Yes its same for both.
I have done a lot of testing on 12S and had no issues, I did find that you have to rerun the motor detection when you change the voltage.
What motor are you using?
Thanks for your reply.
I'm using a generic "1500W" hub motor from ebay. I'll try to attach some images.
It runs good on 4.12 hardware at 10s, 100A battery & 20" wheels. (yes the cooling solution is heavily modified)
Here's a video where you can hear the motor at different currents.
Usually it wont even rotate at 60A, just sits still making all kinds of horrible noises..
Just as a test i tried 100A motor, 120max (10s) and it almost instantly throws an over-current fault code. It's barely able to move on flat ground at those settings.
Could there be a problem with the current sens filtering?
can you send to me the schematic for this version (V2.1)?
Thank you so much.
I think(!) this is the schematic for my version. PCB looks the same at least.
EDIT: Changed the drive link, had the folder restricted
With the latest firmware (5.01?) I've had some odd behaviour with a large outrunner. There seems to be a specific rpm at low speed where sensorless foc completely loses the rotor. I have an encoder so I just make sure that the sensorless rpm is well beyond this point.
I think it is related to recent changes to observer gain scaling.
Dang I must have put something together wrong on the power board of mt A200S v2. About 10 seconds into FOC motor detection and POW, it blew 2 Mosfets of the PCB.
I am building some board at the moment to let me test out the latest firmware in a bunch of situations and see if maybe some tweaks are required.
Nice work on your controller designs. Amazing work.
I've been into high power ebikes for a long time and used lyen infinion controllers back in the day with IRFB 4115 fets at 125v and 75amps. It was awsome on the 12 fet designs and I never had any issues so I was complacent and just stopped researching power systems for ebikes.
Recently, i've rejoined the world of developement and was blown away with what I saw with the vesc project. So I decided to make my own VESC hardware. My hardware is based off the A200s, BESC, VESC, Cheap FOCer and Axiom. I mish mash of ideas and levels of features and safety control. Similar to Axiom and your design I wanted to seperate the control and drive stages. I have done this and so far have already made the control board. Sent out the PCBS and soldered on my components. Everything for the most part seems to be working correctly with stock HW60 FW5.1 But I haven't yet tested the drv side of things.
I had a question about the a200 firmware. I noticed that with your design as well as the besc design. You do not bring communication signal lines to your gate drivers as this is not nessessary for the drivers that you guys are using. My question is, what kind of changes did you need to make to the firmware from VESC? over all what did you need to change outside of current measurement risistor values and input voltage risistor values etc?
Any support is greatly appreciated. I have no intention to commercialize this stuff in any way it is all for learning and if anyone wants my design they are welcome to them .
You basically just need to tell vesc what the hardware is with defines in your config file and your off, its surprisingly simple.
If you get stuck feel free to send me a PM and I will see what I can do to help.
I am working on the V3 design now, was deciding on mosfets to use as I really hate the DirectFets with a passion now.
Here are my findings!
The On Semi (imagine it doesn't say Toshiba in the pic, I got confused!) parts are SO much better its crazy. Unfortunately the 80v one is not out yet, so initial testing will be done with the 60v version. I have basically fininshed the layout of the power board, now moving on to the logic board. Main goal here was to make the heatsinking easy by putting everything on the bottom. While making the board smaller and perform better.
Tech, What has you hating the directfets? Back in the day I use to use a lot of the IRFB 4110 4115 which are dogs these days but the Hexfets were awesome for their time.
Sent you a PM. It be great to connect. vesc hw colaborators are far and few. It be great to have a discussion . Would love your take on what I've been working on as well.
whats about this 75V type: TPH2R608NHL ? Low gate charge 28nC, Rdson is okay 2,1mOhm
I hate DirectFets for a couple of reasons, but the largest one is that you can't inspect them. Most of the solder joints are under the package. Sure if you're doing a massive run of FETs and time to dial in the reflow profile perfectly and an XRay inspection setup to make sure you don't have any voids. It also sucks that you can't hand solder them, you can use hot air but nearly all FET packages can be hand soldered, why make life hard. They've also been awkward to heatsink as the outer case carries current, I know some boards have used that to their advantage but it's more hassle than it's worth.
The main problem is as TheFallen said, inspection and relibability. If you can't see the joint you have no idea without an xray what is actually going on. Also for the price they kinda suck. Sure you can heatsink them easily with the big metal top surface, but the resistance is so high you HAVE to heatsink them or they will melt. With these new fets they manage like 3x the current on a 1in pad, so cooling will be much less of a problem due to the much lower resistance. (Also the 80v ones are Dual cool package, which has an exposed thermal pad on the top!)
They also have much better Cgs / Cgd, which makes them more resistant to self turn on.
Spent a bit more time to do the placement and some layout on the logic board, getting there now. Should be ready to order first prototype at the weekend.
