Diy 10kw charger

[Arlo1]

Luke this one is for you I found this on the diy electric car forums its a open sorce project and yes it is limited by your wall outlet so make sure you have a 60 amp braker on your 240 volt socket http://www.emotorwerks.com/cgi-bin/VMcharger.pl Its about $600 to build one.

 

[Doctorbass]

great find! but it seem to not have any PFC ?

Doc

 

[nieles]

there is an add-on for pfc

http://www.emotorwerks.com/emw3/product/pfc-pcb-add-on-for-emw-10kw-charger/

 

[Njay]

The charger doesn't seems "mains isolated"... is it a problem?

 

[heathyoung]

Wow - thats pretty much what I had in mind, its an unisolated buck converter. Ground the negative.

Design is so-so. Frequency is FAR too low, 10-16Khz - really? Just how big do your want your caps and magnetics to be? You could build it half that size if you ran ~80Khz, using slow crappy IGBT's and underpowered drivers = fail.

So tempting to redesign this. Parts selection is poor. Fashion over function. Ditch the microcontroller - its a wank. Design your inductor PROPERLY for the output voltage you are looking for. Universal = difficult to filter + inefficient. Also difficult to drive, hence the overkill IGBT.

Somebody shot for the stars before trying for the clouds.

 

[liveforphysics]

Yes, I agree. They didn't know jack about power electronics design, and the bling is really not needed.

However! They made a scratch-built functional 10kw charger for EV's, and that's something I haven't done yet, so I say, "Great job guys! You made something to fill a hole on the market, and did it cheap and released the plans openly!"

 

[heathyoung]

True - you are too nice as an engineer

 

[wiredsim]

Well healthyoung, that is how these projects improve.. How far along do you think Linux would be if people didn't get annoyed with how amateur they saw the initial attempts!

Do you think it can be improved piecemeal? If so what would be the first thing you do to improve it?

 

[heathyoung]

Inductor is far too small, it would have limited energy storage. Increase the inductance, and increase the operating frequency. I'm designing a similar charger, (but lower power) and the values I've calculated are as such:

For 80Khz, and 340VDC input, 150V DC output, 10A with 0.2V ripple, you need a 300uH inductor capable of peak currents of 15A, 1000uF of input capacitance, with a ripple current rating of 7.5A, and 2,200uF of output capacitance with an ESR no greater than 50mOhm. The biggest problem at higher currents is not the IGBT, but rather the saturation current of the inductor, and the abilities of the free-wheeling diode - it produces the most heat. Due to the SOAR curves of IGBTs you can snot them fairly hard without failure. Its the diode Vf that produces the most heat.

As you change your voltages, you need to change your inductance as well - a one-size-fits-all approach does not work. Its lazy and poor design. I'll post up the spreadsheet if anyone is interested, the calculations are in the formulas, and if you change values (ie. input/output voltages, frequency) you will see what effect these values have.

 

[Arlo1]

Its open source guys build it better!

 

[jdb]

Wow - thats pretty much what I had in mind, its an unisolated buck converter. Ground the negative.

Actually, the midpoint of the DC link is set to be earth voltage, either through the neutral in 120V input, or the center of the two phases in 240V input. Grounding the negative terminal of the charger's output will short the negative rail caps with no fuse in the path. Both output and both input terminals should be fused.

Pardon my ignorance, but I don't understand what parts of their design are "fashionable" or "blingy". What are you guys referring to? Aside from the fusing violation, I really don't see how it's all that bad. In light of what the auto industry is doing with the SAE J1772 connector system, I think it's awesome!

 

[liveforphysics]

The fancy display pannel is what's blingy, but it at least has a function in the event your EV doesn't insturment its own pack data for some reason.

The lack of fuses doesn't bother me any.

J1772 is a trick to enable billing when charging EVs. It was supposed to be DC straight into the pack like ChaDeMo (which would make it awesome), but instead its just an over priced connector that waits to make sure you're OK ( paid or whatever) before it closes the contactors to your 240vac outlet.

I think this project is great. They just could have made it about 1/3rd the size if they didn't build it like a 1970s radio.

 

[heathyoung]

Wow - thats pretty much what I had in mind, its an unisolated buck converter. Ground the negative.

Actually, the midpoint of the DC link is set to be earth voltage, either through the neutral in 120V input, or the center of the two phases in 240V input. Grounding the negative terminal of the charger's output will short the negative rail caps with no fuse in the path. Both output and both input terminals should be fused.

Pardon my ignorance, but I don't understand what parts of their design are "fashionable" or "blingy". What are you guys referring to? Aside from the fusing violation, I really don't see how it's all that bad. In light of what the auto industry is doing with the SAE J1772 connector system, I think it's awesome!

Yes, brain snap. 'Ground' would be floating at half rectified mains voltage. Not real, 0V ground. Ahem. Hence the reason you DONT use this design if any of your negative battery rails go to the chassis of the vehicle - you would get a rather painful belt off it. Ouch! Some DCC converters do this, expecially if they are unisolated!

Poorly chosen components are the major issue I have with this - grabbing things off the shelf and tacking them together... Microprocessors when there are FAR better things to use - PWM off a micro? Really? Why not use something that was actually designed for the application. Pulse by pulse current limiting is better done in hardware, even using interrupts is poor form. You could even do it with the IGBT driver, IF you specified an adequate component.

Its cute, but needs a serious rethink about the design. Nothing wrong with a pot and a multimeter to set the output. Get a properly designed inductor, and an IGBT thats designed for speed, not for turning kiln elements on and off every 2 minutes. You could make it much smaller, as LFP said - by using modern construction techniques and properly calculated component values. And more modern components.

For example, UCC29910A - its a buck PFC controller. Ideal for this application.

<EDIT> I'm slowly trawling through the ecomodder forum and diyelectric car forums, there is some interesting development, biggest problem these guys seem to have is one is using a 555 timer (god help us) for the PWM source, the other is learning how to program. Neither of them are using proper drivers for their IGBT's. Fortunatly one is doing proper calculations!