SSTC performance issues; audio modulation, ballasting, and noise / Tesla Coils / Forums

Forums

4hv.org :: Forums :: Tesla Coils

« Previous topic | Next topic »

SSTC performance issues; audio modulation, ballasting, and noise

1 2 next

Move Thread

LAN_403

Sigurthr

Mon Mar 25 2013, 06:39AM

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

Hey all, I'm hoping someone might be able to shed some light on an interesting observation.

After many, many unsuccessful attempts to audio modulate my SSTC via the UCC chips' enable pins I decided to go with injecting audio in to the DC bus with a car amp (still in the mail) and a suitable modulation transformer. ***see post script***

I first tried using a much smaller 120VA 120v 1a/12v 10a filament transformer, but the 12v winding was not heavy enough gauge wire and it got quite hot even after very short runs on very low power (<6A draw / 600W on the SSTC). I figured I would either have to find a way to immerse the transformer in oil to help cool it or find another transformer. The DC resistance of the 12v winding is 9.6Ohms - enough that there was significant I2R losses. It dropped output power of my coil to less than half, as expected.

So, I reworked an old 600W HID autotransforming balast into a 1:1 isolation/modulation transformer capable of carrying the >15Arms that my large SSTC draws. Basically the "secondary" winding is nice heavy 12ga which was wired in series with a multitapped 20ga winding. Prior to modding this transformer I had tested it on 120V mains and found that both "windings" were of enough reactance to not trip my breaker. With the tap points chosen so the ratio was 1:1 it operated as a balast and reduced current draw liming power to about 600W, and it did not seem to matter which "side" you had facing mains.

I simply disconnected the heavy winding from the thin winding to isolate the two windings (both are insulated from the core) and chose the two tap points on the thin winding which preserved the same turns ratio between the windings. I measured the DC resistance of the thick gauge winding to be 2.2Ohms.

So I wired it in, in series with the + line of the DC bus between the filter cap bank and the positive rail input on my half bridge. I was expecting a slight loss in output power, as I2R losses and Vdrop would dictate. This is the opposite of what I found! Note that the "primary" side of this transformer is not connected to anything at all right now, as the audio amp is still in the mail! There is NO power being fed in to the thin winding, just the thick winding in series with the DC positive rail. I gained about 3cm of spark length and the coil runs much quieter and cooler, even at full power. I was getting a popping noise when the variac was dialed above about 80Vac which my scope seemed to confirm was caused by the coil temporarily not outputting any signal, momentarily extinguishing the arc, causing noise. I attribute this to not enough filter capacitance (1880uF currently with another 4400uF on the way) as it never happened below about 10Arms of draw on the 120V line. With the modulation transformer installed this now only occurs above 100Vac on the variac, at around 12A of draw on the 120V line. I used to be able to run my coil at only <350W for continuous operation (thermal limitation), now I can run it at <725W and achieve the same level of thermal equilibrium. Has anyone else encountered a similar situation? Am I missing some key factor here? The only wild guess I could come up with is there is an impedance mismatch somewhere and adding the impedance of my transformer in to the equation yields a more efficient transfer of power. I have no basis for this what-so-ever though, hah!

***P.S.: I built several PWM modulators of various schemes to try and audio modulate this coil, but all attempts failed in the same way. Once PRF rose above about 1600Hz the output power (and input draw) of my SSTC dropped off almost exponentially. At 1400Hz I would get a nice banding effect from the output with only about one inch of spark length lost. At 2500Hz I lost nearly 25% of the power. At 20kHz I was down to ~7% output even with the variac cranked all the way. The coil barely drew 1A of power. This trend continued all the way up to a few kilocycles away from f0 at which point increasing the frequency closer to f0 would increase power in the same fashion that it rolled off at above 2kHz. As close as I could get to f0 without using feedback (I'd say no more than 1kHz off) I reached about 90% power but there was a nasty buzzing in the output from the dissonance of the two frequencies and the UCC chips started to heat up. I'm not sure how everyone using the standard antenna feedback in to a schmitt inverter feeding UCC chips gets their audio modulated PWM to work, but I sure as hell could not.

P.P.S.: I have a concern that Vripple from insufficient capacitive filtering on the DC bus will cause an AC signal to appear on the audio amp side of the modulation transformer. Is this nonsense or something I should be concerned with? I don't really have a good way of measuring any point of my SSTC while it is in operation to actually see what the DC bus voltage looks like as both my DMM and scope go haywire when within about 10ft of the coil, even at low power. In theory the grounded metal casing of the scope should act like a faraday cage and protect it, but the trigger goes nuts at about 10ft out and at 12ft my probe reads a 300Vp-p signal even when the coil is running at only 10% power, so I've been too afraid to try bringing it any closer to the coil. I've been doing all my current draw measurements with a Kill-a-Watt meter on the mains input.

Steve Conner

Mon Mar 25 2013, 11:11AM

Registered Member #30 Joined: Fri Feb 03 2006, 10:52AM
Location: Glasgow, Scotland
Posts: 6706

Weird popping noises that suddenly kick in at a particular voltage suggest that RF is getting back into your small-signal circuitry and making it misbehave. Besides making weird noises it could drag the driver out of phase and make your bridge overheat from excessive hard switching.

Moving the wiring around a little can make a big difference, so it's not surprising that adding a big choke made a difference. The real fix is to improve the RF immunity of your driver circuit, so it only reacts to the desired feedback signal.

Sigurthr

Mon Mar 25 2013, 04:00PM

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

Interesting, I hadn't thought about RF penetrating in to the driver above a threshold, but that does make sense (I originally thought if it was lack of filter capacitance it should produce the 120hz hum, not just random pops).

