Why Run SLR Drivers At 50% / High Voltage / Forums

Forums

4hv.org :: Forums :: High Voltage

« Previous topic | Next topic »

Why Run SLR Drivers At 50%

Move Thread

LAN_403

dude_500

Sat Dec 03 2011, 09:17AM

Registered Member #2288 Joined: Wed Aug 12 2009, 10:42PM
Location: Cambridge, MA
Posts: 179

I'm wondering why people like to run SLR drivers at 50% duty cycle at half the resonant frequency. I see that it is good at removing ring-up, but it seems to me that it waste a ton of energy because there is a lot of current that no longer couples to the secondary (awful power factor in the transformer). I attached plots of both this described operational mode, as well as driving it at 99% duty cycle at the frequency of the LC.

The teal line shows inductor current, green shows V1 source current, and red/blue show gate drive. I divided source current by 3 just so that it is visually possible to see both currents without overlap. In actuality, realize source current is scaled by three to make some overlap. (edit: it occurs to me that I divided source by 3 in one picture and inductor current by 3 in the other... and this forum doesn't let me update or change pictures. It should be clear what the graphs show though if you are familiar with the topology)

As can be seen, the source current is entirely in one sign in the 100% duty cycle on f_res drive, which indicates that there isn't waste reactive current going through the transistor bridge. But in the 50% duty cycle drive, 80%+ of the current from the source is reactive, and thus is heating up the bridge through forward voltage drop of IGBT's (or ohmic if using FET's, even worse).

So basically, it seems to me that at 100% on-f_res drive, the entire actual current going through the transistors couples to the output (at least in an ideal simulation), where only a small fraction couples at the 50% duty cycle method and the rest is brought back to the source, but burns up voltage drop doing so.

So why do people do this? Is my analysis flawed in some way, or is there some huge advantage that makes it worth burning up energy in the bridge?

Steve Conner

Sat Dec 03 2011, 11:36AM

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

If you drove it at its resonant frequency, it wouldn't be a SLR any more, it would be some other resonant converter topology. So the question is really, what are the advantages of the SLR?

Well, because it isn't resonant (maybe it should be called the SLQR?) it's easy to control. Each half cycle of drive dishes out a constant amount of charge to the load, so it has the same easy plant transfer function as any other discontinuous mode converter, and it can't resonate itself to destruction if the control loop fails, or you are so scared of feedback math that you never even implemented a control loop.

And, it is a nice match for IGBTs. They happily carry huge peak currents, and they come packaged with fast diodes. So the large amount of reactive power comes cheap. And the SLR drive scheme guarantees zero current switching without any complex PLL-type electronics. You can use slow cheap IGBTs with a simple driver.

You could design a pure resonant converter to perform better into one constant load. But for general purpose hobbyist-type use, the SLR is easier to tame and get results with in practice. I've seen some successful ones made with pulse frequency or delta modulation, as well as capacitor charging supplies for Tesla coils that just run open loop.

I don't know if it is ever used in industry.

dude_500

Sat Dec 03 2011, 06:21PM

Registered Member #2288 Joined: Wed Aug 12 2009, 10:42PM
Location: Cambridge, MA
Posts: 179

I have my project setup for the most basic feedback (is the output voltage below threshold? If so, go). It works remarkably well with only about 5% ripple max at a few amps output on the test load with 100uF filtering, in both 50% SLR and 100% on f_res operation. The voltage even stays well regulated going from load to no load. So I guess there's no reason to not run it 100% for the added efficiency. The transformer's power factor seems to go up from a measured 67% at 100% duty cycle (although simulation seems to indicate that near perfect should be possible) vs. 53% at 50% SLR. Hopefully that will also work well when I hook it up to my multiplier.

Here is a video of the feedback operating in SLR mode. The output voltage is negative polarity, so lower is higher voltage.

Steve Conner

Sat Dec 03 2011, 06:34PM

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

OK, well if you're using this kind of on/off control scheme, you ought to find that the resonant converter loses much of its advantage over the SLR at loads below "flat out". When the converter is off, energy regenerates from the tank into the DC bus through the bridge diodes, in the same way that it does with the SLR. This energy uselessly transferred back and forth is what you called "reactive power".

I use that resonant drive with on/off control myself, as a current limiter for DRSSTCs, and have experimented with it a lot. Never tried making a power supply that way, though. Might be something in it!

I think the ripple could be reduced using noise shaping techniques.

How are you defining power factor for these two systems? If building a DC power supply, a reasonable definition might be something like (Iav*Vav)/(Irms*Vrms) - useful power to the load divided by the "VA rating" required of the high frequency transformer.

dude_500

Sat Dec 03 2011, 06:56PM

Registered Member #2288 Joined: Wed Aug 12 2009, 10:42PM
Location: Cambridge, MA
Posts: 179

Oh, I had not considered the implications of what happens with it goes to an off state. Indeed, you are right that power factor goes down unless it's in the always-on state. Attached a simulation of that at 100% resonance. Indeed, reactive current goes back into the source when it turns off.

The power factor measurements I mentioned in my last post were done at always-on, no pulse regulation.

So it looks like the SLR has a somewhat bad power factor but would be constant over all duty cycles and switching pulse modes, whereas the resonant mode has perfect power factor at always on, but it ranges to really bad over some pulse schemes.

I have been calculating power factor experimentally by having my scope give me <V_trans*I_trans>/<abs(V_trans*I_trans)> where the voltage is measured differentially across the transformer (not including the tank cap, just the primary), and <> is mean. The averaging is done over a small integer number of cycles. Unless I'm mistaken, this is the equivalent to (Iav*Vav)/(Irms*Vrms), correct?

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