DIY Open Source EV Charger

[jyanof]

Originally Posted by MPaulHolmes

Your inductors are working nicely? I can't wait to see your parts list! I want to make one too!!!! Awesome job!

Thanks for all the kind words! I'm pretty excited about getting to this stage. I thought the electrical stuff would really be tough, but it's been stuff like layout and heat dissipation that have caused problems (and I'm a mechanical engineer by day, sheesh!)

I will definitely play around with caps to try and reduce their temperature. Admittedly, I had a medium box fan blowing on the whole thing; they won't have such nice airflow in a real application!

The inductors are working well - this is the only issue (i think): You can tell when the current is high enough to keep the inductor charged during an entire pulse cycle. This occurs around half duty cycle and about 5 amps. At this point, there is always current flowing through the output loop and any slight change in duty cycle results in a very large change in amps. Like, I'll put my fingers on the knob and it'll jump up by an amp. I don't think the sensitivity will be a problem as long as there is enough resolution in the PWM duty cycle (might require adding a crystal like you did? I forget why you added it, but I think one of the results was higher PWM resolution).

Rather, I'm wondering about controlling the current when it is at lower power settings (like, a 2A gassing stage or something). The current will not be constant in the output loop and the sensor would require some sort of filtering in order to read an average current (i don't think we care about the peak current at this point). My current plan (no pun intended) for determining when charging was finished is with an algorithm that looked for a minimum current and stopped the charger.

One solution would be to go to a higher frequency and shorter pulse. There'd be higher switching losses in the mosfet, but it seems to be running pretty cool. It'd also have a side benefit of reducing ripple current which means cooler caps! Hmm... maybe something to look into. I'm at 20kHz now...

Another option would be to add inductance, but more components mean more money.

Anyway, here's a parts list with links for the electrical components as of now:

Inductor (150uH 50A)
E SERIES HELICAL WINDING OUTPUT INDUCTORS : CWS Coil Winding Specialist, manufacturer of transformers, inductors, coils and chokes
Mosfet
IXYS SEMICONDUCTOR|IXFN100N50P|MOSFET, N, SOT-227B | Newark.com
Output Cap
RUBYCON|450USC560M35X50|Aluminum Electrolytic Capacitor | Newark.com
input caps (2x)
CORNELL DUBILIER|380LQ821M400A052|Aluminum Electrolytic Capacitor | Newark.com
diode (2x)
IXYS SEMICONDUCTOR|DSEI30-06A| | Newark.com
Inrush current limiters
Inrush Current Limiters
Bridge Rectifier
40 AMP 600 VOLT BRIDGE RECTIFIER | AllElectronics.com

[arnolde]

Wow, this project is really awesome :-) just what I've been looking for, and was thinking about starting my own, because I also have a problem with paying >2000 EUR for a power supply that generally does less than a 50 EUR PC power supply...

For my 500V battery pack I need a charger that can supply provide 600VDC at 5A. (Although I'd be happy with 2A output for now)

Is there an easy way to adapt your circuit to provide that high a voltage? I was thinking about doubling my 230VAC input with a voltage doubler circuit (simply a rectifier connected differently) so I'd have about 700VDC input. Then limit current by PWM and also I'd like to isolate it, but thats also optional. Also I'll add PFC on the input, one from a PC power supply should do for the beginning.

I assume I'd have less cooling problems with lower current?

Do you have a complete circuit diagram, incl. the PWM setup?

I'd be happy to contribute, i.e. by making profesional PCB's for this project. I'm located in Germany so I can cheaply supply anybody in the EU.

EDIT: I just read the complete thread and was just wondering if you ever condsidered using IGBT's instead of MOSFETs? Like this one: http://www.st.com/stonline/products/...tgf7nb60sl.pdf handles upto 20A and only dissipates (=loses) 25W!!! At 150V/20A thats a loss of 0.83% or an efficiency of over 99%. Dont ask me exactly what IGBT's are, I'm not an EE either, I only remember hearing that they only work over 100V or so, so over 100V they're often used instead of MOSFETs. Also cheaper: This one costs only about 3$, so even if you use 2 in parallel, thats still cheaper than your FET.

