[rkremser]

So New project, Its for a separate application outside of CNC but I don't know a better place to get some feedback. Imagine a 12v motor with peak draw around 270 amps, I'd like to put together a pwm controller for it to deal with speed control. Only has to be one direction so that should make it easier. I am planning on using a micro controller to drive a 555 timer for the pulse train and integrating a 8-seg display to select the speed that the controller will ramp to. interface will be manual button, hold to ramp and hold at max release to freewheel. The control aspect of the circuit isn't much worry to me as i'm accustomed to low power applications but my mind gets stuck thinking about the headache of the 300 amp max capacity i want to design to. More of Mosfet choice and developing a way to dissipate the heat. The duty cycles are going to be very low 1-3 min on with a dwell time around 30+ minutes so that should help with some of the heat concerns. Just looking for some feedback, anything is welcome at all. Thanks

[Al_The_Man]

As a start, there is an old App. note by IR using Hexfets, This is for a 2 quadrant 200a amp drive, with regenerative braking.
The App note number is 941B, and is possibly still available from IR.
Al.

[rkremser]

Can't seem to find the 941b but found 941 (http://www.irf.com/technical-info/appnotes/an-941.pdf) for anyone else interested. I realized I was going to end up running several in parallel and the more i'm looking into this the better i'm starting to feel about it. Prelim costs seem to be on par as well for what i was expecting. At the moment i'm more thinking about the interconnects for this amperage in and out of the circuits not to mention within.

[rkremser]

So anyone deal with the differences between Mosfets and IGBT devices? I've found much higher rated IGBT for the same cost as mosfets. Just haven't used them so i wasn't sure of the tripping points. The other thing i'm concerned is the freewheel diodes to protect the mosfets in the circuit. I assume they either will have to have a 300 amp rating or parallel multiples to get to the same point. Didn't consider the heat of these not to mention the physical layout. I know Mosfets and for that matter have IGBT's have the built in protection Diodes but in most cases people include external diodes, not sure why though.

[extrapilot]

Typically, IGBTs are significantly more expensive than their MOSFET counterparts, except at the high-end of the MOSFET voltage range. But I think you would typically find an IRF-type power FET which is happy in the 40V/70A range for less than $1, where an IGBT might be $5->$20. The upside of the IGBT of course is high voltage capability- which is why they are often seen in automotive ignition systems, etc.

So, assuming all other parameters are equal (except breakdown voltage), if you have a cost competitive IGBT- great. But I expect that paralleling 5-10 NChannel Hexfets from IRF would be far less expensive than the power-equivalent IGBT.

For heat, that is a function of Rdson, and also the time the device is operating in its linear region during the transition from one state to another. And that is a function of your driver stage and your PWM rate. So, you have to do some math there, or call IRF/etc for some help in estimating power dissipation.

On the diode question- most fet body diodes are not rated for high frequency, high power kickback- but more towards basic device protection (limiting exposure of the channel to incidental breakdown voltage). So for inductive drives, often designers will include an outboard diode which is rated for this type of (kickback) abuse. Sizing is perhaps a litte more complicated to calculate, but thankfully, diodes are cheap- easy enough to just throw a higher-capacity diode at this for a one off...

Best,

[rkremser]

Little by little more progress. What is everyone's thoughts on using a mosfet driver ic vs driving them with transistor driven from the pic. I've seen the logic level mosfets but not with the current capacity i'm looking for and would like to drive them harder than the pic can do on its own.

[neilw20]

Don't use IGBTs. At 1.2v ON you will have to dissipate 350 watts at 100% on.
At this voltage you need to use MOSFETs with a low RDSON if you wish to have not much heatsink.
Use one of the IR driver chips that can detect that the MOSFET is saturated,
and just feed the output of the IR device to the gates through individual gate resistors. 220 ohms is a good starting point.
Actually you don't want to switch them on too fast, as their maybe current flowing through a flywheel diode (if you use one and probably will) during the off time. You must make the turn on time is longer than the reverse recovery time of the flywheel diode, else the MOSFET will see a short circuit to supply for the r-r time, and make lots of extra heat.
Use multiple supply capacitors in parallel for low ESR, but because of inductive load, peak currents may be lower than you think. 3 x 2000uF in parallel probably more than sufficient. If they get hot, add more.
Wire all the drains individually to a common point.
Wire all the sources individually to a common point. (negative of supply)
Use individual gate resistors. Important.
Wire common of driver to the common source point with a separate wire.
Do not daisy chain any of the power wiring. You might end up making a big radio transmitter if you do.

