This is the A200S which I have been working on for the last year.
Schematic and BOM: https://github.com/TechAUmNu/A200S
- 18 FET design
- Weight ~800g
- Dimensions 110x105x30mm
- Waterjet 3mm copper bus bars
- Waterjet 10mm thick aluminium heatsinks (Stackable)
- 8AWG cables for input and output, with choice of 8mm bullets or lugs.
- 2x 100v 1000uF High-Ripple United Chemi-Con Capacitors
- 15-75v rating, 16s LiPo maximum
- 200A (13.4kW@16s) continuous for a few minutes
- 300A (20kW@16s) burst for a few seconds
This was after about 30 minutes of absolutely thrashing it. The load was about 9.5kW at 12S. Its quite cool to watch the wire move as the current increases . The beeping is our battery monitor telling us the voltage had dropped below 3.8V from 4.2V so probably need another pack or two in parallel.
They are designed to stack like this:
Nice, how did you go about attaching the water jet copper to the PCB? I tried something similar but had trouble with warping and getting them to solder well.
I just put a bit of solder on each corner of the copper and push them down with my other hand then remove the heat. Warp on the copper is only 0.2mm or so from end to end, so its not a big problem.
Outstanding work, I have been waiting for it!
Will you post schematic,and PCB files?
Will you sell the pcbs?
Best regards, and once again congratulations!
Thanks, its been a long journey of learning!
I will post schematic in the next few days once I have updated it to include all my latest changes.
It was designed in Altium Circuit Maker, so I will release the pcb files on there and post a link to the project since you can't save the files externally to their cloud.
I am selling complete units for £400 and pcbs for £10. Preorder is being placed on the 5th march so you can pm me if you want to get pcbs or a complete one from that batch.
I happen to also be working on a 18 FET 75V 300A design :-) I will share some details if/when I get the parts and get everything running properly.
One thing to consider is wether having 4 electrolytic capacitors externally really helps. I made some analysis and simulations about that a while ago. There is a link in the video description to my ltspice files and SVM simulator if you want to play around with it.
I actually spent quite a long time playing around with that, its true that it doesn't make that much difference, but if people have really long battery wires it does provide at least some protection. The main thing I did was add 6x 4.7uF ceramics right on top of the fets, basically the same as the vesc 6. I wanted to use 10uF ones but they are still not available in 1210. https://www.mouser.co.uk/ProductDetail/Murata-Electronics/GRM32EC72A106K...
The derating becomes so annoying on those high voltage ceramic caps if you dig through the manufacturers website long enough you usually find out they lose 60% of the capacitance anywhere near their rated voltage, if it doesn't blow fets it's working well enough though. I found a few interesting SMT film capacitors a while ago in the 100V range and have no derating at max voltage. The only real draw back was larger size and dificult to solder.
mark, waiting for further progress.
I tried this and they got 9uF at 0V and 1.8uF at 50V. Not bad. https://www.aliexpress.com/item/Chip-capacitors-1206-10UF-106K-100V-X7R-...
It seems quite happy even at 18S :D
This test was very scary since I modified the pulleys myself and the centre hole is not quite right. So the whole table was vibrating across the floor!
The first batch of 20 units and 100 pcbs went in last night. Will be back in April.
Have now added github link for schematic and bom.
I've been following the development of this for awhile now but I am unsure how to order one. Can someone please help me find where I can order a couple?
Hey was just reading through the thread, you are who I would place the order for a complete A200S? Do you have dual enclosures by chance? I'm looking to build a 4WD electric longboard and the motors I want to use need more power than the VESC 6, so I very interested in your product. Please let me know if you do dual enclosures and any other information about your products. Thanks so much for your help.
I would remove your contact details from the last post uigiroux. I have sent you an email about ordering.
The A200S is available to order from me directly (Please send me a PM). There are a few units left from the initial batch at £400 + shipping. Future batches will be £450 + shipping.
I would be interested in this, but I'm curious, this can handle 300 battery amps at 16s max, but what about motor amps? The vesc's so far have all been limited to 120 motor amps. Does this still have that limitation?
