DIY ebike Chargers

[pelle242]

I wonder if buck PFC could be done with zero crossing interrupt and changing the PWM settings every millisecond or so. These CPUs can do a lot with some careful programming. It would probably be a table lookup or some simple calculation would be close enough, or a combination of the two.

Something like this ?:

 

[Alan B]

It may be fine, but it adds parts, cost, complexity and loss. Not sure we need it at 1KW or so.

Would be great to use the already there parts to do the job. Might not be as optimal, but even partial effectiveness would be a help.

 

[Alan B]

I wonder if buck PFC could be done with zero crossing interrupt and changing the PWM settings every millisecond or so. These CPUs can do a lot with some careful programming. It would probably be a table lookup or some simple calculation would be close enough, or a combination of the two.

Something like this ?:

I was thinking of something a bit simpler. Essentially if you have a buck converter that has a significant ratio of bus voltage to output voltage then by varying the timing of the PWM you can draw current from the line over a wider fraction of the phase angle. So you use the parts you already have and just get the zero crossing timing and adjust the PWM through each half cycle.

 

[Alan B]

Found a 450 watt pc supply being tossed out. Might make a good donor for this project.

It has a ball bearing fan. ATX type. The mounting foot is removable, that leaves the box with no brackets hanging out. Cover removes with screws.

Have to do some tests and see if it still works. I don't need much of it to work for our purposes.

Just need to watch out for the 350VDC in there!

Anyone else make progress on this lately?

 

[heathyoung]

I've been too busy with work lately to get much done, but have just about finished the design of the PCB - the majority of it is overspecc'ed to cope with 2-2.5Kw output without cooking itself, and also to interface to a BMS.

 

[Alan B]

I took the 450 watt PC supply apart. I'm having mixed feelings about using it. So much has to be pulled out, not much gets re-used. Still for a prototype it is a nice start.

I will eventually need something that will fit on the Greyborg. The shape will probably be pretty flat, next to the 24 FET motor controller. So it will need to be a custom box, or one of those flat PC supply boxes they use in some small computers.

I picked up a small Arduino Mini Pro that would be more suitable to small packaging of this charger.

 

[magudaman]

I know you guys are past this stage now its seems but this was a pretty cool switching calculator / designer:
http://schmidt-walter.eit.h-da.de/smps_e/smps_e.html

 

[Skippic]

I've been too busy with work lately to get much done, but have just about finished the design of the PCB - the majority of it is overspecc'ed to cope with 2-2.5Kw output without cooking itself, and also to interface to a BMS.

Can you tell us the approximate dimensions and cost of your design?

 

[heathyoung]

OK current BOM of the 'big stuff' (prices are in Aud, pretty much = USD)

Inductor: EK55246-341M-40AH from http://www.coilws.com/ $38.00 (yes, ouch!)

Bias Supply: http://au.element14.com/myrra/47155/power-supply-5w-15vdc-reg/dp/1825783?Ntt=182-5783 (MYRRA - 47155 - POWER SUPPLY, 5W 15VDC REG) $19.02

IGBT Current Sense Resistor: http://au.element14.com/caddock/mp930-0-020-5/resistor-current-sense-0-02ohm/dp/1612016?Ntt=1612016 (CADDOCK - MP930-0.020-5% - RESISTOR, CURRENT SENSE, 0.02OHM, 30W, 5% ) $7.72

Freewheeling Diode: http://au.element14.com/fairchild-semiconductor/rurg8060/diode-soft-recovery-80a/dp/9843930 (FAIRCHILD SEMICONDUCTOR - RURG8060 - DIODE, SOFT RECOVERY, 80A) $7.58

IGBT: http://au.element14.com/fairchild-semiconductor/fga25n120antd/igbt-npt-to-3pn/dp/1095025?Ntt=FGA25N120 (FAIRCHILD SEMICONDUCTOR - FGA25N120ANTD - IGBT,NPT,TO-3PN) $11.45

IGBT Driver: http://au.element14.com/international-rectifier/ir2125pbf/ic-driver-mosfet-500v-2125-dip8/dp/9101721 (INTERNATIONAL RECTIFIER - IR2125PBF - IC, DRIVER, MOSFET 500V, 2125, DIP8) $13.70

