DRSSTC: Some beginners questions / Tesla Coils / Forums

Forums

4hv.org :: Forums :: Tesla Coils

« Previous topic | Next topic »

DRSSTC: Some beginners questions

1 2 next

Move Thread

LAN_403

axelro

Sat Jan 22 2011, 01:07PM

Registered Member #3640 Joined: Sat Jan 22 2011, 12:16PM
Location: Germany close to Heidelberg
Posts: 39

Dear all,
please allow for some beginners questions - hopefully they are not too dump or easy to find via google (at least I tried):

First some basics on configuration
DRSSTC, ~280kHz, 60cm Sec, 5 windings Prim, 26cm Toroid
2 TO147 fast IGBT's (HGTG20N60B3D) in Halfbridge with capacitive voltage divider
Halfway low inductance buildup (I would hope).
Modified Steve Ward driver, optionally + Simplified Predictor in feeback.
14 x 10nF in parallel times 2 as MMC
All pretty prototype like - see the picture.

The coil is hardly pulling any power - max streamer length so far about 15-20cm at 100BPM and approx 100us. I guess 0.5A. Obvious answer is lack of tuning, but I tried hard.

OK, as said - some beginner questions:

1) Increasing streamer length until 120V, then no gain (in original Ward mode)
My conclusion is, the coil is pulling itself closer to resonance with streamers- above 120V it's going into the other direction. So action needed to be to prolongate the primary coil, right? (but behaviour isn't quite consistent - still in the presumed resonance case hardly additional pull of power and with the predictor option, higher voltage increases streamer length).

2) In the streamer (to a wire) one can see brighter pieces. Typically they are at the Toroid, if I let streamers jump from there to a grounded wire - if I have a breakout piece of wire at the Toriod, it's somehow somehere in the middle. Are there any conclusions about resonance one can draw from that phenomenon?

3) As mentioned above - I am using a voltage divider, so on end of the primary is at the divider middle point, the other between the IDBTs. Why is this actually nececessary? One would think the MMC is enough to block DC. I tried w/o divider, clearly it didn't work without the voltage divider properly. Why is this so?

4) Results are better with the little primitive predictor setup (still on breadboard). Overshoot is lower, hower, can't really as nicely tune as Finn. If I use the original Steve Ward setup (switchable with jumpers), I seem to get less output, and slightly off duty cycle.

5) In original Steve WArd feedback, I get sparks between coil and primary over a certain voltage - not in the predictor variant. Still the predictor variant produces longer sparks (at comparable lower voltages) - wierd.

6) Hmm, just thinking: I am driving the IGBT's via the Steve Ward driver with 24V and a 2:1 GDT, gives reasonable square drive. Are the 12V+- at the gate too low to open the guys up properly, perhaps?

OK, any advice appreciated. Thank you
Axel

Gregory

Sat Jan 22 2011, 07:45PM

Registered Member #2922 Joined: Sun Jun 13 2010, 12:08AM
Location:
Posts: 226

you dont need the prediktor controller to this igbts, use the original steve ward. the voltage divider is to put 1\2 and after -1/2 of VCC bus. the MMC isnt to block DC, it occur, but is for make a ressonant LC tank with the primary

axelro

Sat Jan 22 2011, 08:52PM

Registered Member #3640 Joined: Sat Jan 22 2011, 12:16PM
Location: Germany close to Heidelberg
Posts: 39

Hallo Gregory, thank you for your reply.

Sure, I don't need the predictor, and I am sure it doesn't yet quite work as it should, but it's definitely giving me better results then the original ward feedback. In principle, what I did was just simply allowing jumpers to be set either to:
a) a L terminated CT and comparator based feedback with some variable RC delay behind or
b) to be set to the original diode terminated input circuit of Steve.
Behind this piece of circuitry, it still goes into the standard Steve Ward controller.

I think it gives me slightly better results, because the IGBT's are actually a bit slow for this high frequency and the predictor like input gives some phase shift in favor to good switching.

I also do understand that the divider is there to divide - but why does it matter, that is the question. The half bridge switches between 0 and Vcc and so that should be an AC of Vcc/2 via MMC and Primary. Sure, MMC and Primary build a resonant LC, but at the same time it should block the unwanted DC. Still, when I put the output of the primary to ground, my results get worse.

I think I am missing some theory pieces - these I am trying to understand.

