I am not one to argue about something i dont know anything about; but i have seen electric submarine motors hooked to slitters i think that is the spelling anyway they are machines that cut steel of different thicknesses and such. they ran like 13000 volts dont know the amps. This is at Worthington Steel in columbus ohio. I have seen the control boxes explode thankfully i was at the other side of the area of the explosion. some was not so lucky. Any way i would not be one to go electric on something like this. A good fuel cell and a well tuned gas engine is rather safe. I have played with goKarts most of my life. I have two here now. One my son's with 6.5 hp and my and my wife's gokart 250cc water cooled engine with both having a good welded roll cage signs of a better financial life before disability . But if you would like to see a little more about what this fellow is trying to do check out this url http://www.speedace.info/killacycle.htm i have been following this thread thought i would google some info this is on of the pages i found.
To your original question "i need a fet or several that can handle 400 to 500 amps that can be switched with the propellor, do any of you guys know of anything that will do. could i run a bunch of small ones in parallel and have them all switch at once, any info is helpful"
I would definitely say no to paralleling the fet's, because of a potential thermal domino effect that can occur. Using a Multi-Phase PWM scheme similar to how large number of LED's are controlled would be the way to do it to prevent a thermal domino effect, but still dealing with the amounts of current that you mention is going to be very risky.
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔ Beau Schwabe
IC Layout Engineer
Parallax, Inc.
I admire your head start into electronics, keep up the good work and seems to me as though your playing it all safe. I was knocked to death with 300Amps; I was a Nuke welder in the military. Ironic that I was revived by the Doctors with more electricity. Just bringing a real life survivors story dealing with 'Current.' to warn of the dangers.
Ouch! Welcome back to Earth. [noparse]:)[/noparse]
300 Amps was the operating current of the circuit I assume. 300 actual amps through a human body would prolly be a splatter on the wall. Any idea what the voltage was?
Oh, reminds me of the Mythbusters episode where they overcooked some hot water cylinders. They kept the commentary level throughout, to reveal the spectacular results with as much surprise as they themselves got, I think.
Sure did surprised me how explosive one of those could be.
i have a relay that can handle 28 volts and 400 amps, i figured out that i cant get 400 amps with the batteries im using, my question now is, is it possible to build a transistor amplifier that can be switched on and off, the relay once controlled the landing gear motors on large jets, my·uncle got it for me said it was good just it was just past its shelf life, i dont know the switching voltage, but i plan on·hooking a batter and lightbulb up and then another battery and then slowly increase voltage until it switches on, im thinking the switching voltage is between 12 and 28 volts, any suggestions
There is no need to go parallel at all. 200 amps is easy enough with the right mosfet, especially since you are in the prime voltage range of mosfets.
Next is to decide on how strong a mosfet driver you're be needing. As Rayman has said, you are dealing with rise/fall and dwell times. The faster the transitions are compared with dwelling the less heat loss there is and the smaller the heatsink needs to be. That SSR I designed didn't need any heatsinking at all, yet I didn't hammer the slew rate either, because the frequency was less than one hert.
Also, as has been said, you have to deal with tons of switching noise so the appropriate snubbing/filtering would be wise.
Moving on, it's usually a done thing in motor control to use internal current sense with feedback to regulate the current. The feature list grows from there ...
Or you could just go and buy a proven motor drive and wire the Prop into that. Depending on the features of said drive, you can command it with analogue torque (+10-10 volts) or analogue speed (same signal) or digital position (pulse stream) or even using a comport for sending any of these commands to it.
I'll assume you are going to be using a brushed DC motor or motors. You're not after position control so having any feedback device on the motor would be overkill. You could possibly get away with steppers but I suspect you'd get annoyed with them. Most other stuff is mains voltage and higher.
http://www.robotmarketplace.com/products/MAG-S28-400.html···· this is the motor im using, it says 390 amps max, it lists off some motor controllers that can be used with the motor, but they arent rated for that much current, i also have that relay i talked about rated for 400 amps, would it be possible to do some sort of pwm only on a slower scale
A big no for PWM'ing a mechanical relay. Even pulsing heaters at once per 10 seconds is tough on relays. You are wanting at least 1 kHz. The relay would be useful as a main supply for powering up the driver though.
Yep, the motor is brushed, so only two wires to worry about. [noparse]:)[/noparse] The max current, you don't want to exceed. Setup the drive to less than that. In your case you aren't wanting anywhere near that much, so buying/building a smaller drive would be fine.
Your·E-kart would be cool. This idea should be followed through with, but cautiously. Not that 24VDC can kill you, but a few exploding car batteries can seriously mame you for life...
