Better mosfet then eGaN possible ? / General Science and Electronics / Forums

Forums

4hv.org :: Forums :: General Science and Electronics

« Previous topic | Next topic »

Better mosfet then eGaN possible ?

1 2 3 next

Move Thread

LAN_403

Linas

Tue Oct 25 2016, 07:54PM

Registered Member #1143 Joined: Sun Nov 25 2007, 04:55PM
Location: Vilnius, Lithuania
Posts: 721

Hello, for my huge CNC project i need to make quite spindle. That means running BLDC motor with pure sinus generator.

Now i have all controlling figure-out ( with 100ps PWM on fpga ), i am planing to use 200-500KHz AC/DC power supply for motor that could adjust power in voltage in milliseconds, and from that i will be running 3 buck converter at 500KHz-1MHz to get pure sinus after LC filtering, and for mosfet i will generate 0-100% PWM signal based on phase angle. Quick voltage regulation is what allows me to run full sinus PWM signal to Buck converters.

Now here is the problem. Spindle is 1-6kW, and it will run from 100V i think just to keep current low. So i will need around 150V mosfets.

epc-co does have EPC2033 eGaN mosfets that can run at 500KHz with no problem, The problem is that they are bitch to make pcb and solder them. ANd they are so small that i don't
trust them.( so for my project i will run like 9 buck converters with and parallel 3 phases to get current i need )

Can any one show me better mosfet than this one, that could be able to work with very high PWM frequency and have very low on resistance ? Becase i was unable to find any devices that can compete with eGaN, and i really need to get better package for my motor controller. I can have more heat, that is not a problem, but i don't want to use eGaN devices, because they are just so small :(

Blackcurrant

Tue Oct 25 2016, 09:55PM

Registered Member #2989 Joined: Sun Jul 11 2010, 12:01AM
Location: UK
Posts: 94

maybe try

if it's 6Kw then 100V@60A
I think the legs of a TO247 melt at about 70A but you would need to read up on the package type. (don't melt but it seems to be a manufacturing recommendation of about 75A)
Also calculate heat dissipation and gate input drive requirements for these bigger mosfets+clamp circuits for the inductive switching?

Linas

Wed Oct 26 2016, 05:03AM

Registered Member #1143 Joined: Sun Nov 25 2007, 04:55PM
Location: Vilnius, Lithuania
Posts: 721

If you look at the specs, eGaN is at least 3x better than mosfet's you showed me. Not to mention that eGaN ton, toff, rise and fall times are even not specified, because they are in ns range.

Problem is that i have to push 500KHz in order to work, or have some kind different approach.

Hydron

Wed Oct 26 2016, 08:21AM

Registered Member #30656 Joined: Tue Jul 30 2013, 02:40AM
Location: UK
Posts: 208

Saw those at a trade show last year. Weird looking package, but max gate drive voltages aren't so damn low.

DerAlbi

Wed Oct 26 2016, 08:18PM

Registered Member #2906 Joined: Sun Jun 06 2010, 02:20AM
Location: Dresden, Germany
Posts: 727

Hi Linas again.
Let me tell you: you are hunting ghosts

500kHz PWM is not so much of an issue for a normal mosfet. So there is no particular need for super high performance magic mosfets.
Power(=heat) management is far more important in this case. Buck-converters have inherent hard turn off behavior (speaking of the mosfet turning of at full current under full voltage) and soft turn on (no voltage during turn on).
This means switching losses will be quite significant with 500kHz.
Since you use parallel phases (good choise!) you do not need to hande full current on every phase. This releaxes things, still switching losses are high, maybe as high as the actual conduction losses.

If i were you, i would not so much focus on the mosfets inherent capabilities, but concentrate on cooling.

Ballparks:
For a buck converter you can maybe reach 95% efficiency. Given 6kW load, means 5% of that power (=300W) will be loss in your electronics. Wohooo.
Some of this Power will be inductor losses [core losses and skin effect if the indcutor is not optimized for 500kHz; a bought inductor is most likely not high frequency optmized ]. So lets say 50% of that loss is in the mosfets. This still means 150W. Now since you have 9 phases this reduces to totally managable ~15W for every mosfet. But the package needs to be able to handle this kind of loss. TO220 is sufficient.

HOWEVER: i do honestly not beleave that you actually need such high frequency in the first place.
The thing with mechanical stuff is that their time constants are inherently quite long (meaning mass = C, elastisity = L, damping = R). You will not be able to move/vibrate a spindle with 20kHz by motor current control due to the mass of the rotor.
A motor will not only run smooth if the input signal is smooth, but it also runs smooth because the rotor can simply not run not-smooth. (Its a mechanical low pass on its own).
This means you should choose your switching frequency as low as possible (to avoid swithcing losses) to not be audible (maybe 40kHz - 80kHz).
I think your hunt for signal fidelity therefore is misguided.
Using a lower frequency takes a lot of strain off the mosfets due to reduced switching losses. However it will increase the size of your inductors, because you still need filtering (to avoid eddy currents in the motor). But given the inductors power loss is quite high, they must be heavy either way.

