A200S V2 Now available for pre-order here: http://teamtriforceuk.com/a200s-v2/
The A200S is a 200A 16S ESC, based upon the VESC®*-Project.
* VESC is a registered TM, owned by Benjamin Vedder
Printable View
A200S V2 Now available for pre-order here: http://teamtriforceuk.com/a200s-v2/
The A200S is a 200A 16S ESC, based upon the VESC®*-Project.
* VESC is a registered TM, owned by Benjamin Vedder
Features - cost less than a tenner. You'll be very popular ;)
A potted case is a REALLY bad idea! I tried that on a number of IFI Victors a few years ago and they all eventually died. It was traced back to the potting resin expanding at a different rate to the PCB and causing solder joints to crack. A thin conformal coating is a better bet. Room to add our own shock mounts is a good idea.
There's a couple of us trying to build 300A ESCs now.
Personally I'm not a fan of directFETs, can't inspect joints, package is unique to Infineon and they seem to discontinue them quickly. HSOF packages seem much better to me, it's what they use in the Vex BB controllers.
Not sure on the integrated shock mounts. Prefer some flexibility on mounting e.g. mount into a shock mounted electronics box
I'm looking to drive 8019 Neumotors - http://neumotors.com/80xx-series-electric-motors/
I will make the mounting points on the board with plenty of room around them instead, then it gives people mounting flexibility.
So about 350A peak on those motors, I might end up making another board with 6 or 8 fets per phase if there is demand. It really bumps the price up though.
I was thinking that too, only real reason I was going with them was for the ability to stick a heatsink directly onto them, but looking at the HSOF I am going to use this one instead.
http://www.mouser.com/ds/2/196/Infin...-en-767648.pdf
Has almost half the on resistance and bumps the current rating way up for a very small price increase.
Thanks!
Would people prefer it to be
A) Smaller with no 5v Dc-DC. i.e. OPTO
B) Larger with integrated 5v Dc-Dc about 2A (also makes it £10-20 more expensive)
I think its best to be OPTO since people will likely already have a 5v dc-dc and if they don't they can easily add one.
My vote would be OPTO
OPTO sounds sensible to me.
OPTO, but integrated diode on the output works quite well too.
If you're going opto, may as well go with a proper opto isolator to avoid joining ground planes through the RX; I've had signal issues when joining up non opto isolated high power ESCs.
Also, the HSOF packages ideally need thermal vias and to be heatsunk through the board. You can heatsink the case but it's not as efficient. Without a heatsink you'll likely get about 160A cont from 4 of those FETs, so to get to 300A you'll need to dump a lot of heat.
I have added an opto isolator for the reciever input.
For the current design it works out about 110W of dissipation, which should be fine. It will run quite hot 120oC at 300A continuous load. The heatsink is in contact with all power components on both sides of the pcb and some of the board itself next to the fets. Board is probably going to be 2-4 layer with 4oz - 6oz copper and copper bus bars for each phase and power input.
The heatsink is going to be around 1oC/W or better and will form a custom two part machined aluminium case.
Any preference on connector/wire size?
Thinking either 8mm or 10mm bullets on the board or short (<10cm) single 6AWG or dual 8AWG cable.
Was thinking it would be nice if they could bolt together in a stack and have a parallel bus bar that can run between them to cut down on cables. Then a single cable can connect them to the battery. (Or at least that is what I want for Triforce due to the limited space) This option would require to use bullets soldered to the pcb to be practical.
bullet connectors are handy or copper lugs like the big Kontronik ESCs:
http://www.kontronik.com/fileadmin/k...kte/kosmik.jpg
They are slower to work with but very reliable.
That looks really nice! Only problem with lugs is they are relatively heavy for the surface area, and you then have quite a long connection since the lug has to protrude from the case.
I do like the copper lugs
I mean: ESC +5V (+diode drop) -> diode -> external +5V. As found on some VESCs and Ragebridges.
I can't argue about the weight or size, but they are ultra reliable and can actually reduce surface area in other places if you do it right. Having the lugs laser cut to fit your PCB tracks means you can reduce their widths as the lugs are providing the current capacity from the FETs.
I am putting on laser cut copper plate a few mm thick for all the high current areas + 4oz copper on the pcb. Then the heatsink contacts with them through a thermal pad. Most of the heat will be dissipated through that. Which lets me put the FETs back to back and reduce the overall size of the board significantly.
Have now switched to some Fairchild fets instead and dropped the voltage to 80v. This lets me get significantly less power dissipation and hopefully higher currents. (psst They are also a LOT cheaper...)
http://www.mouser.co.uk/Search/Produ...-FDBL86361F085
Will be finishing up PCB this week and then getting a few made up for initial testing.
