Micro Lebowski Controller - DIY 3kw

[ElectricGod]

No leboski your numbers are overly optimistic.
Truth is you forgot the sag... Likely a 100v battery will sag to 90v under load.
Also you will never get a 100v fet to survive at its limits with a 100v battery.

The numbers for this would be 84v absolute max fully charged and that will sag to ~75v under load with a good battery.

That gives you 4500watts which is still optimistic....

Who are you and what did you do with Arlo1?

When did you change your tune about how much current a TO-247 device can carry reliably?

Im still here a TO-247 can do amazing current up to the leg limits. But the 4468 can not. You and I both found the 4468 to be limited around 50 rms amps.

I have some bad ass IXYS parts that do better as well some IR igbts in a 247 package that will do leg limts cold and ~100-110 rms when warm.

The 4468 is a fairly robust mosfet. I run a 12 fet controller that has AOT290's in it at 66 volts and 6o amps at my watt meter. That's just shy of 4kw. My phase amps are currently set to 130 amps. The controller warms up to 85F after several miles of hard acceleration at WOT on a 6kw C80100 outrunner motor. If 12 of the AOT290 can do that, why can't 6 of a much stronger mosfet do more than that? Both are 500 watts of heat dissipation, but the 4468 has 2 micro ohm Rds and 200 amps at 10 volts while the AOT-290 is 3.5 micro ohms, 110 amps at 10v. Double the amps and close to half the Rds ought to be worth a good bit more than 4kw in half the mosfets.

I think I said 6kw earlier...that's a typo. I meant 4kw+. I don't know why I said 6kw. I edited that post.

What's your favorite to-247 mosfet?
I looked at the various IXYS mosfets in TO-247 and everything had higher Rds and similar amperage to the IRF4468. I've looked at their mosfets before and was equally non-plussed. What is so "bad ass" about their mosfets?

 

[Arlo1]

The 4468 is a fairly robust mosfet. I run a 12 fet controller that has AOT290's in it at 66 volts and 6o amps at my watt meter. That's just shy of 4kw. My phase amps are currently set to 130 amps. The controller warms up to 85F after several miles of hard acceleration at WOT on a 6kw C80100 outrunner motor. If 12 of the AOT290 can do that, why can't 6 of a much stronger mosfet do more than that? Both are 500 watts of heat dissipation, but the 4468 has 2 micro ohm Rds and 200 amps at 10 volts while the AOT-290 is 3.5 micro ohms, 110 amps at 10v. Double the amps and close to half the Rds ought to be worth a good bit more than 4kw in half the mosfets.

I think I said 6kw earlier...that's a typo. I meant 4kw+. I don't know why I said 6kw. I edited that post.

What's your favorite to-247 mosfet?
I looked at the various IXYS mosfets in TO-247 and everything had higher Rds and similar amperage to the IRF4468. I've looked at their mosfets before and was equally non-plussed. What is so "bad ass" about their mosfets?

Read my last comment.

I linked one that is better.

I have tested over $10,000 woth of mosfets including about 300 4468 fets. The 4468 is fine if you are cheep. But the IXYS parts will do what they are rated for. The IR parts really struggle to achieve the numbers in the data sheet. I was able to get the same current through 1 IXYS part you can get though 3 4468 parts!

 

[Animalector]

Are they less than 3 times the price? If yes... Win

 

[Arlo1]

Are they less than 3 times the price? If yes... Win

When you look at the whole design spending more to have 1 power switch flow more current then multiple parts in parallel is ALMOST always better.
The less parts in parallel the closer to their limits you can typically run them and the smaller (and usually cheaper) the over all design will be.

 

[izeman]

I just can say that my 6FET did 3kW without sweat. Stayed super cool. Don't see why I shouldn't be able to do 4kW.


Sent from my Redmi Note 3 using Tapatalk

 

[ElectricGod]

I have some bad ass IXYS parts that do better as well some IR igbts in a 247 package that will do leg limts cold and ~100-110 rms when warm.

