Active clamp network - non-ZCS SSTC feedback method / Tesla Coils / Forums

Forums

4hv.org :: Forums :: Tesla Coils

« Previous topic | Next topic »

Active clamp network - non-ZCS SSTC feedback method

1 2 next

Move Thread

LAN_403

LithiumLord

Mon Jul 27 2009, 03:50PM

Registered Member #1739 Joined: Fri Oct 03 2008, 10:05AM
Location: Moscow, Russia
Posts: 261

Why?
Well, the feedback in non-DR systems is always tricky. You go with antenna - but it's just thinking that Isec=dUc/dt, which isn't very nice - on early ringup the signal gets too low to properly track it, later it's a good feedback, but not ideal anyway - and at the point the spark forms the waveform gets messy and we no longer can claim that the current we get here, being a sum of the secondary current, the spark current and an additional displacement due to the capacity being a distributed one, to be the actual one flowing through the secondary alone. Also interference issues - and this is the reason not to use the antenna if extreme accuracy is required. We move to CT - and get the opposing situation - the ringup is lagged because of the current through the CT primary is not enough. The most of the actual duty cycle is ok though. However the startup may get tricky, and functionality on the lower voltages is not possible after some point. Would be nice, however, to somehow combine both methods, measuring the secondary current with something that features the fast response of an antenna.
How?
That's not that tricky in concept - all we need is a driver with the topology alike to an antenna-based TCs where we will just rig the secondary base into the antenna jack and swap the primary phasing. However there is one "small" problem - the common clamping method is based on driving the inrush current back into the PSU - and it is ok for a low-current feedback the antenna is, however I doubt a PSU will take a couple of amps from your secondary just as good ;) So, we need a better clamping method. Zeners are not an option, diode chains are too slow, an NPN amplifying a low voltage of a two antiparallel diode shunt based on their forward voltage drop gets the waveform messed up, Zener+transistor shunt may overheat and is too slow as well... Yes, I checked those out and ended up looking for a better way around. I ended up with this:

The idea behind this concept is that we need to clamp the feedback just like we always did, but somehow dissipate the inrush current. In order to do it, a simple oscillator is implemented. Whenever the voltage across the clamp rails exceeds the desired level by the UCC high threshold, the rising slope of the clamp envelope ends with the MOSFET switch connecting the R2 ballast across the rails. This starts the falling slope that ends when the overvoltage drops below the UCC lower threshold. Therefore the signal gets a nearly triangular envelope, but it is higher then the logic voltage anyway and will be trimmed by the second, passive, clamp! The higher the inrush current through the higher diode of the clamp gets, the shorter is the rising slope and longer is the falling one as the ballast still has to load the current source being clamped. This works up to the point the resistor stops clamping the voltage wthing the desired range and the device enters a constant clamping mode. But this is stll not the limit - in order not to cause a malfunction we now only need to stay withing the voltage limits, and with a 2.7 Ohms it's plenty of headspace ;)
Now about the testing. I assmebled the device straight according to the schematic, it works pretty well. The first tests proved the clamp works well - even with the lower currents it doesn't cause any signal distorsion. After some mess with it I rigged it into the coil and gave it a shot. The failure happened of course, but not in form you could expect - the improved ringup caused the optos to switch the bridge off randomly, I'll do a full-power run as soon as I find a better way around - nothing blew, but the optics are to be improved. Anyway, I managed to get up to like 30% voltage and the clamp performed well, even during the flashovers into random objects. The fun part is that I had the startup osc still there, but it did no good of course as 5v vs 50k wouldn't do much with the secondary I have. However that was not causing any startup issues as the oscillations used to start just on a few volts across the rails. Even more fun - while attempting to run on the small voltage that was like 1/4th of the one I used to get a tiny corona with the antenna feedback in order just to check the traces I was surprized to see like 5cm spark shooting out of the breakout rod. With some messing around just to do a couple of test shots on a higher, at least 80%, voltage, I was pretty amazed to see the coil shooting sparks in a way more channels, like 5+ per bang, some ignoring the breakout rod and just jumping off the rim of my split top load I never seen them to form before that easily. On the lower power I ran it for a while and nothing fried, and keeping in mind that coil runs like 25cm in that mode I can already tell this feedback method is perfect for at least small coils - will tell more as soon as I get the optos cranked up. Drawing arcs in a few ungrounded objects (had no grounded ones so was just messing around) caused the arcs to be bolder then before - feels like the improved feedback surely lacks any older problems, just as expected.
Anyway, requirements and design notes.
1. A good electronics ground is required so that that will also take the secondary current as well.
2. Needs a separate +12v rail regulated or an unregulated ~1A for the clamp supply, expect no reflected voltage unless the loopback diode is installed (later about it)
3. The logic "keep-alive" rail is needed for the initial charge over the clamp capacitor, you can use a divider if have no reference voltage source.
4. Choose VD4 so that it is rated for the same voltage as the actual TC driver needs to exceed the switch-on threshold at all time, 5v seems like a good choice. This part should NOT be below the keep-alive voltage to exclude any unnecessary triggering.
5. Use the pot to set the higher threshold if needed, this also extends the histeresis effectively reducing the MOSFET switching loss. Do not overdo this though as the consequences may get "not good" ;)
6. The loopback diode VD5 is used to drive any voltage that's above the PSU ratng back into the PSU. Use it only in case your unregulated bus is fed from a generic transformer-rectifier-capacitor supply with a large smoothing cap, connect it to the input capacitor otherwise. Use it only in case you know what you are doing it for. The goal of this diode is to protect the device from situation like another coil is working nearby, feeding this one via it's clamps. This will ensure that in case the clamp will start charging above the 12v threshold it will be powered up in order to clamp the current that can damage the unpowered coil.
7. The device needs heatsinking ;)
8. The active clamp network input node described here handles any input current, so you can still use it for any other feedback types if required.