I am unsure of the max motor amps before it explodes but it can measure up to about 500A. However it will take a very short time to reach the temperature cutoff so it won't be practical to use over 200A for any length of time. When the first batch comes back I will run a temperature equilibrium test to see how much continuous current it can handle.
The boards have finished assembly by the fab and are now on their way to me!
nice desing, I am interested in how much of BOM cost does that thick copper make.
Impressive design. Those FET's look monstrous! Can't wait to see this thing in action! Cheers!
That controller looks great. Congratulations. I saw your video also testing 18s. So do you think 75,6V are now also save and tested or would you still remain the 16s limit stated in your first post? I think if 18s are possible this controller could be used in a lot more powerful ebikes that had a problem in the past with the 12s limit from current VESCs.
A couple of people are going to be using them at 18S on ebikes. I have done a decent amount of testing at 12S and some at 18S. During my testing yesterday I put the wrong settings on the board and managed to blow it up on the switch from Hall sensors to sensorless. Also had 0A regen which was stupid. So on the switchover it probably slammed the motor to a stop causing arcs across the current shunt amplifiers and then tracked around the board to the mosfets. The spike was over 120v as the gate drivers also exploded. It was doing 45% duty 164A @ 18S when it hit the magic 8000 ERPM, it was quite spectacular! I will double check everything in the configuration this time before I go through all my spares.
So my advice at the moment is not to use Hall sensors! As the switchover is extremely violent. Also 18S is pushing it so doing anything even slightly dodgy is big no no.
My advice is not to use these FETS with higher than 80V voltage. Had to put a lot of work hours by replacing then every time they blowed up. Then battery pack was on its working voltage ~80V everything worked perfectly (220 battery amps and 450 phase amps), but every time the pack was charged they blowed up, even enormous electrolytic, PET and ceramic cap count not helped (~10000uF). Switched to 120V FETS and all problems is gone. Btw most of the times damage was the same as yours, blowed FETS, MOSFET driver and some tracing gone (and in some cases plasma arc). So bare in mind that it's really dangerous to use them above 80V.
What voltage was your pack when charged?
24s LifePo4 so around 86.4V.
I took a break today from testing to get my head on properly. So spent the night learning to stick weld instead! Its much easier than I thought.
Great work sir!
You had hall sensor switch over at 8k ERPM? I always set it to the lowest RPM sensorless starts tracking well.
That sounds like a much better idea! Thanks
The ina240 can only handle 80V as well, you might find they're failing and damaging the MCU which in turn shorts the mosfets or destroys the gate driver as it dies. I'm not entirely sure how much abuse the ina240 will take or if its prone to sending the entire phase voltage into the MCU if it fails. It might be worth considering isolated current sense and gate drivers due to the large currents and being so close to the voltage limits of non-isolated components, you can protect the phase and dc voltage sense without full isolation using TVS diodes.
Yeah 18s is really pushing what it can do. Hence the 16S lable. I think the people running ebikes at 18S will need to current limit to something sensible like 120-140A. The next rev can use all isolated components as it needs to reach at least 28S.
Which isolated current sense were you thinking of using? Ti and silicon labs have isolated shunt based current sense but the output is differential and needs opamps to convert it to single output. The biggest headache I found with 100V+ is trying to get a power regulator for the low voltage supply, all the buck converters usually max out at 100v-150v and the higher voltage options usually have 80v minimum input.
Probably using the TI AMC1301, still need to find some good fets and drivers. As for power supply I think a separate battery is the best solution. As it removes any complex step down circuitry from the board reducing cost.
Currently using Si8920 isolating op amps, as they are a little bit faster, cheaper and have two ranges of amplification (x16.2 and x8.1). Have a look at them.
Those look great! The driver is quite nice too.
Found a couple of fets. Depends mostly on if it will be a massive board of them or just a few in a compact unit. Cooling will certainly be easier for a big board but weight and size is a problem for certain applications. At the moment I am thinking of doing 150-180V at 100-150A to keep the board size down and not need too many capacitors.
What op-amp setup do you use to convert its differential to single ended 0-3.3v? I built a Si8920 board ages ago but my opamp setup was a annoying in terms of BoM and footprint.
