High efficiency prius conversion

Dear forum friends

I am 35yo electronic engineer and enthusiast. half year ago I bought used prius gen2 which i have converted to PHEV. I have designed and built each part from scratch (charger, bms and converter). system is running fine for last three months so i decided to publish part of my work. In next few days I will publish details and some new ideas for conversion. I will not publish complete schematics as i am using them for other purposes.
however all design principles will be published.

Hope that this thread will be useful for others.

For start let me share some pictures of finished conversion and fuel consumption charts.







Best regards,
Dusan Latinovic



PART 1 - General system description

System consist of following components:​

Charger - 97% efficient charger with PFC. believe or not it is single stage.

BMS - distributed bms, with small board for each of 60 lifepo4 cells. central bms Master unit monitors temperature and it is able to disconnect charger and DCDC converter. BMS master will disconnect if ANY cell voltage rises over 3.5v or goes below 2.75 for more than 10 sec!. it offers ultimate protection.

DCDC Converter - Heart of this conversion. I have spent lot of time optimizing it. Converter efficiency is 98.7% at 240V and goes to 99% at 213V! I named it '99 converter' and it is a reason why this thread is named high efficiency conversion. Actually, energy losses are lower than contactor type conversion which i will prove later . each converter is designed as modular unit with 2kw power. with 2 converters fuel consumption can be less than half and with 3 or 4 you can have some limited EV drive.​

Best regards,



PART 2 - heart of conversion - 99 converter

when i started converting my prius I carefully analyzed all existing solutions

Contactor method
Simply to design, and can provide as much power as needed. drawback: RMS currents and losses are very high in both OEM and AUX battery. Since battery current sensor is not seeing charging current only method to rise SOC is state of charge drift. to provoke state of charge drift you must keep OEM battery above 240V. well, at 240V SOC drift is slow, so for contactor method usual voltages are 245-250V.
At this voltage level OEM battery is simply overcharged, and heats a lot. Without prius excellent battery cooling system, battery would fail in matters of hours.
I have also measured battery self discharge current at different voltages.
at 233V - 80ma
at 234V - 130ma
235V - 180ma
236V - 250ma
240V - 700ma
For high voltages prius Oem battery acts as load. after 236V self discharge current goes rapidly. at 240V almost 200W! is lost as a heat. at higher voltages current doesn't fall below 1A and SOC raises fast. That supports my claim that losses are higher than 2 or 3 DCDC converters dissipating 22wats each .

SOC spoofing method
Best conversion type for prius ii. OEM and Aux batteries are in parallel sharing load. Drawback: it is for prius 2 only, and we don't plan to drive prius 2 for decades.

DCDC converter method
here we have two type of solutions. hymotion goes with converter which is connected to OEM battery after current sensor (like contactor conversion type). RMS currents can be reduced as DCDC converter is current controlled, but voltage must be very high to have SOC drift. Battery stress is large but not as large as with contactor method.

enginer type of conversion connects before current sensor. Mechanism of raising SOC if different. with this connection battery current sensor sees charging current and raises SOC. even with low voltages eg. 236V you can have SOC above 65%. Only drawback is that charging current is limited. prius will throw DTC is current is higher than 16A. I have driven for few weeks with 18-21A, but every two or three days I got DTC. 15-16A proved to be a safe value.

Good thing for converter conversion type is its universality. basically all hybrids and EV can be converted this way with more or less effort.

Here is how I have designed my converter.
Needed output voltage is in 230-240V range. For high efficiency you should start with voltage as close to 240V and then go with simple buck or boost converter. if you start with 48V you will have 4 times more semiconductors, and transformer and you will have to deal with 50-100amp currents on PCB. from my point of view it is just plain wrong.


lets see both converter variants. first buck converter. efficiency will be highest if forward conduction path is active for most of switching period, meaning that duty cycle should be close to 1. this means that battery voltage should be slightly above 240V eg. 250V. problem is with regen current which will go through mosfet body diode. if your PCB, fuses, cables are not designed for that you will have a failure. if voltage is higher eg 270V efficiency will be slightly lower but you wont have problems with regen. In forward path is mosfet switch which should have very low on resistance and is a pricey component. eg if you have 10A output current you will need mosfet with hot resistance around 100mohm to have less than 1V drop.