The specs on those new FETs look great and quantity prices of $4 and below are amazing for a high current FET - its looking very promising! Can you give us a hint on the likely PCB size yet?
Yeah they are pretty sweet. :D
I need to test out thermals first before i can settle on a board size, but it should end up roughly 60x100mm
Might be smaller if it can handle it.
Looking forward to this. FWIW, despite loving compactness as much as anyone, I'd rather have a big ESC that can handle this kind of demand for real (my weapon motor peaks over 300a, though VESC-level efficiency should rein that in). 20% bigger but 20% better thermals is totally worth it at this point. I'm very bored with ESCs that just can't.
Let me know if you need a test system that will happily murder ESCs on demand!
Ok I will make it a bit bigger then, would you object to a small fan to let me push it to the limit? Since these fets are also much much cheaper I could bump up to a 36 fet design again for a £45 increase in BOM cost to £150. That would handle 400A with a 100W dissipation on each phase. At 300A its 50W per phase.
Actually, though I don't have direct experience, fans (assuming fairly standard computer-style fans) always seem to break in combat. They can't handle the shock forces. I'd be inclined to avoid them if possible. The nature of modern compact machines is such that they'd probably be smothered to the point of being ineffective anyway.
In terms of power, it depends on realism. If the 300a setup can truly do 300a, as in, pulling anything up to 300a in a 3 minute fight absolutely cannot kill it, then that's fine. That'd give a weapon 15kw to play with on 12s, which combined with efficient control is plenty for a top heavyweight.
The issue is that the vast majority of hobby level ESCs simply can't do what they claim, and that's infinitely frustrating and getting old very fast. So far it seems the only products on the market that truly, reliably do what they claim come from MGM. The top Kontroniks maybe too. So we don't need another 3-400a esc that will blow at 200a for a few seconds - do whatever you need to do to get beyond "hobby amp" ratings within reasonable cost and you'll be very popular!
OK in that case I will make sure it can 100% handle 300A with no issues. At the 16s maximum it would be 20kW which should be plenty for any of the motors I have seen.
I have 3 of the Fatboy V2 boards which are frankly a joke in comparison to what I have so far. However we didn't manage to blow them up... yet.
Once I have got this board layout finished and got some in and tested to make sure they don't instantly explode I will probably send one to you and Rory to test out.
I 100% agree with Ellis about fans - they always break no matter what their construction or price point. I have tried plenty of modifications but it hardly helps.
Especially when you have 2 of these beasts cable tied to the escs... oops.
https://www.digikey.co.uk/product-de...623-ND/5230581
I have just about finished the pcb layout for my test power board, which I will just wire into the logic on one of the vescs I have laying around.
Attachment 6882Attachment 6883Attachment 6884
I will be getting some copper lasercut to run the high current parts and put a bunch of caps on as well. It is currently 100x100mm so I can get it done at www.seeedstudio.com who have a cheap 4 layer option. So I will have 10 boards for £38.
Will be putting in the order in the next few days so boards should arrive by the end of the month.
This is all getting extraordinarily interesting... keeping an eye on this one!
I have been thinking more about heat sinks the last few days.
From what I know, the maximum time that these would be expected to run at for power is under 5 minutes. So since there is going to be very little airflow inside of bot it makes more sense to use the heat sink as thermal energy storage (ie a big block of metal). I have worked out that it would take 200g of Aluminium 5 minutes to go from 50 - 175oC when the heat dissipated is 80W from the driver. If you also factor in some convection it should be even longer. This would make it possible to make each driver board encased in an aluminium shell which would hopefully provide better protection in combat.
You would just need to watch not to burn yourself on it after the fight lol.
Maybe a dumb idea, but have you thought about heat pipes to help distribute the heat across your aluminium shell?
I did think about that, the mosfets cover the entire surface area and *should* dissipate the same power each, so there shouldn't be any need for heat pipes. If the aluminium heat sink is very thick then heat transfer should not be a problem.
If it was a bunch of very thin fins like in a graphics card then heat pipes would be a necessity. Currently I have two 150x200mm heat sinks for testing which I will sandwidch the power board between. Link
Also going to get a Flir One http://www.flir.com/flirone/ so I can see what is getting hot.
I think I got my power dissipation calculations wrong. So I think it should run even better than expected. New numbers are
16s
15W @ 100A Max Power: 6720W
32W @ 200A Max Power: 13440W
65W @ 300A Max Power: 20160W
128W @ 400A Max Power: 26880W
270W @ 500A Max Power: 33600W
What ambient temperature did you allow for in the calculations?
These are pure heat dissipation so do not include ambient. If you have high ambient then it will be a bit worse, but that depends on the thermal resistance and capacitance of the cooling solution.
Also these are for 100% duty cycle which is worst case scenario.
Should handle 300a with 50oC ambient just fine.