What is the dimentions on this new V3 so far ?
Currently about 91x40x20mm without a heatsink.
Here is a little size comparison between different models of VESC.
Finished the logic board now, so will be ordering first prototypes tomorrow.
Excellent news! Looking forward to seeing how the prototype progresses.
Looking forward to the ne V3. I bought a few of the v2 bare boards and finished logic boards and had issues with the DirectFets too. Still never got a fully working unit even with my reflow oven seeming to work well. Soon as your ready I would like to try the V3 setup please.
With some help from TheFallen we now have a nice spreadsheet with specs of a bunch of different mosfets.
Tonight I worked on the mechanical design of the heatsink and case.
The heatsink is super simple and no weird stuff so it can be easily machined and hopefully cheap! A big thermal pad goes in and contacts the majority of the parts on the bottom of the board.
Then power board goes on top
3d printed spacer and support
Logic board then slots into the hole
And finally the top cover goes on to keep dust and stuff out and hold the logic board down. 6 M2.5 bolts hold it all together.
4x 4.5mm holes for mounting.
Awesome ..., waiting eagerly for protos
very nice Projekt! I am Building a High Power Ebike and Looking for a small Powerfull controller. I would like to use your V3 because it will fit into my frame. When do you think the Controller could be ready?
Count me in to buy one of the first versions please.
Today the parts arrived! Woooo
I have assembled the power board now. Will do the logic tomorrow.
On the mechanical side, the MJF cases I ordered arrived and look sweeeet!
This is how I think the cables will likely go, up to 4x 12awg XT60 input, 8awg output
Sooo tiny. Only two sets of shunts?
They are low side shunts, so shouldn't see as much current, I can switch them to copper ones if needs be. Or maybe lower value as currently they have higher resistance than the fets!
Only 2 sets as for low side shunts, the 3rd current can be calculated quite well. I have the same arrangement on the A100S and it works fine (better than phase shunts IMO)
Dont you see any problems with using 2 shunts when you are running at high duty cycle (90-99%)? At high duty cycle the switching can be too close to the timing of the adc measurement, which can either lead to noisy measurements or faulty measurements in case of too much filtering (the signal will never reach its actual value in time). Do you have an RC filter on the current measurement?
In vesc firmware only 2 current measurements are used. If 3 current measurements are available, then the 2 measurements on the phases with the lowest modulation will be used, this will allow up to 100% duty cycle (only limited by gate bootstrap).
As I see it these are the advantages and disadvantages of using phase shunts compared to low side shunts:
- Possibility for harder filtering because current is always flowing in the shunt (not dependent on dutycycle/modulation). This is why 1kohm + 15nF (cutoff @10.6kHz) works on vesc 6 and 75300.
- Possibility to sample current twice as fast (compared to low side shunt) -> better control at high speeds.
- Double losses in shunts as current is always flowing in the shunt (compared to half the time with low side shunt).
- Higher price.
Do you have any other reasons why you think low side shunts are better?
I haven't seen those issues when running my smaller A100S with 2 low side shunts. If anything its significantly less noisy than phase shunts even with no filtering. You don't have the common mode pwm to get filtered out.
Main reasoning for low side is it allows an extremely compact layout, and produces less heat than phase shunts.
Price is another issue. For high voltage shunt amplifiers you have non isolated ones up to 80v which are not cheap. Then after that you need isolated and it adds a bunch of extra parts to the bom, all adding cost.
Aim of this hardware is to make it crazy compact and as cheap as possible (within reason).
Got the pcbs all soldered up and managed to test so far current handling in DC mode and spin a motor in BLDC mode. FOC not working correctly as I made mistake on logic board and connected one of the sides of the shunts to ground, need to add mod wires to fix it tomorrow.
So far I have tried running at 165A, it managed about 5s before the tracks got too hot. Which is surprisingly long time given its a 1oz/0.5oz board! I have designed 8 layer board now with 4oz copper on all layers. The mosfets didn't really change temperature much.
Assembled the logic board, IMU works, BLE took a day to figure out why the inbuilt blackmagic programmer didn't recognise it. Turned out I needed to add the ID code for the processor to some list so it knew what type it was. You might notice the dual gate drivers, planning to test how that goes tomorrow (so far only tried with 1 driver fitted)
This is the rise and fall time with 1 fitted and 0R gate resistor
3d printed rubber spacer holds it in place and helps seal the unit against ingress of dirt.
I ordered heatsinks, which are now on the way (might arrive early next week)
Finally, I have been trying out casting the silicone rubber wire guard with a 3d printed mold. Seems to work quite well. So will order a resin printed one to get it nice and smooth.
Looks very promising! What's the 3D printed rubber material you used for the spacer, TPU?
I sent you an e-mail by the way, maybe you could have a look.
The material is a bit like TPU, https://www.3dprint-uk.co.uk/flexible-rubber-like-material-for-3d-printe...