How do I know what value of inductivity (coil) I need? I can see that for my 5A output current I'll obviously need one with at least 5A rating, but how many mH?

About adjusting the current in fine tuning: I recommend looking at PWM controllers that are meant for SMPS design, like the US2842 (http://www.st.com/stonline/products/...99/uc3842b.pdf), they sense the current themselves and auto-regulate to keep it constant, and on the current adjust knob you can add resistors in the top and bottom of the pot (wherever necessary) to keep the full swing range of the pot inside your desired current limit range. That way you should be able to comfortably adjust between 0-25A in 0.1A steps with a normal trim pot.

As for coils, would these be ok, do you think?
http://www.vishay.com/docs/34022/ihv.pdf (30A, 150µH)

Last but not least, here is a circuit I found of a commercial ATX power supply, that may provide some interesting ideas: http://www.smpspowersupply.com/ATX_p..._schematic.pdf

After reading this paper http://www.irf.com/technical-info/wh...oosewisely.pdf which compares MOSFET and IGBT, I guess it really doesnt matter, while MOSFETS seem to me to have a very slight advantage.

[arnolde]

No answer yet on a Sunday? Awe, c'mon, I wanna get started NOW :-D
I've made a partial list of stuff to order on Farnell. I'm still undecided on what kind of current sensing transformer to use - I have a 50V/50mV shunt but I'd rather have isolation in place. Which makes me wonder... since the secondary circuit is DC, even if pulsed, is that suitable for sensing current magnetically?

EDIT: Maybe this would be suitable? Costs under 5$.
http://docs-europe.electrocomponents...6b8008f208.pdf

[jyanof]

For my 500V battery pack I need a charger that can supply provide 600VDC at 5A. (Although I'd be happy with 2A output for now)

Is there an easy way to adapt your circuit to provide that high a voltage?

A 500V Pack! Wow! What motor controller are you going to use? Even the zilla has a max of 400V!

Anyway, this part of the charger would be easy to adapt. It's essentially a buck mode charger and'll convert any voltage to a lower voltage. You'd need mosfets/diodes/caps that can handle the high voltage (and maybe check that the inductor laminations can handle it too?). As you stated though, your current requirement would be much lower, so it might be relatively easy to find something suitable.

I assume I'd have less cooling problems with lower current?

As far as heat generation, you're looking for something that'll handle 4 times less current, so that's 16 times less heat generation (assuming the same Rds_on). Rds_on will likely be higher for a high voltage device, but not 16x! Diode losses would be proportional to current as well, so 4x less heat with 4x less current.

I think you're issues will be finding a source with a high enough voltage. Besides your voltage doubler method, you could use a boost circuit before the buck circuit to get to a high enough voltage. A PFC controller would do this while timing the PWM to increase the power factor. Here's some light reading:
http://services.eng.uts.edu.au/~venk...3/ch07s3p1.htm

I just read the complete thread and was just wondering if you ever condsidered using IGBT's instead of MOSFETs?

From what I understand, IGBT's have a constant voltage drop like a diode does. I think they're good for high current high voltage applications (like the zilla), but mosfets might be better for this project.


Do you have a complete circuit diagram, incl. the PWM setup?

Currently, I'm just using a 555 timer as the PWM source. I'm reading through the Atmega stuff and trying some things out on the STK500. I'm learning slowly... Here's the 555 circuit:
DPRG: A Simple PWM Circuit Based on the 555 Timer

About adjusting the current in fine tuning: I recommend looking at PWM controllers that are meant for SMPS design,

I started out looking at actual Lead Acid charge controllers. Some draw backs, I think, are:

1. they all use a resistive device as a current sensor (which is why I initially had a shunt that was going to be amplified with an opamp)
2. they all have unfamiliar ways to tune the feedback loop so that it is stable... it's been a while since I had a control theory class.
3. in the end, we'd be adapting something to work in an application it wasn't designed for. This brings a lot of unknowns and challenges that we'd have to deal with. At this point, programing a micro seems like it'll provide the most flexibility to adjust the charge profile to whatever we want it to be and eliminates all the band-aids to get a different product to work.