Please give IXYS part numbers on those badasses

Just spend some time on their web page. http://ixapps.ixys.com/DataSheet/DS100198(IXFK-FX420N10T).pdf
http://www.ixys.com/ProductPortfolio/PowerDevices.aspx

Weird...I got on their site and used their parts tool to look at all the 100 volt mosfets. This one was NOT in the list when I searched. I guess their search tool was glitching.

IFX420N10...
100 volt
2.6 m-ohm Rds
160 amps...leg limit
1670 watts of dissipation
115ns TDoff

Yeah...that's a pretty good mosfet.

 

[liveforphysics]

https://www.digikey.com/product-detail/en/vishay-semiconductor-diodes-division/VS-FC420SA10/VS-FC420SA10GI-ND/7427232

Due to being already isolated from the base, and supporting DIY friendly interconnects, those IMHO justify the cost of $24 a device.

A few of these in parallel can be better, but requires the meaningful added Rth of the isolator for the heat sink and special clamp bars and drain interconnect (or aluminum PCB bolted to heatsink).

https://www.digikey.com/product-detail/en/infineon-technologies/IPB017N10N5ATMA1/IPB017N10N5ATMA1DKR-ND/5213959

Banks of the Ti nexfet are likely able to surpass TO247 packages at this moment.

 

[liveforphysics]

MTI200WX75GD

That part I could see being a 1 chip power stage replacement for everything in the sub 3kW game in ebike.

 

[amberwolf]

THe series has been used in a Lebowski powerstage design, by Bobc,
https://endless-sphere.com/forums/viewtopic.php?f=30&t=87104&p=1275806&hilit=ixys+MTI%2A#p1275806
I have one of the boards, and one of Whereswally606's brainboards, but havent' yet got all the parts collected to build them, and hadn't ordered the powerchips yet.


Is it safe to assume any of the chips in that series "should be" usable in place of any other, physically/electrically, so I could build up my boards with the one you reference instead of the lower-power one he'd been planning? (mti145wx100gd)

EDIT: looks like the one you reference is only 75v; the one he'd planned is 100v? I guess more current is probably more important for my purposes, since neither would allow 28s anyway.



Some other references to the series:


https://endless-sphere.com/forums/search.php?keywords=ixys+MTI*&terms=all&author=&sc=1&sf=all&sk=t&sd=d&sr=posts&st=0&ch=300&t=0&submit=Search



The chart from Ixys (sorry I couldn't get the tabbing right):

PartNumber 	Division 	VDSS max(V) 	ID25 Tc=25°C(A) 	ID90 Tc=90°C(A)	RDS(on)maxTj=25°C(mOhms)	tf typ(ns) 	tr typ(ns) 	RthJC max(K/W)	Config	Package Style
										
MTC120WX55GD-SMD 	IXYS POWER 	55 	150 	120 (80°C) 	3.1 	100 (125°C) 	110 (125°C) 	1.3 	Triple phase leg 	ISOPLUS-DIL
MTC120W55GC-SMD 	IXYS POWER 	55 	150 	120 (80°C) 	3.1 	100 (125°C) 	110 (125°C) 	1.0 	Six Pack 	ISOPLUS-DIL
MTC120WX75GD-SMD 	IXYS POWER 	75 	180 	128 (100°C) 	3.1 	- 	- 	0.7 	Triple phase leg 	ISOPLUS-DIL
MTI200WX75GD-SMD 	IXYS POWER 	75 	255 	190 	1.1 	55 	70 	0.85 	Triple phase leg 	ISOPLUS-DIL
MTI85W100GC-SMD 	IXYS POWER 	100 	120 	90 	3.2 	40 	55 	1.2 	Six Pack 	ISOPLUS-DIL
MTI145WX100GD-SMD 	IXYS POWER 	100 	190 	145 	1.7 	40 	75 	0.85 	Triple phase leg 	ISOPLUS-DIL
GMM3x60-015X2-SMD 	IXYS POWER 	150 	50 	38 	24 	100 	50 	1.0 	Triple phase leg 	ISOPLUS-DIL

 

[Njay]

https://www.digikey.com/product-detail/en/vishay-semiconductor-diodes-division/VS-FC420SA10/VS-FC420SA10GI-ND/7427232 (...)