HV Enthusiast

Mon Jul 27 2009, 05:19PM

Registered Member #15 Joined: Thu Feb 02 2006, 01:11PM
Location:
Posts: 3068

Just a suggestion if you want to get more responses . . .

Most people here will read about the first two sentences of any post to see if it interests them. If you write a book (like you have here), people aren't going to take the time to read the entire thing.

I would recommend adding a abstract or introduction in the beginning limited to about 2-3 sentences top stating the problem and what you propose in solving it.

Then in the end, put a quick summary, again 2-3 sentences, showing what you came up with and how it solves the problem stated in the beginning.

To be honest, a post like this, i just skip right over as i don't have time to read a whole page of.

Also, the subject "Active Clamp Network" doesn't really do much to tell the reader what the post is about.

Just some suggestions so you get more interest in your thread.

LithiumLord

Mon Jul 27 2009, 05:53PM

Registered Member #1739 Joined: Fri Oct 03 2008, 10:05AM
Location: Moscow, Russia
Posts: 261

Heh, in fact both the abstract and conclusion are there, however I got a couple of principles - I don't want to post any unfinished research or it ends up in a classic troll thread like "I want to run a TC using a potato as a spargap", and I prefer to present all the info I obtained in my own experiments both to exclude any misunderstandings and any delays in the work of those trying to test a concept - in fact, it is best to get feedback instead of questions you should have answered straight off. Actually, was aiming for this to look more as an article on the feedback method I developed, however the title, yes, is "a bit" unclear, will fix now :)

LithiumLord

Thu Jul 30 2009, 01:00AM

Registered Member #1739 Joined: Fri Oct 03 2008, 10:05AM
Location: Moscow, Russia
Posts: 261

I guess this one will work as an excuse for doubleposting. After a few tweaks of the optocoupling stuff and implementing a dutycycle adjustment pot (no 15k one so I ended up with 4k7 - will replace as soon as I get the proper one) I finally was ready for the real first light with the new feedback. And now - fun pictures :)

Yeah, yeah, very impressive - I can light up a neon bulb with an SSTC. However what is really interesting is the fact that this was a ringup test to find out the minimal input voltage, and at this picture the variac was set for only 5v on the bus!
Ok, a bit more, 30vdc now - with antenna feedback I could see a tiny corona on this voltage at best.

The feedback seems to be very fast according to those pics, now let's go for 120vac.

As seen, the ringup is fast enough to make very branchy sparks, so it's time to finally crank it up to the top. 240VAC!

Woah! Primary(!) racing arc!

As you can see, the spark appearence is much different from the original, and reduced duty cycle is not the reason - the higher dv/dt even makes the topload gap (I use split toroids to exclude heating) break out frequently, like I never seen with the antenna! Wonder what other advantage will I get from this later - so far I need to swap the pot with a bigger one to get back to the bigger duty cycle and do some long-term heating tests with bolder arcs.
And a video, of course:

Mates

Mon Aug 03 2009, 09:49AM

Registered Member #1025 Joined: Sun Sept 23 2007, 07:53PM
Location: Czech Rep.
Posts: 566

Hi Ivan,
Actually, Iâ€™m a bit surprised that there is no rather wild discussion about this interesting SSTC feedback project. People on the forum are just copying two or three basic designs with minimum changes making the whole SSTC adventure more or less boring... Finally something new appeared but it seems nobody is interested...