As I always say though, these are the reasons I'm headed down this road and should only be a consideration for others. Another option might be better, but we won't really know until someone tries it out! (I thought those initial ST mosfets would be good due to their 60A rating... little did i know that the non-isolation would cause electrical problems and electrically isolating them would cause heat problems)

Feel free to try other things out and let us know how it works!

As for coils, would these be ok, do you think?

Here's a good resource for many buck-converter parameters:
http://www.google.com/url?sa=t&sourc...tq1ee3ryq5Qhpg

I actually sized my inductor from an equation in a TI application note. See equation 12 on page 11. I selected it to have constant current at 2A minimum at 100khz. Not that I'm at 20 khz (to reduce switching losses), I don't have constant current until about 5 amps...
http://focus.ti.com.cn/cn/lit/an/slua019/slua019.pdf

As for isolation, I've struggled with this one. It appears the high power chargers such as the PFC-20 and Russco's are not isolated. My design won't be isolated - the positive rail is connected to B+ while B- and the (-) rail are switched on and off with the mosfet. I think the only way to isolate them would be to use a very large isolation transformer which would be large and heavy and expensive. Consequently, it is greatly recommended to isolate your traction pack from the vehicle and clean the batteries frequently to reduce leakage paths.

No answer yet on a Sunday? Awe, c'mon, I wanna get started NOW :-D

ha! I love the enthusiasm! I was fishing this weekend which provided many hours to think about how to program this sucker. DIY stuff can get addicting! Do it up, and let us know how it works!

[arnolde]

Hi,

I use a standard industry 3-phase 400V inverter for a standard industry 400V 3-phase motor :-) 400V times sqrt(2) is 560 so thats the DC voltage on the DC circuit. luckily 500V is sufficient, maybe even 450V would have been enough, but high voltage = low current and higher possible motor rpm.

Thanks a lot for the links, I'll order all the parts today and hopefully tomorrow I can start wiring things up. I cant wait to show all the German smart-asses on the electronics forum how easy it is to do what they say is "impossible for under 2000 EUR" or even "impossible even just to draft a circuit for less than 500 EUR" - harharhar :-) But I owe it to you, I had similar ideas but since you've already proved it will work in principle, that gives me the extra motivation to actually do it. Thanks a lot for the pioneer work!

I hope to get all components from stock. I use rs-components and Farnell for my electronics orders. So even with orders from the USA, it's here in 3 days :-)

You say you "originally had a shunt" for current measurung? What do you use now?

Take a look at this: http://www.st.com/stonline/products/...99/uc3842b.pdf see figure 19 on page 9, looks rather simple to me? Maybe not quite as simple as a 555 circuit but at least its self-regulating!

Consider safety issues with using just a µC for the PWM. Imagine your program locks up (bug in software, or even a hardware problem with the µC, or overheating) and leaves the duty cycle "ON". It will roast your batteries. You need to implement several safety features to prevent that from happening.

[jyanof]

For current sensing up til now, I've used a handheld DC clamp ammeter. It's pretty slick and has made my life a lot easier in many situations. For the future, I picked up a hall effect sensor to measure current:
838-L03S050D15

I haven't hooked it up yet, but that's the plan.

Yeah, the 555 timer setup definitely is still a 'dumb' configuration. The chip you show looks intriguing - what are you planning to use? This still needs hardware compensation to adjust the feedback loop to make it stable. Maybe it's easy to do, but it scares me! I'm hoping that it's easy to accomplish with software in the micro.

I agree about the safety issue! I bet that including a fuse in the output and a breaker in the input will solve some problems. There's going to be A LOT of current if it freezes on, so something will blow before the batteries get fried. In fact, the mosfets failed ON during one of my previous tests with the ST mosfet - The rectifier blew and shorted and then the main breaker (which is a dedicated 240V line) tripped. So, every electrical component (except the caps) failed.

I've debated putting hardware over-current protection like Paul has in his car controller. I don't think it's necessary since a charger is a much more benign application than a car controller - less changes in current, voltage, load, etc. I think just a fuse is good enough over-current protection.