This note in the datasheet is a bit confusing:

Maximum continuous current admitted 100 A to do not overcome the maximum temperature of terminals

 

[amberwolf]

They don't want you to melt the legs off, so 100A would be the max you could keep it at without doing that. For how long, I don't know. Probably depends on your cooling setup.

 

[Njay]

I realize that amberwolf, the thing is that it seems a value rather low for the apparent beefiness of the terminals.

 

[ElectricGod]

Too bad that TO-263 is specifically surface mount. It makes replacing them much harder if you don't want to reflow the entire board...which I generally don't want to do. This package minus the screw tab is identical in size to TO-220. I'd like like to see some of these new mosfets in TO-220.

 

[Animalector]

I'd suggest you should not need to replace them.

install temperature sensors for compensation, limit current within limits of the device and it'll be safe. Replacing them should not be a consideration in the design in my opinion

 

[ElectricGod]

I'd suggest you should not need to replace them.

install temperature sensors for compensation, limit current within limits of the device and it'll be safe. Replacing them should not be a consideration in the design in my opinion

I'm going to guess that you've never had a bad mosfet before or have one die. I have...twice. One was an IRF4110 in a sinusoidal Grintech about 2 years ago and the other was two AOT290 just a few weeks ago. 2 mosfets in the same phase died inexplicably. The controller arrived new, ran for less than an hour under no load and suddenly had a blown phase. There were no shorts or stupidity involved. The mosfets very likely were marginal and just hadn't died until I started bench testing. I was able to quickly replace the dead mosfets and the controller has never failed again. Same for the bad IRF4110 in the Grintech 2 years ago. IF they were SMT instead of TO-220...well that would have been a giant pain to replace them. Bad mosfets happen and I do consider replacing them a necessity from time to time. AND people do stupid things like short phases together or their motor fries...whatever...and mosfets die in the process. Easy replacement IMHO is a requirement and NOT ever optional. I blew up an expensive 160 amp Castle RC controller 2+ years ago. The mosfets overheated and died. The controller was 200 feet in the air and I didn't have any form of telemetry on board. I had no idea the ESC was so hot. The mosfets died and the plane lost power. Later looking it over and all those SMT mosfets that were blown gave me major indigestion. I tried unsoldering them without re-flowing to no avail. The board, top and bottom was packed full of mosfets. Re-flowing in a frying pan or toaster oven was a marginal option at best. I'll reflow large LED's and if I had a controller with SMT mosfets on it, I'd probably try to reflow the bad ones off the board, but a simple soldering iron and 3 legs to unsolder is SOOOOO much easier and quicker to do. IMHO, there's not really a question here....I'll take TO-220 or TO-247 over SMT mosfets every day of the year!

 

[Animalector]

Yeah I agree with what you're saying to a point.. But.. Quality control during production plays a part. hobby ebike controllers are cheaply made from cheaply sourced parts. very little quality control. I'll bet a High-end manufacturer will be batch testing their mosfets, miller plateau matching them, blah blah.. whatever they do.. Then, they'll have temperature compensation set 'very' conservatively, they'll have current limiting in place to ensure it will never exceed it's capacity to deliver, and they'll be operating at voltages well within the stated limits for the device. From a design perspective, for an OEM product, replacing the FETs won't be a consideration.

The situations you've described are hobbyists extracting more from their device than what it was designed to deliver.

But I do agree, that TO-220 are easier to replace..

