A question about power amplifiers (and/or SMPS design)

[MysticalDork]

Hi! I'm an electronics hobbyist, and I'm in the process of designing and building a tabletop solid state tesla coil, to demonstrate wireless transmission of power and electric fields, etc.

Since if all goes right this coil should be capable of running full-CW mode, I figured it would be a fun idea to add audio modulation. There are two methods I plan to use:
A) Simple low voltage PWM interruption: This will allow simple integration and crappy audio quality for a minimum of effort.

B) I also plan to integrate (probably at a later date, I need to see plasma!) a method to modulate the main input rail's voltage (~170V, 2-5A rectified mains) to vary the input power, and hence the breakout size and allow reasonably high-quality modulation. The problem lies with this second method, as I have no idea how to modulate a 170V supply rail without a giant audio transformer and a separate large power amp feeding it.

I figure I only need about 15-25% modulation depth, so I've been toying with the idea of trying to use something like a TPA3123D2 class-D amplifier module, somehow fed into or modulating the rail. I'm trying to keep the electronics payload of this project as small and compact as I can.

This is the circuit I'm currently working off of, with some minor tweaks. (Yes I'm aware the 3904 and '06 are swapped.)

Thanks in advance for the help. First time posting, so if I've made a goof please be gentle.

 

[Tony Stewart]

How do you plan to receive all this power? or at least verify you are radiating energy and at what frequencies?
By interference? or Radio or you really just want to create high voltage arcs?

 

[MysticalDork]

The simplest way is just an antenna, but it's also pretty trivial to make a resonant tank circuit to receive at a specific frequency. I have an oscilloscope, which I'll use for tuning and testing. The resonant frequency I'm shooting for is 4.096 MHz. High voltage arcs are definitely a nice thing to have.

 

[Tony Stewart]

The xtal wont pull up more than 400 ppm and not 2000 ppm and that would be with ~1 pf instead of 20.
Maybe in series mode or another Xtal. designed for 4.096

The Secondary transformer can be pulled up from 3.6 to 4 by reducing the winding capacitance or using Litz wire for lower inductance.
But the Q may not much more than a few hundred.

The upper circuit may have bias issues with variable MOSFET Gate thresholds spec'd from 2 to 4V and the bias resistors are fixed, so when clock has not started or stopped for any reason the FET may be stuck on.. A polyfuse is a handy thing to have.

 

[MysticalDork]

Awesome, thanks for the pointers. I'm planning on using a self-contained oscillator module and a 4000-series buffer to amplify the clock signal rather than using a discrete crystal, for simplicity's sake. IIRC, the reason for the 3.7 fRes on the secondary is to allow tuning by removing turns. The schematic isn't mine, so some things have already changed.

Any ideas about the input voltage modulation?

 

[Tony Stewart]

modulation for what purpose? Voice? just modulate FM with switched cap on discrete crystal oscillator.

 

[MysticalDork]

The modulation is for a plasma speaker, and class-E amplifiers don't play well with FM since they've got a tuned low-bandwidth resonant topology.

 

[Tony Stewart]

for a Q of 100 at 4MHz you have a bandwidth of 40KHz.
Is this for high quality with noise immunity? WHen you ask a question. all the relevant details are essential

 

[MysticalDork]

This isn't for anything other than tinkering and fun, really. I've seen a couple people that used FM with a class-E tesla coil, and it reduces the efficiency and causes marked fet heating. I figure it's more efficient to have the coil run at it's preferred resonant point and just modulate its amplitude by modulating the supply rail.

 

[MysticalDork]

Bump. I guess I should retract all the stuff about tesla coils, since the application isn't important. All I need is help designing a power supply (/amplifier?) that will take a standard line level audio signal and ~150v rectified mains, and output a fairly clean, linear and high current signal that varies from ~110 to ~150-170 volts. No zero crossing is required, this is all on the positive.

 

[dr pepper]

A Tl494 smps chip can be xtal controlled instead of its built in osc, if you put audio in on the deadband i/p you can very effectively control the power to the coil, the freq will be modulated but only by the bandwidth of the applied signal so a high Q coil ought to work.

 

[MysticalDork]

I like the way that idea has pointed me. How difficult would it be to use the TL494 just as a SMPS, and just design the SMPS to be a very low-delta dc-dc converter? 150v in, 100-150v out determined by the deadband input? Essentially a really high power class-D amplifier (audio is converted to variable dead-time digital, then converted back to an analog signal) I think that would work, but what would I use as the main transformer?

 

[dr pepper]

Yes you can.
Dead time and pwm are essentially the same thing, its just a way of forcing a smps to vary its o/p.
I dont think modulating the power rail to your circuit would be very efficient, a little like supplying a fixed power supply with a variable power supply, just have 1 smps that does the job. Its quite possbile to get high quality audio this way.

To accomplish what you say, or how I unterpret it I'd probably put an audio transformer in series with the rail to the tc, and then feed the transformer primary with high power audio, similarish to how an am transmitter works, as I said though its a waste of circuitry.

 

[ronsimpson]

I can't find the right part. This is not the right part but......
Vin if from the rectified power line and filtered. 140 to 150V
Vcc is low voltage.
The two MOSFETs chop the power line and L1/Cout filter back to DC.
With most PWM buck down configurations it is possible to add "audio in" to the feed back loop.

 

[MysticalDork]

Ahh, OK. I'll look into integrated-switch DC-DC converters. Thanks!

 

[MysticalDork]

dr pepper Thanks for the input. I realize that the most efficient method would be to integrate the modulation into the main TC amplifier, and the reason I'm not doing that is that I've already finished designing and am almost finished building the main TC amp. I figure it's easier (if a little less efficient) to just have an "add-on" power modulator than to change the whole TC circuit.

 

[dr pepper]

Ok fair nough.
One thing if your existing driver uses feedback then it'll try to counteract any supply rail modulation, the driver needs to be ratiometric, ie if the rail volts changes so does the o/p.

 

[MysticalDork]

Yep, the driver I'm using has no feedback, just a crystal oscillator. All but the last stage runs at 12v, and the last stage is just a big SiCFET and a DC blocking cap across the primary with the HV rail across that. The FET is driven through an isolation transformer, so no danger of line noise or whatever messing with the LV side of things. I designed this thing as a "board sandwich" with one being the power supply, one being the main amplifier, one being the modulator (future) and all of them just stacked together with long nylon standoffs.

 

[dr pepper]

Dont forget to show us it when its done.

 

[ronsimpson]

A while back I made a supply much like this.
I rectified and filtered the power line to get 140VDC. The built a small isolated 0 to 30V DC supply that sat on top of the 140V. In this case we never needed to go below 140 and never above 170. So the switching supply was very simple and small. (low voltage parts)