Looking at the specs quick·12 or 16 of this·StripFET Power MOSFET in a parallel modified, multi-point-PWM·H-Bridge (we can·call it·an HP-Bridge) config should give you the current capacity that would exeed the motors rating of 390A by a factor of nearly 2. Why have a motor controller twice as powerful as the motor? Reliability. That way your not stressing the FET's hard even at a full power stall. Of course you would limit the·HP-bridge output with·a certain·duty cycle to prevent damage to the motor. When you design electronic circuits to handle twice the currents you need, you reduce the heat stress on the components and increase the reliability of the device. These pages of an H-bridgedesign is an exelent reference and resource on how they work and go into multiple ways to control it. The StripFET can handle 280A continuous @25C & 200A continuous @100C with a pulsed current of 1120A @24V! The continuous rating at 100C would seem a wiser choice for a more reliable design purpose.
The datasheet says that PTOT = 300W That's only 12.5A @ 24V, If i had to guess, that would be the non-heat sunk rating. 200A/24V @100C·= 4800W! (no way can that be right...) It's a bit funny that the Nominal electrical characteristics of this FET·are also the same as the·ABS maximum ratings... So your motor can handle up to 390A @ 24VDC? That's 9.360KW! The design x2 rule is simple - choose components with twice the power rating you require (or design with twice as many than you need), and since that motor you have at 390A/24V is making·12.5HP (converted),·I doubt you'll be running the full 390A full time. A known fact is it only takes 17HP to keep a pickup truck moving 55MPH on a level road. So depending on gearing and not pushing the motor to its ABS limits, on my Birel racing kart chassis should push 50MPH+. A side note about aerodynamics, when you have to push through a fluid (like air), resistance to motion is on a logarithmic scale, not linear. My '99 KX 125cc shifter kart·has been clocked over 120MPH but is pushing around 50HP.
Back to the controller design. With such a high current application, it's going to want to generate ALOT of heat so if you go with a design that limits the power to 75% of what the motor can handle, you can reduce your costs a bit, make it more reliable, and have less heat related problems.If you were to·design the controller around a MAX of 600A @ 24VDC, then limit the duty cycle·you won't have to worry about making an expensive puddle of molten copper windings. So if you design the bridge to output 600A MAX @ 24VDC and limit the HP-bridge·with a 50% or less duty cycle, which would produce around·300A, then the FET's·will only work 50% as hard as the motor working at 75-80%, (will also give you tunability, ie. you can turn the juice up some without worry).
At 100C those StripFET's are rated at 200A, so you'd need only 3 of these FET's per corner of the H-bridge for a MAX of 600A! That's 12·needed (with reversing & regen/braking), 9 needed with only forward & braking, and only·3 needed for foward only (no braking or reverse). The following features should be considered:
Heat sinking - Air or liquid?·You would have to determine how much heat is being disipated by each of the FET's, here's a good page for determining that.·Going with my original thoughts, something this powerful would dictate to me that i make my power rail/heat sinks·from square copper tubbing (easy to solider together and solider to, the best choice for heat sink/power rails) and pump distilled water or coolant through them via hoses and back through a radiatorlike the ones you can get for a liquid cooling system for your PC.
Symetry - Wire size is important with high powered electronics but where they tie into the power rails is also an important design factor. The more symetrical your connection with the power rail is the more balanced the power will be across a given set of paralleled FET's.
Wire sizing - care should be taken as for choosing primary wire size from your batteries to your rails, and to your motor, as in if you expect a 10ga wire to handle 150A your in for big disapointment (and maybe a fire). Most car stereos above 1000WRMS will use at LEAST 4ga, most use 0ga primary wire. It's because of the voltage drop through 10-20ft of it while drawing close to 100A for a 1000W system.
Scale - while you can't drive a hundred 20A MOSFET's at the same time, mainly·because the gate capacitances would be ridiculous to try to drive, you can scale (parallel a few high current ones) your power side to a certain extent without adverse side effects.
I know theory looks great on paper, so I'm going to make a version with the other motor on that site you got your motor from.
Some good links -
Liquid Cooling!·- That's right kids, you can be the first on your block to have your FET's cooler than Cucumbers!
H-Bridges & You...·very handy information on multiple techniques on how to control an H-Bridge for Reversing, speed control, ramping, braking, even regenerative braking! All done with PWM... Trivial for the prop.
so basicly your saying that i need to parallel enough of the fets so that it can handle the amps, then i need to make an h bridge, and in this way when i turn on one corner no current flows until i turn on the other current, so all the first fets could be on, and i coul use my 400 amp relay to turn on the other corner, when all fets are on i could then turn them all off and on in a duty cycle that would not allow them to work at full force, which in turn limit my motor, but not to the degree of uselessness, and the initial surge of current to get the motor spinning is dispersed evenly through the fet corner of the h bridge, and then the motor wont draw so many amps cause its turning, right?
Close. . . put the relay down and step away from the relay. · You won't need the relay, take a look at the picture. With the H-bridge wired like this you have alot of control over what it can do. to make it spin either way you would turn on ONLY two corners (upper left & lower right) or (upper right & lower left), but NEVER turn on the same side!!!!! Turning on both left or both right will create an electrical fire!!! You may be able to save your assets by installing (3) 150A automotive circuit breakers in parallel (but i'm not sure if that would protect to 450A).