I thought about the micro freezing at some PWM state and never shutting off. That could fry the batteries if it stays on all night. I guess robust code could help here. I also thought about having a manual timer switch (something like a dial you see at spas) that disconnects the charger after a certain amount of time. You'd have to be a little smart about how long you're charge should take, and it still may fry the batteries for a little bit, just not as long. If I had lithiums, I'd be really concerned about this.

Any other ideas? I haven't thought too much about failure protection.

[MPaulHolmes]

I don't think the microcontroller freezing would be a problem. You can just have a watchdog timer set so that the program restarts in like 1 second if the microcontroller freezes up. Then the program just goes right back to its charging algorithm. I guess you would have to make sure that you start the charging algorithm sort of passively, to see if the battery is mostly charged already? So that if it hangs repeatedly and keeps starting over, it wouldn't be hard on a fully charged battery.

By the way, I never knew what watchdog was until like 1 month ago. The watchdog is a separate timer that keeps counting even if the microcontroller freezes up. If it ever reaches its max, it triggers a reset of the program. So, you put code in that makes sure the watchdog keeps starting over so it never reaches its max value (that you choose). That way, if the code stops running for some reason, the watchdog triggers a reset. It's awesome! ya!

[arnolde]

Thanks a lot for the literature :-)

Okay I guess a dedicated PWM controller is too complicated (or in fact, too time consuming to read all the design sheets and decide on a design to use) for me right now, too. Since the critical part is the power switching part, I think I'll also use a 555 to start with until I've proven that the whole thing works as I plan. Still nobody seems to beleive I can build a 3000W PSU for under 500 EUR so I'm itching to prove them wrong ;-)

I'll draw up a circuit and order components now. The 2 coils will come from USA (150µH, 30A, 16 EUR each) and the capacitors are also on the expensive side (400V/560µF, almost 10 EUR each)

Say, how do you power your NE555? Seperate battery/PSU or from the high voltage? I think I'll add a little 5$ transformer to provide dual 15V, then I can use a hall sensor too later if necessary.

I think I'll start with 48kHz PWM frequency (3,3k / 10nF)

Say, those transformer like things, are those your 150µH coils? According to the datasheet they should look quite different... http://www.vishay.com/docs/34022/ihv.pdf

[jyanof]

Originally Posted by arnolde

Thanks a lot for the literature :-)

Say, how do you power your NE555? Seperate battery/PSU or from the high voltage? I think I'll add a little 5$ transformer to provide dual 15V, then I can use a hall sensor too later if necessary.
...
Say, those transformer like things, are those your 150µH coils? According to the datasheet they should look quite different... http://www.vishay.com/docs/34022/ihv.pdf

I'm actually using a separate power supply for LCD monitors. I got it on ebay pretty cheap:
AC Adapter Power Supply for LCD monitor TV+Cord 12V 5A - eBay (item 260377413127 end time Jun-12-09 12:29:57 PDT)

I like it because it'll work on 120V or 240V and up to 5A (not that I'm anywhere near 5A). it's gotta be some sort of SMPS, maybe a flyback transformer? I haven't taken it apart.

The inductors are the two square-ish looking things with the copper coil in the middle. See the link to coilws.com in an above post... The big black transformer in the background on the table is a multitap transformer I set up as an isolation transformer so I could still use the scope.

Why are you getting such high amp rated inductors? Something smaller might suffice? Of course, over-designing might help you be successful the first time!

[arnolde]

Actually, you are using these exact same coils in your own circuit! (At least according to the diagram you published). Thats one factor, and the other factor is, these are the only coils >10A and anywhere near 100...500µH that my suppliers can supply.

Of course, I only really need ones for 5A output. But then, I'm not sure if the "DC rating" is the measure to go by, since the current is not steady. For all I know, the current peaks while "sloshing back and forth" could be 2 or 3 times that. So yes, I'd rather have it a little overdimensioned. Also maybe I'll want to use this circuit someday to charge other voltages at higher currents. i.e. a 144V battery could be charged with 30A :-P