Tested the dual driver setup today, seems to work perfectly and halves the switching speed compared to using 1 driver. Deadtime is now at 500ns which is very respectable (VESC6 is 360ns, 75/300 is 660ns).
Moved on to testing the current measurement in DC mode now I have added mod wires to get round the shorted vias on the logic board. Works fine now.
Then started looking at FOC,
The board detects my 150cc rotomax (with a tweak to the detection function that improves detection on all hardware) at 5mOhm, 1.8uH so good test motor.
I can run foc_openloop at 100A with no problems, and motor spins perfectly after detection :D
Now I am happy enough to go finish off the pcbs for the proper boards and get them ordered. Hopefully they don't take too long to arrive, but they are 8 layers with 3/4/4/4/4/4/4/3oz copper so probably like 2 weeks or something. Then we can see what it can REALLY do.
In other news, the heatsinks arrived and look gorgeous
Getting better with the silicone part also, added some little tabs to hold it in place, although I think it needs to be a bit thicker where the wires pass through so it can't pull away so easily.
Looks very nice. Extreme compactness!
What are the dual drivers? Two drivers in parallel?
Regarding the dead time, interestingly, I saw no difference tweaking it in the firmware. I Set it originally to 660. Controller worked fine. Then I dropped it to 360. It still works fine.
How do you validate the proper deadtime value? My gate drivers have 19ns propagation delay and mosfets have 90ns gate . ON gate resistance is 12.2mOhm (2.2 + 10) and Off gate is goes through a schottky diode and 4.4mOhm total resistance (2.2 + 2.2). Gate drivers are hardware set to 100ns minimum deadtime.
Yeah its two FD6288Q in parallel, which gives a super compact 3 phase, 4A ish driver for only about $0.6. https://lcsc.com/product-detail/Others_Fortior-Tech-FD6288Q_C328453.html
Pretty sure you could actually run more in parallel to get higher currents as well.
The 10v supply is a chargepump from the 5v, and they can run in parallel also.
To work out the proper deadtime value I set it super high (5us) then measure time taken to turn off and turn on to the threshold voltage then take turn on time away from turn off time (so you start turning on before its actually finished discharging the remaining charge). In this case of 500ns I was being a bit conservative and could set it much lower if I wanted without shoot through.
Paralleling looks interesting. Do you parallel both inputs and outputs of the gate drivers? I wonder if some rise/fall time slight mismatch between two drivers may affect the signal.
Thanks for the tip on DT calculations.
I also found this source about DT calculations from Infineon pretty informative:
Heatsinks really look good. CNCelos seems to be based in Portugal, how come?
It probably has some odd affect / glitch as the outputs are not perfectly matched, but there is so much capacitance on the gates it doesn't really matter. Would be more of a problem with lower gate charge mosfets I think.
Yeah Portugal, I was originally going to get them anodised black, but after seeing them I quite like the contrast of the black case and heatsink. (also cheaper :D)
Didn't know they was in Portugal until they shipped them!
I like your new design. I am looking for 140v300a continues VESC based controller. Are going to make it a bit more powerful?
That won't be possible in this form factor unfortunately, the max is probably like 100v / 150A or something.
These are the currently avaliable fets in this package >60 (most are not released yet)
Just ordered the proper power board PCBs, apparently 3mm, tg170, 8 layer, with 4oz copper is not cheap, who would have thought!!
Yess! Looking forward to it.
I have a general question, I found the documentation for the V1x versions of the A200S on github, but nothing more recent. Is the documentation of the more recent boards open source somewhere as well?
Yeah its all on my circuitmaker https://circuitmaker.com/User/Projects/Euan-Mutch
Great news that it may be able to handle 100v, that's a huge deal for me personally and many others I'm sure. And an impressive feat in such a small footprint.
Thanks for posting these updates! Will look forward to when the new PCBs arrive.
Thanks for the link to the circuit layout!
Looks like the v3 will fit my application if I can mount the logic board separately and use a cable to connect the two via the pico-clasp 20-circuit connectors. Can you foresee any timing issues or otherwise with changing the circuit length in this way?
Not sure how it will handle moving the logic board away, would probably still work if you could shield the cable.
The interboard connector is this one https://www.mouser.co.uk/ProductDetail/Samtec/LSHM-110-030-L-DV-A-N-K-TR...
Same on both sides as its hermaphroditic. Maybe someone already makes a cable that can connect, or could maybe do a pcb that let's you attach some off the shelf shielded cable?
Edit: samtec makes cables for this. https://www.samtec.com/products/hlcd#
Thanks, that's very considerate of you to put some thought into this, and recommend the cable supplier!
Good to hear you think it may be possible with a shielded cable. Would just be mounted alongside the power board, so not too far away, but near the stator, so will have to see if it's too noisy.
All the best!