Anyway, say the upper left and lower right make it go foward, if you turned those off and turned on the upper right and lower left then the motor would spin the other way (reverse!). Also, if you were to turn on both upper OR both lower it would act like using the regular brakes. When you short curcuit the leads on a motor it acts just like a brake, this is because of an opposing magnetic field generated by the motion of the spinning armature·and permanent magnets. Look at howstuffworks.com to see how a brushed DC motor work for more.
Straight ON or OFF a little to coarse to be drivable? Then why not introduce Pulse Width Modulation to the gates? Then you could read a potentiometer and scale it to PWM output to the gates through a driver IC. You can switch to foward, then push on the "gas" pedal that's conected to the potentiometer and have linear control over the output. So now your moving pretty fast, hit the brake pedal gently that's connected to the regular brakes but also to a potentiometer, and the prop is set to only give a certain·PWM pulse length to the upper two gates so you slow down gradually to nearly a stop.
Stuck and need to back up? hit the reverse switch to engage the opposite two gates you used for foward and push the "gas" again easy, clever you, you set the maximum PWM duty pulse length so you can't back up too quickly. I think you really aught to have a look at those links i gave.. took nearly three days worth of hunting for them.
but one of the concerns with that is that if you use paralleled fet and say you use 3, and it just so happens that one triggers a hair early, then the one that triggered early would take the front of the amp surge and it would go pop, or in this case booooooooooom, thats why i only want to use one switching element, do you know of a fet that can handle 400 amps, you pulled up some pretty big ones
- The switching speed is limited. This means there is a period where one transistor can lead the other but still not be taking 100% of the load. Also, extreme short durations of overload is acceptable. Once they are all turned on MOSFETs nicely share the load. Unlike BJTs.
- There is an easy balancing trick for firing parallel MOSFETs. Throw in a series resistor on each Gate. This allows equal and simultaneous Gate charging currents to flow.
I presume the above applies to IGBTs also, even though they are BJT based. This is because the IGBT doesn't saturate. Which is inturn because of it's internal MOSFET frontend.
Maybe that's a big clue to how MOSFETs operate at the fundamental level.
evanh said...
Throw in a series resistor on each Gate. This allows equal and simultaneous Gate charging currents to flow.
*NOTE: series resistors are fine on the gates, but spec that they should be precision 1% tolerance resistors, or a resistor network pack (even better) available in different configs, resistances, and number of R's per pkg. Look it up·at mouser.com And dont forget the clamp diodes!!!!!!!
*NOTE2: putting a resistor in series with a gate would only affect it's total turn on/off time, it won't actually balance multiple fet turn on times. It will only reduce your gate voltage and reduce the upper switching frequency. I just simulated a few·quick virtual MOSFET's·in parallel with Multisim 2001, and that was the conclusion.
I put enough time on R&D on this already.. like two days pining the web for that info. ST.com has both of those StripFET's available, mouser.com stocks one of them. I would highly recomend you contact ST about the suitability for this application and if so how many they would recomend for a paralleled H-Bridge configuration. Don't forget to include your voltage and current requirements 500A @ 24VDC. Why would you e-mail ST about these FET's for your app? Just in case I missed something in the datasheet... I'm not perfect. I'm another hobbyist like you, but that may soon change, since i'm on the verge of being laid off... AGAIN. Autobody sucks!
The datasheet does not specify a TON or TOFF range...· so i guess 10 of these FET's would likely turn on at the same time, give or take a few nS. What I did look at was ISD @ 100C (standard ratings not absolute) but the VSD confused me a bit. I guess the absolute ratings are one at a time not all at once.
Component suppliers have agents scurrying around every city. But for a prototype just order some from a catalogue. Here, in New Zealand, we use Radio Spares or Farnell for most one-off orders. As well as local shops for the common stuff.
Here's a quick selection I made. As you can see, those are nearly all from International Rectifier. When you throw in every manufacturer the selection is huge.
PS: I'd be more inclined to choose the STV270N4F3 over the STV300NH02. The STV270 is designed for 24 volt systems where as the STV300 is for 12 volt systems.
A quick theory on paralleled MOSFET's and thermal balance affect on ISD across the group.
As·TJ increases, so does RDS. Thus if one FET is sinking more current TJ will increase with RDS and the current will naturally seek out the next least-resistive path, likely the next FET in the group with a lower RDSON which will pass the current more easily than the hotter one having a higher RDSON. I believe in this theory and that is why i suggest power connections to each power rail to be centered and if multiple connection points are used, for them to be symetrical. Then heat should spread linearly from the center of the parallel group instead of from one end.
A side note about material selection for power rails: As current increases, voltage will decrease slightly due to natural resistance in the metal. Even pure Silver (Au) has electrical resistance, albeit it's neglegable for 99% of electronic designs. These facts also contribute to current balancing in a paralleled group of MOSFET's.