I thing that the presented coil has an extraordinary performance (despite it looks a bit crappy). It also means that the point of high quality SSTC giving long term performance and not killing the MOSFETs is not so much in a perfectly made topload, primary and secondary but in a good driver...

I plan to start a new SSTC project soon. It will have a conical secondary made of traffic cone and Iâ€™m seriously thinking of using your feedback design. Can you provide a complete scheme, including the power section?

Good luck in your further work...

Cheers Mates

Nicko

Mon Aug 03 2009, 01:15PM

Registered Member #1334 Joined: Tue Feb 19 2008, 04:37PM
Location: Nr. London, UK
Posts: 615

Mates wrote ...

People on the forum are just copying two or three basic designs with minimum changes making the whole SSTC adventure more or less boring...

Being fair, my qualitative view is that most people actually have little if any electronics knowledge, and that copying a known working design suits their purposes just fine. Knowing how to wire up a 555 does not an EE make - a few years at university or similar is required - this sort of stuff is a bit cerebral. For that matter, even though my degree is in EE (a good few years ago), it was nearly all LV, so when I started on this route (not that long ago) I had to start learning the HV side from scratch - not easy and a long long way to go.

When messing with HV and stuff like this, for most its the destination that matters, not the journey; for the curious amongst us, it can be the other way round (with apologies to RW Emerson).

wrote ...
Finally something new appeared but it seems nobody is interested...

I am!! Whilst the SW & other popular designs have merit, they're certainly not the end game... Finn's doing great stuff, as, I suspect, are many many others, quietly, in the background....

wrote ...
Good luck in your further work...

Absolutely!

Cheers

Steve Conner

Mon Aug 03 2009, 02:39PM

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

I just didn't really understand the original post at all, it seems a lot of trouble to go to for nothing. Why extract more power than you need through the CT, only to think up a complicated way of getting rid of the excess power?

I prefer to use a CT with two back-to-back fast rectifiers as a burden, as shown in this schematic:

The non-linear behaviour of the diodes gives good sensitivity for weak signals while clamping the feedback signal to about 1V peak-peak at high power. I tested it over a wide range of input currents, and the phase lag at low currents was hardly noticeable.

1V p-p of signal may be a problem for the standard 74HC14 circuit (unless you bias the input properly, which nobody bothers to do) but for my 4046 PLL circuit it was plenty. 1N5819 schottkies worked too, IIRC.

LithiumLord

Mon Aug 03 2009, 07:45PM

Registered Member #1739 Joined: Fri Oct 03 2008, 10:05AM
Location: Moscow, Russia
Posts: 261

Well, I guess it's time for me to fight for the honor of my creation lol.
As for HC14, we are kinda stuck for sure, as even with biasing the slopes will be present, also it's pretty hard to get a CT to operate well on very low currents. Should you exclude the CT, and additional effects will kick in as well, primarily the fact you can't bias the secondary itself that easily (we are talking about non-DR coils here). As for the slopes, according to my own tests such a circuit produces some long ones, which may have consequences.
Now the power waste. Let's take a sample coil, say T secondary turns, k coupling, P power, N primary turns and V input voltage, running CW on an air load. The current to be consumed is, apparently, P/V. This current, as we take it, is a pure active one, eg the one we eat off the filter caps and never give back. Apparently the secondary current will be in phase with it, and with the spark load act like if the coupled fraction of the secondary would be burdened with active resistance, with the current floating there be equal to (P/V)*N/(T*k). This current will make the clamp dissipate some power, we will simulate it as a Zener, therefore it will dissipate (P/V)*N/(T*k)*Vcc. Now let's calculate the power ratio which is equal to P/((P/V)*N/(T*k)*Vcc)=V*T*K/(N*Vcc)
Now, let's try a simulated example. We have a 1000T secondary 0.5 coupled to a 15T primary running 300w off 300vdc bus. The active current will be 1A and power ratio will be equal to 2000. The dissipation will be @150mW. Should it be a 2kVA coil, and we still get 1W dissipation, kinda nice!

Firnagzen

Tue Aug 04 2009, 01:52AM

Registered Member #567 Joined: Tue Mar 06 2007, 10:55AM
Location: Singapore
Posts: 147

... As opposed to? Some basis of comparison would be nice.

LithiumLord

Tue Aug 04 2009, 09:50AM

Registered Member #1739 Joined: Fri Oct 03 2008, 10:05AM
Location: Moscow, Russia
Posts: 261

As the counterexample form the last post I used a shunt of two antiparallel diodes as is clear reading down the thread ;) For topology comparison, tuning tricks, tests and design notes see the first post here ;)

1 2 next

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