Three Phase Electronic Speed Control In-Flight Blow-Out. / General Science and Electronics / Forums

Forums

4hv.org :: Forums :: General Science and Electronics

« Previous topic | Next topic »

Three Phase Electronic Speed Control In-Flight Blow-Out.

previous 1 2 3 4 next

Move Thread

LAN_403

hen918

Tue Jun 09 2015, 09:19PM

Registered Member #11591 Joined: Wed Mar 20 2013, 08:20PM
Location: UK
Posts: 556

DerAlbi wrote ...

... your estimations are somewhat way off...

first of all: No one, absolutley no one uses 32A rated Mosfets to PWM a current of 60A. ...

They do if the datasheet says this: "Pulsed Drain Current (t = 100 Î¼s) I(DM): 400A"
I wouldn't use those MOSFETs, but the Chinese making a three phase h-bridge and driver capable of running a brushless motor at 330W for $14 would do!

Also, the switching time will be in the region of about 200ns, as evidenced by my 'scoping of the output of my very similar, but functioning, ESC
EDIT: I would post a screen shot of my DSO, but the forum doesn't like Bitmaps.

Patrick

Tue Jun 09 2015, 10:55PM

Registered Member #2431 Joined: Tue Oct 13 2009, 09:47PM
Location: Chico, CA. USA
Posts: 5639

I use PNG files instead of BMP.

Ash Small

Wed Jun 10 2015, 10:25AM

Registered Member #3414 Joined: Sun Nov 14 2010, 05:05PM
Location: UK
Posts: 4245

hen918 wrote ...

but the forum doesn't like Bitmaps.

You can use MS Paint (or any other graphics software) to save a BMP as a JPG, then post it on the forum.

DerAlbi

Wed Jun 10 2015, 10:28AM

Registered Member #2906 Joined: Sun Jun 06 2010, 02:20AM
Location: Dresden, Germany
Posts: 727

Hen, you should not scope the output of the circuit, you should measure the gate voltage directly. The length/duration of the Millerplateau is the lossy part of the switching. The Output voltage can be slow by whatever reason which does not mean that the current is still flowing... that only may show that they used a deadtime of ~200ns for the halfbridge which is reasonable.
Make sure you use a 1:100 probe because the probe capacitance (1:1 and 1:10) will/can significantly worsen your measurement at the gate.

They do if the datasheet says this: "Pulsed Drain Current (t = 100 Î¼s) I(DM): 400A"

That is single impulse only and in no way repetetive. Also chinese engineers know that.

Also think about the 60A again: thats totally a stupid assumption. If a halfbridge is PWMed with 0.5 dutycycle and your RMS current is 30A you are doing something wrong! Under full load (acceleration) the PWM goes to 100% dutycycle and your RMS current is suddenly 60A then.... THIS IS OUT OF SPEC! Again and again. These assumptions are nonsense.
And IF they apply actually, then - as said multiple times before - the load is way too much for the halfbridge, and no heatsink will save the Mosfets long term.
There simply is no argument. And as long as you stay in spec, we have shown that the disipated power is way within reason and not even limited by the mosfets but the FR4. Even if the disipated power is underestimated by a factor of 2 it would be totally fine....

Patrick

Wed Jun 10 2015, 03:52PM

Registered Member #2431 Joined: Tue Oct 13 2009, 09:47PM
Location: Chico, CA. USA
Posts: 5639

As soon as the rain and wind stops, ill make a flight. Its totally repaired.
I got the labels back on, so hopefully i spelled everything right. i have been to college you know.

One of the important things i may not have mentioned, was that this ESC isnt directly in the windstream. but the previous configurations didnt show signs of overheating in this similar place. so once again there seems to be a significant difference.
I should have mentioned that for DerAlbi's benefit.

.

...

Thu Jun 11 2015, 03:36AM

Registered Member #56 Joined: Thu Feb 09 2006, 05:02AM
Location: Southern Califorina, USA
Posts: 2445

@hen918 200ns for the switching time sounds pretty reasonable for a device like this, the I admit the 1us figure is quite conservative, but I also generally prefer to over estimate thermals since its better than having the device explode when I was hoping it would work

The assumption of a 50% duty cycle also falls into this same category, its tempting to do thermal calculations for a best case scenario (like a perfectly matched massless motor which only pulls full current at high duty cycles) but designing them for a reasonable worst case usually results in widgets that last longer. To each their own.

@Patrick- I didn't catch that on the picture in the first post. Is it only the tail ESC that is mounted on the body (from the picture on the first post it looks like the other two are mounted in the prop wash). It would make sense that it runs a lot hotter if that is the case, and in that case the little heatfin you added may not actually be enough depending on how much air is flowing through the middle of the copter. It actually does not take much airflow, I have found air speeds of just a few m/s will drastically improve the thermals of a heatfin. You might take inspiration from the controllers sold for r/c car duty, an equivalently sized/priced car ESC will have a substantial heatsink on it, ex

DerAlbi

Thu Jun 11 2015, 12:25PM

Registered Member #2906 Joined: Sun Jun 06 2010, 02:20AM
Location: Dresden, Germany
Posts: 727

Hens 200ns are not representative as i said before.. you have to look at the Millerplateau and not the output voltage. This only tells us that the deadtime of the halfbridge is >200ns which implies that the actual switching time is way shorter again.
Your 50% were not the problem. The problem was, that you compensated the 50% by doubling the current which is a bad thought. With 50% dutycycle the current will not rise higher and surely not double. .
It is ok to be conservative.. assuming "doubling" within quatractic relations can be quite over-conservative and using 50x longer switching tims is also a bit unreasonable but it is ok if you do it that way in your hobby builds.

HOWEVER such estimations cannot be used to dicredit the work of an actual engineer who just had more knowledge than you just by looking at a photo and starting the wild, uninformed, unreasonable guessing.

Patrick

Thu Jun 11 2015, 03:47PM

Registered Member #2431 Joined: Tue Oct 13 2009, 09:47PM
Location: Chico, CA. USA
Posts: 5639

ok lets just make the flight, ill report the results here soon, ive got the whole machine repaired. Ive just been exhausted from work but will have time soon.

hen918

Thu Jun 11 2015, 04:55PM

Registered Member #11591 Joined: Wed Mar 20 2013, 08:20PM
Location: UK
Posts: 556

Whatever the heat MOSFETs were dissipating, components being ran at (or near to) their maximum ratings never last long

DerAlbi: I'm not blaming the designer, I'm not even blaming his boss who said "ahhh, so the maximum for these devices is 32A, let's slap a 30A label on it and sell them like that." But then I'm not blaming anyone, because $14 is cheap enough for me (or most people) not to care. You get what you pay for, irrespective of what it says on the label.

DerAlbi

Thu Jun 11 2015, 05:13PM

Registered Member #2906 Joined: Sun Jun 06 2010, 02:20AM
Location: Dresden, Germany
Posts: 727

I cant see how a 30A device in a "FlyingThing" can be maxed out... The absolute most unrepeatable current peaky may not exceed 30A... so the actual load should be way way lower on average. If not, its the buyer who decided wrong and again not the engineer nor his boss

I think there is a problem with the philosophy between engineer and consumer... Engineer discribes maximum rating, consumer takes it as average... doesnt work that way

previous 1 2 3 4 next

Moderator(s): Chris Russell, Noelle, Alex, Tesladownunder, Dave Marshall, Dave Billington, Bjørn, Steve Conner, Wolfram, Kizmo, Mads Barnkob