While this theory and facts can be dismissed for 3 or 4 devices in close proximity, a paralleled group of 10 or 20 might have use for this theory and would lead to a conclusion of having those 10-20 FET's spread across multiple power rails, effectively acting as one FET device. Of course all those gate capacitances are now multiplied by the number of devices.
i got to looking at this one, and employing the things you guys have told, i think im going to buy a bunch, now, you said with several of them i could do regenerative breaking, which is something i havnt much looked at, how would that work,
also where did you get that one program (Multisim 2001)
thanks for all your help i really appreciate it, i went to google and spent 2 weeks trying to find one that would fit, then somebody said "parallel them" and somebody else said " dont they will explode" i did talk to one of my professors about paralleling them and he said that if you heat sink them enough, and dont stall the motor that they should (the way he said should scared me) be ok,
about the power rails i was thinking of just paralleling up some 12 gauge wire i have laying around or just casting a bus bar in my forge
and i know what it feels like to get laid off, my dad just got back to work, we are still feeling the money pinch, and tell me the economy is just fine (lookout i might start to rant, im good at it)
LMAO, googled 40KW BLDC motor and came across this forum, you could get lost in there for hours lol
Check out this motor!! (i know your past this phase - pun intended) 10KW!! and it's about the size of a Mopar alternator LOL... Your motor is 9.36KW MAX
Regen braking? read the pages located in the H-bridges & you link i gave before (hover your mouse over the colored text and click) regen braking is discused as a side effect of one of the controll methods on one of the three pages.
ER uh, I just got one of those erie sickening feelings, I don't think the Amperage specs of any of the MOSFET's we're looking at were rated that kind of Amperage @ the MAX voltage. I believe the MOSFET 220A continuous monster I found was only pushing 1.5V when it got that 220A rating. think about it for a minute, 220A @ 24V? That's 5,280 Watts power disipation in a surface mount? Now 220A @ 1.5V = 330W of power disipation in a surface mount sounds more realistic. LMAO, I thought something was "off", nope, just me. I'm e-mailing ST about these StripFET's anyway and if they will fit into what you and I want them to do. I have confidence that the 220A @ 1.5V scenereo is the case, and if that's the case, thinking your going to buy a bunch is a very good statement:
With the StripFET I picked (and if the 220A rating actually comes from only 1.5V), for a controller capable of 440A @ 26V, you'd need:
26V/1.5V = 17.33333333 (18 StripFET's just to make up the voltage rating and still have the 220A current rating) and that's provided you can run these in series across the rails.
Oh it get's better... for the controller to be rated at 440A @ 26V you will need to parallel that so you have 2 rows of 18 series/paralleled[noparse][[/noparse]1] FET's each! That's 36 "220A" FET's... per corner. 144 StripFET's total for a true 4-corner H-Bridge design. and each of these "quarters" will have between 9-11 mOhms (0.009-0.011 Ohms). Sounds like nothing till you put 350A through the corner. 350A2 x 0.0108 = 1,347.50 Watts worth of heat to disipate when the Junctions are already·over ~100C.
[noparse][[/noparse]1]Series/paralelling is quite simple in practice, but a bit difficult to word so how about a picture? Vertical they are in series and horizontally they are in parallel. With very high powered car amps this kind of speaker connection to multi-KW subwoofer amps is done exactly this way (usually for competition). The purpose of paralleling the FET's is to add current capacity to the system and to balance the current overt the group of FET's, and you'd series them for the system to handle more voltage, as in the calculation above. The picture is one corner of an Experimental 11KW H-Bridge design. To keep gate capacitances low and managable, the majority of the gates are tied to the gate array input. You can have the GATE ARRAY on all the time, this will not induce power to transfer from the rail to the motor, for that... You'll also need to apply a PWM signal to the input labled PWM Gate for power to transfer from the rail through the·PowerFET array to the motor.
This design allows you to drive·4 gates instead of 72 gates in the traditional H-Bridge, allowing for higher PWM frequencies, lower costs because you only need a single gate driver IC to drive the whole controller from a·MCU. The GATE ARRAY input has an added feature of giving the designer a safety for the circuit, by·not driving·the GATE ARRAY (keeping the power rail from transfering power to the mid rail), false pulses arriving at the PWM Gate cannot activate a potentially dangerous situation. However not foolproof, if the gate array input and a pulse arrives at the PWM Gate the whole array will conduct, and in an H-Bridge deign, if a lower corner and upper corner on the same side of the motor conducts you will then be directly shorting the power rails which would be fataly bad.
* A quick note about the picture - Motor terminal A = Mid rail
It's going to be an expensive exercise me thinks. I forgot to mention that some of those more powerful parts, ie: your chosen IRF2804, far exceed the package current rating.
At any rate here's an old work horse, L292. You'll be able to pick up many of the common features of a motor drive from this.
Ah, Rinks, the Vds of 1.5 volts is worst case ON state condition. It's not the maximum switching voltage. That STV300 is spec'd for switching a 280 amp device at up to 24 volts.
