DIY Open Source EV Charger

[jyanof]

These are the inductors I'm using:
E SERIES HELICAL WINDING OUTPUT INDUCTORS : CWS Coil Winding Specialist, manufacturer of transformers, inductors, coils and chokes

I had actually looked at the Vishay inductors you have, but at the time, all the suppliers I looked at were sold out of them. These guys were the only other supplier I could find that had something similar at a reasonable price.

I do notice that mine get pretty hot when conducting 20A. Of course, they are rated for operation up to 200C, which is insanely hot!

[arnolde]

Remember, anything that gets hot, is actually wasting energy and reducing the efficiency. I'd try raising the pwm frequency if I were you, 20 khz is really pretty low, there's a good reason why smps's are usually between 50 and 200 khz. Maybe the switching losses will get higher then, but I assume the other losses (in coils and capacitors) will get lower and result in overall better performance.

I just kicked myself for ordering those expensive capacitors, thank god I only got 2. Later I found a surplus store that sells slightly smaller ones for 2$ each. I think I should have wound the coils myself, too, I'd just need to know how thick the wire is and how many turns, and what material to use for the core (if there is one). Well, next time I will.

Actually I think that with enough searching for low-cost parts, this thing can theoretically be built for under 100 EUR in parts (including µC, display and all). I wonder if there's any chance of getting it EMI certified...

Update: So, I finally got my final order sent out. Altogether about 150 EUR worth of parts, from 3 suppliers. I hope I didnt forget anything important :-) I decided to leave out the little 15V power supply for now, since I have lots of dc-regulated plug-casing-type PSU's around. I also looked at making a printed PCB but at over 40 EUR each, I'll save that for the next revision.

I'm still confident I can get everything into a PC power supply casing. From surplus stores I can get one with a temp-regulated fan... and just throw out the contents, and use the case and fan :-)

2.EDIT: Farnell sucks!!! They had the inductor listed as "min order 1 unit, in stock, delivery about 3 days, overseas processing fee 20 EUR". Now they called me and told me the min order is 10 units and it would take 8-9 weeks!!! Bah. I guess I'll have to wind my own, there's no way I'm gonna wait 8 weeks!

[arnolde]

Problem solved: I looked up how to calculate a coil, and found that I need only 100-200 turns of wire (for 150-300µH) on a 30mm diameter, 60mm long non-conducting (plastic) core. If I use a ferrite core I supposedly only need 50 turns (4x more inductivity). So I just bought 100m of 1.5mm copper wire (gives me 1.75mm² which is good for over 20A current) and will wind my own. For the 20 EUR incl. shipping the wire costs me, I can make 5 coils, whereas I would have only gotten one for that price by ordering it.

[arnolde]

I was just working through this paper http://powerelectronics.com/mag/606PET25.pdf and writing an Excel sheet for all the calculations. It seems that the lower the gate current is, the higher the switching power loss of the MOSFET? (See Page 52, Eq.12) In my case, if I did everything right, the loss is 60W with 100mA gate current and only 6W with 1A gate current! Now the NE555 probably cant serve with 1A can it? Maybe thats why most people use special MOSFET driver chips, which take a TTL input and whomps the MOSFET up and down at 1A...

Another interesting thing I noticed, is that my output caps were way overdimensioned. Since my voltage difference Vin-Vout is only max 100V and the current is <6A, and I can allow several volts of ripple, the formulas only call for 17µF! That will make a nice snug design :-)

[MazdaMatt]

Arnolde - mosfets don't ACTUALLY act as a switch, they are more of a dimmer. WHen they are fully on (Rdson) or fully off (R = infinity), there is not much heat generated, but when they are switching the resistance swings from infinity to Rdson and in between in burns a lot of power. The "level" of the dimmer is proportional to how charged the gate is. Think of the gate like a capacitor. If you half-fill the capacitor, it has a semi-high resistance that results in major heat generation. So the idea is to fill up and dump out that capacitor as quickly as you can...

HOWEVER - if you fill it and dump it TOO fast, then you generate a very large current change in a small period of time (aka a large dI/dt). The voltage across those coils = L*dI/dt where L is the inductor rating. So if you switch a 1A charge off in 1uS, you are generating 1V for each uH of inductance. You have what, 150uH? So if you switch off 5A in 1us you generate 750 volts. There is also a "ringing" that occurs in fast switching.

So you have a trade-off, but you sure as hell aren't switching fast enough out of a 555. And those gate drivers... more like 15A of gate current for a short period of time.

Find the gate capacitance of your mosfets and figure out how much gate current is required to fill it up in 1us. Once you have this, then you can check for ringing and voltage spikes and back it off... or shoot for 2us for safety...

[arnolde]

Thanks for the nice explaination!
The MOSFET I'm using (STP10NK60ZFP) has a CRSS (Gate capacity?) of 37pF.
I'm now looking for the right formula to calculate the charging time/current...

[MazdaMatt]

Arnolde, was there a rating in "C" rather than "pF" for the gate? Could you link a datasheet plz?

[MPaulHolmes]

Digi-Key - 497-5892-5-ND (STMicroelectronics - STP10NK60ZFP)


http://www.st.com/stonline/books/pdf/docs/8526.pdf

[MazdaMatt]

Paul, you're a machine. The gate charge is 50-70nC (a coulomb is a direct measure of the number of electrons... with some huge scalar like 8.3x10^8 or something... maybe that's the speed of light... i dunno... been a while) Anyway, 1C/sec is 1A.

So, 0.050uC in 1uS = 0.05A... Does this seem wrong? Paul, are these mosfets little wussy things compared to your controller?

Oh yeah, it is 0.05A PER mosfet, so if you have 10, then you need 500mA... which strikes me as being really small.

[MPaulHolmes]

yes, wusses. They are rated for 10 amps (compared to 130 amps in the controller)

The gate charge is about 5 times less than the motor controller mosfets' gate charge. But they are only 200v, and the charger ones are 650v