Linas

Thu Oct 27 2016, 11:11AM

Registered Member #1143 Joined: Sun Nov 25 2007, 04:55PM
Location: Vilnius, Lithuania
Posts: 721

Well, it is a good place to start by using right devices from the beginning.

I had lot of thinking and get the idea how to simplify things a bit.

My goal is pure sinus driver with ability to have 1-2KHz sinus wave. So what i can do is to use 3 phase shifted DCDC converters with synchronous rectification, run this device at 200-400KHz with some SiC mosfet on 220 or 3 phase power input, so i have small power dissipation over large area.

1477566679 1143 FT178055 Untitled

In synchronous rectification i will use multiple eGaN mosfet's because i don't think i was be able to have batter heat performance . As a bonus, this system should be very rubust, because H bridge will not see any nasty loads and back EMF, that would kill Buck converter, and rectification mosfets can be very high power and they are so easy to drive.

From controller point of view, all i need is to have 3 analog sinus generators that connects to comparator, and LTC3722 will do the rest.

DerAlbi

Thu Oct 27 2016, 07:20PM

Registered Member #2906 Joined: Sun Jun 06 2010, 02:20AM
Location: Dresden, Germany
Posts: 727

Linas, think about what you are actually doing maybe another way:
You could also call your setup a "Class-D Audio-Amp" which shall produce a good since wave. I hope you agree with that. Motor <-> Speaker. Both inductive Energy converters.

Now look at Class-D Amps. Their design problem is also the high switching frequency, because they are actually just step down converters with an active diode (sound like your design, right?)
Now.. the typical tradeoff in these Amps is "Switchingfrequency = HighestFrequency*10". Leaving an audio-amp with 200-300kHz operation frequency which is deemed sufficient to produce a high quality signal.

Applying the same rule to your design you would end up with only 20kHZ switching frequency which would produce a clean enough sine wave given the right filter. However i would say this frequency can be audible and might lead to audible resolution problems on lower RPM, so i would suggest a higher frequency. As said before: maybe 40kHz - 80kHz.
This must be a tradeof beween inductor size and switching losses and at this power levels this should be an educated decission instead of a wild guess.
I mean.. whats your (engineering)argument to go for those insane high frequencies?
Dont tell me its signal fidelity, because my next question would be why a motor would need that

Some additional thoughts:
The problem is that your mosfets may have nice characteristics, but 7mOhm is not that low for that kind of load (high current). The power dissipation capability of the package does not seem good, except you manage to put an SMD-heat sink on top (but i dont think thats enough).
So the road will lead to other mosfets. To combat the very possible worse switching behavior of other mosfets, use a lower frequency.

Linas

Fri Oct 28 2016, 05:14AM

Registered Member #1143 Joined: Sun Nov 25 2007, 04:55PM
Location: Vilnius, Lithuania
Posts: 721

yes, you are right, i will not going put high frequency just because i want clean sinus. I will use as low frequency as i can.

All i want is quite motor for spindle, and BLDC is a way to go, Servo would be better, but they are high voltage devices usually.

i will make 100-300W single phase and see how it works

Sulaiman

Fri Oct 28 2016, 10:15AM

Registered Member #162 Joined: Mon Feb 13 2006, 10:25AM
Location: United Kingdom
Posts: 3141

Linas,
I suspect that you have a plethora of CNC related thoughts bouncing around in your head
I think that you should step back from your project and specify what you want your CNC project to do,
over-specifying requiremnts costs a lot of investment.
e.g. spindle motor 1 - 6 kW
your choice of spindle motor power affects its weight and maximum cutting speed
so all three axes need to be engineered to match the spindle motor....

drive frequency 1 - 2 kHz .... 60,000 to 120,000 rpm !
(the highest rpm spindle motor/drive that I remember us repairing was 60,000 rpm ... scary !!)

There are reasons why high speed spindle motors are long and thin - inertia and centrifugal forces.

Have you eliminated simple inverter drives like this,
Â£135, in a box, 1-ph in, 3-ph out, 24,000 rpm, voltage/current/temperature monitoring, programmable ramping, RS485 comms..... ......

(one per axis plus at least one spare)
It would at least remove some of the design/build/modify burden
and be a lot easier to maintain if inherited :)

at the top end, servo- or positional-feedback allows thread cutting etc.

I have negligible machinist skills, I've not designed or built much for years, my electronics knowledge is dated,
so just thoughts based on recent repairman experience.

Linas

Fri Oct 28 2016, 01:22PM

Registered Member #1143 Joined: Sun Nov 25 2007, 04:55PM
Location: Vilnius, Lithuania
Posts: 721

2Khz for BLDC is not 120krpm, you forgot that BLDC can have 10-40 poles, that means driving frequency have to be much higher by the number of poles. 2Khz for 20 poles are just 6Krpm

1 2 3 next

Moderator(s): Chris Russell, Noelle, Alex, Tesladownunder, Dave Marshall, Dave Billington, Bjørn, Steve Conner, Wolfram, Kizmo, Mads Barnkob