SOT-227b Based 6 FET XIECHANG

[12p3phPMDC]

Here's some pics of the 6 fet I've been working on.
It's getting closer.

I used 3/8" wide by 1/8" thick copper bar as the bus bar connections.
I had to connect the gate drive from the 6 fet controller via wire.
It ain't pretty, but it's all I could do for now.

Eventually, I'd like to create a board that will adapt the fet array to the 6 fet,
and then after that a BLDC motor control reference design board from ST Micro
adapted directly to the 6 fet array. This will be truly open source as ST gives you the code and schematics for free.
but, one step at time.

But, for now I figured the best thing to do is to stick to the XIECHANG because of the knowledge base.

I have a couple of options for heatsinking.
One is a largest copper heatsink I could find. 7mm ~1/4" thick solid copper, 3" x 4" x 1" thin fins tall.
This allows you to pack in 6 SOT227's within the confines of the sink.

THe other is an aluminum extrusion enclosure. I'm going to use this first for packaging reasons.
It's pretty tall, but it'll allow everything to fit.

This should allow for a relatively easy DIY 100V 200A fet board, mostly because there is no board.
The biggest drawback is the cost of the SOT227s. I found a source for $12 a fet. That is the cheapest I could find.
More typically they are $20 to 30 to $40

One of the coolest features is that the SOT227b ISOBLOC package is that it is isolated on the sink side from the drain
via a built in Ceramic Insulator between the die and the base metal.
This allows you to bolt these things down without insulating mica or kapton or whatever. I need a little arctic silver or another
high quality grease and it should have excellent low thermal impedance.

I did a tl783 high voltage regulator mod to allow the input to go up to 100V.


View attachment 6



View attachment 3
View attachment 2

 

[liveforphysics]

Nice job

It seems they have a wide range of FETs in the 227B package.
I assume this is the part you're using?

http://ixdev.ixys.com/DataSheet/99239.pdf

 

[BatteryMooch]

I have a couple of options for heatsinking.

One is a largest copper heatsink I could find. 7mm ~1/4" thick solid copper, 3" x 4" x 1" thin fins tall.
This allows you to pack in 6 SOT227's within the confines of the sink.

THe other is an aluminum extrusion enclosure. I'm going to use this first for packaging reasons.
It's pretty tall, but it'll allow everything to fit.

Nice job! I love the SOT227 package.

IMHO, stick with the Al enclosure unless you have a fan impinging directly down onto the copper sink's fins from very close. The Al sink is made for convection cooling without a fan (widely spaced fins) and will still work fairly well without one. It will work much better if there's a fan on it, but every heat sink will do that.

However, the close spacing of the copper sink's fins means that there's almost no air movement between them. Without a fan, that sink won't perform much better than a solid block of copper, which is miserably (once the copper reaches thermal equilibrium, i.e., when it heats up).

 

[12p3phPMDC]

Thanks Gentlemen...

I got some samples of an ST and VISHAY (IRF Fets) sot227b fets.
Those are the ones shown in the pics.

I did get the $12 fets from IXYS. and yes, they have a huge offering of SOT227s. They are very similar to the one you picked.
The silicon is rated higher at 360A. In addition they make some bad-a$$ IGBTs in a SOT-227 pack also isolated.
I found a 200V 400A (200A pins) IGBTS for about $20 per.

I bought the IXFN360N10T. If anybody wants some.. I bought 20 from Digikey which was the minimum order and a small lead time quoted at 10weeks, but was actually about 4.
I'll probably sell 6 or 7 of them.
If anybody can find a better price on a high voltage high current SOT227, I'll be suprized and delighted.

PM me. If there is enough interest, we should try to get the cost down further and get a bunch more.
I think these would work great for a soft-start master switch, but folks balked at the idea earlier.
Tiberius super-diodes too...

As Jeremy pointed out, the biggest challenge to current is the thermal. :?
The pins are rated for 200A, so until testing begins, I'm going to assume this to be the limit.

Thanks for the advice Camlight. In addition, the AL extrusion is much more turnkey for
the enclosure in general, as I would have to fab a top side enclosore for the copper sink.

Jeremy is working on some great 6/12 fet designs right now, too. Hopefully, it'll help knock the cost
of a high power 6 fet down. But I think the SOT227 approach is easier, but will definitely cost more.
This place is turning into FET central... I'm excited to see all the controllers being beefed up around here
and look forward to what Jeremy comes up with! and LYEN too.

My next steps

1. On/off Switch
(What do you guys think, just use the battery connectors or is this a nice feature? I can't make up my mind)

2. Hall sensors on the HXT
-60 degree sensors would be the cleanest, I'm thinking I want to put them in the slots.
-GWHYs 17.14 seems to work great for him
-Need to review BigMooses flux mapping.

3. Connectorize hall wiring,
4. Connectorize throttle wiring.
5. Drill out end panels for grommets
6. Heat shrink terminals/general insulation
7. Clip leads...
8. lots of little things.

