Steve Ward's Mini SSTC problem / Tesla Coils / Forums

Forums

4hv.org :: Forums :: Tesla Coils

« Previous topic | Next topic »

Steve Ward's Mini SSTC problem

previous 1 2 3 next

Move Thread

LAN_403

101111

Fri Oct 24 2008, 02:59PM

Registered Member #575 Joined: Sun Mar 11 2007, 04:00AM
Location: Norway
Posts: 263

About MOSFET heating, there was a thread about that and they concluded with that this is normal. I tried to dig up the thread but didn't find it.

I believe that the 5ohm resistor have something to do with the ON/OFF time but I cannot confirm this.

Now I get it.. It is simple, when the GDT switches it charges the gate on the MOSFET, the higher resistance on the gate the longer turn on time. The higher resistance the slower you will charge out the get and that means the longer off time.

Correct me if I am wrong.

Brad

Fri Oct 24 2008, 04:44PM

Registered Member #1632 Joined: Mon Aug 11 2008, 08:53PM
Location: Plainfield, IL
Posts: 12

AH! That makes sense, so it is a little bit of a MOSFET safety precaution? I think I will install them! As far as the diode getting destroyed by a large pulse, would a poor secondary RF ground cause this, because I'm not convinced my ground is the best, I'm using the water pipes for the hot water heat in the house (not our water supply, just home heating). I have a good RF ground for my SGTC outside, but I'm building this SSTC inside so it is not convenient to use my outdoor RF ground. Perhaps the lack of the 5ohm resistors with running high duty cycle high pulse width caused this to happen, my arcs were screaming when this happened, the pitch was so high and so loud I could hardly stand it!

Brad

Curtis

Fri Oct 24 2008, 07:40PM

Registered Member #883 Joined: Sat Jul 07 2007, 01:02PM
Location:
Posts: 16

hello. i am new to this also but have read ALOT about theory. The resistors slow down the turn on and off of the mosfets. this generates dead time and "softens" the switching transition. Depending on the transistors used it is advisible to put a diode in parallel with the resistor so that the turn on is slow but the turn off is fast.

The diodes are freewheeling diodes. the catch the current that could be backflowing from the resonator. Mosfets have an intrinsic bod doide that serves this purpose but it is tpicall a ver crapp doide. It is best to bypass this diode with a high quality fast diode. IGBTs can have or not have a built in diode but it is still best to bypass this with a discreet device.

Brad

Fri Oct 24 2008, 08:38PM

Registered Member #1632 Joined: Mon Aug 11 2008, 08:53PM
Location: Plainfield, IL
Posts: 12

Thats a great idea, I will definitely put a diode in parallel to the resistors. Do you think my diode died because of hard switching because of my lack of a resistor on the gate, or do I possibly have another problem looming somewhere? It sounds like removing them is not the best thing to do, should I replace them with a beefier diode, or just the same one and see if the resistor on the gate makes the difference?

Thanks for all the help and advice guys!

Brad

101111

Fri Oct 24 2008, 08:59PM

Registered Member #575 Joined: Sun Mar 11 2007, 04:00AM
Location: Norway
Posts: 263

About your RF earth: I believe that it should be connected to earth, but if you are unsure there is no simpler way to find out then to measure it, simply either turn off the main fuses and check with an ohmmeter to another earth in a mains socket or measure volt from one random phase to the pipe, in theory it should give about half of your mains voltage.
I remember some about that use of free-wheeling diodes in antiparrarel like on Steve's schematic is only needed when you are out of tune or drawing arcs to ground.

And thanks Curtis, I didn't know this. There is not too long ago I started to understand the ways of the Tesla coil.

The Diode Curtis is talking about should be wired like in the picture.

Brad

Fri Oct 24 2008, 09:04PM

Registered Member #1632 Joined: Mon Aug 11 2008, 08:53PM
Location: Plainfield, IL
Posts: 12

Excellent, that sounds great, I don't have any spare diodes for MOSFET protection, but I do have some for the gate drive, I am gonna order some for the MOSFETs but for the time being, I will cut them out, install my gate resistors and gate diodes and see what happens!! It should make for an interesting weekend. I'll let ya know how it turns out!!

Thanks a bunch!

Brad

LithiumLord

Fri Oct 24 2008, 10:16PM

Registered Member #1739 Joined: Fri Oct 03 2008, 10:05AM
Location: Moscow, Russia
Posts: 261

It's EXTREMELY unrecommended to fire up a coil like that. Use at least UF diodes, but I'd recommend either a low-voltage Shottky + UF bypass, or a high-voltage Shottky one - makes fets' life easier ;)
Something feels very wrong. The UF failure is a very rare case, even with a direct drive circuit with deadtime I use in my coil I have not seen my UFs heat up much, although the low-side ones are placed on small ]-shaped heatsinks, just as calculated beforehand. Also you have no deadtime in the circuit, so heat is definitely not a reason. So you either had an overvoltage due to excessive inductance or your diode managed to open in a close-to-shootthrough condition and has failed due to a repetative pulsed overcurrent. Check everything.

