i connect the GDT.

[futurist]

The signal looks completely fine

[Mads Barnkob]

You only have minimal overshoot/ringing and some droop, you could try 2-3 more windings on the GDT to counter this, but there is really no need to, as futurist says, it looks just fine

[Tardief]

Thank you.
I thought I need a  squre wave without any deformation.

[Tardief]

Unfortunately I am still not sure about the best GDT option.
We made some new measurements.

This was made with 15 turns.



The other with 20 turns.


The waveform looks better with 20 turns, but if I am right than the more turns will reduce the current and that can be a problem.
Should we stick with the 15 turns to avoid that?

Which is the better?
Using one core for driving all of the IGBTs or using two of them?

[Mads Barnkob]

From my DRSSTC1: http://kaizerpowerelectronics.dk/tesla-coils/kaizer-drsstc-i/

Code: [Select]

Its a Epcos ring core, material: N30, good up to 5MHz, Aemm^2: 95,89, AL: 5750nH

Inductance with 10 turns: L = AL * N^2 = 5750 * 10^2 = 575 uH

Peak current: Ipeak = (Vin * t * D) / Lmag = (24 * (70000/1000000) * 0,5) / (575 * 10^-6) = 1460 mA

Irms = Ipeak * 0,577 = 842 mA

Minimum number of turns needed to avoid saturation

t, 50% duty cycle = (1 / 70000) / 2 = 7,3*10^-6

Nmin = ( V x t ) / ( B x Ae ) = (12 * (7,3*10^-6)) / (0,2 * (95,89*10^-6)) = 4,6 turns

Current needed to drive a single 60N60 IGBT gate

I = Qc / t = (146*10^-9) / (1/70000) = 10,22 mA, including magnetising current, double this figure.
So with more turns you will lower your peak current, but also have less voltage drop, but remember that a few Volt droop is not a problem. It is better to have a fast turn-on without too much ringing, than a nice looking too slow turn-on with too high switching losses.

There is also the IGBT gate drive calculator, I do not recall if we talked about that: http://kaizerpowerelectronics.dk/calculators/igbt-gate-drive-calculator/

[Tardief]

Hi,
We are currently testing the driver with some capacitor(instead of the IGBTs).
What should be the open time for a CM600 H-bridge?
The capacitor was 68nF. The GDT had 13 turns with a core from Eastern Voltage.
We got more than 700ns rise time.  We used a driver from Gao. Without load the rise time was 40ns.

[Mads Barnkob]

If you scroll down to "Soft switching and zero current switching" on this page: http://kaizerpowerelectronics.dk/tesla-coils/drsstc-design-guide/igbts/

You can see that Kizmo has switched a CM600 on in just 100 ns, but that is due to resonant switching, so it seems good with the rule of thumb of 10 times faster in resonant switching, that you have 700 ns in a simulated hard switching setup.

[Tardief]

Hi,
After the exams we had some time to work on the coil. 



We have already made some tests but we are not sure in some things.
The current limit was ~300A. The IGBTs stayes cool.
The load was resistive at the firs image. A 4kW halogen light bulb.











Cyan = Vce, Blue = primary current.
On the Vce we measured  ~15V drop. 

We tested the IGBTs with a welding transformer. Here are the results:
At 17A the Uce was 1,06V
30A - 1.19V
40A - 1.27
50A - 1.35
60A - 1.41
70A - 1.46
80A - 1.5
90A - 1.55
100A - 1.59
It seems good, but then what did we measure?

And what do you think about the spikes?  (we used UD2.7 from Loneoceans) We tried to reduce them with phase lead.

I hope we can show you huge sparks soon.

[Mads Barnkob]

First I would like to say, really nice job on the topload, secondary and inverter box, it all looks very professional!

Let me guess, you have been testing at low input voltage?

I have described the problems with low voltage testing IGBTs here: https://highvoltageforum.net/index.php?topic=184.msg1203#msg1203

[Tardief]

You are right. The Bus voltage did not go beyond 100V. Thank you! 

What do you think about the voltage drop?