555 Timer for bootstrap: http://au.element14.com/intersil/icm7555ipaz/ic-timer-cmos-7555-dip8/dp/9663762 (INTERSIL - ICM7555IPAZ - IC, TIMER CMOS, 7555, DIP8) $1.26

PWM Controller IC: http://au.element14.com/texas-instruments/tl494in/ic-controller-pwm/dp/1470529 (TEXAS INSTRUMENTS - TL494IN - IC, CONTROLLER, PWM) $1.25

Comparator for Constant Current: http://au.element14.com/national-semiconductor/lm358an/op-amp-dual-low-power-dip8-358/dp/9486810 (NATIONAL SEMICONDUCTOR - LM358AN - OP AMP, DUAL LOW POWER, DIP8, 358) $1.76

Input capacitors - note ripple current ratings!: http://au.element14.com/kemet/alc10a331dc400/capacitor-330uf-400v-35x35/dp/1679496?Ntt=1679496 (KEMET - ALC10A331DC400 - CAPACITOR, 330UF, 400V, 35X35) $15.45 X 2 (or 3)

Inrush Current Protection Thermister: http://au.element14.com/ametherm/ms32-10015/thermistor-series-ms32/dp/1653452?Ntt=1653452 (AMETHERM - MS32 10015 - THERMISTOR, SERIES MS32) $8.34

Current Sense Resistor: http://au.element14.com/bourns/pwr4412-2sdr0050f/resistor-0r005-1-5w/dp/1435985?in_merch=true&MER=i-e1a0-00001051 (BOURNS - PWR4412-2SDR0050F - RESISTOR, 0R005 1% 5W) $1.97

Note that my design already reuses the bridge rectifier, common mode chokes (massive), and fuses + holders of the original Vectrix charger.
These would add a cost since you need at least a bridge rectifier, and common mode chokes if you don't want a transmitter.

Figure a nice 35A bridge rectifier. Common mode chokes - need to be able to handle 15A. Same place for these as the inductor.

PCB size is 150 X 200, this could be reduced - (those caps take up a lot of realestate on the PCB unfortunatly - the output cap could be reduced or omitted, but at the expense of a higher ripple and lower regulation - since a battery is across the output, this is not too major a consideration.) Going smaller on the input capacitance would require caps with a massive ripple rating - or you will boil them.

 

[Alan B]

OK current BOM of the 'big stuff' (prices are in Aud, pretty much = USD)

Inductor: EK55246-341M-40AH from http://www.coilws.com/ $38.00 (yes, ouch!)

Bias Supply: http://au.element14.com/myrra/47155/power-supply-5w-15vdc-reg/dp/1825783?Ntt=182-5783 (MYRRA - 47155 - POWER SUPPLY, 5W 15VDC REG) $19.02

IGBT Current Sense Resistor: http://au.element14.com/caddock/mp930-0-020-5/resistor-current-sense-0-02ohm/dp/1612016?Ntt=1612016 (CADDOCK - MP930-0.020-5% - RESISTOR, CURRENT SENSE, 0.02OHM, 30W, 5% ) $7.72

Freewheeling Diode: http://au.element14.com/fairchild-semiconductor/rurg8060/diode-soft-recovery-80a/dp/9843930 (FAIRCHILD SEMICONDUCTOR - RURG8060 - DIODE, SOFT RECOVERY, 80A) $7.58

IGBT: http://au.element14.com/fairchild-semiconductor/fga25n120antd/igbt-npt-to-3pn/dp/1095025?Ntt=FGA25N120 (FAIRCHILD SEMICONDUCTOR - FGA25N120ANTD - IGBT,NPT,TO-3PN) $11.45

IGBT Driver: http://au.element14.com/international-rectifier/ir2125pbf/ic-driver-mosfet-500v-2125-dip8/dp/9101721 (INTERNATIONAL RECTIFIER - IR2125PBF - IC, DRIVER, MOSFET 500V, 2125, DIP8) $13.70