There is a very good description in the link below that I strongly suggest reading and understanding ... · http://www.mcmanis.com/chuck/robotics/projects/esc2/FET-power.html · ·
RinksCustoms, ·
That schematic that you posted won't work without severe voltage drop effects dictated by the bandgap
voltage required across the Gate-Drain junction to turn·each transistor on in the first place.· The Vgs
will be higher the closer you are to the negative rail, and the Vgs will be lower the closer you are to the
positive rail. ·
If the transistor Chain does turn on, you will definitely have a thermal event starting with the transistors
that are closest to the positive rail and then migrating towards the negative rail.· This is because the Vds
will be higher the closer you are to the positive rail and the Vds will be lower the closer you are to the
negative rail.· As a result, the higher Vds voltage will generate more heat initiating the thermal event.
Depending on which way the transistor fails... Full "ON" or full "OFF" will determine if the event stops
there or avalanches to the next available series transistor in line. · ·
As it is, the thermal mechanical limitations of the TO-220AB package pins can only handle 75 Amps, but that
rating is just at the verge of the pins physically melting.·· The cross-sectional area of the pins according
to the data sheet worst scenario is about .53 square millimeters (1.15mm x .46mm)... working this backwards
and trying to determine an equivalent wire diameter... ·
A = pi * r^2 ·
.53 = pi * r^2
.53/pi = r^2
.169 = r^2
r = .411
diameter ·= .822 mm · http://www.powerstream.com/Wire_Size.htm· ·
This equates to a wire gauge of about 20.· Under conservative guidelines using 95 Circular Mills per Amp, a 20
gauge wire can deliver about 11 Amps.· For a TO-220AB package to be rated for 75 Amps, you would have to be using
14 Circular Mills per Amp. To put this in perspective a FUSE, depending on the metal used, is typically rated in
the same ball-park ranging from 10 to 20 Circular Mills per Amp. · ·
1973.55 CM (Circular Mill) = 1 square mm (milli-meter)·· ·
...So the leads on a TO-220AB transistor are about 1046 CM· (.53 x 1973.55 = 1045.98)
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔ Beau Schwabe
IC Layout Engineer
Parallax, Inc.
Post Edited (Beau Schwabe (Parallax)) : 9/25/2008 4:10:39 PM GMT
Also, Beau, I'd expect the current carrying capability of a TO-220's pins to be lower than 75A since the pins are generally tin, not copper or aluminium.
(Of course, the conductivity problems could be solved if you could cool the tin leads with liquid helium to make them superconducting. This would undoubtedly be A, very expensive; B, hazardous for the silicon semiconductor and C, if the semiconductor did work, it would generate enough heat to stop the tin from superconducting)
Comments
Just for information
I am not one to argue about something i dont know anything about; but i have seen electric submarine motors hooked to slitters i think that is the spelling anyway they are machines that cut steel of different thicknesses and such. they ran like 13000 volts dont know the amps. This is at Worthington Steel in columbus ohio. I have seen the control boxes explode thankfully i was at the other side of the area of the explosion. some was not so lucky. Any way i would not be one to go electric on something like this. A good fuel cell and a well tuned gas engine is rather safe. I have played with goKarts most of my life. I have two here now. One my son's with 6.5 hp and my and my wife's gokart 250cc water cooled engine with both having a good welded roll cage signs of a better financial life before disability . But if you would like to see a little more about what this fellow is trying to do check out this url http://www.speedace.info/killacycle.htm i have been following this thread thought i would google some info this is on of the pages i found.
Badger
as i said just for information
To your original question "i need a fet or several that can handle 400 to 500 amps that can be switched with the propellor, do any of you guys know of anything that will do. could i run a bunch of small ones in parallel and have them all switch at once, any info is helpful"
I would definitely say no to paralleling the fet's, because of a potential thermal domino effect that can occur. Using a Multi-Phase PWM scheme similar to how large number of LED's are controlled would be the way to do it to prevent a thermal domino effect, but still dealing with the amounts of current that you mention is going to be very risky.
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Beau Schwabe
IC Layout Engineer
Parallax, Inc.
I admire your head start into electronics, keep up the good work and seems to me as though your playing it all safe. I was knocked to death with 300Amps; I was a Nuke welder in the military. Ironic that I was revived by the Doctors with more electricity. Just bringing a real life survivors story dealing with 'Current.' to warn of the dangers.
300 Amps was the operating current of the circuit I assume. 300 actual amps through a human body would prolly be a splatter on the wall. Any idea what the voltage was?
Sure did surprised me how explosive one of those could be.
Next is to decide on how strong a mosfet driver you're be needing. As Rayman has said, you are dealing with rise/fall and dwell times. The faster the transitions are compared with dwelling the less heat loss there is and the smaller the heatsink needs to be. That SSR I designed didn't need any heatsinking at all, yet I didn't hammer the slew rate either, because the frequency was less than one hert.
Also, as has been said, you have to deal with tons of switching noise so the appropriate snubbing/filtering would be wise.
Moving on, it's usually a done thing in motor control to use internal current sense with feedback to regulate the current. The feature list grows from there ...
Or you could just go and buy a proven motor drive and wire the Prop into that. Depending on the features of said drive, you can command it with analogue torque (+10-10 volts) or analogue speed (same signal) or digital position (pulse stream) or even using a comport for sending any of these commands to it.