Best regards
Axel

Gregory

Sat Jan 22 2011, 10:59PM

Registered Member #2922 Joined: Sun Jun 13 2010, 12:08AM
Location:
Posts: 226

When the top igbt is high the top capacitor put vcc/2 in the LC, and when the bottom igbt is high, the bottom capacitor put -vcc/2 in the LC. Why minus? beacuse the current in the LC with the bottom capacitor is reverse. See this document, it can help you:

axelro

Sun Jan 23 2011, 08:07PM

Registered Member #3640 Joined: Sat Jan 22 2011, 12:16PM
Location: Germany close to Heidelberg
Posts: 39

Gregory,
again appreciate you trying to help. I am just not a 100% sure I understand in which respect it answers question 3. Have a look at below, please.

About question 3)
To get closer to the root of the question, I know simulated it with LTSpice (attached). There is a shortcut acress C3 which one can open/close to see the difference between using the divider or going to ground right away. Interestingly, if I believe the simulation, I do not need the divider. Still last time I tried, it behaved differently. I need to try again.

Edit: I just tried, and in confirmation of the simulation, it seems to work just fine without divider - putting the one side of L to ground directly. No clue what I did wrong the first time.

About question 6)
I know increased the input to the driver from 24V to 30V - through a 2:1 GTD I have now +- 15V at the gates of the half bridge. No significant difference. I guess it doesn't hurt though, so I'll keep it.

Anyone able to advise for the other questions ?

Regards
Axel
]halfbridge_variant_with_and_without_divider.zip[/ file]

Gregory

Sun Jan 23 2011, 09:45PM

Registered Member #2922 Joined: Sun Jun 13 2010, 12:08AM
Location:
Posts: 226

3 - The bridge can work with not the divider, but don't use with not! The two capacitor balance the voltage and make all much more stable! I think that is dangerous for all run with not the dividir with only the bottom capacitor...

6 - 15v is better than 12v, "the more, the better" hehehe. I think that is more efficient use a 1:2 GDT with 12v at the primary. It will give 24v at the gates. The igbts will survive more current with more voltage at gates. I'm using 26v at my gates with no problem... This is called gate overdrive, reduce the life of the igbt, but with a little interrupter duty cicle like a DRSSTC and as you don't need 100 years of igbt duration, I think that overdrive is the best.

Sorry my english. I'm brazillian. you can add me at facebook or msn! We can speak better there!

axelro

Sun Jan 23 2011, 10:08PM

Registered Member #3640 Joined: Sat Jan 22 2011, 12:16PM
Location: Germany close to Heidelberg
Posts: 39

Hallo Gregory,
on 3) - most likely you are right, but still trying to understand why this is so.

on 6) Actually it's the other way round - I am having 30V on the full bridge stage of my driver (ala Steve Ward), which is +- 30V then on the primary side of the GDT. The GDT is 2:1, so it turns into +-15 V at the secondary of the GDT. The advantage is, that this minimizes leakage inductance and doubles the amperage. So it can load/unload even IGBT bricks pretty quickly. Of course this isn't really necessary for the small TO247 IGBT's I am using, but I am planning to use bigger ones in some later version (check my picture - there you'll see the future power stage, so far unconnected behind the small Half bridge).

Btw - your English is fine and much better then my Portugese!

Gregory

Mon Jan 24 2011, 01:53AM

Registered Member #2922 Joined: Sun Jun 13 2010, 12:08AM
Location:
Posts: 226

Thanks!

3 -
Sstc Html M6a57bad0

This have two steps:

1- The top mosfet turn on and charge the capacitor through the primary.

2 - The bottom mosfet turn on and discharge the energy in the primary.

But now think in this with another capacitor, the top capacitor that don't appear in this picture. The system get balancied, with exactily 1/2vcc at the bottom capacitor. Try imagine the diferences!

6 - If you will use bigger IGBTs than use a 2:1 GDT but with 40v at primay to get 20v at secondarys... 15v is very low to drive high currents at the igbts

axelro

Mon Jan 24 2011, 09:05PM

Registered Member #3640 Joined: Sat Jan 22 2011, 12:16PM
Location: Germany close to Heidelberg
Posts: 39

Gregory, I am trying to imagine. Let's assume 100V total voltage.

With a divider I am putting the LC combination from +50V to -50V - so it's a total of 100V difference.

Without a divider I am putting the LC combination from +100V to 0V - so again it's a total of 100V difference.

Should be the same electrically, unless there are unknown effects. The simulation seems to show the same.

Gregory

Mon Jan 24 2011, 11:37PM

Registered Member #2922 Joined: Sun Jun 13 2010, 12:08AM
Location:
Posts: 226

But the the system run more stable with less current at each IGBT, but with the same total power.

1 2 next

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