555 Timer for bootstrap: http://au.element14.com/intersil/icm7555ipaz/ic-timer-cmos-7555-dip8/dp/9663762 (INTERSIL - ICM7555IPAZ - IC, TIMER CMOS, 7555, DIP8) $1.26

PWM Controller IC: http://au.element14.com/texas-instruments/tl494in/ic-controller-pwm/dp/1470529 (TEXAS INSTRUMENTS - TL494IN - IC, CONTROLLER, PWM) $1.25

Comparator for Constant Current: http://au.element14.com/national-semiconductor/lm358an/op-amp-dual-low-power-dip8-358/dp/9486810 (NATIONAL SEMICONDUCTOR - LM358AN - OP AMP, DUAL LOW POWER, DIP8, 358) $1.76

Input capacitors - note ripple current ratings!: http://au.element14.com/kemet/alc10a331dc400/capacitor-330uf-400v-35x35/dp/1679496?Ntt=1679496 (KEMET - ALC10A331DC400 - CAPACITOR, 330UF, 400V, 35X35) $15.45 X 2 (or 3)

Inrush Current Protection Thermister: http://au.element14.com/ametherm/ms32-10015/thermistor-series-ms32/dp/1653452?Ntt=1653452 (AMETHERM - MS32 10015 - THERMISTOR, SERIES MS32) $8.34

Current Sense Resistor: http://au.element14.com/bourns/pwr4412-2sdr0050f/resistor-0r005-1-5w/dp/1435985?in_merch=true&MER=i-e1a0-00001051 (BOURNS - PWR4412-2SDR0050F - RESISTOR, 0R005 1% 5W) $1.97

Note that my design already reuses the bridge rectifier, common mode chokes (massive), and fuses + holders of the original Vectrix charger.
These would add a cost since you need at least a bridge rectifier, and common mode chokes if you don't want a transmitter.

Figure a nice 35A bridge rectifier. Common mode chokes - need to be able to handle 15A. Same place for these as the inductor.

PCB size is 150 X 200, this could be reduced - (those caps take up a lot of realestate on the PCB unfortunatly - the output cap could be reduced or omitted, but at the expense of a higher ripple and lower regulation - since a battery is across the output, this is not too major a consideration.) Going smaller on the input capacitance would require caps with a massive ripple rating - or you will boil them.

Excellent! Is there anything for inrush current limiting, or is this not required?

 

[heathyoung]

Yes, its in the BOM:

http://au.element14.com/ametherm/ms32-10015/thermistor-series-ms32/dp/1653452?Ntt=1653452 (AMETHERM - MS32 10015 - THERMISTOR, SERIES MS32) $8.34

Its expensive because its the right size for the job, and a quality part. Limits the IRC to 30A if I remember correctly (been a while since I did the calcs)

 

[Alan B]

Excellent. Missed it in there. Thanks.

 

[heathyoung]

Here is the PCB as it stands.

Component values for 240V, 10A input, 150V, 12A output (1800W). Would scale with current values to at least 15-20A. (yep, thats 3KW).

I have not yet built this, its a work in progress (ie. layout is a bit so-so) - no line filtering (athough there is room in the bottom LH corner, which would normally be removed for the Vectrix charger board).

There could be bits wrong, I'll have to build a prototype and let the smoke out

Some points: The resistor values in a few places need tweaking (WAG's), the BOM is far from complete (left out the output caps - doah). The second RURG8060 can be omitted in cases where the battery is not permanently connected (like it is in the vectrix).

Everything floats at half mains potential. This isn't a project for the tame-of-heart.

 

[Skippic]

Thank you healthyoung for doing all this. It looks like without you we would stand little chance. For those who don't want to calculate the price for the mentioned components it is $142.95.

Will this require active cooling?

 

[heathyoung]

The plan is that it won't require active cooling, but will require a good heatsink.

I'll do the heat calcs for the heatsink later. The inductor is overkill so it doesn't require active cooling.

 

[Skippic]

Sounds great! Just thinking of a way to make it adaptable to different power and space requirements of all users. From what I understand the central part of your PCB should be very similar for all applications and the giant capacitors and inductors will depend on the power needs. I know you are not in that stage yet, but once tested it might be useful to have the controlling part on a small PCB with connections for the big components external to the board.

