[Stiive]

Haha, didn't think of using a laptop charger.
However, considering these don't give really give me the voltages I want, what did you mean by this comment?

Originally Posted by PStechPaul

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.

[PStechPaul]

Originally Posted by Stiive

Haha, didn't think of using a laptop charger.
However, considering these don't give really give me the voltages I want, what did you mean by this comment?

I'm assuming you have multiple 12V batteries wired in series so you can tap off of the intermediate connections for multiples of 12V. And 144V is at the low end of rating for most AC-DC switchers, corresponding to about 100 VAC.

I thought you needed 15V at 1A for the gate drives, and 3.3V at 500mA for the processor. I see now that you need 3.6V. But that should be standard as well: http://www.mpja.com/36V-22A-Desktop-...info/18357+PS/, or you can use a 5V supply and a linear regulator. You could use a wide input 12V nominal DC-DC converter for both, and the 15V high side gate drive power supply should be connected to the top 12V battery to minimize the common mode voltage. And the 3.6V can be a 2W DC-DC on the bottom 12V battery.

For the same output power, a DC-DC converter or an AC switching supply will draw less current at higher voltage. So the 15W gate drive supply will draw about 1.5 amp on 12V but less than 120mA on 144V. And the 3.6V supply needs only about 2W, or about 180mA from a 12V battery, and only about 15mA from 144VDC. At such low currents, you don't really need to worry about balancing the loads on the battery pack.

The current sensor supply may not need to be isolated or floating if it is a toroidal CT. There are all sorts of power supplies that can provide the 12V needed: http://www.mpja.com/12V2A-Hitron-Pow...info/18595+PS/

[stickytechnology]

Originally Posted by PStechPaul

I'm assuming you have multiple 12V batteries wired in series so you can tap off of the intermediate connections for multiples of 12V. And 144V is at the low end of rating for most AC-DC switchers, corresponding to about 100 VAC.

This probably isn't a good idea, unless you charge individually as well...
How are you planning on charging your 12V aux battery, Stiive? This is the big reason I kept my pack voltage under 360V (96 LiFePO4), as you can use off-line converters for the 12V supply. The internal topology of a wall wart is generally a full bridge followed by a flyback, so they do work on DC, as PStechPaul mentioned.

[Stiive]

Originally Posted by PStechPaul

I'm assuming you have multiple 12V batteries wired in series so you can tap off of the intermediate connections for multiples of 12V. And 144V is at the low end of rating for most AC-DC switchers, corresponding to about 100 VAC.

Yeh, that doesn't sound ideal, especially because I'll prob be testing with LiPo's and i don't want to cycle a few more than the others.

[Stiive]

Originally Posted by stickytechnology

How are you planning on charging your 12V aux battery, Stiive?

This project isn't actually for a EV atm, but if i'm going to make the hardware open-source might as well make it versatile

Originally Posted by stickytechnology

The internal topology of a wall wart is generally a full bridge followed by a flyback, so they do work on DC, as PStechPaul mentioned.

Guess I assumed they would just use the known frequency AC to flux the transformer, but that means it would only work in 1 country.



I have some 650V 75A trench IGBTs, perhaps ill just make my own DC-DC converter. If i convert to say 15V, then I can use some isolated DC-DC converters (e.g. 15V to 3.3V) to get my other voltages.
Not the cheapest way, but should be relatively easy

[stickytechnology]

Originally Posted by Stiive

This project isn't actually for a EV atm, but if i'm going to make the hardware open-source might as well make it versatile

What's the project?

Originally Posted by Stiive

I have some 650V 75A trench IGBTs, perhaps ill just make my own DC-DC converter. If i convert to say 15V, then I can use some isolated DC-DC converters (e.g. 15V to 3.3V) to get my other voltages.
Not the cheapest way, but should be relatively easy

I'd probably go higher in voltage than that, and not so beefy in terms of current. Something like this:
http://www.digikey.com/product-detai...53-5-ND/725341

Remember that if you use the flyback topology (simplest, as you only need one switch) the switch rating has to be at least: V_in + (V_out * turns ratio) + √(L_leakage I_in^2 /C_switchnode) so for a 400V input, a 20 turns ratio, and a 15V output, the first two terms are already 700V. The third term is impossible to calculate, as it depends on the leakage inductance and the parasitic capacitance of the switch node, which you won't know until you build the circuit, but probably count on another couple of hundred volts.