Pity the 405 doesn't support differential signals. Can probably get a small triple/quad opamp chip to do it.
Thinking of using some INA826 to convert it to single ended.
There's a converter in some of the AMC1301 reference designs, I'm not sure if the same circuit works with the Si8920 as the voltage output is different? Why Ti didn't just make the AMC1301 with single ended 3.3v output in the first place I really don't know. The higher end MCUs with differential analogue inputs usually all have sigma delta modulator support removing the need for the AMC1301 in the first place.
My design is exactly the same as PaltaTech 150kW diff to single-ended design, only using different OpaAmp (LMV751) and conversion ratio is 1:1 (using 10k resistors). Have to add 10pF cap parallel to feedback resistor because there was some small ringing (nothing serious). Everything works perfectly, within 2-4% error around 200 battery amps. Hope it helps a little bit.
Been doing some more testing today using 12S, seems that if I run at high current it loses tracking at about 7300 ERPM with about 120A . Which causes the motor to stop suddenly. This is probably what caused the exploded fets in combination with the hall sensors. I am unsure why it is doing this? It has no problem running up to 95% duty cycle without the load on it. So must be current related. Like motor saturation or something. I tried setting the stator saturation compensation to 10% and 15% with no effect. Any ideas??
Update: So I decided to run foc detection again, this time with the load attached to measure the R and L values. The resistance stayed the same and inductance went down quite a lot. Now when I run the motor it spins up to the same RPM before stopping, but this time with an ABS_OVERCURRENT fault.
Turning the abs current limit up to 300A now allows the motor to spin to about 8500 ERPM before tracking is lost again.
By increasing the stator saturation to 20% I can push the motor current to 160A before it fails, but the ERPM is still the same so that really doesn't help.
If I try the same test on BLDC then it fails at the same 8500 ERPM.
Beginning to think its just the motor is really bad at high current lol. Annoyingly I don't have any other large motors that can be attached to my test rig.
Even weirder. If i switch to just using Hall sensors and set the switch erpm to 100000 basically always on. Then it fails at 140A. If I double the load to half the ERPM, it still fails at the same current! Weird.
Update: Same test in BLDC sensored mode fails at 2000ERPM so the above test doesn't really prove anything. I think I am happy to ship them out now. With some guidance on settings at 18S.
Main things are:
On 18S please don't go over 80A motor current unless you are 100% confident with your motor settings. If you really want to test this then go for it but make sure your motor runs PERFECTLY on 12S beforehand. Even at very high current and duty. Since there is no margin for error at 18S any mistake in settings could destroy the controller.
12S seems to be very safe, been slamming the motor to a stop and losing tracking at 150A with no ill effects.
Battery current reads half what it should read. So set your limit at 50% what you actually want for both battery current and battery regen.
Why is battery current incorrect? Sounds like there's something going wrong with current sense.
Are you using same burned PCB shown above? How does your current sampled data looks like?
Same design older board. With the same firmware and layout.
There are no filters on the current sense lines. Although I use a different amp than the vesc 6. Maybe i should try modding one with the filter. Cut off frequency of it was 300khz which seems a bit pointless.
I will put a scope on the current sense and see what it looks like.
It might be that the motor current is significantly higher than shown.
It might be the solder connecting to the shunt resistors is reducing their resistance quite a lot. I will try with a clean board with no extra solder.
You can try sample current in VESC tool under sampled data tab to see if your phase currents differs from each other.
To measure shunt resistance simply connect power supply which has current control, measure voltage drop at given current and calculate their resistance. Simple. From experience can say that added solder to the shunt resistor gives no more than 5-10% error, so in your case don't think that this is an issue, but you can give a try.
When I put a 3A limited power supply across the shunts only two seem to respond. Almost like its not running with 3 shunts.
Really need to see phase currents in vesc tool, it should be obvious if there's a problem there.
Well that was the issue thanks guys! It was reading almost exactly half the current on each phase!! Now runs great at 200A for ages even without the heatsink :D
Ran for a 30s and it went up by 20oC!
What was the cause? Wrong value in firmware?