Brad

Sat Oct 25 2008, 01:58AM

Registered Member #1632 Joined: Mon Aug 11 2008, 08:53PM
Location: Plainfield, IL
Posts: 12

Yes, there are problem, I just checked everything on the power side and heres the list. Both MOSFETs are dead, one fast diode survived, bridge rectifier is dead, all capacitors lived. I'm gonna check my wiring again and make sure everything is hooked up correctly. The strange thing is that it worked so well when it worked! When you say that I may have too much inductance, does that mean I should try removing turns from my primary? I have 10 turns right now on a 4" form. My secondary is ~ 13" of 30 awg wound on a 3" G-10 1/16" wall tube. I am new to all of this SSTC stuff, as far as dead time goes, will the resistors that I am now putting on the gates create some dead time, or is the dead time you speak of much longer than the extra time to charge the gate through the resistor? I'm gonna get back to my circuit and get back to work.

Thanks!

Brad

LithiumLord

Sat Oct 25 2008, 02:59AM

Registered Member #1739 Joined: Fri Oct 03 2008, 10:05AM
Location: Moscow, Russia
Posts: 261

Heh, under inductance I meant stray one, one that mounts up on all your wires etc. The best way is when your bridge is connected with extremely short thick wires, and film caps are rigged straight to the mosfet terminals, as well as diodes.
DO NOT reduce primary turns before you achieve a perfect operation of the electronics! You'll get stuck not knowing what led to a blow in that case - either your electronics are faulty or you just pushed it over the edge. The concept has numerous downsides, so counting on it isn't a good idea - everything should be tracked with a scope carefully in this case, then make crashtests slowly pulling the power up untill you feel it's enough. ADD A FUSE! A fuse between the el-caps and the film ones will exclude excessive damage in case anything goes wrong, especially if it's arc-proof.
The resistors are not exactly what I mean, but the resistor-diode circuit will make the rise and fall times different, injecting some delay to prevent the shoot-through condition while having some switching loss increase as a downside. (what I meant as higher dead-time was an actual delay created on the generator level, it was necessary as I use a GDT-less concept where each gate has it's own driver to control it, that is very good for achieving low switching losses and preventing the shoot-through condition, however I'd not advice that trick for the beginning as it's a bit complex and is really worth it only if you need to make not very powerfull fets drive much current into the coil).

So far, do the following:
1. Make all the connections in the power part as short as possible
2. Add an EMP shield! It probably will not affect much in this schematic, but anyway it's a very good practice, very nice finished look and makes one thing less to fear of.
3. Add a fuse.
4. Check rise/fall times with a scope, adjust them so the injected deadtime would exceed the mosfet delay.
5. Check gate ringing!
And optionally - you can load your halfbridge with a lightbulb and ballast it with a shunt between one side of the caps and fet terminal. Starting this up with an external generator instead of feedback and preferably lower voltage supplied to the bridge will allow you to get a scope trace of the bridge falling current, the smaller spikes you'll see are generated by the drive, but if they would grow up to feeding voltage and get a wider look it's a definite a sign of a shoot-through condition. The test is not really necessary if you are sure in your traces obtained from gates of unloaded fets, but you can try it just to make sure. If anything looks wrong there, first post it here - maybe it's nothing to worry about.

Good luck!

Nicko

Sat Oct 25 2008, 06:11AM

Registered Member #1334 Joined: Tue Feb 19 2008, 04:37PM
Location: Nr. London, UK
Posts: 615

Hi,

I've just built one of these and it worked perfectly first time.... because I tested each part in turn to eliminate potentially fatal and expensive mistakes up front. I also run at 230VAC mains (in the UK), so had to make some minor changes to some components... otherwise is just as Steve Ward designed it,

You will need a 'scope (good to a few MHz, say 10 and very preferably a dual trace one) and a signal gen (a 555-based oscillator will do). With no HV applied, inject a 5V 100+kHz signal into the aerial and 'scope the signal path through to the GDT. Unsolder & check the outputs of the GDT - they should be in anti-phase with a good square-wave shape and about 10V peak to peak with not too much ringing or roll-off. The signals on the gates of the FETs MUST be in anti-phase (thet's why you need a 2 channel 'scope), else the FETs WILL DIE. If the GDT signal is not a pretty clean square wave, your GDT core may well be the wrong material. Not all cores are ferrite, and not all ferrite cores make a good GDT at a couple of hundred kHz (which is where your coil will be running - mine runs at up to 450kHz depending on configuration). Use the recommended cores - I've just switched to using FERROXCUBE TN23/14/7-3E25 because the one everone seems to like (Fair-Rite type 77 or 78) are difficult to find in the UK.

I run with 2SK3680-01 FETs, but I've used IRFP450s too which have a far higher RDSon and they were ok. My gate resistors are 4R7 3W, tank cap is only 220uF. No diodes on the gate resistors.

Keep all high current leads SHORT AND FAT (I use 14AWG minimum). I recently used a current transformer on the primary and it peaks at about 40A with a recognisable square wave - my primary was similar to yours - 7 turns of 14AWG on a 4" former. Remember, white plastic throughout, no black cable ties or pipe. Secondary was 14" on a 2.5" former.

I built a safety box (my first TC and I was worried!) and the coil looked like this.

Last night I change the primary to 7 turns on a 3" former, and its still absolutely fine. I use an IR temp probe (from Fluke) to check the headsink temp without touching it, and the FETs are only running very slightly warm (about 5C above ambient).

DO CHECK THAT THE FETs & DIODES ARE INSULATED FROM THE HEATSINK! If they are shorting to each other, they will die.

HTH

Nick

previous 1 2 3 next

Moderator(s): Chris Russell, Noelle, Alex, Tesladownunder, Dave Marshall, Dave Billington, Bjørn, Steve Conner, Wolfram, Kizmo, Mads Barnkob