 

[heathyoung]

Potentially yes - the main board will allow many different combinations of output voltages and currents - the driver cannot be too far away from the IGBT though.

There are a heap of different inductor styles, and these could be adopted later to suit.

I'm starting to get the parts together, fortunatly the vectrix charger has some nice 400V low esr electrolytics I can use for the test board.

 

[IBScootn]

Heathyoung,

Glad to see you making progress on this design. I've been looking for a charger for my e-motorcycle pack (24x 40AH GBS lifepo cells) that can dump up to 30A into it. Why, 30ish amps, 32A is the max recommendated current for these cells. 70A max J1772 charging stations are popping up everywhere; see: plugshare.com. Currently, I'm charging at 9A, and would like triple that current to reduce opportunity charge times. I located a J1772 to Nema 14-50 adapter cable, now just need a high current charger. At around 84V, what is the max current one could hope for out of this design?

Good luck,

IBS

 

[heathyoung]

Heathyoung,

Glad to see you making progress on this design. I've been looking for a charger for my e-motorcycle pack (24x 40AH GBS lifepo cells) that can dump up to 30A into it. Why, 30ish amps, 32A is the max recommendated current for these cells. 70A max J1772 charging stations are popping up everywhere; see: plugshare.com. Currently, I'm charging at 9A, and would like triple that current to reduce opportunity charge times. I located a J1772 to Nema 14-50 adapter cable, now just need a high current charger. At around 84V, what is the max current one could hope for out of this design?

Good luck,

IBS

Once again, your biggest issue will be the current handling capabilities of your inductor - at 32A your peak inductor current will be 36A, but more worrying is the 16A(!) ripple current rating on the input caps - You would be pushing your luck with 15A ripple current rating (at 30A you would need 14.89A).

To a lesser (but not insignificant) degree - you will need more heatsinking on the flyback diode and IGBT due to Vf losses.

But my concern would be boiling those input caps dry. If you are running 240V - it works out better. 13.4A (hmm. I should redesign the board for a voltage doubler config).

 

[IBScootn]

Yes, the J1772 to Nema 14-50 adapter will give me only 240V (no 120V option) and max 50A.

The J1772 charging stations are popping up everywhere around me, and I would love to take maximum advantage of them (especially since most are a free service at this time). The thought of a full charge in under 1 hour near one of my destinations would completely untether my e-motorcycle and range anxiety would be a thing of the past.

So 32A would be awesome; but 25A would put a big grin on my face too.

Thanks,

 

[cor]

But my concern would be boiling those input caps dry. If you are running 240V - it works out better. 13.4A (hmm. I should redesign the board for a voltage doubler config).

Hi Heath,
The voltage doubler does not need a different layout, only a switch (or jumper) and having the input caps split in two series 200V instead of parallel 400V ones.
The voltage doubler simply hardwires one AC input (neutral) to the middle point between the 200V caps, so each half phase charges alternating the top and bottom cap through one diode of the bridge. This is used in many devices designed to run at 230V and with a switch to double voltage for USA operation.
Success,
Cor.

 

[cor]

Hi Heath,
I noticed that you use plain bridge rectifier for the input, which will be a problem at higher power levels. Not only the fact that you start tripping breakers but also it is hard on the input caps.
My preferred topology is where the entire charger is based on a single PFC stage, so only one (FET) switch is needed and essentially there should not be a need for large caps if the supply is always feeding the battery.
Normally the PFC has an inductor that is shorted to ground with a FET, then during flyback it charges an output cap to 400V, allowing for universal (90-260V AC) input. It is possible to get a 150V current limited (CC-CV charger profile) with still only this one stage and based on the simple UC3854 design, by adding a second winding to the inductor to give an isolated output at 150V from the switching 170-340V input during flyback. (In essence it is a mix between PFC and flyback converter due to the adding of the second winding). This has the benefits:
1. isolation - no grid voltage by accident on your bike frame!
2. PFC so you can pull max power from a circuit without tripping breakers and burning up extension cords.
3. Easy to change output voltage by changing the winding of the secondary.
4. No big caps needed. The PFC input cannot use input caps anyway and since the battery is directly connected to the secondary output, there is no need for bulk caps on the output either. Most batteries can take the AC pulsing that is typical for the output of the PFC and flatten it out to give a good voltage feedback.