 

[Doctorbass]

Nice Job!..

Let us know the measured performances!

Also They are alot easier to replace than multiple soldered to-220 when some blow!

I saw that they are rated with 7.6 mohm rds on.. that mean at 100A, each of these will dissipate 75W.. or 37W at 50% duty?.. Am i wrong? that would be alot of heat!

Doc

 

[12p3phPMDC]

Heres the datasheet.

The IXFN360N10T has Rdson @ 2.6 mOhms, so cut your numbers by a 1/3.

And check out the Rthjc of 0.18 C/W and an Rthcs of 0.05 C/W.

You have a good point regarding ease of repair/service, just screws.
But, you have to pay for this ease of service.

I also included a few other datasheets for reference.
The ST part and the Vishay. They are all comparable, but the IXYS is better and Cheaper!!!

View attachment (IXFN360N10T.pdf
View attachment fb180sa1.pdf
View attachment ste250ns10.pdf

And here's little CADdoodle that uses flag terminals and cable bussbar as an altenate design.

 

[Doctorbass]

Oh.. Sorry I took the number of the datasheet that Luke posted:

http://ixdev.ixys.com/DataSheet/99239.pdf

so 2.6 ohms is good enough taking account of the Rth. Nice!

To mount them on a narrow controller with the style of case of the infineon, it could be great to mount 3 on each side o the case.. .like one side positive and one side negative.. 3 phase per side. Bus bar at the top and the rest of electronic on the normal baord location..

Doc

 

[liveforphysics]

Since digi-rip-off charges $12/fet for 20 of them, if previous digi-rip markup percentages are an indicator here, I think they could be had for $6 apiece somewhere with a more direct buying path from IXYS.

I know with the IR parts they sell, I can generally buy them for 1/2-1/3rd of the prices listed on digi-rip.

 

[12p3phPMDC]

I know, that's what I thought, but I went to inside reps for Arrow, an IXYS distrubotor they can't get the IXFN360N10T, and
the closest match was $19 per.

I couldn't believe that Digikey had the best price.

I agree, We gotta figure out how to buy Direct.

 

[BatteryMooch]

Heres the datasheet.
The IXFN360N10T has Rdson @ 2.6 mOhms, so cut your numbers by a 1/3.
And check out the Rthjc of 0.18 C/W and an Rthcs of 0.05 C/W.

Be careful about the Rds-on specs, they're only valid for a 25C junction temperature for very short bursts. At, let's say, 125C (reasonable for a hard working FET), the Rds-on rises to about 4.7mOhms...almost double.

But, you're right, the thermal resistance numbers for that FET are incredible!

 

[12p3phPMDC]

Be careful about the Rds-on specs, they're only valid for a 25C junction temperature for very short bursts. At, let's say, 125C (reasonable for a hard working FET), the Rds-on rises to about 4.7mOhms...almost double.

But, you're right, the thermal resistance numbers for that FET are incredible!

I hear you, but that story is the same for any fet.... And you're right they are great thermal numbers. Even better considering no insulating layer is required between the sink and the fet. :wink:

I got some leads on direct pricing from IXYS or ARROW ...although a group buy would be required, which at this point, I don't think there is enough interest.
~$10 per fet @$2500 minimum order, or ~250 pcs. I'll check on minimums through distribution. IXYS has 900 or so in factory stock and it's a brand new part.
Buying direct from IXYS requires credit checks...etc... so ARROW would be the way to go. Just have to get them to add the part.

But all in all, digikey doesn't seem to bad on this one. Maybe it's because they didn't stock it. Also found out it's a brand new part here to stay, so somehow, IXYS
is cranking these things out for less. Gotta love economies of scale.

 

[Thud]

Bump,
Update please.

 

[12p3phPMDC]

Like others on this forum, I want an independent power board. The idea is to be able to drive the power board with small signals, like TTL, or maybe
the existing gate drive on the XIECHANG.



I've been chipping away at a PCB that has 14A peak gate drive chips (the 5 pin through hole part next to each fet).
I need to put down some capacitor footprints next to the gate drive chips. I'm thinking that I'm going to use low esr surface mount film caps.
The inductance to the gate should be low and this IXDD414 chip can drive a class 9 size fets or IGBTs at high switching speeds.



Analog devices makes a really nice integrate ISOLATED DC/DC converter and half bridge/phase leg drive chip with ISOLATED floating high side drive.

http://www.analog.com/static/imported-files/data_sheets/ADuM6132.pdf
I have more routing to do, but it's coming together.




Below this circuit board lies the 1/8" thick copper bus bar that carries the real bulk of the currents. It
will all bolt together with 4mm screws and 10-24 brass screws in the copper bars.

This design should make it easy to swap between IGBT's like the IXGN400N30A3 all
in an ISOTOP SOT-227b package.

http://ixdev.ixys.com/DataSheet/DS99592B(IXGN400N30A3).pdf

or 100V 200A fets. Man, IGBTs are impressive

THe IGBT claims 300V 400A, so derated to 200V 200A package limited with good cooling, and you can
drive 200V x 200A = 40,000 W.

