Wanted: 500V 30kW charger / Prius G3 + logic board ? [SOLD]

[addvalue]

Looking for an affordable solution to build a dedicated 126S, 530V, 60A HV battery charger. No special requirements on hardware, be it re-use OEM, re-purpose power supplies or chinese import as long as it is proven technology with a known can-bus protocol to operate by can-bus as a continuous CC CV charger. Edit: To be powered by 3-phase 230/400VAC from grid or generator. Edit2: updated title to indicate interest in a Prius Gen 3 with logic boared.

[mjc506]

Prius inverter boost converter + 3-phase rectifier? Plus some micro to control the boost converter PWM.

[addvalue]

Prius inverter boost converter + 3-phase rectifier? Plus some micro to control the boost converter PWM.

Thanks that is a good suggestion. I read up on the Prius inverter developments as much as I could but didn't find any examples of actual charger implementations. Could you point me to any examples?

[mjc506]

No examples (yet) that I know of I'm afraid. You're basically varying the PWM on the boost converter to regulate voltage (current) into the battery - you'll need to monitor pack voltage, current and probably temperature too. A reasonable charging curve would be constant current up to your target 'fully charged' voltage, then constant voltage there and allow the current to decay to some small value. (Turning it off as soon as it hits target voltage will result in being less than fully charged as the pack 'recovers'). Plenty of enhancements you can do in software if desired (estimating pack resistance, thus state of health, coulomb counting for true capacity, 'overshooting' the target voltage taking into account pack resistance to charge a little quicker at the top end...)

[bexander]

I have tested using the Prius Gen3 as a charger with celeron55:s SW and it works but it is loud and most likely creates a lot of interference on the power grid. I did some changes to the SW but I could never get the current to be stable enough for my liking.

Here is the thread:
viewtopic.php?t=825

And the SW created by celeron55:
https://github.com/celeron55/prius3charger_buck

[addvalue]

I have tested using the Prius Gen3 as a charger with celeron55:s SW and it works but it is loud and most likely creates a lot of interference on the power grid. I did some changes to the SW but I could never get the current to be stable enough for my liking.

Thanks, missed that thread amongst all others related to chargers. I just read up on those ten pages and think I can follow a bit but it is still hard to grasp the real potential of these inverters. One thing that caught my eye was the suggestion to use a thyristor to lower the rectified input. That made me think I might not even need an inverter to charge the 126S battery (350V - 530V) from a 3-phase generator that wouldn't care much about the power factor. I wonder what could be achieved with a bridge rectifier and a SCR. Or some direct control over the generators output voltage.

That aside, what voltages and power did you run it up to? From 3-phase I assume. And did this Gen 3 project come to a halt mid 2022 or is it still alive elsewhere?

[addvalue]

----you'll need to monitor pack voltage, current and probably temperature too. A reasonable charging curve would be constant current up to your target 'fully charged' voltage, then constant voltage there and allow the current to decay to some small value. ---

At first I'll probably get away with a CC CV curve and some can-bus commands for the BMS to use to monitor and control the charger.

[bexander]

I ran a 96S pack connected to the battery side of the inverter and supplied 400V AC 3-phase on the MG1-output. That allows the diodes in the IGBT to do the rectifing so you get ~600VDC at the motor side of the inverter. The SW handles precharge of capacitors and over all contactor control and also EVSE.
Not using this a charger any more.

I just relized that in your case, with that high voltage, it is no longer possible to do it as above. The battery side of a Prius Gen3 can handle 400V. Maybe 470V if you modify/remove the DC/DC.

Another option is to use 9pcs of Eltek FlatPack HE2 48V 3000W power supplies connected in series. That will give you 400-518V output voltage and 27kW capabillity or 62A. I run 7pcs for my 96S pack.

[addvalue]

Another option is to use 9pcs of Eltek FlatPack HE2 48V 3000W power supplies connected in series. That will give you 400-518V output voltage and 27kW capabillity or 62A. I run 7pcs for my 96S pack.

The funny part is that I am already building a charger with 9 R4850G2's in series, at 58,5V 3kW per unit gives 526,5V 27kW. But I worry running into unknown safety issues and/or control issues putting so many in series. HV isolation and voltage fluctuations between units for instance. How did you deal with that, do you experience any issues keeping the system stable over the full charge range. Seen any effects of units starting to fight each others CC CV control?

[bexander]

Well, in the FlatPacks case they are specified 500Vdc to earth.

So should be almost within spec with 9 in series.

I have 7 isolated CAN buses, one for each FP to control them. One of the FP are doing the current limitation and the others just tags along and supplies a specific voltage.

More info on the FP communication are found here.
viewtopic.php?t=1351

And my entire charger code as of today.

CHG3_1_37.ino

(44.41 KiB) Downloaded 45 times

I have no experience of the R4850G2's but should work in a similar way.
It really a problem, charger wise when you need voltage above 450Vdc.

EDIT: No issues with the FP fighting each other. I do keep them within there voltage operating window at all times.

[addvalue]

I have 7 isolated CAN buses, one for each FP to control them. One of the FP are doing the current limitation and the others just tags along and supplies a specific voltage.
EDIT: No issues with the FP fighting each other. I do keep them within there voltage operating window at all times.

This good to know, I was thinking along the same lines, not sure if I need to use 9 isolated can busses, I am trying to reverse engineer how three units communicate with a central control unit in the original 1U rack (48V parallel but shouldn't make a difference on the can-bus comms). I hope to figure out whether there is some can ID negotiation going on at startup / hotswap, should be really. And if not then just broadcast instantaneous control voltage and current to all might work.
Edit: Do I understand correct you set six units at C >> maxC and V = stringV/7 + dV (say dV is 2V) to stay outside the current throttling range and the last unit at C = maxC and V = stringV/7 - 6*dV ?

[bexander]

Regarding the CAN-bus, it is referenced to the 48V negative output in order to get reliable communication. When they are connected in series all units have different potential at there 48V negative output hence can-isolators are needed. I do not know if the R4850G2 can better handle all CAN-buses connected in parallel?

I start all FP at 44V, slowly ramp up the "master" to 48V since it can't do current limiting to the set current below 47V voltage. Then slowly increase the 6 "slaves" up to the same voltage as the master and then increasing the master one step, then slaves one step, all while monitoring actual voltages and current.