144Vdc EV Charger

HighHopes

10 kW
Joined
Mar 28, 2013
Messages
929
This thread will document my EV charger design & build. Since my high performance drive for induction motor operates on 144Vdc I figured I would need a custom designed EV charger as well to match the voltage level.

At this moment in time I am not planning on bi-directional operation, this design is for charge mode only.

6.6kW, 144V EV Charger, Concept Topology:
Front End: 2-phase interleaved power factor correction with voltage boost pre-regulator
Power Supply: Phase shifted full bridge with zero voltage switching and galvanically isolated output by means of a transformer
1 EV Charger TopologyR0.2.jpg

Link to High Performance Drive for 3-phase AC Induction Motor, which is still in DRAFT state: http://ivanbennett.com/forum/index.php?topic=112.0

Disclaimer:
View attachment Disclaimer.txt
 
Reserved for hardware design files

Power Factor Correction: 2-phase Interleaved Boost
DC/DC High Frequency Converter: Phase Shifted Zero Voltage Switching Full Bridge
Power Supply: flyback DC/DC (HV to 12V, 1A), control power 5V, 3.3V (digital/analog)
Digital Signal Processor: TI Piccolo controller 280x5

DRAFT SCHEMATIC:
View attachment EVCharger_5kW144Vdc_RevB.pdf
Update Feb 11: oops, diode D14, D15, D123, D24 should be deleted. thx for pointing out the mistake

Magnetics Design: TBD
 
Reserved for software design files

PFC + PSFB: TBD (plan to heavily adapt from texas instruments controlsuite)
pinout summary:
View attachment Connector Pinout V0.1.xlsx
Texas Instrument's isolated control card:

This card is for sale on Texas Instrument's website. It's pins are mapped to the 100pin DIMM that is on the EV Charger schematic (i.e. see "connector Pinout" above. I plan to use the on-board isolated USB connection to communicate with traction pack BMS.)

Battery Charger: Algorithm constant voltage (CV) and constant current (CC) as per profile suitable for large traction pack. EV Charger to allow for feedback from traction pack BMS system (health, state of charge, enable/disable, etc.)
 
i have some design work to clean up before posting first draft. will get back to it soon (don't hold your breath)
 
Excited to see it my friend!
 
Great stuff, would be good to install it in a bike, then only a power cord is needed to charge. Or a smaller version, because 5kW is probably too much for a ~2kWh battery (depends on the vehicle anyway).
I believe it is very useful to integrate the BMS and the charger circuit.
How do you plan to limit the charge current? On the usual cheap BMS-s there is no current limit, just an on/off MOSFET, and the current limitation is in the charger.
 
well, that's a good question. for me this is a two part project. first is the power supply which is a 5kW, 144V with very little current ripple capable (and good power quality on the input so it doesn't kill your house's power). i feel very comfortable managing this first part on my own. the second part is the charge profile along with generic protection and then more specialized protection related to multi cell battery packs. the second part of this project, i'll be honest, is a new for me, so any help out there would be appreciated. historially all my power electronic sources has been from ground power or a generator, i never had battery as a source in my project.. so charging a battery, is new. but that's exactly why i liked this project, gives me something to study and learn which is interesting while i simultaneously help others with their motor drive designs on this forumn. we all try to help as we can :)

i just found this a few minutes ago, here is a paper written on the above which as you can see is nearly 100% the same as what i showed above in concept drawing for the power supply. http://www.delta-q.com/home/serve_p...n-Board3.3kWBatteryChargerforPHEVApplication_
 
This thread - http://www.diyelectriccar.com/forums/showthread.php/sic-llc-modular-charger-design-162082.html may also be useful - Using a PFC front end and an LLC topology.
 
Forum member avandalen has reverse engineered a low power ebike charger, schematics are on his page:
http://www.avdweb.nl/solar-bike/electronics/portable-lightweight-lifepo4-ebike-battery-charger-800g.html
In this circuit the constant current generator is built-in with the feedback from the output current, so a second stage current generator is not needed.
If the charger is controlled from the BMS, then any current profile can be implemented - this is not usual, but has advantages.
Avandalen has also a general introduction to BMS; most BMS-es apply passive (resistive) cell balancing:
http://www.avdweb.nl/solar-bike/electronics/bms.html
 
Great to see your first thread on ES ;)
 
heathyoung, thx for the link. my opinion the charger in that design is not made to be a product, though it may actually be functional. that is, it might charge a battery but it would never pass a realistic specification of performance.

peters, looks like the design you linked would be acceptable for low power, i would consider using this design for an ebike. its not useful of course for my purpose cause i'm at considerably higher power so i concern more for safety/protection plus i am focusing on higher power quality at both input and output than what their design will give.

