drunkencat and steveo new controller amp monster :twisted:

[drunkencat129]

oke guys i got a new controller from work is a massive power house lol no more of thoes small fets look at theas babys lol
the model on the fets are irfp2907
here is the pdf for the fets
http://www.datasheetcatalog.org/datasheet/irf/irfp2907.pdf

ps the pics are cut off go to the link to see full pic

http://i43.tinypic.com/16ibgg3.jpg

http://i40.tinypic.com/35mds12.jpg


nice thick wires as well

http://i41.tinypic.com/a4wzu8.jpg

 

[drunkencat129]

wooot 209A jesus lol at 75v and there 4.5 mohms

 

[Jeremy Harris]

Not bad, but remember that these FETs have a package limited current of 90 amps, so the 209 amp rating is a bit of a moot point. The limit is the bond wire fusing current, inside the device.

Overall, IRFB3077s are a fair bit better for a 75V device, as they have a lower Rdson (3.3mohm max, 2.8mohm typ.), have the same 75V rating but have a package current limitation of 120 amps.

Jeremy

 

[CamLight]

Overall, IRFB3077s are a fair bit better for a 75V device, as they have a lower Rdson (3.3mohm max, 2.8mohm typ.), have the same 75V rating but have a package current limitation of 120 amps.

Jeremy

You'd have to balance that lower Rds-on of the IRFB3077 with the IRFP2907's better junction-sink thermal resistance though (0.9C/W vs. 0.56C/W).

 

[karma]

nice also work well in amplifiers do we get a video of this controller in Action?

cheers

 

[drunkencat129]

at 133v maby working on it tonight maby later on in the week or i fi can find some 200v ones

 

[Jeremy Harris]

You'd have to balance that lower Rds-on of the IRFB3077 with the IRFP2907's better junction-sink thermal resistance though (0.9C/W vs. 0.56C/W).

Very true, but there will still be more heat to get rid of (and hence more power wasted) from the higher Rdson device.

One thing I'd really like to know is what the most common FET failure mode is in controllers. MY gut feeling tells me that it's probably junction over-temperature causing them to blow, rather than an absolute voltage or current overload. Most controllers don't seem to have a particularly good thermal path from the device to the case, as far as I can see. Certainly the two I've taken apart have had mountains of heatsink compound, have both used thick silcone insulating pads and have not had very good contact between the spreader bar and the case.

Jeremy

 

[CamLight]

You'd have to balance that lower Rds-on of the IRFB3077 with the IRFP2907's better junction-sink thermal resistance though (0.9C/W vs. 0.56C/W).

Very true, but there will still be more heat to get rid of (and hence more power wasted) from the higher Rdson device.

One thing I'd really like to know is what the most common FET failure mode is in controllers. MY gut feeling tells me that it's probably junction over-temperature causing them to blow, rather than an absolute voltage or current overload. Most controllers don't seem to have a particularly good thermal path from the device to the case, as far as I can see. Certainly the two I've taken apart have had mountains of heatsink compound, have both used thick silcone insulating pads and have not had very good contact between the spreader bar and the case.

Jeremy

True, there's more power to move away from the junction. But the lower thermal resistance reduces the temperature rise on the more thermally efficient chip. As an example, here are the numbers at 50A.

IRFB3077 power dissipated = 50A x 50A x .0033 ohm Rds-on = 8.25W
IRFP2907 power dissipated = 50A x 50A x .0045 ohms Rds-on = 11.25W

So far it seems like the IRFB3077 will run cooler due to the lower power level that needs to be dissipated (due to the IRFB3077's lower Rds-on value). But, what is the actual junction temperature rise?

IRFB3077 junction temp rise = 8.25W x 0.9C/W = 7.43C temp rise at the MOSFET junction
IRFP2907 junction temp rise = 11.25W x 0.56C/W = 6.30C temp rise at the MOSFET junction.

Due to its much lower thermal resistances, the IRFP2907 will actually run cooler than the IRFB3077 even though the IRFP2907 has a higher on-resistance. Other factors can change this, like gata capacitance which might slow down one or the other MOSFET (heating it up more), but it's often more complex than selecting the lowest on-resistance MOSFET when trying to keep a controller running cool.

IMHO, it's a combination of two factors that blow out controllers; over-voltage and over-temperature. Since protecting against over-temperature is a lot more expensive, I agree with you that over-temperature is probably the leading cause of MOSFET failure in a controller. Especially since so few controller designers know much about calculating the thermal efficiency of their products or even how to properly measure the case temperature of an operating MOSFET.

Sil-Pads, and other silicone or rubberized thermal pads, have HUGE thermal resistance ratings! Hardcoat anodizing the heat sink is the way to go for achieving insulation between MOSFETs. But, it's much more expensive than using those pads. Using too much thermal compound is another frequent sin. Add onto that the fact that most MOSFETs are mounted with screws (not clips), which often raises up one end of the MOSFET, and it gets even worse. And as a final blow to the health of those poor MOSFETs, they don't even use Belleville washers under the screws. This allows the screw to loosen over time due to thermal cycling and the MOSFET is then not firmly against the heat sink.

