A possible homemade CC/CV bulk charger?

[ARod1993]

So I've been looking at some of the charger diagrams, and I've drawn up plans for a reasonably safe and efficient high-power (1-5kW) CCCV bulk charger almost entirely out of "dumb" parts (inductors, capacitors, resistors, MOSFETs, gate drivers, and some op-amps and comparators for feedback purposes; the idea would be to take in either 120V or 240V 60Hz line power, then throw it through a line filter, an inductor for basic PFC correction, and a bridge rectifier (a center tap on the inductor and a switch lets you use the bridge rectifier as a voltage doubler) to get 350VDC or so. That would then be fed through a full-bridge converter (see Chapter 7 of Power Electronics by Daniel Hart for a diagram) to get a smooth controllable DC voltage out. The DC voltage out would have a very high-value resistor divider (to measure voltage) and a current sensor across the positive lead, with two separate feedback networks (one for voltage, one for current) and a mux tied to a comparator that would switch between current control and voltage control when the voltage across the leads hits the charge voltage of the pack.

I've put a rough draft of it here; some of the component values are obviously incorrect (namely the L and C values on the full-bridge converter and on the buck converter driving the 12V supply rail for the gate drives) and I haven't fleshed out the feedback network yet. The plan is to come back when I'm a bit more awake and finish this up, but I wanted to submit the rough draft to you guys for comment. I've posted a screenshot of the circuit diagram so far below:

 

[heathyoung]

Looks good, stabilising your feedback network in a SMPS at these current levels is fun.

Are you doing current or voltage mode? There are quite a few good quality current mode controllers out there.

 

[liveforphysics]

Sounds like a fantastic project! It would be great to have an electric motorcycle sized DIY charger!

 

[ARod1993]

Thanks! The idea was to have current- and voltage- referenced feedback networks in parallel, both feeding into a multiplexer, with the mux output being fed through a comparator-based PWM generator to the gate drive. There would be a potentiometer used to set current-mode cutoff voltage; that would be part of a resistor-divider feeding into a comparator referenced to the clean 12V coming out of the 7812. When the voltage output required to maintain charging current reached or surpassed the set top-up voltage for the pack, the comparator output would flip; that would also flip the line selector signal on the mux and switch the feedback network in use from current mode to voltage mode.

The idea would be to provide a constant-current 20-50A (controllable via a potentiometer tied into the current feedback network) bulk charge current up to the top-up voltage of the pack (designed to be controllable from about 40V up to 100-120V for the guys running 24-30s LiPo) and then a slow trailing off of the current down to zero (or until the pack gets unplugged). The original inspiration for this came out of my friend DGonz's motorcycle project; he spent $200 on a cheap Chinese 1.2kW charger and had it blow up in his face the first time he plugged it in because he reversed the input leads (blue and brown instead of black and red). He's been trying to debug the damn thing for a couple of weeks now and still can't entirely figure out what blew up, so I've been toying with the idea of building him one from scratch.

 

[liveforphysics]

(designed to be controllable from about 40V up to 100-120V for the guys running 24-30s LiPo)

This would also charge any model of Zero or Brammo.

 

[ARod1993]

(designed to be controllable from about 40V up to 100-120V for the guys running 24-30s LiPo)

This would also charge any model of Zero or Brammo.

Niiiiiice

 

[circuit]

So I've been looking at some of the charger diagrams, and I've drawn up plans for a reasonably safe and efficient high-power [..]

Are you sure the thing you did there with mosfet driver is correct? Have you used drivers before?

 

[ARod1993]

So I've been looking at some of the charger diagrams, and I've drawn up plans for a reasonably safe and efficient high-power [..]

Are you sure the thing you did there with mosfet driver is correct? Have you used drivers before?

I've used IR2125 driver chips before (power electronics lab course last fall) and at this point I basically have the driver circuit for those memorized. I don't know the IRS21844s as well because I've never used them before, but the fact that they drive both the high and low side FETs from a single signal with included shoot-through prevention via resistor-programmable deadtime makes them really useful.

