PDA

View Full Version : Brushless ESC overheating at low throttle question.


H20Cat
09-04-2003, 09:56 PM
Is it common that a brushless ESC's get hot at low to mid throttle. My aveox ESC says that the ESC should be cooled very well especially when running below full throttle.

Can someone give a qick reason for this, I thought that the more throttle, the more heat.

Thanks.

AndyKunz
09-05-2003, 06:57 AM
It's because the FETs are spending more time in their (very inefficient) linear mode. Brushed ESCs do the same thing.

One way to reduce heating is for the control to use a faster slew rate on the FET gate. The problem is that this induces harmonics that cause RF interference.

Andy

Twinpowered
09-05-2003, 08:31 AM
Andy,

Would it be possible to change the rate depending on stick position? That would make the ESC more efficient at low power, while keeping it efficient at high. Can't the interference be filtered or limited by allowing only limited rate changes?

You talked about changing the (fixed) rate in the thin gap thread, so perhaps this will be one of your next steps in refining your design? If that is possible of course...


:cool:

AndyKunz
09-05-2003, 09:15 AM
What you are talking about is what we call "fixed on time/variable frequency" instead of "fixed frequency/variable on time."

There are advantages to both, and what you mention would be an advantage only to reduce heat and improve efficiency of the FET. The problem is now that instead of filtering one constant frequency, there are 213 frequencies to filter.

Andy

H20Cat
09-05-2003, 11:18 AM
Thanks Andy, I was trying to figure out why my ESC overheated yesterday... I was running too slow because of the chop. I had to keep it about 1/2 throttle or else I would get airborn. It ran perfect about a week ago at mostly full throttle.

CM007
09-05-2003, 12:22 PM
I thought awhile ago someone posted that, half throttle with BL is different than with brushed? Temperature wise that is.

CM

AndyKunz
09-05-2003, 12:49 PM
It is different, but not that there are no similarities.

On a BL control, each FET can be on at the very most 33% of the time. Usually we leave what we call the high side turn on for the full period of a commutation. The low side is pulsed in order to control the RPM.

Because the high side gets such a long on-time followed by a long off-time, it helps it to keep that side of the ESC cool. That's good, because it allows the use of lower-quality P-FETs to achieve identical results, or if using N-FETs then you need a voltage booster to turn them on.

On the low side, though, each FET still gets a maximum of 33% on-time, but at part throttle it spends more time in the linear range. This causes it to heat up.

A brushed ESC only has low-side FETs, and it behaves the same way, except that it never gets to cool off 67% of the time like they do on a BL.

So they are similar in principle, but dissimilar in specifics.

Andy

Twinpowered
09-05-2003, 03:12 PM
Ouch. 213 filters would make a very big and expensive ESC I guess. But that's not what I exactly meant: I was thinking more like a "gearbox": 2 or 3 frequencies, for low/(middle)/high speed settings, and changing timing for everything in between. That should keep the component count down and improve low speed efficiency: ideal for a fun boat.

I'm working on a design like that, and I must say: it really makes you appreciate a good ESC. It also makes me wonder why I'm doing it: your price is very low considering the parts and labour involved. But it is great to be messing around with electronics, and I learn a lot of it (what you can do and most of all what NOT to do :D).

What you are talking about is what we call "fixed on time/variable frequency" instead of "fixed frequency/variable on time.
There are advantages to both, and what you mention would be an advantage only to reduce heat and improve efficiency of the FET. The problem is now that instead of filtering one constant frequency, there are 213 frequencies to filter."

Ray Bidwell
09-08-2003, 10:50 AM
A good example of overheating at low throttle is the SAWS if they did not take off at full throttle the esc would overheat before they even got to the line.

Twinpowered
09-08-2003, 03:12 PM
Ray, could it be that the reduced cooling as a result of low speeds is causing this? I think given the amount of time from take-off to line the FET's inefficiency alone can't cause such serious overheating. SAW setups are by nature hot ones, but I could be wrong.

Above all: what we were talking about was low to medium throttle for a prolonged period of time. When racing, the throttle is expected to be at or at least near full throttle all the time. In a race setup, cooling system, ESC, motor, prop are all tuned for that usage.
As Andy stated: At partial throttle, the ESCs MOSFET's are less efficient. But also not unimportant: the air- or water cooling is less efficient (lower speed: less flow). And worse: because you're using less Amps, you can abuse the ESC for a longer period. So we have a hotter ESC with less cooling running longer. Read: major problems.

I don't race, and do more than an occasional "slow drive-by". Given the info above you can see why I'm looking for a solution, apart from my "build from scratch" and "wanna know how it works" mania.

Andy:
Do you think I could be onto something, or would the RFI cause too much problems? Reliability will suffer anyhow of course, because there are more components that can fail. Would love to hear your opinion, because you obviously know more than anyone about this subject.

AndyKunz
09-09-2003, 07:01 AM
Originally posted by Twinpowered
Andy:
Do you think I could be onto something, or would the RFI cause too much problems? Reliability will suffer anyhow of course, because there are more components that can fail. Would love to hear your opinion, because you obviously know more than anyone about this subject.

We use the "3 speed" type format in some industrial situations, mostly to provide more smoothness or precision at certain speeds. It has its place, to be sure. The problems of EMI/RFI on a factory floor are not nearly as worrisome though. You can do shielding and localized filtering as required. You can hook up a spectrum analyzer and see it under operating conditions. Can't do that with a model. It is easiest to select an appropriate frequency (for instance, stay away from 455kHz and 10.7 MHz and related frequencies), get rid of harmonics by controlling switching time, and do a bang-up job keeping the noise in the box.