Output current limit adding to the UC3854 reference design adds one transistor and a few resistors.
This design will also integrate well with Richard and Gary's BMS when you add a TL431 (the OVP point) driving an opto-coupler to the Vsense input of the UC3854 which will also give the needed isolation to separate the output from the input voltage.

The inductor with a secondary winding will likely be a dedicated item, but this charger should be super simple, small due to absence of big caps and flexible in output voltage. I have used a plain PFC from a Dell server supply as a charger for my other EV, delivering 1400W at up to 385V with high efficiency.

 

[pelle242]

Yes, the J1772 to Nema 14-50 adapter will give me only 240V (no 120V option) and max 50A.

The J1772 charging stations are popping up everywhere around me, and I would love to take maximum advantage of them (especially since most are a free service at this time). The thought of a full charge in under 1 hour near one of my destinations would completely untether my e-motorcycle and range anxiety would be a thing of the past.

So 32A would be awesome; but 25A would put a big grin on my face too.

Thanks,

Have you read up on J1772? According to wikipedia you need to interface a data signal with the right protocol for it to work.
http://en.wikipedia.org/wiki/SAE_J1772

Not saying it cant be done but the power conversion is probably the easy part..

 

[IBScootn]

No, the J1772 to 240V conversion is the easy part; it is available from: modularevpower.com. Check out their eBay store. They have a variety of adapter boxes or you can just use their cable with AVC1 control board. Plenty of designs out there if you want to build your own control board too.

So, all we need is a cheap bulk charger now.

 

[heathyoung]

Hi Heath,
I noticed that you use plain bridge rectifier for the input, which will be a problem at higher power levels. Not only the fact that you start tripping breakers but also it is hard on the input caps.
My preferred topology is where the entire charger is based on a single PFC stage, so only one (FET) switch is needed and essentially there should not be a need for large caps if the supply is always feeding the battery.
Normally the PFC has an inductor that is shorted to ground with a FET, then during flyback it charges an output cap to 400V, allowing for universal (90-260V AC) input. It is possible to get a 150V current limited (CC-CV charger profile) with still only this one stage and based on the simple UC3854 design, by adding a second winding to the inductor to give an isolated output at 150V from the switching 170-340V input during flyback. (In essence it is a mix between PFC and flyback converter due to the adding of the second winding). This has the benefits:
1. isolation - no grid voltage by accident on your bike frame!
2. PFC so you can pull max power from a circuit without tripping breakers and burning up extension cords.
3. Easy to change output voltage by changing the winding of the secondary.
4. No big caps needed. The PFC input cannot use input caps anyway and since the battery is directly connected to the secondary output, there is no need for bulk caps on the output either. Most batteries can take the AC pulsing that is typical for the output of the PFC and flatten it out to give a good voltage feedback.

Output current limit adding to the UC3854 reference design adds one transistor and a few resistors.
This design will also integrate well with Richard and Gary's BMS when you add a TL431 (the OVP point) driving an opto-coupler to the Vsense input of the UC3854 which will also give the needed isolation to separate the output from the input voltage.

The inductor with a secondary winding will likely be a dedicated item, but this charger should be super simple, small due to absence of big caps and flexible in output voltage. I have used a plain PFC from a Dell server supply as a charger for my other EV, delivering 1400W at up to 385V with high efficiency.

Yep the design is a little 'brute force' in its approach - hence the large low esr caps.

Using a boost topology PFC in a flyback configuration is an interesting approach - but a bad idea at high power - core losses are going to be a bitch.

There are non-isolated buck PFC's around, but the chips are difficult to lay your hands on!

I do have the original transformer from the vectrix charger, it was in full bridge forward converter configuration.

ST AN1060 makes for interesting reading - 'Flyback converters with the L6561 PFC controller'. They have a high PF battery charger with CC/CV mode.