Also, you're right; I frocked up the connection to the full-bridge. M1 and M4 should use the same gate driver outputs, as should M2 and M3; the original configuration would do absolutely nothing when powered up. I fixed the error on the schematic just now; thanks for the heads-up!

 

[circuit]

[..]

I've used IR2125 driver chips before (power electronics lab course last fall) and at this point I basically have the driver circuit for those memorized. I don't know the IRS21844s as well because I've never used them before, but the fact that they drive both the high and low side FETs from a single signal with included shoot-through prevention via resistor-programmable deadtime makes them really useful.

Also, you're right; I frocked up the connection to the full-bridge. M1 and M4 should use the same gate driver outputs, as should M2 and M3; the original configuration would do absolutely nothing when powered up. I fixed the error on the schematic just now; thanks for the heads-up!

This can not work as well. You need separate drivers for each half bridge.

 

[ARod1993]

[..]

I've used IR2125 driver chips before (power electronics lab course last fall) and at this point I basically have the driver circuit for those memorized. I don't know the IRS21844s as well because I've never used them before, but the fact that they drive both the high and low side FETs from a single signal with included shoot-through prevention via resistor-programmable deadtime makes them really useful.

Also, you're right; I frocked up the connection to the full-bridge. M1 and M4 should use the same gate driver outputs, as should M2 and M3; the original configuration would do absolutely nothing when powered up. I fixed the error on the schematic just now; thanks for the heads-up!

This can not work as well. You need separate drivers for each half bridge.

OK; I thought that doing that would just result in halving the drive current to each FET. Rookie mistake Also, just updated the schematic with the additional gate driver.

 

[ARod1993]

So I'm trying to simulate the basic full-bridge converter in LTSpice and I'm getting results that don't make sense; the simulator always gives me zero voltage across the load resistor with the exception of what appear to be random spikes. I've verified that I'm getting bidirectional current through both inductors that comprise the transformer, and I'm getting some sort of current through the main inductor, but no voltage across the load resistor; also, does anyone know why it takes so long for LTSpice to simulate what you're doing?

 

[nieles]

nice project!!

you are missing the bootstrap capacitors for the highside mosfet drivers (between vb and vs)

something for later, it would be nice to sense the output voltage at the the end of the lead (at the plug)
so you compensate for the voltage drop from the fuse/charging leads

 

[ARod1993]

Thanks! I'll definitely keep that in mind for later iterations; the other thing I was considering once I get the damn thing prototyped would be to encase it in some form of removable shock-absorbent foam to enable people to safely mount it directly to the bike without needing to worry about damage. Also, I updated the schematics and added the bootstrap caps to both gate drivers; the new URL for the circuit is here. I've added a fully regulated 12V line powered by a separate internal DC/DC converter and regulated by a Type III compensation circuit with an error amplifier; that should give clean voltage-controlled DC for the gate drives (and in later revisions might allow for some means of charging auxiliary 12V units off the provided 12V line.

 

[nieles]

since the mosfet driver you are using has a Vss and COM, it might be better to keep them separated for better handling.

use the COM for low side return (high current)
use Vss for the digital input side.

 

[ARod1993]

Fixed (I think?). I'm trying to figure out how to do that, especially since both the 12V line and everything past the full bridge are isolated from mains ground and I'm trying to decide whether to leave all that floating. I feel like that's a good idea for people who want to do naughty things like series/parallel connections, but I'm unsure how to deal with ground as such when doing that.

 

[nieles]

I think the vss and com do not need to be connected. If you look at the data sheet, the functional diagram (page 5) it shows a vss/com level shift block. Not 100% sure what the function is of the level shift, not enough experience with these devices.

 

[ARod1993]

I'm just unsure what COM and Vss are supposed to be connected to other than ground.