Hi guys,
I'm working on higher voltage ESC to replace my older 100V (500A) ebike controller as there is plans to change my 24s (a123 20ah Lifepo4) battery pack to 24s Li-ion battery pack. Project details so far:
- 18 x IPB044N15N5 150V 174A 4.4mOhm fets
- 6 x 680uF 160V EKXJ161ELL681MM50S caps
- 72 x 10uF 25V X7R ceramic caps combined to withstand 200V (total capacitance 11.25uF)
- 120V (with possibility to increase up to 130V with some later tests)
- 3 separate boards (control, driver, power)
- Insulated Metal Substrate (IMS) for power board for better heat transfer
- Aluminum CNC'ed case
- 8 AWG power cables
Progress so far is traced PCB boards which are ready to be manufactured, just waiting for all the components to see if everything fits and there is no problems with their patterns. As this is just a hobby progress is just a little bit slow but I'm getting there, so will update later with all changes, costs and photos of the project.
See you sooner or later,
Martin.
Nice project, same schematic as BESC or using different drivers and current sense?
Thanks. Haven't seen BESC schematic, but I'm using isolated current sense amplifiers (SI8920) and isolated power supply for them and for high side driver (FAN7391) supply.
Boards have arrived. Had some time to assemble them and do some testing. Already noticed some strange behavior in current measurement. Will do some more testing shortly. Here are some photos of recent progress:
P.S. Sharp eye's owners could notice shorted FET gate pins in power PCB, it's fixed.
Aren't your ceramic caps a bit far away from the fets?
For your current measurement, I had an issue with hand soldering those shunt resistors that if you got any solder on the top or basically anywhere but the interface to the pcb then they had incorrect measurements. Sometimes off by quite a lot.
Also how is the current sense tracks routed? Looks like you have SH+ on one side and SH- on the other? Remember that those tracks are a differential pair, so if you don't need them parallel they will get different interference currents across them and your measurement will be wrong.
This is how I did mine,
Hi TechAUmNu,
Can't answer for this moment, will do some testing and measurements in a short time. In previous version even don't used ceramic caps (just a few wima mkp10 0.1uF) and traces was even longer, haven't had any big issues back then. Have traced beefy current lines between caps and FETS, so inductance mus be minimal. It is hard to put ceramic caps closely to the FETS when using IMS.
Yeah those shunts are sensitive for soldering them right, had some issues in previous versions, but it's easy to measure their real values accurately and do some corrections.
I think that the problem is somewhere else with current measurement, maybe in to small signal filtering or maybe have to add coaxial cable for current signals between boards. Other case could be new opamps (for conversion from diff to single ended signal) that I have changed from my last ESC version, will try swapping the old ones back. Of course there always could be some errors in schematic.
If it helps to understand my problem, have noticed that battery current is negative and motor current is positive. Spinning motor to the same direction it sometimes spins to one direction and sometimes to the other (looks it doesn't know in which direction he is spinning) phase voltage measurement related?!
Will do deeper analysis and let you know the results, in that time if you have any ideas that could help let me know!
Thanks, Martin.
Found a problem, voltage reference (for current measurement) op-amp was ringing and was messing current readings, changed it to other model and fixed the problem. Now working on decreasing step down converter noise on VCC, it could be the case for op-amp ringing.
May I ask you what motor you are using?
I used to run 500a Kelly controller but on 48V and was accelerating in line with 250cc-400cc motorcycles.
Hi,
I'm using QS 205 50H V3 T3 motor. Pushing to it 200 battery amps (BMS shows 225A) and 450 phase amps at 82V. It's a rocket ship, can do wheelies from 40 km/h and race some motorcycles too. Max speed that I have reached was 95 km/h.
Martin.
Hi Martin,
That's scary.
What is the wheel od?
What is motor turn number?
What battery setup feeds this?
Thanks
Amiran
I'm using 18 inch 125cc motorcycle wheel.
Motor is 11Kv (3 winding turns - 3T):
http://www.cnqsmotor.com/en/article_read/QS%20Motor%20205%2050H%20V3%20Motor%203000W%20Electric%20High%20Power%20Bike%20Spoke%20Hub%20Motor/259.html
Battery used is a123 20ah lifepo4 prismatic cells (capable of 500+ A) connected in 24s configuration: https://www.buya123products.com/uploads/vipcase/468623916e3ecc5b8a5f3d20825eb98d.pdf
Hi Martin,
May you send your schematic to me?
