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

previous 1 2 3 next

Move Thread

LAN_403

TwirlyWhirly555

Wed Feb 10 2016, 06:34AM

Registered Member #4104 Joined: Fri Sept 23 2011, 06:54PM
Location: Uk .
Posts: 122

Ive made a few FM modulation based ones using a VOC/PLL ,

This was the first one I did -

Runs abit short as the 1.2Kw input was heating the secondary allot , frez was around 500Khz

Sigurthr

Wed Feb 10 2016, 06:48AM

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

I did write up a bit about the subject a few years ago, but I don't think it contains anything that isn't covered in this thread so far, and the info there is a bit out of date and doesn't touch the topic of modern semiconductor class-H. Feel free to take a look if you like:

I don't have any recordings of coil operation from then. That was before I had a good camera with high enough noise immunity to operate near a coil. My first few cameras were destroyed by TCs, so there isn't a lot of videos even on my YT about TCs. I was working on putting a formal design for FMSSTC out there but couldn't in good faith endorse use of the 4046 given its failings, and I never found a beginner-suitable alternative (the LTC6990 is quite fragile).

If you like, feel free to check out my conglomerated SSTC data file. It's not related to anything other than Class-D, but it shows the basic build parameters I recommend to new solid state coilers. Several people have successfully built TCs based on my files, even if they're really nothing special, heh.

As for your other questions; I assume they are unrelated? (No IGBTs are going to work anywhere near 4MHz). I'll answer them separately.

Using an XO works fine for Class-E, as would any oscillator, as the Class-E coil is a non-tracking tuned resonant design. You have to pick the DS capacitor to critically damp the primary circuit to minimize loss in the transistor. Since any change in f0 of the secondary (via loading) would result in a necessary change in the primary resonance (if f0 tracking were part of the design), and therefore a change in the DS capacitance which cannot be *practically achieved*; the system runs open loop by nature.

There is of course the idea of using variable capacitance and some kind of intelligent control loop with feedback to monitor the primary and secondary circuits and adjust accordingly, but I don't think anyone has done it due to the poor ROI for such a huge undertaking.

Anyway, since you can't vary the DS cap, there's no sense in using a PLL (at least not with it's PhaseComp, you can use it for its VFO), as you don't want to change the drive frequency, and ideally you don't want the secondary f0 to change either. This of course means the schmitt trigger feedback system isn't applicable too.

If you're wondering PLL vs Open-Loop-Oscillator vs Schmitt Trigger for standard Class-D SSTC then it is a much less involved topic. PLL gets you roughly equal resonance tracking as a Schmitt trigger feedback will. The slight propagation delay in your signal line is negligible to a well designed and built SSTC. There are of course ICs and switches that have "huge" delays but you just avoid them.

PLL: Pro; resonance tracking, ground strike tolerance, FM modulation capability. Con; Increased difficulty and parts cost/count, can't drive coil away from resonance. Also, an unbelievable nightmare to get working in any of the discovered but officially undocumented failure modes (this is in reference to the 4046, there's tons of unofficial documentation here on the forums). You have to build your driver specifically to the resonator as there will be a limited range of tuning available. You could of course implement a digital PLL, or roll your own from discrete/logic to circumvent the issues with the 4046, but this adds a ton of complexity.

Schmitt: Pro; resonance tracking, low cost, low parts count, ease of use. Con; ground-strike intolerant, can't FM modulate, can't drive coil away from resonance.

Open Loop (oscillator); Pro; FM modulation capability (if using a VCO), can be driven in sidebands, intrinsic negative feedback of coil output power if users get too close (great for displays around kids), low cost and complexity. Cons: no resonance tracking, tuned operation, hard switching (if not class-E). You have to build your driver specifically to the resonator as there will be a limited range of tuning available.

As for tips for a PLL FMSSTC; have LOTS of capacitors and resistors on hand to swap in and out of circuit to get the PLL frequency settings correct. If you use the 4046 you really need to nail the vco range and locking range just right; not too big or too small. I've used a ratio of phi (1.618) to the secondary's actual f0 successfully in the past but it isn't easy to replicate. Using phi keeps harmonics out of the lock range. If you are going to inject audio into the VCO bias point, you have to get the audio levels just right or it can destabilize the PLL lock. It's well worth looking at the threads here that document and explain the failure modes of the 4046 and ensuring your bridge is tolerant of those conditions. Even in a perfectly built coil the conditions which cause 4046 failure can arise.

Chris_Knight

Wed Feb 10 2016, 06:56AM

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

"..." and Sigurthr -

Sorry, I was being unclear. As Sigurthr guessed, those two questions are completely unrelated. I've been juggling a lot of different ideas here, and in my other thread I proposed an experiment comparing the audio performance of a 4mHz class E AM, a PLL FM with FGH40N60SMD at potentially 400kHz, and a class D with the same resonance and bridge with PWM modulation.