At the end of the day, i'm not going to be seriously stressing this controller (if it ever gets built) so I'm happy to try the SMT fets.
Andy

 

[ElectricGod]

Yeah I agree with what you're saying to a point.. But.. Quality control during production plays a part. hobby ebike controllers are cheaply made from cheaply sourced parts. very little quality control. I'll bet a High-end manufacturer will be batch testing their mosfets, miller plateau matching them, blah blah.. whatever they do.. Then, they'll have temperature compensation set 'very' conservatively, they'll have current limiting in place to ensure it will never exceed it's capacity to deliver, and they'll be operating at voltages well within the stated limits for the device. From a design perspective, for an OEM product, replacing the FETs won't be a consideration.

The situations you've described are hobbyists extracting more from their device than what it was designed to deliver.

But I do agree, that TO-220 are easier to replace..

At the end of the day, i'm not going to be seriously stressing this controller (if it ever gets built) so I'm happy to try the SMT fets.
Andy

When I swap out mosfets I test all of them and look for the best of them of the stock I have on hand and then collect together the best matched of those mosfets. So called "high end manufacturers" are cutting corners all the time too. Everyone wants to build their stuff for as little effort and cost as possible. Do you get a bit more "bang for the buck" or "better engineering" if you buy supposedly "high end" stuff. I sure hope so! Otherwise, what's the point?

I'm not sure what a "hobby ebike controller" is. Is that the $35 special on ebay? Is that a home built controller? Are ASI FOC controllers or Sabvoton controllers categorically ""hobby ebike" controllers? That's a generic term that can mean anything and nothing.

I have several supposed "high end" controllers. They run motors. Do they do a spectacularly better job than some Chinese controllers I have? No not really....just a little bit better.

On both of the controllers that I have that had bad mosfets in them. One uses legitimate Infineon mosfets and the uses legitimate AOT mosfets. Both controllers were bench testing motors under zero load...just motors free spinning and then the legit infineon and AOT mosfets die? In either controller battery amps were no more than 6 amps which is so ridiculously low that there is no chance EVER that the mosfets were remotely close to an overload situation. Nothing you said about people over watting their controllers had anything to do with what happened to my 2 controllers. Bad components happen. Stuff fails sometimes for no good reason. I don't know what your back ground is, but I've been doing electronic component lever repair since 1988. I've seen lots of supposedly "High end" and well engineered electronic equipment fail for no good reason. Expecting that running components conservatively is not a guarantee that they will not fail. I'm still going to say that easily replaceable components that are more likely to fail is always the best policy.

From a manufacturers perspective, they want to sell stuff they make. They don't care that you can't repair their stuff. In fact they prefer that you can't repair it so that you have to buy another one! From their perspective, SMT components are perfect for them! Also, SMT components are easily machine placable so that saves them manufacturing costs. In anybodies attempt at cutting costs, this is always a good thing for them and bad for people like me. I don't want to throw away an expensive electronic device when I can otherwise repair it. I'll take TO-220 or TO-247 mosfets over SMT mosfets every day of the year for this reason alone. SMT parts make repairs for whatever reason 10X harder to do. Trust me, I KNOW this is a fact! Your logic is flawed in this regard. SMT is good for manufacturers and BAD for you if/when you need to repair something. Does that mean I won't use things that have SMT components on them? No way...that's stupid. But at the same time, if I can avoid it for certain types of parts, well I'm going to avoid it...every time I can. Can SMT power mosfets be reliable? Sure they can, but they also take the brunt of abuse in any controller there is no matter who makes it or how well it is designed.If something is going to fail, it's the mosfets that are going to pop first.

I want to build a Lebowski controller so that I can build the power stage how I want it built. No thank you to SMT mosfets!

 

[amberwolf]

In either controller battery amps were no more than 6 amps which is so ridiculously low that there is no chance EVER that the mosfets were remotely close to an overload situation. Nothing you said about people over watting their controllers had anything to do with what happened to my 2 controllers. Bad components happen. Stuff fails sometimes for no good reason. I don't know what your back ground is, but I've been doing electronic component lever repair since 1988. I've seen lots of supposedly "High end" and well engineered electronic equipment fail for no good reason. Expecting that running components conservatively is not a guarantee that they will not fail.

I expect ESD during handling, either component or board or system level, is a large part of why stuff fails "randomly".