Do you think shielding the driver in a grounded project box with a cutout for the antenna will be sufficient? I hope I don't need to start installing chokes and low pass filters all over the place.

doctor electrons

Mon Mar 25 2013, 07:57PM

Registered Member #2390 Joined: Sat Sept 26 2009, 02:04PM
Location: Milwaukee Wisconsin
Posts: 381

Not sure if this is a part of your problem or not but in addition to what Steve said.
At RF board traces and other wiring/components can form capacitors. Parallel traces have been
known to become a problem if they aren't far enough apart, less so if they are parallel on opposite
sides of the board (top and bottom).
I don't see any reason that you shouldn't shield your control electronics, there is really no negative side to
doing so. Again, like Steve said, moving things around can help. If you do decide to shield the goodies you can
use thin copper sheet available at most hardware stores. You can also use screen, i have some fine copper
screening that i use for just that. Keep in mind when using the screen that there is a method for calculating the
size of the openings in the screen based of frequency. Something to do with bandwidth and the hole size.

Sigurthr

Tue Mar 26 2013, 06:54AM

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

I changed drivers to a nice dual layer PCB which has most of the second layer as a ground plane, placed the entire driver inside a grounded metal enclosure, moved all small signal side wiring away from the RF side of things and I'm still getting the popping effects above 100Vac or so. They sound less harsh but it is still there. Fets are running pretty happy still too.

This new driver board (Designed by Polonium-210) seems to have an issue with starting the oscillation though... I have to touch the antenna to kick start things, which often results in a painful RF burn if I'm not holding something metal in my hand to make the actual connection.

I'm going to rebuild my own driver in a small enough form factor so that I can put it in a shielded metal case as well, then use this one from Po210 as a backup driver.

Physics Junkie

Tue Mar 26 2013, 07:42AM

Registered Member #7267 Joined: Tue Oct 16 2012, 12:16AM
Location: Detroit, Michigan
Posts: 407

I know what you mean by having to touch the antenna. I had similar noises with my small sstc's and i dont use shielded enclosure. All I did to fix it was use a CT instead of antenna. It seemed like touching the antenna or grounding the opposite end would kick start it like you said.

Steve Conner

Tue Mar 26 2013, 11:25AM

Registered Member #30 Joined: Fri Feb 03 2006, 10:52AM
Location: Glasgow, Scotland
Posts: 6706

All I'm saying is, the first time I tried to build a solid-state Tesla coil, I learned a lot about EMC really quickly.

When you say the metal enclosure is grounded, what is it grounded to? The obvious answer is the ground plane of the driver board.

But then what is THAT grounded to?! The not-so-obvious answer is that there needs to be a high frequency return path from the driver board ground plane to the DC bus negative rail, and the whole lot needs to be referred to the ground wire of the outlet too.

A lot of RF current couples through the interwinding capacitance of the GDT, and that needs a path back to where it came from, otherwise it'll kick the driver's ground plane up and down relative to the rest of the world. The antenna has no way of telling this from the legitimate signal. If the antenna stays still and the groundplane moves, that's the same as the groundplane staying still while the antenna moves.

So to recap, the DC bus negative rail should be grounded to the chassis and the outlet green wire (through a small Class Y rated capacitor if isolation from the line is required) and the driver board's ground plane should also be grounded to the same point.

I never liked antenna feedback. I dumped it in favour of a Faraday shielded CT.

Sigurthr

Tue Mar 26 2013, 03:39PM

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

Interesting, I did not think to include the negative dc rail in the grounding scheme. Ill have to do this next time I work on the coil and report back. As it stands the ground plane and enclosure are grounded to the variac case which is connected to mains earth. Secondary lv side is connected to both mains earth and a dedicated rf ground rod array. I didn't know it was ok to have the negative dc bus connected via ground to the lv end of the secondary.

I'm all for switching to CT feedback but I don't know what specs to look for in a CT or how to properly burden it.

Physics Junkie

Tue Mar 26 2013, 06:59PM

Registered Member #7267 Joined: Tue Oct 16 2012, 12:16AM
Location: Detroit, Michigan
Posts: 407

Take a look at the feedback circuit on ward's UD 1.3

Load with zeners and ultrafast diodes clipping the sine wave. Resistor limits current to the inverter.

For the CT, I use a blue epoxy coated ferrite core by epcos N87 version. Manufacturer code is B64290L0674X087.
The fair-rite type 77 is good too. I've also used their green epoxy coated core type 77. Code is 5977002721
I've seen a couple different ways of winding the cores. I've never tried cascaded method but I've read its optimal. I usually do 70 or 100 turns of 26awg magnet wire.

Sigurthr

Wed Mar 27 2013, 08:17AM

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

Well I bonded all grounds together, including the secondary lv end and performance dropped by about 25%. I also had a strange issue pop up; after shutting off power to the variac the coil stayed on at low power! I even yanked the plug but there was still breakout. I segregated the negative dc bus from the secondary lv end and everything went back to normal. Somehow power was being fed in from somewhere around 30Vac judging by arc size. Bonding the dc bus to the driver ground plane had no apparent positive effect. In fact it made it harder to start oscillation.

My amplifier came in today and works well. Modulation transformer seems to work good too. Tested the rig on a zvs driver and got audio out. Damned thing won't work on the TC though. I got brave and scoped the dc bus with the modulation present and all looked good; ac audio signal superimposed on a dc voltage. For some reason when TC oscillation starts the audio signal vanishes from the DC bus though. At first I though it was swamped by the Vripple so I tried a 10:1 modulation transformer and got a 150Vp-p audio signal which was way stronger than the Vripple, but again as soon as TC oscillation started the audio vanished. I can only assume it was the amp going in to some kind of shutdown mode. My dreams of an audio SSTC seem to be slipping away.

1 2 next

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