For boost converter situation is a bit different. Mosfet is in freewheeling path of DCDC converter which is active less than 20% of time. In forward path is diode with voltage drop in range 0.8V. it means that you can go with lower rated cheap mosfet and still have high efficiency.

There is a drawback for boost converter also. Output voltage should be higher than AUX battery voltage. if it is lower inductor will saturate and converter will behave like diode. that will also stress cables fuses, diode, pcb etc. For that reason i choose 60 cell pack with 192V nominal voltage. if OEM battery voltage goes below that during high load, additional current will go from AUX battery directly and it wont be so high to damage anything (at 192V oem battery is supplying lot of current).

Inductor is hand wound on e55 ferrite core (largest core i found with PCB bobbin and i wanted to keep everything on PCB). wire is 2.1mm thick and inductor resistance is only 22mohm . mosfet used is infineon 20A coolmos, and diode is 30a fast recovery type. Currents on PCB are below 12A so usual connectors, fuse holders, fuses can be used. rated converter output current is 9A total losses are around 22-23W and converter is passive cooled. Converter is current controlled so it can be paralleled. unit is 19cm wide 14cm long and 6cm high (1,5lit volume with heatsink)

I am getting a little tired from typing and have some problems uploading diagrams so i will continue tomorrow.​

Best regards,
Dusan Latinovic​

here are converter diagrams. I am using boost converter. There is only one power transistor and one power diode bolted to heatsink (see attached picture). also there is a picture of 3 parallel converters currently powering my car

​

PART 3 - charger

at the beginning of this part i will put a disclaimer. diagrams here is for knowledge only and everybody is solely responsible for itself. playing with high voltage and with mains can be dangerous.

Attached is a simple schematic. at input we have diode bridge followed by PFC regulator which boost voltage to some 380V for universal input. we have a large DC link capacitor used for energy storage and for holdup requirements. at the end is a buck converter which reduces voltage to required 200-210V or any other below 300V. this is non isolated converter. it is much simpler than isolated and have better efficiency. it can be used only in systems where HV section is completely isolated like in Prius. some sort of inrush current limiting is needed.

In next post I will describe one novel approach for battery charger design, and show how this system can be improved.

Best regards,
Dusan Latinovic​

here is how i have realized my 230V charger. also there is a diagram for 120V charger.
what is novel here? well if you use PFC regulator front-end you should use large DC link capacitor. its main purpose is energy storage, since instant power drawn from line will vary from 0 watt to approximately twice of output power. it also provides holdup time. that is necessary for computer PSU and etc, but for charger holdup time is not important. if you go one step further you can say that even a energy storage capacitor is not needed. AUX battery with its constant voltage will act as energy storage. In this case single switching stage can be used for both power factor correction and charging. energy is transferred directly to battery and it varies with instant line voltage. small output capacitors (2x220uF) are there just to filter high frequency ripple.

only drawback of this solution is that battery current is not constant and it varies with instant power. Battery RMS current will be bit higher than average. My current charger provides 5A (1KW). Rms charging current is little below 7A. Hardly a problem for 40ah lifepo4 pack .

Advantages are numerous:
power factor is good usually over 0.9 (depending of your control technique). with typical charger without pfc power factor is around 0.5-0.55. this means that for 1kw charger almost 2kva socket is needed. my charger draws only 1100va.
being single stage non isolated charger efficiency is extremely high. at 230V it is over 97%. if you omit output diode and with better mosfet it can be over 98%!.
There is no large inrush current! since there is no large electrolytic capacitors on input no inrush limiter is needed. on my charger i use 4 x 2.2uF capacitors for ripple filtering on input.
No load consumption is 0.5W, since there are no relays, fans and etc!
Chargers can be paralleled. in Europe with standard 230V and 16A socket up to 3 chargers can be used.



This is a 120V schematic. principle is same.

In next period I will take photo of charger inside. it is packed in same casing as DCDC converter and it is passive cooled.

One question for people from America - how many of You have 240V socket?

Best regards,
Dusan Latinovic​

 

charger pictures



Charger on its position, also fuse and main switch



Bms cell boards