[stickytechnology]

Originally Posted by PStechPaul

It's just a simple push-pull square wave driving a transformer with a center tap. I have just 8 turns of #10 AWG on each primary, and the secondary is 100 turns of #18. The core is a 500 VA Powerstat (toroid). At the original design frequency of 60 Hz it was probably 0.25 volts/turn. But at 500 Hz I figured 2.5 volts/turn (RMS). Probably about the same for a square wave. So 8 turns should be about 20 volts peak and at 2 kHz up to 80 volts. So it should be fine for 2-4 batteries in series, or up to 56 volts. But my MOSFETs are only rated to 100V so I need to limit this version to three batteries, or 42 volts. You get twice the applied voltage during the off cycle.

The secondary just goes to a full wave bridge, with two large capacitors in series, so by connecting their common point to one side of the bridge, it becomes a voltage doubler, with about 320V output. I'll add a switch so it will still give 320V with two batteries. The main difficulty is transients during the time both drivers are off. I have TVS diodes across the MOSFETs and I may add a snubber across the transformer.

I'm generating the square wave with a PIC16F684, and the dead time is about 1 uSec. I have a simulation in LTSpice. I can post the file if interested, but I probably have to copy it as text into the post. Any way to post a file?

This seems like a good application for a resonant forward converter, eg Royer oscillator:
http://en.wikipedia.org/wiki/Royer_oscillator
http://www.neon-john.net/Induction/Roy/Roy.htm
like you might find in a CF light power supply. There are big efficiency gains in resonance, because you can switch at zero voltage and/or current.

[Otmar]

Originally Posted by MPaulHolmes

The line filters are just off the shelf components that are about $0.80 each.

I like the idea of using line filters as cheap transformers! I also use one in a simple buck converter in the Zilla for the main power supply. The advantage is that it runs from 25V to 450V input and puts out a coarse 15V at 500mA. It is not isolated but I don't mind since I like keeping the main processor referenced to B- and isolating other things as needed including the throttle connection.

Details on the buck converter can be found in the old design idea here, I don't think it is on the power integrations site anymore. Of course power can be increased with a larger switch and inductor.
http://www.nalanda.nitc.ac.in/indust...werInt/di1.pdf

[PStechPaul]

Originally Posted by stickytechnology

This seems like a good application for a resonant forward converter, eg Royer oscillator:
http://en.wikipedia.org/wiki/Royer_oscillator
http://www.neon-john.net/Induction/Roy/Roy.htm
like you might find in a CF light power supply. There are big efficiency gains in resonance, because you can switch at zero voltage and/or current.

In the Wiki there was reference to a comprehensive article by Jim Williams: http://cds.linear.com/docs/Application%20Note/an65f.pdf

It seems the Royer oscillator is designed for DC to AC conversion with a sinusoidal output, but this is not what you want for a DC-DC converter. It also requires a specially built saturable core transformer.

The square wave push-pull converter is ideal for DC-DC conversion because the output from the FWB is already almost pure DC, with just some gaps during the switching dead-time. This is only one or two microseconds, so the energy storage capacitors can be very small. The efficiency is determined by the on resistance of the MOSFETs, the core losses of the transformer, and the copper winding losses. All of these can be made almost arbitrarily low, at the expense of size, weight, and cost. So a 90-95% target efficiency seems reasonable enough and the overall efficiency of the system will be mostly determined by the batteries and the motor.

I did not see any efficiency ratings for the Royer oscillator "Roy", and, again, it produces AC, probably at high frequency for induction heating purposes. But I still need to test my converter at higher power levels. The original design only provided about 250 watts, which was pretty much the safe limit for a single 17 A-H SLA battery.

[Stiive]

Well so given everyone is using a different type of DC-DC, lets deduce which will be best for my project :P

These are the different DC power requirements.

Gate Drivers: +15V +-4% with minimum peak 1A*3gates (=3A). Rise time<50ms. No isolation required
(http://www.semikron.com/products/dat...L_L6100204.pdf)

MCU: +3.6V 150mA. Isolation preferred - high accuracy/noise immunity required for ADC ref
(http://www.st.com/internet/com/TECHN...DM00037051.pdf)

Current sensors: 12-15V +-5% +-15mA. High noise immunity i guess
(http://www.lem.com/docs/products/htb...v12%20e%20.pdf)


To limit noise/EMI/cost in the inverter, I'm thinking boosting the 12V battery voltage over bucking the DC bus - but am happy to explore both

[PStechPaul]

If you have a 360V battery pack, you can use an ordinary 240 VAC switching supply which has a 350 VDC internal link. Even if your battery pack is higher, you can just tap off enough of them to get the DC voltage needed. 144 VDC would be fine. A 15V 1A switcher is maybe $20. And same for the 3.3VDC supply.

Best to use standard off-the-shelf components wherever possible.

[Stiive]

Got some part numbers?
Concerned with some of the ones I have researched previously not having a wide enough HV spread.

Bus voltage will most likely be ~ 300-420V, but was hoping to have a versatile system that will allow me to test at lower voltages (say <100V while I iron out the bugs)

[PStechPaul]

Here is a 3.3V supply that's good for 90-277 VAC, which is equivalent to 127 to 391 VDC. 300 mA (1W), $16:
http://www.digikey.com/product-detai...0-5-ND/2652153

And 15V at 267mA (4W), 90-264 VAC (127 to 373 VDC), for $24
http://www.digikey.com/product-detai...068-ND/2256248

Others:
http://www.newark.com/tracopower/tml...dc/dp/51R5391?
http://www.newark.com/tracopower/tml...vdc/dp/51R5387

For 36-75 VDC input:
http://www.newark.com/tracopower/tes...-3v/dp/51R5134
http://www.newark.com/tracopower/tes...15v/dp/51R5137

[Stiive]

Thanks for that! No wonder I didn't find these earlier, I assumed these AC converters wouldnt work with DC input, but I guess cus these are variable freq AC they can though. None of them seem to show a schematic of the internals... Do they normally rectify the ac then flyback/forward/push-pull?

Very low efficiency, but that doesnt bother me too much.

[Stiive]

Cant seem to find one of these switching supplies with 15V output at the current I need

[PStechPaul]

You should be able to find an open frame switcher with 15 volts at 1 amp, and they make much larger ones as well. But for gate drive, you really only need like 1 amp peak, to charge the IGBT gate capacitance, so just hang a decent size low ESR cap on the output and you're good to go.

All switching supplies AFAIK have a bridge rectifier at the input and a large filter capacitor rated at about 400 VDC. You can just connect the DC to the AC input as long as the diodes are conservatively rated. If you have an open frame supply you can just bypass the rectifier, although a diode might be a good idea to keep the supply capacitor from draining when your main battery pack sags under heavy acceleration.

You could even use one of these with 1.6A for only $6:
http://www.mpja.com/15VDC-16A-Regula...info/18724+PS/
or if you want 2.8 amps for about $25:
http://www.mpja.com/15V-28A-40W-Powe...info/16002+PS/