The legs are copper. With a tin tinning maybe. The high rating of 75 amps on the TO-220 legs for their size will be because of the short length and assumed good heat-sinking. BTW, It is a good idea to have good heat-sinking of the pads that the legs are soldered to, not just the body.
This topic interests me as I have an application similiar to this that requires large currents and mosfets switching them.
I've tried to look for ways, outside the box, that would reduce the power dissipated during switching and at the same time, attempt to reduce the awful domino effect many of you have described. What I've come up with, is actually pretty simple. Add a small inductor between the mosfet drain and the power source. This inductor should obviously have some pretty good current handling ability. You would also need a diode in antiparallel with the inductor to absorb some of the inductive "bounce".
My thinking is this. As the mosfets turn on, the inductor will reduce some of that initial current surge as the mosfets start to turn on at different times. This SHOULD reduce the domino failure of the mosfets. Now, when they all start turning off, you may still have that problem of excess power dissipation as one starts to turn off faster than the others. To take care of the inductive shock, the diode should be rated to absorb that extra power.
Good? Bad? I haven't tried it yet since school has been eating up exactly 98.63% of my time with the other 1.37% going towards sleep/eating.
Beau, That IS a very good breakdown(no pun intended) of the thermal mechanics of what goes on with mosfets. I also recommend the reading.
RinksCustoms,
I think the 1.5V that you are refering to isn't the maximum voltage that can be applied with the 220A, but the drop produced when 220A is put through it. 1.5V / 220A = 0.0068 ohms... I think that was the Rds of the fet if I'm not mistaken. But yes, you would have 330W being dissipated at that current level. However, if you just directly parallel enough fets for the job, I think it should be good. I'd recommend finding how many fets you'd need for the job, then doubling that for safety(or more if you're not into gambling).
thank you beau for that read, i get what he's saying,
thank you rink for the schematic, but i was thinking more like 4-8 fets, and i noticed that one of the fets that i saw from ixys (dont know if thats right) was a smd type, and i didnt think that was right either so i didnt bother to look at it, i dont see how that could be possible without melting legs
i do plan on doing a liquid cooling system, if i had a scanner or a simple cad system i would upload my idea, but in my best description it would be a block of metal (by the way, should i use aluminum or copper for this if its a heatsink) and then drilling a TON of holes in it and then connecting the holes with tubing, and having a pump connect to this tubing system, and then i·would run distilled water or a different coolant if i can find a cheap one, and then mounting the·2·fets·(per block)·on the top of the metal block, all the tubing would go past a fan array with a condenser shaped tubing system before entering a fairly large holding tank, i would have the "cool" water comming out of the bottom and the warm water coming in the top
also i dont think i will be using 20 gauge wire since i have some power line wire "hanging" around, its really thick, i dont·know what gauge it·is, and i have some 1/2 copper·stranded copper wire in·my·barn·(when the tree fell it pulled it out of the house)
one quick question, when fets at this many amps get overloaded to the melt like resistors or explode like caps
To the quick question - Explode - With force! I've had a few explode in my face, and if it weren't for the eye protection, I'd be out 2 eyeballs. I got a few little shards wedged in my cheek that drew some blood. Stay safe!
In theory either meltdown or explosion can happen. However, the vertical arrangement of the two transistors from one power rail to the other lends itself to catastrophic failure. When one of the two transistors fails it will always go short circuit. This can be while the opposing transistor is still active or not.
If the failure occurs while the other transistor is fully turned on then the failing transistor will be the one to explode leaving the other one perfectly okay. If, however, the transistor fails while the other one is not fully turned on then there is a good beat the failed transistor will stay shorted and in one piece, then when the other one turns on it will be the one to explode.
Thanks for the link Rinks. I just had a quick look at the instruction manual, excellent read! They've made use of a rather neat motor design, it's series wound in some fashion, possibly a Universal or something that has huge speed range. This combined with the block of relays and allows them to build the amplifier with effectively a single transistor. Saving much space/cost/weight and putting all power through only one active part. This also explains why they don't need a lot of cooling and are getting away with 16 kHz modulation frequency. Very smart design. [noparse]:)[/noparse]
Philldapill said...
I've tried to look for ways, outside the box, that would reduce the power dissipated during switching and at the same time, attempt to reduce the awful domino effect many of you have described.
It's not that big a worry with MOSFETs. With BJTs it was a nightmare.
Philldapill said...
What I've come up with, is actually pretty simple. Add a small inductor between the mosfet drain and the power source. This inductor should obviously have some pretty good current handling ability. You would also need a diode in antiparallel with the inductor to absorb some of the inductive "bounce".