If you use 6 SUP75N06-08 Siliconix MOSFETs in parallel they will need 30w heatsink for each one.
The 75 is 75A, N is N-channel, 06 is 60v and -08 is 8 milli-ohms.
Make sure you select a device with very low RDSON otherwise you will need a lot of heatsink.
That is based on 50 amps per device, with an RDSON of 0.008 ohms
3 3"x3" finned heatsinks will be about right. 2 devices per heatsink.

[rkremser]

As for the flywheel diodes, i've seen a lot of diodes with the surge rating around 6x the continuous surge rating which i understand why but given the low duty cycle can i get away under rating the continuous current. I"m having a hard time finding any diodes for reasonable prices that are 80Amps + to parallel. I"m not to keen on trying to parallel more than 4 of these things and trying to keep this controller size down including heatsinks.

As for component selections up to this point
Mosfets - 511-STB230NH03L TO-247 package Rdon 3.6 mOhm
*this choice was mainly due to the package to make it easier to mount to heatsinks
I wasn't sure the best way to manage mounting higher capacity options with powerso or dpak packages.

Mosfet Driver - MAX4420EPA+
Flywheel Diodes - ?
Controller - PIC (not sure which yet dependent on features i want to add)

[extrapilot]

Well, the freewheeling current/power calculation has some dependencies, so it is hard to know what your requirements will be without having that data. One option would be to look at an avalanche-rated fet which can tolerate the sorts of power you will see. A part I have used in high-current PWM applications is the IRFB4110, which has some good characteristics as regards body diode. It may be that a couple of these in parallel can provide you the power control and back-emf management you require without additional tvs-type diodes. Certainly, it would be a good idea to call IRF and detail out the specifics of the load in question- they have good engineering in the support department.

As for drivers, I prefer IRF's parts, but as long as it meets the drive requirements, its all good.

On the PIC, yea, there are a ton of options. But some of the motor-drive-targeted devices have native capability (i.e. 3-phase), which makes it easier to manage and less likely that a code error will trigger some cross-conduction event- which with 200A/12V would be particularly ugly.

Best,
R

[rkremser]

Ok not sure if anyone is still following this one or not but i'm against a brick wall with the freewheel diodes. Of course I redefined my controller and was looking at much higher power levels, Shooting for 800-1000Amp max to prevent controller blowouts. Of course i'm fine in regards to paralleling the mosfets but the diodes have me at a complete loss. The motor that will be driven is going to be a series wound dc motor, any shortcuts that would limit the requirements for the didoes? I was thinking that since the field coils will be de-energized the generator effect of the motor will not be present but the collapsing fields will of course still be there. Any suggestions are welcome, just looking for something to get me pointed back in the right direction.

[extrapilot]

Have you considered an active clamp approach?

[Mariss Freimanis]

You can use MOSFETs as synchrous diodes. A MOSFET has an intrinsic drain to source diode (same rating as the MOSFET) and MOSFETs themselves are perfectly happy to conduct in either direction. Turn the MOSFET on while the diode is supposed to conduct; the MOSFET Rds shorts out the diode to greatly reduce forward voltage drop heating.

We do this in our motor drive bridges. At 7A the intrinsic diode Vfw is 0.9V and dissipates 6.3W. With the MOSFET on (Rds 0.035 Ohm typical), the dissipation is 1.7W, a fair improvement.

Mariss

[TOTALLYRC]

You can use MOSFETs as synchrous diodes. A MOSFET has an intrinsic drain to source diode (same rating as the MOSFET) and MOSFETs themselves are perfectly happy to conduct in either direction. Turn the MOSFET on while the diode is supposed to conduct; the MOSFET Rds shorts out the diode to greatly reduce forward voltage drop heating.

We do this in our motor drive bridges. At 7A the intrinsic diode Vfw is 0.9V and dissipates 6.3W. With the MOSFET on (Rds 0.035 Ohm typical), the dissipation is 1.7W, a fair improvement.

Mariss

Sneaky, very sneaky.
That is why I only think of electronic circuit design.

If I read you correctly, the mosfet bypasses/shorts out the diode and eliminates all/most of its resitance from the circuit when it would normally be conducting and then you turn the mosfet off and let the diode do its job when needed.

Mike

[Mariss Freimanis]

Sneaky? I like that.:-)

What we do is sell resistors, capacitors and semiconductors. We buy them from our vendors, mark up the prices by an obscene amount and then resell them to you. How can we do that? I connect the parts in unusual and sneaky ways before we sell them to you.:-)

Mariss