Replacing controller mosfets to higher quality?

R

rg12

100 kW
Joined
Jul 26, 2014
Messages
1,591
Hey guys,

I did some water cooling project for a 45A Greentime controller with a soldered shunt and one guy commented that instead of the water cooling I could use lower ohm (higher quality) mosfets.
All I know is that Greentime uses 4110 mosfets (whatever that means (they brag about it)).

Any thoughts?
 
Alan B said:
Drain-to-Source Voltage V, RDS(on) mΩ, Id A, Power Dissipation W TC=25°C
IRFB4110 100 3.5 120 370
AOT290L 100 2.7 140 500
CSD19536KCS 100 2.3 150 375
Click to expand...

Ok, I did some reading and I don't understand the last number on each row, is it wattage?
I want to over current a controller so should I just go with the CSD as it has the lowest RDS? but why does it say 375 at the end of the line which is almost the same as the 4110?
 
Those are just some summary numbers from the datasheets, look at the full datasheets if you want more data. Those are just some examples of FETs that have been discussed or are being used. Power Dissipation is not what you want to be doing with your FETs, anyway. That's mostly an indication of the thermal path capacity. :)
 
Alan B said:
Those are just some summary numbers from the datasheets, look at the full datasheets if you want more data. Those are just some examples of FETs that have been discussed or are being used. Power Dissipation is not what you want to be doing with your FETs, anyway. That's mostly an indication of the thermal path capacity. :)
Click to expand...

Why wouldn't I want power dissipation?
and my knowledge is kind of limited so a datasheet will only get me even more confused.
So out of the two you mentioned, which is the most recommended for an over current project? and by how much (in percentage (language I understand)) will it be better?
 
Of course things are always a little more complex in real life and the numbers will only get you so far. Another thing that has to be considered is the gate charge. If the gate charge is too large, it will slow down the switching speed but it also depends on the gate driver used. If the proposed replacement part has a lower gate charge or at least is pretty close, it will be fine.

Switching losses and resistance losses both contribute to heating. Using a lower Rds reduces the resistance losses. Slower switching will likely increase switching losses. If you found a part with 1/2 the Rds, the heating would be less but not 1/2 due to switching losses.

The other critical factor is heat dissipation. If the controller has adequate cooling, you can run it way harder. Having a higher than needed voltage rating will make it more immune to voltage spikes. Water cooling might be overkill, but you can also look at improving passive air cooling (better air flow, better heat sink fins, conduction to frame, etc.)

The AOT290L has a proven track record, but would only reduce the heating by a little.
 
fechter said:
Of course things are always a little more complex in real life and the numbers will only get you so far. Another thing that has to be considered is the gate charge. If the gate charge is too large, it will slow down the switching speed but it also depends on the gate driver used. If the proposed replacement part has a lower gate charge or at least is pretty close, it will be fine.

Switching losses and resistance losses both contribute to heating. Using a lower Rds reduces the resistance losses. Slower switching will likely increase switching losses. If you found a part with 1/2 the Rds, the heating would be less but not 1/2 due to switching losses.

The other critical factor is heat dissipation. If the controller has adequate cooling, you can run it way harder. Having a higher than needed voltage rating will make it more immune to voltage spikes. Water cooling might be overkill, but you can also look at improving passive air cooling (better air flow, better heat sink fins, conduction to frame, etc.)

The AOT290L has a proven track record, but would only reduce the heating by a little.
Click to expand...

So what is the benefit of the better mosfets if it only reduces heat by a little?
Noise? smoothness?
 
Less heat is about it. Better FETs won’t blow up as easy too.
It won’t change how the controller behaves.
 
FET IR heating would be reduced by (3.5-2.7)/3.5 = 23%, and the dissipation rating of the AOT290L is higher so it can handle more heat, or the same heat with more thermal margin. The PhaseRunner uses AOT290L's, which is one reason the little 6 FET controller can handle as much power as it does while being very small in size.

FET IR heating may be the major source of heat in the controller. The other is switching losses which are harder to estimate, but should be less in a well designed controller. If IR losses weren't important in the controller, they would not bother to parallel FETs. Paralleling FETs reduces IR losses but not switching losses (providing the FETs are well driven).

Changing from 4110's to 290L's would be like adding 23% more parallel FETs to the controller. Since parallel FETs don't share current perfectly, this is about as much improvement as going from 12 to 18 FETs, but without the increase in size.
 
Alan B said:
FET IR heating would be reduced by (3.5-2.7)/3.5 = 23%, and the dissipation rating of the AOT290L is higher so it can handle more heat, or the same heat with more thermal margin. The PhaseRunner uses AOT290L's, which is one reason the little 6 FET controller can handle as much power as it does while being very small in size.

FET IR heating may be the major source of heat in the controller. The other is switching losses which are harder to estimate, but should be less in a well designed controller. If IR losses weren't important in the controller, they would not bother to parallel FETs. Paralleling FETs reduces IR losses but not switching losses (providing the FETs are well driven).

Changing from 4110's to 290L's would be like adding 23% more parallel FETs to the controller. Since parallel FETs don't share current perfectly, this is about as much improvement as going from 12 to 18 FETs, but without the increase in size.
Click to expand...

but 12 to 18 fets is 50% gain
 
No, it is not a full 50% increase, because the FETs don't share current equally, so the designer cannot increase current that much. He is only allowed to take about half of the new FETs because of unequal current division. Now some hotrodders ignore this, and they often blow up controllers. Designing for more current and power is not quite that simple.
 
Alan B said:
No, it is not a full 50% increase, because the FETs don't share current equally, so the designer cannot increase current that much. He is only allowed to take about half of the new FETs because of unequal current division. Now some hotrodders ignore this, and they often blow up controllers. Designing for more current and power is not quite that simple.
Click to expand...

oh ok got it, thanks alot :)
 
You must log in or register to reply here.

Similar threads

L
Joe's E-bike build
2
Replies
46
Views
4,253
lightbulb010
L
jetpackjbd
    • Like
1000 Miles In - JoyKie 350W 36V Mountain Bike Review (Risky Cheap Chinese Bike Purchase)
Replies
10
Views
564
2old
2
mxlemming
    • Like
GaN and IMS MESC controller boards
2
Replies
31
Views
3,393
mxlemming
mxlemming
Z
    • Like
    • Doin' Science
Custom Pedal Assist Controller... Update from last year!
Replies
8
Views
1,015
zachleedogg
Z
R
    • Like
A critique of mid-drive riding recommendations (and hub drive criticisms)
Replies
15
Views
5,082
Bullfrog
B
Back
Top