DIY Open Source EV Charger

[arnolde]

Lol, 14A peak current, I bet its overkill ^^

I'll think I'll try the IR2101S.

[NiHaoMike]

Originally Posted by arnolde

Beware of cheap film caps, they tend to "fail short" (if they fail) and explode. Electrolytics apparently dont have that behaviour. (At least according to the powerelectronics.com article)

I have seen a lot of electrolytics fail (mainly from excessive ripple current). Most of them just pop out the "pop stars" on the top, but a few were blown right off the board. I have never seen a polyester film capacitor fail in such a dramatic manner.

[arnolde]

Well, I dont know, I was just repeating the advice in the article.

I picked up my scope today (which I had lent to my Dad), tomorrow the NE555 (which I stupidly forgot in my thursday order) and driver chips will be delivered, I cant wait to start assembly :-D

EDIT: Now this is cool... just found an optocoupler with MOSFET driver output :-) http://docs-europe.electrocomponents...6b807796fc.pdf

[jyanof]

hey arnolde (or anyone?)

I'm thinking a little far ahead, but how do you plan to isolate the power section from the control section while still measuring battery voltage?

I've been going back and forth about using an optoisolator between control board and mosfet driver input and finally went through the mental exercise of adding it in. Now, the switching section is isolated, but I still need to measure battery voltage.

I had planned on using a resistor divider to bring the voltage down and then route it to an analog input on the micro (likely with an opamp to amplify the range to get higher resolution). In order to do this, I'd need to have a common ground between the battery (-) and the control board is tied to the output loop and is no longer isolated.

Maybe that's ok? Thinking about it now, I think optoisolators are still necessary since the power section ground ( neg from the rectifier) is at a different voltage than Batt(-) due to the voltage across the inductor.

i'll just have to try it! Unless there's an isolated way to measure voltage?

[arnolde]

In fact, there is: With a voltage-to-frequency converter. Those chips simply convert a voltage (like 0...10V) to a square wave TTL signal (like 0...10khz) and you put that through an optocoupler and hey presto, there's your isolated measurement. Same thing with isolating the current from a shunt. Measure the shunt voltage (0...100mV) and that way you can isolate DC current. I'm planning on making a small module with a shunt, a voltage divider, 2 v2f converters and an optocoupler to include in the battery package. That way I can add a µC-controller voltage/current dispplay in the cockpit without worrying about long wires or isolation.

[jyanof]

very interesting! would you filter the output with an RC circuit to get a more constant voltage?

do you happen to have a part number in mind?

[arnolde]

Filter? Of course not, you want a clean square wave signal, you dont measure the voltage (it's not PWM) but you measure the frequency, and calculate the voltage from that. It's much more precise, and easy to do with a µC.

Personally I got some MCP3909 because they go from -1V to +1V, but then I found out they don't put out the signal in such a nice way as I want. I dont know any others to recommend, just call up any decent electronics website (newark?) and search for "volt 2 freq" and there will be about 10 types to choose from, and read the datasheets...

The MCP3909 seemed ideal because they each have 2 A/D inputs, already laballed "current" and "voltage" but it seems the 2 normal freq.outputs are not seperate outputs of these seperate values, but in fact the chip calculates "power" from the inputs and puts that out instead. On the SPI output however, it will output both in "clear". But there is more decoding necessary on that one, and it's meant for higher frequencies (1mhz) and Im not sure if I can go slow enough. Anyway I'm a little confused by the datasheet and so I'll just have to try them out.

Another approach, of course, would be to just stick a µC with A/D ports onto the measurement circuit (instead of the v2f) and output direct values in clear text via opto-isolated RS232 serial, and that can be sent to a) the charger logic and b) a cockpit instrument. Actually.... hmmm... that sounds even better. I think I may well just do that. I just thought of that myself :-) only drawback is, that part then isnt isolated from the pack so you have to disconnect it when messing around with it (updating firmware, etc.)

[MPaulHolmes]

great idea! Hey, why does the pack have to be isolated from the charger? There is no motor and no switching 500amps or whatever.

[arnolde]

Say Jyanof, how did you get those nice scope readings? When I hooked up my scope to the input diodes, it blew my stray-current-protector (I'm not sure what the english word for that is) in my fusebox (every German household has one), because the scope ground was basically connected to a power line. So I unclipped that and just left the scope ungrounded and only used the center pin, but those readings were ok but not brilliant. Did you find a different way?

[jyanof]

Originally Posted by MPaulHolmes

great idea! Hey, why does the pack have to be isolated from the charger? There is no motor and no switching 500amps or whatever.

i think isolating the power section of the charger from the control section will reduce noise in the control section. it may not be 500A, but 30A is still significant. I didn't isolate it when i ran it with just the 555 timer supplying the pwm, but I wasn't doing any ADC conversions either.

another issue is that the battery (-) is not always at the same voltage as the rail (-) from the output of the rectifier diode. or, maybe it's never at that voltage... anyway, I think isolating the voltage measurement may make things easier.