Andy

Twinpowered
09-09-2003, 02:27 PM
Andy,
Thanks for the reply, I'll give it a try. Don't think I'll ever know half a much as you about the subject, but so far it's given me a much better understanding of advantages and pitfalls of existing ESC's. So even if I fail miserably I have won something.

Your suggestions are noted, most of them are understood. Will try to comply. One question though: Bang-up job is NOT understood. What does it mean?

Sorry for my ignorance, but my English is a bit rusty at best. Don't worry: I'll keep practicing.

Ray Bidwell
09-09-2003, 04:31 PM
The problem in the saw setups is that you are using large props x937 was most popular. These props cause you to draw a huge amount of amps until they plane out and it is this large amp load that would cause the esc to overheat. You also want a large runup to the line so you were going about 1/16 mile from the line turning then a 1/16 mile to the line then 1/16 of a mile in the trap.

AndyKunz
09-10-2003, 07:08 AM
Bang-up is an idiom for best quality. In our shop we call it "Old World Craftsmanship" or "German Engineering." It refers to meticulous examination of every detail.

Andy

Twinpowered
09-10-2003, 01:57 PM
Ray,
Thanks for explaining. It's now clear: you would need a lot of headroom for a SAW setup (major Amps), and the cooling system would need enough mass (= peak cooling capacity) to stow away the heat generated temporarily. People with this problem could try a bigger heatsink, preferable with lower C/W value, so heat is better dissipated later on. Lower C/W would also raise continuous current capability of the ESC. Not a task for the faint-hearted though: you need to know exactly what you're doing, or smoke is all you get.

Also explains Andy's success: copper is great at conducting heat, and his design has plenty of headroom (450A peak). Quality did it's job too I guess.

Andy,
Since I'm in the old world, and Germany is right next to us, I'll try to live up to the "old" standards. ;)
MOSFET's I intend to use are not German though. Datasheet says: made in the USA. Well, they'll will have to do I guess. :)

AndyKunz
09-10-2003, 02:43 PM
I've had MUCH better success with IR than I had with the french FETs.

Andy

Twinpowered
09-10-2003, 04:31 PM
Andy,
That's a typo right? You must have known the French are famous for their fries, not their FETs. ;)

But serious: So far the IRL3713 (2.6mOhm typical) and ISL9N302AP3 (1.9mOhm typical) are my best bets. That last one is made with PWM control in mind, so very well suited.
Current capability should be within an amp of each other (i.e. 21-22A per FET). IRL3713 is much cheaper, so I'll propably use those. Fairchild is nicer though: they've sent me 4 to evaluate. I'm sure Fairchild 'll send you 6 if you would like to test them.
Both are relatively new designs, and they're hard to get, esp. the N302. Bulk is a little easier, but unfortunatly even then most distributors don't carry these types (not over here at least).

With 6 FETs planned on the final version, I should have plenty of room for error on that side, though I sure hope I never need it.
Now only a microPIC to attach them to, and some other stuff of course... and an awful lot of work. Why oh why did I start this? :D



Note to casual reader:
It does NOT pay to build only one high-end ESC. Andy does it way cheaper, and better. You have to use bulk quantities to beat Andy's price on parts alone, and even then it's a heck of a job to build a really good one. Believe me, I know.

***Edited by moderator for language content***

AndyKunz
09-10-2003, 05:34 PM
I have a few tubes of 3713's here, and get several tubes every month automatically. They weren't too hard to get - try Future.

You'll have to buy a whole tube though.

Andy

Jay Turner
09-10-2003, 07:47 PM
The problem in the saw setups is that you are using large props....these props cause you to draw a huge amount of amps until they plane out and it is this large amp load that would cause the esc to overheat That's part of the problem, but the racers using large cell counts - over 12 - found that if they used much part-throttle to run out and get lined up with the lights, they only got one and one half passes before the controller shut off due to overheating. That means a wasted chance at a record. If they used a minimum time at part-throttle, they could get 2 to 3 passes with the same setups. Part throttle was the killer.... :hammer:

Twinpowered
09-11-2003, 02:34 PM
Andy,
You're good. Future is exactly the one I meant. They have a sales office over here too, and they're about the only distributor over here that has them on stock.

And yes, I'll have to build 8 ESC's to make it worthwile. Because I use twin motors/engines in virtually every setup I 'll use them all sooner or later. Will be later I guess: designing and building a hull from scratch takes me quite some time. Better speed up the building program or build some Liberty ships ;)

Jay,
Looks like cooling is a major problem. Has anyone tried the following:
connect the heatsink of the ESC to a heatpipe, and let that transfer the heat to a copper plate. You can mount that plate underneath the hull, in a way that it's level with the rest (don't need to introduce drag). You'd have to mount it on a place where it gets wet of course, and in such a way that it can expand and shrink depending on temp. The amount of extra cooling should be substantial, esp. for the weight. Perhaps this could be even weight-neutral, as the heatsink of the ESC can be smaller (needs trial and error or very good calculations).

Just a thought though, never tried this myself. We use these techniques in PC's, why not in models?

P.S. To moderator: sorry, my English isn't perfect. Trying to improve. Can you mail me about my errors? :o