However, I've been thinking. The PLL chip works up to around 2.2Mhz, right (I read that somewhere in the forum..) For a fairer comparison between all three, and assuming I switched over to high speed MOSFETs, how fast could I run the fullbridge? From what I understand, anything over 1Mhz is a real challenge for a large power fullbridge.

Is there a real difference in hiss between 400kHz and 4Mhz? What would you say would be a good compromise?

_______________________________________________ _________________________________________

Wow, thank you both for such detailed and helpful information. I will certainly read the documentation you linked.

"..." - Looking at that first thread, the resonance was only 319kHz? I'm surprised to hear that IGBTs failed, I know loneoceans has used IGBTs in his 450kHz QCWDRSSTC

Sigurthr

Wed Feb 10 2016, 07:01AM

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

I did find a video of my FMSSTC actually. This is from before I rebuilt it, so the audio is actually a bit better now, and I lowered the power a bit to 285W. In the video it is operating at about 400W though.

As far as the hiss goes, it's a combination of power and frequency. I haven't noticed much difference in actual sound texture between 200KHz and 700KHz. If you have better modulation depth the audio shines through the hiss, if you can't modulate as strongly or if you increase output power to increase volume instead of increasing modulation depth you get more hiss.

I've never made a coil above 1.3MHz, and that was a real challenge. Not only does the bridge have to dissipate more power, but the gate drive requirements become far more difficult. UCC chips are only good to about 1MHz for a half bridge, and I wouldn't use a full bridge above 500KHz (same gate drive currents in both scenarios) without changing gate drivers or doubling/tripling up on them.

Chris_Knight

Wed Feb 10 2016, 07:14AM

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

Thank you for that detailed explanation. It'll take some more reading for it to fully sink in.

My goal is audio quality similar to this. I know class e AM modulation can achieve this, but is this doable with FM PLL?

Here is my planned secondary for the PLL and the PWM. The class e will probably be a coated toilet paper roll similar to Richie or a small pvc pipe, I'm not sure which one is preferable.

Which would you consider more difficult and time consuming to build: PLL or class e?

...

Wed Feb 10 2016, 08:50PM

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

It is hard to read the diagram shown on that coil, but judging by the lack of heatsinks I would guess that coil is most likely being frequency modulated, or possibly being PWMed like the original plasmasonic from eastern voltage research (note that his scheme is not the same as the PWM which was mentioned earlier, his coil worked by modulating duty cycle of the drive to the high/low side transistors; at maximum output each one was on roughly 50% of the time, and at low output each was on for something like 20% of the time).

Going to a higher switching frequency reduces the background hiss, and at around 5-10MHz the coil becomes completely silent so long as you can maintain a stable discharge (no flickering). In my experience there is not a huge difference between 400kHz and 1MHz, but it takes a good bit more effort to build a coil that runs at 1MHz than at 400kHz. With the improvements in the available mosfets its certainly not impossible to hard switch at 1-2MHz, but you are going to pay a good bit more for the switches and gate drivers for a given coil power.

It is absolutely easier to build a hard switched coil than a class e coil, because the tuning is quite tricky on a class E. Furthermore, it is not trivial to audio modulate a class E coil because they need to be tuned to each specific operating point, so even with an additional bus modulator it is not easy to maintain class E operation over a wide variation in output power.

Sigurthr

Thu Feb 11 2016, 01:27AM

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

Hard switched is way easier. Hard switched without messing with the 4046 is even more easy.

The video you linked shows a coil with very low bass response, resulting in a plasma tweeter type effect, rather than a full bandwidth audio reproduction.

Perezx

Sun Feb 14 2016, 08:39AM

Registered Member #54263 Joined: Thu Jan 15 2015, 09:54AM
Location: Perth
Posts: 35

Couple years ago I made simple audiomodded direct-feedback SSTC.
Interrupter was built using 555 timer with basic frequency around 44 KHz, so it was silent (well, white noise).
Audio signal was sent to "5" (control) pin of 555, forcing it to change both frequency and duty cycle.
This resulted in strange phase and frequency modulation, although, sounds very good.

Chris_Knight

Wed Feb 17 2016, 06:14PM

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

Hey "..."

So I've been thinking, and I think the class d audio amp as a class e tesla coil power supply is actually a good idea I want to try. I've been working with reaching's class e audio modulated SSTC, but I'm not completely sold on the linear regulator modulation.

Where would I get a DC coupled class d amplifier? Other than reading about class d audio amps, I haven't had a lot of experience

Perezx

Sun Feb 21 2016, 01:32AM

Registered Member #54263 Joined: Thu Jan 15 2015, 09:54AM
Location: Perth
Posts: 35

You can just use 555 or 494 which allow frequency and/or duty cycle control

previous 1 2 3 next

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