If the charger is controlled from the BMS, then any current profile can be implemented
this is exactly the thing i spent a couple hours trying to understand. the link you provided did not actually say this anywhere that i read. i was all over TI's website, could not find this. now i am looking at other BMS suppliers to see.. where does it say that the BMS is involved with the charge profile? for example, if i just send battery pack 144V with some ripple and whatever current gets pulled gets pulled. will BMS turn this into the exacted needed profile to recharge each string of batteries said BMS is responsible for? not sure of that, i still think the main DC/DC converter (PSFB in my case) has to set the CV and CC needed by the batteries and the BMS just manages when to shut off charge, what happens after charger is disconnected (like balancing) and monitoring battery health while in use.
 
BMS subject is new to me, so excuse if i say some stupid stuff.

this is the part number i have been looking at for BMS in an electric vehicle large battery pack. part # BQ76PL536A. what i am trying to figure out is if this part somehow provides the actual charge profile or no, it just monitors the battery while some other dc/dc converter (i.e. the single main PSFB shown in first post picture) manages the actual charge profile.

whoaa... digikey lists this for $12 per. how many do you need for a 20kWH pack? ummmm.. this is getting into stupid levels of cost is it not?
 
Yes it is clear your design is for high power, I was just trying to show on the example that CC is generated by the charger circuit, not by the the BMS.
HighHopes said:
i still think the main DC/DC converter (PSFB in my case) has to set the CV and CC needed by the batteries and the BMS just manages when to shut off charge, what happens after charger is disconnected (like balancing) and monitoring battery health while in use.
Yes, it is exactly how the usual charger-BMS combos work.
where does it say that the BMS is involved with the charge profile?
Nowhere, it was just my idea, sorry.
Normally the balancing works with a hysteresis voltage control by the BMS: when a cell from the string reaches the top voltage (e.g. 4.2V), then the uC on the BMS cuts off the charger current with a MOSFET (in series with the whole pack) and enables the bleed resistor connected parallel with that cell. Then waits until the cell voltage falls to 4.1V (or similar) and then enables the charge current again for a short time, until the cell is 4.2V again. During that short ON period the charge of the other cells is also increased a little bit.
But from my experience there is a potential problem if the charge current is very high: there is some latency from a BMS module (an IC with 4..8 cells) detecting a cell at 4.2V to the uC cutting off the charge current, because shifting the information through the BMS modules and processing by the uC takes some time. If the charge current is high and the pack is already being balanced for some time (some cells are very much at the top), then due to the latency the cells at the top voltage can get to a higher voltage (e.g. 4.3V) very quickly, that is out of spec. To avoid this, it would be useful to reduce the charge current in the balancing phase, because then the raising of the cell voltages is slower.

On your block diagram there is a CAN/SPI comm. between the BMS and the charger and I thought you also intend it to control the charger current.
 
Ok phew, thought I didn't understand something! I hope latency is not a problem cause were talking critical function. I'll keep an eye on it. I like your idea of letting bms manage the profile I think this is not hard to do but it means fairly high expense cause u need a bms every 6 cells. This $ seems crazy to me, makes me wonder for alternate no bms solution or many wires but one dedicated uC.

The high speed SPI is so there is status and health feedback direct from battery to main power supply. Looked like easy to implement and could be useful. It's just architecture right now, during detail review the final schematic will be tweeked.

Now I go to read how PSFB can do CC charging. CV is a given, for CC probably need large output inductor is my guess.
 
ok, i posted first draft schematic of main power supply. this is how i normally design, i do the archetecture as a detail design. what i mean is i draw out the schematic and i don't worry too much what resistor value is what, or what capacitor value is actually needed. i just throw it together paying some attention. i make notes as i go, which i also left on the schematic. this is just my method how i like to design stuff.

so first task.. for a brave soul.. calculate the power dissipation for the dual mosfet driver shown on the PFC sheet. i want to check that this part will be around 60 to 70degC without fan cooling and i do not have a good feeling so thought it should be double checked. this mosfet driver (reference designation U2 in schematic Rev A) is driving two mosfets (part number shown) and is supply by 12V and switching at 100kHz. who wants to take up this challenge? analysis should be done in excel or preferably MathCAD. (feel free to PM me if you need some hints).
 
so first task.. for a brave soul.. calculate the power dissipation for the dual mosfet driver shown on the PFC sheet.
this a really important part of product development. to *know* what you're doing. power dissipation is a key factor because it helps you size your power supplies, heat rise in the box (for which you then calculate a thermal management solution), component sizing (0805 resistor is rated 0.125W for example so its good to know if you're above that number).

so hopefully there's someone out there that read this and is interested to get involved. don't worry if you're new to analysis, take a part. jump in. PM me if you want to talk out the analysis first.

as members provide analysis we can post the results here for discussion and then collect the results on the first couple of messages to go along with the design. it could be interesting.

without the analysis, the design is a shot in the dark :)
 
Control power supply is done but not posted yet. It's a simple fly back based on power integrations chip. 12v, 10W. From here it goes 12v to 3.3v . I just blue sky estimated 10W, I will go back during detail design phase to work out the actual (and tweak design as needed).