The life of a controller MOSFET is a tough one.

 

[bobmcree]

since the subject of running at 150v is bound to come up again i am answering a private message from steveo here:

the greatly improved thermal conductivity from the device junction to the heat sink in the TO247 package makes it the logical next step for ebike controllers, so it is good to finally see one using them. we went through this awhile ago, and decided the IRFP4568 would be the best option for a drop-in 150v replacement. With an Rds-on of .0048 ohms coupled with the better conductivity between the larger case and the heat sink and the junction to case improvement from the larger die the benefits of the TO247 4568 vs the irfb4115 are obvious, but the larger package does not easily fit most controllers. it looks like half the locations are not populated, so assuming everything else will take 150v i would go with the 4568. I can get 100 pcs @ 3.74 if you want to get a couple of other people together.

there is a TO247 package version of the 4110, but the junction to case resistance is just as high as the TO220 part, indicating they just remounted the same die in a larger package.

 

[Jeremy Harris]

True, there's more power to move away from the junction. But the lower thermal resistance reduces the temperature rise on the more thermally efficient chip. As an example, here are the numbers at 50A.

IRFB3077 power dissipated = 50A x 50A x .0033 ohm Rds-on = 8.25W
IRFP2907 power dissipated = 50A x 50A x .0045 ohms Rds-on = 11.25W

So far it seems like the IRFB3077 will run cooler due to the lower power level that needs to be dissipated (due to the IRFB3077's lower Rds-on value). But, what is the actual junction temperature rise?

IRFB3077 junction temp rise = 8.25W x 0.9C/W = 7.43C temp rise at the MOSFET junction
IRFP2907 junction temp rise = 11.25W x 0.56C/W = 6.30C temp rise at the MOSFET junction.

Due to its much lower thermal resistances, the IRFP2907 will actually run cooler than the IRFB3077 even though the IRFP2907 has a higher on-resistance. Other factors can change this, like gata capacitance which might slow down one or the other MOSFET (heating it up more), but it's often more complex than selecting the lowest on-resistance MOSFET when trying to keep a controller running cool.

IMHO, it's a combination of two factors that blow out controllers; over-voltage and over-temperature. Since protecting against over-temperature is a lot more expensive, I agree with you that over-temperature is probably the leading cause of MOSFET failure in a controller. Especially since so few controller designers know much about calculating the thermal efficiency of their products or even how to properly measure the case temperature of an operating MOSFET.

Sil-Pads, and other silicone or rubberized thermal pads, have HUGE thermal resistance ratings! Hardcoat anodizing the heat sink is the way to go for achieving insulation between MOSFETs. But, it's much more expensive than using those pads. Using too much thermal compound is another frequent sin. Add onto that the fact that most MOSFETs are mounted with screws (not clips), which often raises up one end of the MOSFET, and it gets even worse. And as a final blow to the health of those poor MOSFETs, they don't even use Belleville washers under the screws. This allows the screw to loosen over time due to thermal cycling and the MOSFET is then not firmly against the heat sink.

The life of a controller MOSFET is a tough one.

We're in danger of violently agreeing on this - perhaps time to shift to a "thermal efficiency" thread? It could be useful for those looking for high power and reliability.

Jeremy

 

[CamLight]

We're in danger of violently agreeing on this - perhaps time to shift to a "thermal efficiency" thread? It could be useful for those looking for high power and reliability.

Jeremy

Very interesting idea! Will anyone read it though?
It will very quickly get very technical and we'll need a way to have, perhaps, a running set of "bottom-line" recommendations for all those who just want to know about a particular MOSFET, mounting method, heat sink calculation, etc. These recommendations can then become a repository for the conclusions made in other threads?

 

[steveo]

since the subject of running at 150v is bound to come up again i am answering a private message from steveo here:

the greatly improved thermal conductivity from the device junction to the heat sink in the TO247 package makes it the logical next step for ebike controllers, so it is good to finally see one using them. we went through this awhile ago, and decided the IRFP4568 would be the best option for a drop-in 150v replacement. With an Rds-on of .0048 ohms coupled with the better conductivity between the larger case and the heat sink and the junction to case improvement from the larger die the benefits of the TO247 4568 vs the irfb4115 are obvious, but the larger package does not easily fit most controllers. it looks like half the locations are not populated, so assuming everything else will take 150v i would go with the 4568. I can get 100 pcs @ 3.74 if you want to get a couple of other people together.

there is a TO247 package version of the 4110, but the junction to case resistance is just as high as the TO220 part, indicating they just remounted the same die in a larger package.

Hey bob

I couldn't wait; I ordered a small order at avnet 6$ a piece not to bad.

I will start the modding hopefully end of next week!

thanks for you help. I will let you all know my results.

-steveo