Thank you so much.
Quang
Hi nami,
Schematic is not debugged fully it's not in final revision. Have put a lot of effort in this version so I'll think about sharing schematic.
Back to ESC. Low inductance (low resistance as well) motors doesn't go hand in hand with high voltages. Have done some testing with different voltages and ESC is working very well up to 80V with 920kV motor.
Tried 120V - motor spun up, lost sync and bam - have FETs blown on two phases. Have done some calculations and it seems that have exceeded FETs pulsed drain current. Is it possible to that ESC haven't managed to control the current then it lost sync?
Need to determinate minimum requirements for the motors to fit with this ESC.
Work in progress!!
Martin.
You may not want to do that with your caps.
The series string length will add a relatively large amount of inductance to the capacitors and the lack of balancing resistors can lead to voltage inbalances in the capacitors leading to premature failure.
Consider using flim capacitors, attaching them to the power stage board and routing cutouts for them in the above mounted control board instead
Hi bgdwiepp,
Have thought about using film capacitors, but they are way to large for slim design that I'm aiming for. Not counting them out just yet, maybe will figure out how to fit them as I'm already redesigning driver board.
Yeah, have changed all these caps just to two stacked caps in series, added a tvs diodes near them and three more pairs at the end off the board (will try to shorten traces in new board):
Thank's for good advices, always waiting to hear more.
Martin.
I think your effective capacitance at your supply voltage would be quite low with ceramics - you wouldnt need very big film capacitors at all to exceed that. Something like https://www.digikey.com.au/product-detail/en/kemet/C4AQLBU5100A1XK/399-1... across the supply going to each phase would be significantly more capacitance at a similar esr/esl to the ceramics, they could be mounted in a similar fashion to your electrolytics.
I know these may not be viable for your form factor, but you can get much much better results for current sensing with hall effect current sensors, this will reduce the need for the isolated dc modules and get you better frequency response. There are other added benefits too, some of these have a fast overcurrent output you can use to control your fet drivers for hardware overcurrent protection.
For higher efficiency you should use a fet driver with a lower propagation delay, better delay matching and more output drive, this will allow a smaller deadtime and thus lower reverse recovery losses in your fets, one example driver is ucc21540 from TI, silabs also has a similar part albeit not as good/fast.
If you drop the isolated power supply and use a bootstrap solution for high side drive, look very closely at the diode capacitance and reverse recovery charge, at higher frequencies and voltages this can lead to massive losses.
Also, at these voltages, remember clearance distances, iirc over 150v at standard atmospheric conditions you need about 50mil of space between traces on external layers, i am not sure you have that - check out the saturn pcb toolkit for more info.
Thanks for advises,
Have changed MOSFET drivers to UCC21520A as you suggested for better propagation delay and as they have shutdown pin have too added overvoltage protection with really fast response time - 80ns (driver delay 30ns + comparator delay 50ns). UCC21520A pinout is same as UCC21540, so can change it to later if needed, but it's more expensive. Also it's good to have hardware dead time in case of wrong software implementation.
As for the rest design I'm sticking that was done before, maybe adding a few 1uF film capacitors as there is enough space for them. Have calculated that at 120V ceramic caps capacitance should be about 33uF.
Hall effect current sensors is really beefy and very expensive so have considered them out really long time ago. Could be possible to mount simple hall sensors on phase rail and measure current that way, but then they have to calibrated each time and from practice I can say that it is hard to isolate them from external noise sources.
Minimum clearance between high and low voltage traces is minimum 50 mils in power board and 80 mils in driver board.
Planing too order new revision boards soon.
Martin.
Hi Martin.
Thumbs up for your project! If this project goes through, you will have something quite unique on the market.
I am building go karts, and we need a high voltage high amp vesc 🙂 looking forward following your project
Hi,
Project looks great. Have you got the design debugged and working?
Thanks so much!
Hi
great job, you finally got a stable operation of the board
my congratulations on the project