Now I am working on combining the digital control into one DSP. Right now the software is done, but it requires two DSP based on how the code was written. But DSP has more than enough horse power and oithink enough ADC and PWM channels to manage as a single uC. Course this means @zombiess (software champ!) will actually have to do something. By software done I mean everything except charge profile and specific safety feature related to application such as charger shall not run if car is In motion or what ever. Actually that's a good activity for u zombiess, start brain storming all the interface functions needed. That way i can make sure we have enough GPIO. We can decide soon if the control power supply also needs an external +5 V dedicated to external sensor or otherwise power. I don't like using local power supply to manage external things, I keep the two things Isolated.

OK, enough rambling for now I have to go back to work
 
Bugger… I was all set to post of the first complete draft this morning. But where did all my schematic updates go to? Did I somehow forgot to save it? Grrrr

Now I Don't even remember what I changed. Lol.
 
oh sweet.. found the original software used by TI to combine into one controller.
i am currently working on analog feedback path. i am putting some effort into output voltage measurement cause battery are sensitive to this so i want good quality measurement. then i will work on adding some communication so feedback can come from BMS. then revise the 100 pin connector's pin-out (make sure i have mapped signals according to where software is expecting them to be) then the architecture will be done and i can post next revision.

there is list of features that is starting to grow that i will need to add, application specific. such as:
-add LCD display (show battery health, state of charge, health, time, power usage, temperature)
-some relay driver so i can isolate
-need also some way to avoid having a big spark when you connect charger to car battery, any idea how to do that?

what else???
 
revised schematic today (see post #1) and also added a pinout summary (see post #2).
i decided not to post the software links as its not really necessary. software will be build like this,
PFC control will be based on TI Controlsuite: C:\ti\controlSUITE\development_kits\ILPFC_v1.0
PSFB control will be based on TI Controlsuite: C:\ti\controlSUITE\development_kits\HVPSFB_v1.1

of course the trick is to combine these into one DSP as was done in TI software ACDC (see TI website, search for sprc711)

there are some slight differences in my design versus what the software is expecting. i need to go in detail of the software to make sure its OK. it would be great if the "software person" did this task as it is more of an overview of the software flow and make sure the signals are available that are needed. make sure current sensors are located in proper spot. that the sensor location support constant voltage and constant current. this will take me a while.

my next task is to add the interface for BMS communication & LCD. need to also add some other things like contactors for isolation and a few others. i made all sorts of notes as shown on the draft schematics, i sort of record my thoughts right on the schematic. so there will be at least one more draft "archetecture" schematic before it goes for review. then after revised on to the detail analysis (i.e. pick out exact resistor, cap values (especially important for control loop stability) and analysis for power dissipation, stress test, monte-carlo variability etc.)

@zombiess, haven't forgotten about the caps. :!:
 
Exciting to see it come together! A design from the guru himself. :)
 
Each battery chemistry has its own preferred charge profile but they all follow about the same thing. The battery is charged in four phases: trickle charge, precharge, constant current, and constant voltage so the EV charger algorithm has to be able to produce each state.

the first 3 modes are variations on current control. so the output voltage is variable while the current is controlled. this makes sense because we want to monitor the voltage as the battery gets charged so we know when it is fully charged (probably the BMS will also indicate when battery is fully charged too). so the "variable voltage" output, actually this is the climbing battery voltage. this is all well and good, but now you have to look at your algorithm with respect to where the current sensor is located (transformer primary) and make sure you can achieve the fidelity and stability needed.

once battery is at its proper voltage, then charger switches over to constant voltage mode.

here is a charge profile for a Li-ion (i plan to use LifeP04 or whatever else is out there by the time i finish the project who knows!?). Blue is current, red is voltage. there are some differences when dealing with large traction packs because you might have a damaged cell or whatever but its basically like this.

TI_ChargeProfile.jpg
 
hmm.. how come some attached jpg image show in the post and some are click to view?
 
There are many charge profiles, even some with 5 or 6 stages. In practice, if youre charging at a reasonable rate for the cells anode at that temp, the battery doesn't seem to care what your charge profile looks like.

To the cells, all the current ripple you want to feed it from roughly chopped up DC spikes or whatever is fine, a handfull of studys indicate it may be an extremely minor lifecycle benefit. If you have just CC to some CV upper limit with ugly current ripple, this tends to offer excellent cell longevity. Did the profile you have come from A123? It may well be iron phosphates have some reason for recommending it, I don't have much experience with them. Typically a many stage profile is only needed if looking to push the envelope of riding the safe anode ion acceptance rate of the anode with minimal margin. Is this charger for a small pack that it can push to beyond max constant charge C-rate?
 
Back
Top