Class H Topology for Audio Modulated SSTC / Tesla Coils / Forums

Forums

4hv.org :: Forums :: Tesla Coils

« Previous topic | Next topic »

Class H Topology for Audio Modulated SSTC

1 2 3 next

Move Thread

LAN_403

Chris_Knight

Mon Feb 08 2016, 09:10AM

Registered Member #58280 Joined: Sat Jan 09 2016, 06:48AM
Location:
Posts: 43

Richie Burnett (GeordieBoy) in a couple threads several years ago mentioned the possibility of using Class H for efficient and low distortion audio modulation on a SSTC. Potentially, this would be able to produce speaker quality audio at high inefficiencies. PWM/FM modulation via the interrupter, the traditional method for audio modulation on SSTCs and DRSSTCs produce distorted and harsh output, and are not efficient. AM modulation is able to create high quality audio, however it suffers from low efficiency and is impractical at high powers.

Has anyone taken up his suggestion? Any thoughts?

Aforementioned threads:

Class H radio modulator:

Sigurthr

Tue Feb 09 2016, 12:45AM

Registered Member #4463 Joined: Wed Apr 18 2012, 08:08AM
Location: MI's Upper Peninsula
Posts: 597

I've long thought about it and wanted to do it, but my builds don't tend to have the budget required for Class-G/H. I'd love to see it done and some finalized designs put out.

By the way, FM (not PWM) modulation via main signal feedback path works fantastically when you use a VCO (be that standalone or inside a PLL). The fidelity is amazing and efficiency isn't really much worse than running CW. You do have to use a sideband though I've found, as setting the VCO directly to f0 causes destructive interference from the two sidebands cancelling out parts of the audio. My current SSTC works via this principle, and I've found there's a certain frequency offset needed to achieve wide enough audio bandwidth without loss of fidelity. I've wondered if the resulting bandwidth in the sideband is tied to the coil parameters or is a constant of the topology, but have no way of investigating. My coil uses the Upper Sideband, with a 7.1KHz offset from f0, and actually produces higher output power with audio modulation and the carrier set into the resonators USB than the same coil set without modulation to run at the natural f0. This contradicts the idea that modulation is pushing the drive frequency closer to f0.

...

Tue Feb 09 2016, 04:13AM

Registered Member #56 Joined: Thu Feb 09 2006, 05:02AM
Location: Southern Califorina, USA
Posts: 2445

While that does work (and is how my coils work), a proper bus modulator works much better in terms of available audio output (or stated differently, to how high of an output amplitude is achievable for the same level of distortion). This is doubly important, since trying to get higher amplitude audio by putting more power into the coil increases the hiss, which necessitates further increasing the input power, etc. To reduce the hiss your only real solution is to move to higher frequencies, but with the usual hard-switched frequency modulated coil it is not practical to increase the switching frequency above a megahertz or so before you start dissipating more in the bridge than you would in a well designed bus modulator

I have settled on a ~500kHz switching frequency and a killowatt or two of input power, which when combined with an optimized topload (with many breakouts to maximize audio level and audio/hiss ratio) gives acceptable room filling sound, but is still outpaced by a decent set of laptop speakers in terms of generating intelligible sounding audio.

One technique that I have wanted to try is using the usual frequency modulated coil, but adding a feedback loop which regulates the coil power (which should be roughly proportional to the air displacement) to be proportional to the drive signal. Hopefully such an approach would allow more or less full modulation depth with minimal distortion, in theory it should be possible to do even better than a bus modulated coil.

I had drawn out plans to use a microprocessor to do the feedback, using the internal PWM peripheral running as a constant duty cycle/variable frequency source (using dithering and sub-clock edge generation to get acceptable frequency resolution) to directly drive the bridge, and using a/d converters to measure the setpoint and bus current. This would allow for easy addition of further corrections into the loop to compensate for nonlinear streamer growth and non-flat frequency response, and let you add in safeguards to keep it from freaking out if you ever try and go past the resonant frequency or demand negative sparks, etc. Unsurprisingly I never did get around to building it...