I'll assume you are going to be using a brushed DC motor or motors. You're not after position control so having any feedback device on the motor would be overkill. You could possibly get away with steppers but I suspect you'd get annoyed with them. Most other stuff is mains voltage and higher.
Evan
Yep, the motor is brushed, so only two wires to worry about. [noparse]:)[/noparse] The max current, you don't want to exceed. Setup the drive to less than that. In your case you aren't wanting anywhere near that much, so buying/building a smaller drive would be fine.
This motor is crazy! The largest brushless motor I have ever found. Max power is 6.5 Kw! This controller would go good with the motor. www.castlecreations.com/products/phoenix_hv_series.html It would be easy to control with the prop because it has the servo input thing (forgot what it's called). Anyway, with 4 SLA's you'll probably get 6-7 horsepower. This should be good for a small go-kart.
Looking at the specs quick·12 or 16 of this·StripFET Power MOSFET in a parallel modified, multi-point-PWM·H-Bridge (we can·call it·an HP-Bridge) config should give you the current capacity that would exeed the motors rating of 390A by a factor of nearly 2. Why have a motor controller twice as powerful as the motor? Reliability. That way your not stressing the FET's hard even at a full power stall. Of course you would limit the·HP-bridge output with·a certain·duty cycle to prevent damage to the motor. When you design electronic circuits to handle twice the currents you need, you reduce the heat stress on the components and increase the reliability of the device. These pages of an H-bridge design is an exelent reference and resource on how they work and go into multiple ways to control it. The StripFET can handle 280A continuous @25C & 200A continuous @100C with a pulsed current of 1120A @24V! The continuous rating at 100C would seem a wiser choice for a more reliable design purpose.
The datasheet says that PTOT = 300W That's only 12.5A @ 24V, If i had to guess, that would be the non-heat sunk rating. 200A/24V @100C·= 4800W! (no way can that be right...) It's a bit funny that the Nominal electrical characteristics of this FET·are also the same as the·ABS maximum ratings... So your motor can handle up to 390A @ 24VDC? That's 9.360KW! The design x2 rule is simple - choose components with twice the power rating you require (or design with twice as many than you need), and since that motor you have at 390A/24V is making·12.5HP (converted),·I doubt you'll be running the full 390A full time. A known fact is it only takes 17HP to keep a pickup truck moving 55MPH on a level road. So depending on gearing and not pushing the motor to its ABS limits, on my Birel racing kart chassis should push 50MPH+. A side note about aerodynamics, when you have to push through a fluid (like air), resistance to motion is on a logarithmic scale, not linear. My '99 KX 125cc shifter kart·has been clocked over 120MPH but is pushing around 50HP.
Back to the controller design. With such a high current application, it's going to want to generate ALOT of heat so if you go with a design that limits the power to 75% of what the motor can handle, you can reduce your costs a bit, make it more reliable, and have less heat related problems.If you were to·design the controller around a MAX of 600A @ 24VDC, then limit the duty cycle·you won't have to worry about making an expensive puddle of molten copper windings. So if you design the bridge to output 600A MAX @ 24VDC and limit the HP-bridge·with a 50% or less duty cycle, which would produce around·300A, then the FET's·will only work 50% as hard as the motor working at 75-80%, (will also give you tunability, ie. you can turn the juice up some without worry).
At 100C those StripFET's are rated at 200A, so you'd need only 3 of these FET's per corner of the H-bridge for a MAX of 600A! That's 12·needed (with reversing & regen/braking), 9 needed with only forward & braking, and only·3 needed for foward only (no braking or reverse). The following features should be considered:
I know theory looks great on paper, so I'm going to make a version with the other motor on that site you got your motor from.
Some good links -
Liquid Cooling!·- That's right kids, you can be the first on your block to have your FET's cooler than Cucumbers!
H-Bridges & You...·very handy information on multiple techniques on how to control an H-Bridge for Reversing, speed control, ramping, braking, even regenerative braking! All done with PWM... Trivial for the prop.
Power & heat disipation in MOSFET's for the layman·- aka, Thermalresistance for Dummies
StripFET 1·220A/40V @ 100C - choice!
StripFET 2 200A/24V @ 100C
SBR Schotty 60A Rectifier Diode - think of it as genetic-electron engineering.
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
E3 = Thought
http://folding.stanford.edu/·- Donating some CPU/GPU downtime just might lead to a cure for cancer! My team stats.
Anyway, say the upper left and lower right make it go foward, if you turned those off and turned on the upper right and lower left then the motor would spin the other way (reverse!). Also, if you were to turn on both upper OR both lower it would act like using the regular brakes. When you short curcuit the leads on a motor it acts just like a brake, this is because of an opposing magnetic field generated by the motion of the spinning armature·and permanent magnets. Look at howstuffworks.com to see how a brushed DC motor work for more.
