DIY Open Source EV Charger

[jyanof]

Originally Posted by MazdaMatt

Mike, please fill me in on what way a lower frequency increases the size of components. Which components?

Not mike, but this is the problem I'm running into... somewhat. I don't really know if it's a problem.

I sized the inductors so that there would be constant current (filtered output) in the output loop at a min current of 2A. However, that was at 100kHz. I lowered the switching frequency to 20kHz to reduce switching losses due to the initial heat problems. Now, at 20kHz, I don't have constant current in the output loop until I get up to about 5A. If I wanted constant current at a lower amperage, I'd need larger inductors.

Additionally, lower frequency requires more ripple current to be supplied by the input caps. More ripple current means more capacitors or larger capacitors or more expensive capacitors.

Anyway, I don't know yet if the un-filtered output at low current is a problem. I foresee it being an issue if I can't get a good current measurement below 5A. Hopefully I can filter the current reading from the current sensor to get an average current. (my hall effect DC clamp meter does this, so I should be able to as well?)

[jyanof]

Originally Posted by MazdaMatt

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.

Matt and arnolde, I'm just being picky here, but it's not the 150uH inductors that cause the ringing. The current in those inductors do not change from full current to zero current. The 'flyback' diode allows the current to continue through the output loop, thus the current in those inductors do not change (much).

I guess there is a finite time in between the mosfet turning off and the diode becoming biased which I think can cause some voltage spikes. There's also a finite time for the diode to turn off, but that's another story.

On the other hand, the input current IS getting switched on and off from zero to full current. This is the main source of Drain/Source ringing - you don't want ANY inductance in the input loop (and since zero inductance is impossible, you'll at least want to minimize it). Minimize parasitic inductance by minimizing trace/wire lengths and keeping the caps/diode/mosfet close together.

[jyanof]

BTW - i'm slowly learning this micro controller programming stuff. So far, I've gotten an LED on the STK500 to dim and brighten using PWM!

I'm curious how fine of a PWM resolution i'll need. For example, if I increment the duty cycle by 1, will the current jump by .3 amps or 30 amps? The knob on the 555 timer was very sensitive once there was constant current in the output loop. If high resolution isn't needed, then the PWM can be run at a higher frequency.

So, i think my next test will somehow rig a couple knobs to change duty cycle. maybe one will have a macro effect to get the charger going. The 2nd knob would increment the duty cycle by only 1 or 2. I could test at a high resolution and see what happens and possibly extract that to lower resolutions. i haven't really worked out the details yet...

[rsandberg]

I happened upon this article today and thought about the ripple you are experiencing (and maybe even the low current settings).

Multiphase buck converter. 3/4 of the way down the page.

Buck converter - Wikipedia, the free encyclopedia

Hope this helps,
Rick

[arnolde]

jyanof: You might want to try reducing the 10k pullup resistor on pin 7 of the NE555 to 2.2k or 1k, that will increase the current that acts on the MOSFET gate and shorten the switchon phase. That should reduce some switching loss, then maybe you can increase the frequency again somewhat.

The lower your freq is, the more your caps and coil get stressed. (=you really need bigger ones at 20khz)

[NiHaoMike]

Here's an observation I made while working on an Antec Truepower Trio550: while tracing the input circuit, I noticed that the preregulator did not use electrolytic capacitors on the input. (I got that power supply for free because the fan doesn't work. I found that the fan does work when connected directly to the output, so I just did that. Now I'm checking to see if it can be modified to have a built in UPS.) Instead, it used a 1uF, 400v polyester film capacitor. A quick search revealed that such capacitors cost only $1 each in small quantities. Since the supply is rated for 550W and an EV charger that operates on a standard outlet would be 1875W max, only 4 of those capacitors would be needed. Might as well make that 5 or even 10 for more margin. There'll be a lot of ripple, but that doesn't matter for charging batteries. (The PC power supply used a large electrolytic on the output to filter it.) It might even be possible to eliminate the output capacitors since buck converters don't ripple too badly on the output and the pulses might actually keep the batteries desulfated. Just a RC filter for the voltage sense to keep the pulses from interfering with the voltage sense circuits.

[jyanof]

Originally Posted by arnolde

jyanof: You might want to try reducing the 10k pullup resistor on pin 7 of the NE555 to 2.2k or 1k, that will increase the current that acts on the MOSFET gate and shorten the switchon phase. That should reduce some switching loss, then maybe you can increase the frequency again somewhat.

The lower your freq is, the more your caps and coil get stressed. (=you really need bigger ones at 20khz)

arnolde, I was using the NE555 to provide a signal for the mosfet driver. The driver and gate resistor took care of the rise and fall times of the mosfet.

i agree that the components are undersized at the lower frequency. If i can, I'm gonna shoot for higher frequency operation.

[jyanof]

Originally Posted by NiHaoMike

...Instead, it used a 1uF, 400v polyester film capacitor. ....

I've heard of mixing different types of caps to meet all the demands of the system. Like, using film caps with low ESR for providing most of the ripple current and large electrolytics for smoothing the rectifier output. I think it's definitely an option I'm going to look into and it's less worrisome to know they're so cheap!

[arnolde]

Originally Posted by jyanof

I was using the NE555 to provide a signal for the mosfet driver.

Oh... you never mentioned that can you tell us what MOS driver you're using and what other components are used? As you can see from my circuit posting, I just stuck your posted circuit together with the NE555 example you pointed me to, I never knew there was more in between...

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)

[jyanof]

IXDD414PI

I'm sure it's overkill, but it works!