Please put in your ideas on this so we can move it forward..
Help Me finish this, so that we can make boards that plug into the existing 6 fet board!!!!

Machined copper bus bars will make it easy to bolt together
a beefy copper bus system.

View attachment 1


THe extrusion I have can easily be adapted for water cooling as well.

Questions:

Can we drive the IXDD414 with the existing gate drive ciruits on the 6 fet board, which in turn can drive
some monster transitors?

 

[12p3phPMDC]

I found a great new shunt from Vishay that makes it
easy to integrate with my bolt down design.



WSM5515 shunts are available in several sizes.
L1600 = 0.00016 Ω
L2000 = 0.00020 Ω
L2500 = 0.00025 Ω
L3000 = 0.00030 Ω
L5000 = 0.00050 Ω

http://www.vishay.com/docs/30135/wsms5515.pdf

 

[SilverSurfer]

If we could get a good low price Id be interested in joining a group buy. SS

 

[LightningEVAustralia]

Excellent But why stop there? Think bigger. Like 600v 600A IGBT's :lol:

I have 2 Toshvert AC controllers, they have 600v 150A IGBT's and driver boards, combined output is 200Kva.

Think of the possabilities if one could just use the $25 6fet with a simple mod to drive Hi powered IGBT's, Instant $5000 controller, suddenly EV's will become affordable.

Ofcourse lets start with the small one and go from there, I'm in.

 

[12p3phPMDC]

Yes, yes,

you bring up one of my dilemmas.

I'm thinking that it may be a good idea to just do a phase leg board driver.

There are a bunch of options, but the ADUM6132 seems like such a good fit...I have to
check on the isolation...what kind of voltages can you run with this little guy?

If we do a phase leg board, then you keep adding fets to your hearts content. Then connecting
everything in a 3 phase bridge becomes your next problem.

I'm going to use 6 ga wire for this system. It could be bumped up.
The ISOTOP package is only good for 200A, so
eventually you have to parallel them...

 

[12p3phPMDC]

Thanks for the link and the image Camlight.

 

[BatteryMooch]

My pleasure.

 

[rkosiorek]

my goodness. months of ho-hum stuff and then within days 2 threads to get exited about. this one and the one with the handcycle.

i'm interested in how this comes out. especially the isolated drivers thing.

rick

 

[12p3phPMDC]

I need to send an email to e-crazyman, and see if he'll build the 6 fet with "no loads" on the gate drive section.
That way, we don't have to spend alot of time reworking boards, and just hook the microcontroller outputs
to the ADUM6132 fet drives.

 

[Microbatman]

Very interesting.
Like the idea of being able to do away with having 12 to 18 FETS in a row.
And at the same time increasing reliability and robustness.
Really like the way these are designed. Keep us posted about progress.
While we are on the subject of alternatives to widely accepted controller items.
Any ideas on the economics and reliability factor (less failure due to heat) going from a hall based sensor to an optical based sensor?

 

[12p3phPMDC]

Thanks Microbatman,

For 100V to 200V , 200A size controllers, the ISOBLOC package is hard to beat.

If we make a modular system, with the fet drive mounted right on top of the package, in a 2 fet phase leg configuration,
then they can stack 6, 12, 18 fets, and drive huge currents, 200A, 400A, 600A....
I think that coming up with low inductance, distributed capacitance, bussbar system is key to the super high current stuff.
It's time to rethink this board. 2 phase leg boards would also make a great H bridge.

A short length of ribbon cable with connections is probably the easiest way to connect the brain board to the brawn board.
It's easy to peel off individual stands of a neatly gathered strip and feather them out to the TO220 fet connections.
This brawn board should also be able to interface with a CASTLE CREATIONS or Turnigy brain board...
But these controllers would still have ~50V upper limits. But it would be cool to run it sensorless as well.

re: economics or reliability of optical
Hmm... haven't thought about it much... Optical seems like a neat way to go, and I think most packaged encoders are optical,
A standardized encoder module could be a cool thing... maybe optical is the way to do that ?
Right now, I'm for whatever is easiest on this front, therefore, I'll be going with hall sensors.

 

[Ratking]

Hello

Are there any way to just use the board from a rc controller to controll the mosfets?
I understand that the drive dont handle many large fets, but they should be able to drive one or two 200amp fets?

Thanks

 

[12p3phPMDC]

Ratking,

Sure, you just need to strip the fet drivers off an RC controller and drive the new fet bridge with appropriate FET drives.
As easy as that sounds, it's a fair amount of work until we come up with logic level FET Bridge

As an example, most of the BLDC development kits allow you to run sensored or sensoreless, by changing jumpers.

http://www.st.com/mcu/modules.php?name=mcu&file=devicedocsmdt&FAM=15&DEV=ST7MC-KIT/BLDC.

The dev kits have the same problem when driving big fets quickly.