Straight ON or OFF a little to coarse to be drivable? Then why not introduce Pulse Width Modulation to the gates? Then you could read a potentiometer and scale it to PWM output to the gates through a driver IC. You can switch to foward, then push on the "gas" pedal that's conected to the potentiometer and have linear control over the output. So now your moving pretty fast, hit the brake pedal gently that's connected to the regular brakes but also to a potentiometer, and the prop is set to only give a certain·PWM pulse length to the upper two gates so you slow down gradually to nearly a stop.
Stuck and need to back up? hit the reverse switch to engage the opposite two gates you used for foward and push the "gas" again easy, clever you, you set the maximum PWM duty pulse length so you can't back up too quickly. I think you really aught to have a look at those links i gave.. took nearly three days worth of hunting for them.
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
E3 = Thought
http://folding.stanford.edu/·- Donating some CPU/GPU downtime just might lead to a cure for cancer! My team stats.
- The switching speed is limited. This means there is a period where one transistor can lead the other but still not be taking 100% of the load. Also, extreme short durations of overload is acceptable. Once they are all turned on MOSFETs nicely share the load. Unlike BJTs.
- There is an easy balancing trick for firing parallel MOSFETs. Throw in a series resistor on each Gate. This allows equal and simultaneous Gate charging currents to flow.
Maybe that's a big clue to how MOSFETs operate at the fundamental level.
*NOTE2: putting a resistor in series with a gate would only affect it's total turn on/off time, it won't actually balance multiple fet turn on times. It will only reduce your gate voltage and reduce the upper switching frequency. I just simulated a few·quick virtual MOSFET's·in parallel with Multisim 2001, and that was the conclusion.
I put enough time on R&D on this already.. like two days pining the web for that info. ST.com has both of those StripFET's available, mouser.com stocks one of them. I would highly recomend you contact ST about the suitability for this application and if so how many they would recomend for a paralleled H-Bridge configuration. Don't forget to include your voltage and current requirements 500A @ 24VDC. Why would you e-mail ST about these FET's for your app? Just in case I missed something in the datasheet... I'm not perfect. I'm another hobbyist like you, but that may soon change, since i'm on the verge of being laid off... AGAIN. Autobody sucks!
The datasheet does not specify a TON or TOFF range...
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
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) : 9/25/2008 1:20:55 AM GMT
Here's a quick selection I made. As you can see, those are nearly all from International Rectifier. When you throw in every manufacturer the selection is huge.
As·TJ increases, so does RDS. Thus if one FET is sinking more current TJ will increase with RDS and the current will naturally seek out the next least-resistive path, likely the next FET in the group with a lower RDSON which will pass the current more easily than the hotter one having a higher RDSON. I believe in this theory and that is why i suggest power connections to each power rail to be centered and if multiple connection points are used, for them to be symetrical. Then heat should spread linearly from the center of the parallel group instead of from one end.
A side note about material selection for power rails: As current increases, voltage will decrease slightly due to natural resistance in the metal. Even pure Silver (Au) has electrical resistance, albeit it's neglegable for 99% of electronic designs. These facts also contribute to current balancing in a paralleled group of MOSFET's.
While this theory and facts can be dismissed for 3 or 4 devices in close proximity, a paralleled group of 10 or 20 might have use for this theory and would lead to a conclusion of having those 10-20 FET's spread across multiple power rails, effectively acting as one FET device. Of course all those gate capacitances are now multiplied by the number of devices.
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
E3 = Thought
http://folding.stanford.edu/·- Donating some CPU/GPU downtime just might lead to a cure for cancer! My team stats.
i got to looking at this one, and employing the things you guys have told, i think im going to buy a bunch, now, you said with several of them i could do regenerative breaking, which is something i havnt much looked at, how would that work,
also where did you get that one program (Multisim 2001)
thanks for all your help i really appreciate it, i went to google and spent 2 weeks trying to find one that would fit, then somebody said "parallel them" and somebody else said " dont they will explode" i did talk to one of my professors about paralleling them and he said that if you heat sink them enough, and dont stall the motor that they should (the way he said should scared me) be ok,
about the power rails i was thinking of just paralleling up some 12 gauge wire i have laying around or just casting a bus bar in my forge
and i know what it feels like to get laid off, my dad just got back to work, we are still feeling the money pinch, and tell me the economy is just fine (lookout i might start to rant, im good at it)
Check out this motor!! (i know your past this phase - pun intended) 10KW!! and it's about the size of a Mopar alternator LOL... Your motor is 9.36KW MAX
National Instruments Multisim 10.1!!! (haha·I have·ver 6)
The pictures you see are exported from TinyCAD
Regen braking? read the pages located in the H-bridges & you link i gave before (hover your mouse over the colored text and click) regen braking is discused as a side effect of one of the controll methods on one of the three pages.
ER uh, I just got one of those erie sickening feelings, I don't think the Amperage specs of any of the MOSFET's we're looking at were rated that kind of Amperage @ the MAX voltage. I believe the MOSFET 220A continuous monster I found was only pushing 1.5V when it got that 220A rating. think about it for a minute, 220A @ 24V? That's 5,280 Watts power disipation in a surface mount? Now 220A @ 1.5V = 330W of power disipation in a surface mount sounds more realistic.