In any case, class H for the modulator does not make a whole lot of sense, you may as well just run class D--it is not like the audio quality of your coil is going to be improved by added an intermediate linear regulator.

Another test I have been wanting to try building a simple bus modulated coil using a class D audio amp as the modulator. My coils have input impedances on the order of 10 ohms (depending on input voltage), so they are actually relatively well matched to a normal audio amp. The tricky part would be finding an amp which is DC coupled (depending on the exact topology it may or may not be possible to modify one for DC output). Alternatively you could build a bias tee of sorts using a large inductor in series with your DC bus to isolate it, and capacitively coupling the audio in. My back of the envelope calculations showed that a gapped MOT with 2 primaries in series (impedance of about 25ohms at 100Hz and while still able to handle a few amps of DC) in combination with a 1000uF capacitor (impedance 1.5ohms) would do the trick. The good news is that suitable amps are becoming quite cheap, ex

would be sufficient for a small coil and costs $35.
One of these years...

Sigurthr

Tue Feb 09 2016, 06:04AM

Registered Member #4463 Joined: Wed Apr 18 2012, 08:08AM
Location: MI's Upper Peninsula
Posts: 597

You know on the subject of Class-D... I have never once, in over a dozen CW coils gotten a single attempt at PWM via interruption (enable lines on UCC chips) to produce even remotely intelligible audio. Every single try resulted in heavily distorted garbled audio. Even pure sine inputs resulted in nasty output.

Also, the coil's power was always a fraction of what it should have been, and I don't just mean because of the average duty cycle of the interrupter signal. I once did a frequency sweep for the PWM carrier from 15KHz up to the coil's f0 (which was I think 435KHz on that coil) and there were a few maxima and minima nodes at subharmonics of the f0, where it is obvious the modulation signal's harmonics are exciting the coil, but the phase mismatch between the feedback and the modulation harmonics (or modulation itself when it is input at the f0) caused just as much if not worse distortion. At best I think I found a single maxima node around 107KHz (or somewhere close, it's been years now) where output power was roughly 15% of normal. That's how bad it was.

The frustrations with the above is what led me to move into FM driven coil. (And by the way "..." I too settled on a max f0 of 500kHz @ ~1.25kW for FMSSTCs, so kudos to like minds and all that.) One nice thing is that my 350KHz FMSSTC coil produces harmonics that are receivable on standard AM radios which carry the high fidelity audio. So in a way I've got my own high power am band transmitter - even if it really is AM-via-slope-detection-of-FM, but since the resonator makes a poor antenna the transmissions are limited to my house (probably being extended beyond the immediate magnetic field of the coil via coupling to mains wiring, otherwise I'd see the signal drop off just a room or two away).

So I'm not sure that class-D on the DC bus would work out as well as Class-H given my exceptionally poor experiences with Class-D.

Getting back to AM modulation, I did once try to do high level AM via a modulation transformer, a la the old style radio transmitters. I couldn't find a transformer with a low enough impedance to the coil's f0 though (as the current through the DC bus is pulsating at the coil's f0, and the modulation transformer's secondary is in series with the DC bus, you can't impede the current through the transformer much and preserve output). I tried the modulation setup on a ZVS flyback driver where it worked beautifully (I reversed the transformer winding ratio so the voltage swing was in line with acceptable limits of the ZVS).

There too you need a fair amount of modulation power, but the nice thing is you can just use a car audio class-D amp because the modulation transformer's primary winding can be such that it blocks the class-D carrier.

Chris_Knight

Tue Feb 09 2016, 08:27AM

Registered Member #58280 Joined: Sat Jan 09 2016, 06:48AM
Location:
Posts: 43

EDIT: Aha forgot about harmonics...