It isn't the only reason, by far, but it's a big one in stuff I saw when I worked at Honeywell CFSG, when they did failure analysis on various things (mostly in flight displays and computers).

Another big one is mechanical damage during component handling and installation, primarily during manual assembly stages. Sometimes soldering process damage (overheating, mostly). Again, stuff I saw at H-CFSG.

 

[liveforphysics]

Just as perspective, it takes ~7amps from the pack (28s) to maintain 660amps RMS in the phase leads or deathbike (stalled rotor).

 

[Animalector]

To be fair I've only been doing component level repair since 2004.. but much of that has been surface mount components. Even surface mount igbts for washing machine drives...

I wonder what the data shows on FET failures for the Bosch, Yamaha and Shimano mid drives.. Toyota Prius controllers even?

Agreed manufacturers dont care how you use their device, but FETs are not a serviceable parts. They're not designed to be maintained or replaced periodically as far as I know. Well, not like a bearing or a bushing or a fuse..

Anyway. I get your point, ebike controllers fail. Often it's the FETs.. Solution... use FETs that are easy to replace. I'll still try these SMT ones though..

Andy

 

[ElectricGod]

Just as perspective, it takes ~7amps from the pack (28s) to maintain 660amps RMS in the phase leads or deathbike (stalled rotor).

Good point...It's insane how little current it takes from the battery to create so much current in the phases in a dead locked situation.

Of course in both of my cases the motor was unloaded and spinning freely and then suddenly NOT spinning when the mosfets shorted internally and locked the phase. I bet since the motors were spinning their own armature mass only that phase currents were similar to battery current.

 

[ElectricGod]

To be fair I've only been doing component level repair since 2004.. but much of that has been surface mount components. Even surface mount igbts for washing machine drives...

I wonder what the data shows on FET failures for the Bosch, Yamaha and Shimano mid drives.. Toyota Prius controllers even?

Agreed manufacturers dont care how you use their device, but FETs are not a serviceable parts. They're not designed to be maintained or replaced periodically as far as I know. Well, not like a bearing or a bushing or a fuse..

Anyway. I get your point, ebike controllers fail. Often it's the FETs.. Solution... use FETs that are easy to replace. I'll still try these SMT ones though..

Andy

You probably have lots more access to SMT repair tools than me. I no longer do regular component level repair except for myself. As a result, my budget for the specialized tools needed for SMT repair is a bit limited. If you have the tools needed, well that makes a big difference in SMT repair. If I can't reflow in an old aluminum pan or a toaster oven, it's probably not going to happen anymore. That makes me much less inclined to go with SMT power components.

You bring up a good idea! I'd like to see those failure rates too! Of course most folks aren't going to pay the money it costs to by a Bosch mid drive when for half that amount, they can get 2-3X the wattage elsewhere. From a personal perspective...mine...they probably don't build anything I would want anyway...no matter how reliable it is. Proprietary solutions are exactly that...proprietary. That generally means you get what that manufacturer wants you to have and nothing more. Where's the fun in that? Ever watch EV commercials on youtube or where ever? You have some dude or girl riding along doing 15mph with a big stupid grin on their face on some commercially made massively under powered EV. I can see that they are moving along slowly and the size of the motor and battery bay tells the story quickly that the EV is about as fun as plain oatmeal. Those proprietary solutions are a great way to see what way under powering something will get you and how long it will last, but then it will also be uber boring. That's fine for some folks, but for people that know better and have higher expectations...not ever. I suspect that since you are on ES pretty regularly like me that you aren't going to settle for plain oatmeal.

So anyway back to the Micro Lebowski...
I've been trying out various controllers over the past few years to find a worthwhile FOC controller that is highly hackable. Most are locked down by the manufacturer and they wont give you the firmware to hack. That effectively means it is built to their designs and NOT hackable. Controllers like the Lebowski, various VESC options and the Power Velocity controllers however are highly hackable and therefore good options for people like me that don't want to settle for what some manufacture says is what I want.