With the StripFET I picked (and if the 220A rating actually comes from only 1.5V), for a controller capable of 440A @ 26V, you'd need:
26V/1.5V = 17.33333333 (18 StripFET's just to make up the voltage rating and still have the 220A current rating) and that's provided you can run these in series across the rails.
Oh it get's better... for the controller to be rated at 440A @ 26V you will need to parallel that so you have 2 rows of 18 series/paralleled[noparse][[/noparse]1] FET's each! That's 36 "220A" FET's... per corner. 144 StripFET's total for a true 4-corner H-Bridge design.
[noparse][[/noparse]1]Series/paralelling is quite simple in practice, but a bit difficult to word so how about a picture? Vertical they are in series and horizontally they are in parallel. With very high powered car amps this kind of speaker connection to multi-KW subwoofer amps is done exactly this way (usually for competition). The purpose of paralleling the FET's is to add current capacity to the system and to balance the current overt the group of FET's, and you'd series them for the system to handle more voltage, as in the calculation above. The picture is one corner of an Experimental 11KW H-Bridge design. To keep gate capacitances low and managable, the majority of the gates are tied to the gate array input. You can have the GATE ARRAY on all the time, this will not induce power to transfer from the rail to the motor, for that... You'll also need to apply a PWM signal to the input labled PWM Gate for power to transfer from the rail through the·PowerFET array to the motor.
This design allows you to drive·4 gates instead of 72 gates in the traditional H-Bridge, allowing for higher PWM frequencies, lower costs because you only need a single gate driver IC to drive the whole controller from a·MCU. The GATE ARRAY input has an added feature of giving the designer a safety for the circuit, by·not driving·the GATE ARRAY (keeping the power rail from transfering power to the mid rail), false pulses arriving at the PWM Gate cannot activate a potentially dangerous situation. However not foolproof, if the gate array input and a pulse arrives at the PWM Gate the whole array will conduct, and in an H-Bridge deign, if a lower corner and upper corner on the same side of the motor conducts you will then be directly shorting the power rails which would be fataly bad.
* A quick note about the picture - Motor terminal A = Mid rail
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
E3 = Thought
http://folding.stanford.edu/·- Donating some CPU/GPU downtime just might lead to a cure for cancer! My team stats.
At any rate here's an old work horse, L292. You'll be able to pick up many of the common features of a motor drive from this.
·
http://www.mcmanis.com/chuck/robotics/projects/esc2/FET-power.html
·
·
RinksCustoms,
·
That schematic that you posted won't work without severe voltage drop effects dictated by the bandgap
voltage required across the Gate-Drain junction to turn·each transistor on in the first place.· The Vgs
will be higher the closer you are to the negative rail, and the Vgs will be lower the closer you are to the
positive rail.
·
If the transistor Chain does turn on, you will definitely have a thermal event starting with the transistors
that are closest to the positive rail and then migrating towards the negative rail.· This is because the Vds
will be higher the closer you are to the positive rail and the Vds will be lower the closer you are to the
negative rail.· As a result, the higher Vds voltage will generate more heat initiating the thermal event.
Depending on which way the transistor fails... Full "ON" or full "OFF" will determine if the event stops
there or avalanches to the next available series transistor in line.
·
·
As it is, the thermal mechanical limitations of the TO-220AB package pins can only handle 75 Amps, but that
rating is just at the verge of the pins physically melting.·· The cross-sectional area of the pins according
to the data sheet worst scenario is about .53 square millimeters (1.15mm x .46mm)... working this backwards
and trying to determine an equivalent wire diameter...
·
A = pi * r^2
·
.53 = pi * r^2
.53/pi = r^2
.169 = r^2
r = .411
diameter ·= .822 mm
·
http://www.powerstream.com/Wire_Size.htm·
·
This equates to a wire gauge of about 20.· Under conservative guidelines using 95 Circular Mills per Amp, a 20
gauge wire can deliver about 11 Amps.· For a TO-220AB package to be rated for 75 Amps, you would have to be using
14 Circular Mills per Amp. To put this in perspective a FUSE, depending on the metal used, is typically rated in
the same ball-park ranging from 10 to 20 Circular Mills per Amp.
·
·
1973.55 CM (Circular Mill) = 1 square mm (milli-meter)··
·
...So the leads on a TO-220AB transistor are about 1046 CM· (.53 x 1973.55 = 1045.98)
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Beau Schwabe
IC Layout Engineer
Parallax, Inc.
Post Edited (Beau Schwabe (Parallax)) : 9/25/2008 4:10:39 PM GMT
Also, Beau, I'd expect the current carrying capability of a TO-220's pins to be lower than 75A since the pins are generally tin, not copper or aluminium.
(Of course, the conductivity problems could be solved if you could cool the tin leads with liquid helium to make them superconducting. This would undoubtedly be A, very expensive; B, hazardous for the silicon semiconductor and C, if the semiconductor did work, it would generate enough heat to stop the tin from superconducting)