As an example, the offline supplies we were talking about earlier (wall warts) usually have 700 or 900V switches, for a 360V input rail.

[PStechPaul]

I'm dealing with the inductive spike issue in my DC-DC converter design. I'm using a 2 kHz square wave with 1 uSec dead time, and at 20 volts I was seeing the 40 volts as expected but also a spike up to about 80 volts, and even higher as I raised the supply voltage. I was able to reduce it somewhat by adding snubbers, but it was still not enough to guarantee that I'd stay within the 100V Vds specification of my HUF75645 MOSFETs. I had one fail as I was doing some of the testing, and it seemed to be especially bad after I added fast 9A UCC27321 gate drivers.

In my simulation, I found that I could use 100 ohm gate resistors and add 0.1uF capacitors from gate to GND, and it reduced the initial startup spikes as well as the steady-state. And the efficiency was still pretty good. The MOSFETs were only dissipated about 5 watts each, with 350 watts output. I have not yet implemented this in the actual circuit, but I will.

[Stiive]

Originally Posted by stickytechnology

What's the project?

Just to get the motor spinning at the moment

Originally Posted by stickytechnology

I'd probably go higher in voltage than that, and not so beefy in terms of current. Something like this:
http://www.digikey.com/product-detai...53-5-ND/725341

Remember that if you use the flyback topology (simplest, as you only need one switch) the switch rating has to be at least: V_in + (V_out * turns ratio) + √(L_leakage I_in^2 /C_switchnode) so for a 400V input, a 20 turns ratio, and a 15V output, the first two terms are already 700V. The third term is impossible to calculate, as it depends on the leakage inductance and the parasitic capacitance of the switch node, which you won't know until you build the circuit, but probably count on another couple of hundred volts.

As an example, the offline supplies we were talking about earlier (wall warts) usually have 700 or 900V switches, for a 360V input rail.

Thanks for the formula, that one is new to me. However, I was thinking of using a simple forward buck converter topology to step down to 15V as I don't need isolation or therefore a transformer. From there I can use isolated DC-DC to get the other voltages.
I think in this case a diode should keep the voltage seen by the switch in check - do you have any calcs for this topology?

Yeh the current rating of the IGBT is a bit high, but I have a few of them lying around... but then again, how do i provide the 15V required for the gate driver without the output of the buck converter? Might have to source something else that can run off the 12V battery.

Guess I should look for the ideal solution, not the "found lying around my room" solution :P

[Stiive]

Thanks for your info Paul

Originally Posted by PStechPaul

I'm dealing with the inductive spike issue in my DC-DC converter design. I'm using a 2 kHz square wave with 1 uSec dead time, and at 20 volts I was seeing the 40 volts as expected but also a spike up to about 80 volts, and even higher as I raised the supply voltage.

Am I right in understand you are boosting voltage 2:1? What will this boost converter be used for?

Originally Posted by PStechPaul

In my simulation, I found that I could use 100 ohm gate resistors and add 0.1uF capacitors from gate to GND, and it reduced the initial startup spikes as well as the steady-state. And the efficiency was still pretty good. The MOSFETs were only dissipated about 5 watts each, with 350 watts output. I have not yet implemented this in the actual circuit, but I will.

Out of curiosity, what electronics simulator are you using?

[PStechPaul]

I am boosting 24VDC from 2 batteries to about 300VDC for a VFD so I can power a 240 VAC 3 phase motor. Previously I had the output connected in doubler mode, so I got 300 VDC from just one 12V battery. You can see it in action where I take a ride on my electric tractor, using just one 17 A-H battery, and drawing only about 250 watts.
http://youtu.be/y0qWY4bVnEA

I'm using LTSpice, which is totally free and very powerful. If you use Linear Technology components, they have just about every model as well as basic circuits. Here's the schematic, from a screen capture:



And a screen shot of the simulation showing some of the waveforms:



If you want to use this tool, just download it from:
http://www.linear.com/designtools/software/

You can also load the ASC file for this DC-DC converter:
http://www.enginuitysystems.com/pix/...CT_Doubler.asc

It's a simple text file.