Avalanche

Tue Feb 09 2016, 12:44PM

Registered Member #103 Joined: Thu Feb 09 2006, 08:16PM
Location: Derby, UK
Posts: 845

I would have thought that whatever AF modulation method you choose for a TC, getting a good result will be difficult because a TC has such a high Q-factor. Ultimately, any method is trying to achieve amplitude modulation of the output voltage, and will result in sidebands (TC resonant frequency + and - the audio frequency, say 15kHz). If the roll-off is sharp (which I would expect it would be) then any modulation will be nowhere near linear. Has anyone experimented with this, and/or analysed it?

...

Tue Feb 09 2016, 05:03PM

Registered Member #56 Joined: Thu Feb 09 2006, 05:02AM
Location: Southern Califorina, USA
Posts: 2445

It is not as bad as you would think, since when under heaving streamer loading the coil Q is actually quite reasonable--on the order of 10 for the coils I have built. This in combination with the resonant frequency way above the audio (500khz vs 20khz) means that the energy stored in the resonator does not really fight against you.

Sigurthr

Wed Feb 10 2016, 02:06AM

Registered Member #4463 Joined: Wed Apr 18 2012, 08:08AM
Location: MI's Upper Peninsula
Posts: 597

Again, agreed with "...". On my current FMSSTC I've got 14.2KHz bandwidth on the USB. This actually works out as plenty for audio because being in the USB means the roll off is on the high side of the audio. Normally the high-pass effect of the resonator fights bass and enhances treble reproduction, but since I'm in the USB bass is enhanced and treble is attenuated resulting in a very clean sound reproduction.

Now, with doing AM, be it via Class-H or Modulation Transformer (which really is just a magnetic version of class-H as far as the end device is concerned) the modulation signal bandwidth is not limited by frequency, but rather amplitude.

You can of course increase the signal beyond 100% normal output with Class-H, but you cannot decrease below the threshold where there is breakout and/or feedback. This means that your zero-audio bias point for the modulation is very flexible and allows tremendous dynamic range with class-H. You are limited to a maximum of 100% normal output for the bias point when using magnetic modulation though as you can't pass a DC offset to the audio like you can on class-H.

Chris_Knight

Wed Feb 10 2016, 03:46AM

Registered Member #58280 Joined: Sat Jan 09 2016, 06:48AM
Location:
Posts: 43

Sigurthr do you have any documentation? Sounds interesting and I'd love to see the similarities and differences between yours and "..."'s

I have some other questions if you have a bit of time:

What are the advantages of using the 4.096Mhz crystal as an oscillator for class E vs PLL or Schmidt trigger?

I'm thinking about creating my own FM modulated PLL SSTC. I'm using a full bridge of FGH40N60SMD IGBTs. Any design tips? I've read Richie, Steve Conner, and several 4hv posts.

...

Wed Feb 10 2016, 06:15AM

Registered Member #56 Joined: Thu Feb 09 2006, 05:02AM
Location: Southern Califorina, USA
Posts: 2445

I would not recommend hard switching any IGBT at these frequencies, let alone the FHG40n60. I tried running my coil with more modern IGBTs and despite their improved ratings compared to the mosfets they would always fail after a few hours of operation, even with TVS diodes directly across the source/drain. Depending on the power level you want you can get away with fairly reasonable mosfets, I ultimately switched from the IXYS monsters I started with to the FCH104N60F (or something similar) and they are doing just fine at ~1kw power levels.

reference to the first audio modulated coil I built

I have since found that adding a topload helps coil performance a bit, in both by increasing power consumption and increasing spark length. I have since built 2 twins to that coil which are essentially direct clones, but with cheaper mosfets and tuned to run at 1kw to avoid blowing breakers.

A youtube video of the most recent coil:

Also a comparison of it running at full power on halfwave/fullwave rectified mains:

and

And the usual sparkler video:

1 2 3 next

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