Yep, that can certainly reduce noise. Forget the diode though, the inductors will discharge into the motor just fine. And the inductors go on the load side rather than the supply side.
Philldapill said...
I've tried to look for ways, outside the box, that would reduce the power dissipated during switching and at the same time, attempt to reduce the awful domino effect many of you have described.
It's not that big a worry with MOSFETs. With BJTs it was a nightmare.
Philldapill said...
What I've come up with, is actually pretty simple. Add a small inductor between the mosfet drain and the power source. This inductor should obviously have some pretty good current handling ability. You would also need a diode in antiparallel with the inductor to absorb some of the inductive "bounce".
Yep, that can certainly reduce noise. Forget the diode though, the inductors will discharge into the motor just fine. And the inductors go on the load side rather than the supply side.
Evan
Actually evan, isn't the the clamp diode(s) there to sink the exessive Back EMF from damaging the FET's· since most fets have integrated clamping diodes that are just "weak" to deal with a heavy inductive load like say·a 9.36KW motor?
Comments
I've tried to look for ways, outside the box, that would reduce the power dissipated during switching and at the same time, attempt to reduce the awful domino effect many of you have described. What I've come up with, is actually pretty simple. Add a small inductor between the mosfet drain and the power source. This inductor should obviously have some pretty good current handling ability. You would also need a diode in antiparallel with the inductor to absorb some of the inductive "bounce".
My thinking is this. As the mosfets turn on, the inductor will reduce some of that initial current surge as the mosfets start to turn on at different times. This SHOULD reduce the domino failure of the mosfets. Now, when they all start turning off, you may still have that problem of excess power dissipation as one starts to turn off faster than the others. To take care of the inductive shock, the diode should be rated to absorb that extra power.
Good? Bad? I haven't tried it yet since school has been eating up exactly 98.63% of my time with the other 1.37% going towards sleep/eating.
RinksCustoms,
I think the 1.5V that you are refering to isn't the maximum voltage that can be applied with the 220A, but the drop produced when 220A is put through it. 1.5V / 220A = 0.0068 ohms... I think that was the Rds of the fet if I'm not mistaken. But yes, you would have 330W being dissipated at that current level. However, if you just directly parallel enough fets for the job, I think it should be good. I'd recommend finding how many fets you'd need for the job, then doubling that for safety(or more if you're not into gambling).
This Monster of an·EV controller still intrigues me of it's power density! 640KW Peak!!! Interestingly enough, it mentions coolant and using IGBT's..
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
E3 = Thought
http://folding.stanford.edu/·- Donating some CPU/GPU downtime just might lead to a cure for cancer! My team stats.
thank you rink for the schematic, but i was thinking more like 4-8 fets, and i noticed that one of the fets that i saw from ixys (dont know if thats right) was a smd type, and i didnt think that was right either so i didnt bother to look at it, i dont see how that could be possible without melting legs
i do plan on doing a liquid cooling system, if i had a scanner or a simple cad system i would upload my idea, but in my best description it would be a block of metal (by the way, should i use aluminum or copper for this if its a heatsink) and then drilling a TON of holes in it and then connecting the holes with tubing, and having a pump connect to this tubing system, and then i·would run distilled water or a different coolant if i can find a cheap one, and then mounting the·2·fets·(per block)·on the top of the metal block, all the tubing would go past a fan array with a condenser shaped tubing system before entering a fairly large holding tank, i would have the "cool" water comming out of the bottom and the warm water coming in the top
also i dont think i will be using 20 gauge wire since i have some power line wire "hanging" around, its really thick, i dont·know what gauge it·is, and i have some 1/2 copper·stranded copper wire in·my·barn·(when the tree fell it pulled it out of the house)
one quick question, when fets at this many amps get overloaded to the melt like resistors or explode like caps
By the way, science geek, what school are you at?
If the failure occurs while the other transistor is fully turned on then the failing transistor will be the one to explode leaving the other one perfectly okay. If, however, the transistor fails while the other one is not fully turned on then there is a good beat the failed transistor will stay shorted and in one piece, then when the other one turns on it will be the one to explode.
Yep, that can certainly reduce noise. Forget the diode though, the inductors will discharge into the motor just fine. And the inductors go on the load side rather than the supply side.
Evan
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
E3 = Thought
http://folding.stanford.edu/·- Donating some CPU/GPU downtime just might lead to a cure for cancer! My team stats.
Post Edited (RinksCustoms) : 9/26/2008 3:50:36 PM GMT