Scoped the gate driver on the EB236, scope pics, poor driver

[zombiess]

Not sure if anyone here has taken the time to go through and make these measurements before, I know I've searched and I didn't see them so I thought I'd post up my info I gathered today while checking out the controller. This controller is heavily modded, details covered in this thread http://endless-sphere.com/forums/viewtopic.php?f=2&t=37166

I built an EB236-AS-1 from scratch with some thermal mods and also extra heat sinking by tying all the tabs together with buss bars. It uses 6 individual 0 ohm gate resistors to each FET gate.

Now on to the scope info. Hooked it up to monitor the driver and see how slow the switching performance is. Here are the pictures I took of my scope while testing. I'm using a 510 ohm resistor for R115 and the on board switching power supply was producing a solid 20V. The regulators were outputting 5V and 15V respectively. Maximum gate drive voltage seen was 15V. Not sure if the R115 resistor being so high has any effect on this, I didn't bother measuring the 15 regulator for ripple current to see if it was holding up to the current load being placed on it by the drivers due to being run on 12S instead of 30S with such a large limiting resistor. I'll probably try to check that tomorrow to verify that a 510 ohm R115 resistor isn't causing me any issues with the lower voltage I'm testing at.

This board was built with IRFB4115 FETs all miller plateau matched at 4.0V Vgs to maximize synchronous switching to maximize current sharing.

First thing I discovered is that PWM switching frequency is 15,875hz. This is the really ugly gate drive signal at part throttle (probably caused by me under powering the on board power supply) showing the total time on and time off giving a total period of 63uS. It should look more like a square wave.

At 100% throttle there is zero PWM occurring. This is phase A at 100% throttle showing the 15V gate signal and commutation period. Wheel was going 26.1 MPH on 46.1V of battery power.

This is a close up of the gate turning on, you can see the Miller Plateau occurring right around 5V which is right around where I measured it on this bank when i was testing the FETs. Spec sheet shows it happening around 6.0V but these seem to happen 1V lower. It takes about 2.6uS to reach the miller plateau. Hangs out on the plateau for 2.2uS then continues on to fully turned on around 7.0V. I actually measured the full turn on right around 5.0V. It takes about 5.6uS to go from 0V to 7V and 4.8uS to go from 0V to 6V and 2.5uS to go 0V to 5V which should be close to full on based on what I measured on many of the IRFB4115 FETs I have and sampled randomly, unless I measured incorrectly.

The total time to transition from 0V to the full 15V gate drive voltage is a long 27uS.

Turn off time is about 4.6uS to drop from 15V down to 0V and about 2uS to drop to the point where it should be off.

I'm wondering if the switching speeds in the IRFB4110's might be a little slower because they have about double the Qg nC (total gate charge) causing them to switch slower. I did measure the IRFB4110s I have appear to fully turn on at 4.0V Vgs so they might not switch as slow as I think.

I hope this info helps some of you guys building controllers and playing with the Xie Chang boards. Conclusions is the EB2xx driver leaves a lot to be desired, but also to be noted is the devices appeared to perform much better than their spec sheet suggests. I'm guessing this better performance is one of the reasons we can get away with abusing these controllers so much.

Question for those in the know, since the resistance is linear in relation to Vgs and the resistance is really high at low Vgs the FETs won't produce much heat since they are essentially current limited. Is this why the controllers work as well as they do even with such slow switching times or am I missing something? Any suggestions to improve the switching times of the stock config without substituting an entire new driver board? The 6 individual gate resistors are already at 0 ohms would bridging them int a pure parallel config help any? I really should test at a higher supply voltage or bypass the R115 resistor to verify the on board power supply isn't current limiting the driver circuitry and scope it again to see if there is any difference.

I'm no expert on any of this so please don't take this as 100% correct, I'm only posting what information I have measured and know so hopefully some real experts such as Rhitee05, Bigmoose or Electroglide will chime in with their thoughts.

 

[Arlo1]

Found it. lol. I dont understand the 27us 0-15v full on.??

 

[Lebowski]

Oh man that's even worse than total utter cr@p ! Looking at the pictures it takes 14 usec to properly
switch on (to 10V gate-source). That should happen in around 300 nsec ! You should swap out for
NCP5181 gate drivers, maybe use 2 or 3 per 6-FET bank, then you should easely reach sub u-sec
figures...

These are for single 4115's driven by a ncp5181,
top picture shows a 2 u-sec pulse at the motor, the second one a 9 usec pulse under load:
http://www.endless-sphere.com/forums/viewtopic.php?f=30&t=35387&start=105#p540239

 

[zombiess]

Found it. lol. I dont understand the 27us 0-15v full on.??

If you are reading the graph the same way I am, that's what it says. I need to bypass the R115 limiting resistor at 46V and see what happens or try running it at the full 125 volts. If I see a change I might drop R115 down from 510 to 330 ohms which I've been told is good for 125V use. I might have too much current limiting going on (I really don't like the shape of the PWM pulse, not even close to square after it gets past the Miller Plateau around 5V (I'm pretty sure this is a current limiting issue), but the only way to tell that is to put it back on the bike and make some more measurements while making some changes which I'll probably do tomorrow with the same setup before I put this controller into it's box and play with it. It's super easy to gather this kind of data when you have the proper tools, just a few mins of hookup and playing around. To me it's worth it. To others like you it still provides some valuable data to look at.

I wonder if these controllers work because they are big, dumb and are only used for short bursts, much like Lukes suicidal bicycle of doom. No way I would attempt to run my 18 FET controller at 100A continuous or even 85A continuous without some sort of forced air cooling. I'm planning to scope the signals on one of the EB3xx boards at some point in the near future to see what they look like, I'm guessing the drivers are improved a bit.

These are the numbers I got, love 'em, hate 'em it doesn't matter, they are just data points. I searched and found no info like this so I decided to gather my own data. I'm not against these controllers at all, they suit lots of people quite well. I think with a few tweaks the performance could be improved by 50% or better. I've been curious about these numbers ever since I starts researching drivers and FET power stages.

 

[zombiess]

Oh man that's even worse than total utter cr@p ! Looking at the pictures it takes 14 usec to properly
switch on (to 10V gate-source). That should happen in around 300 nsec ! You should swap out for
NCP5181 gate drivers, maybe use 2 or 3 per 6-FET bank, then you should easely reach sub u-sec
figures...

These are for single 4115's driver by a ncp5181,
top picture shows a 2 u-sec pulse at the motor, the second one a 9 usec pulse under load:
http://www.endless-sphere.com/forums/viewtopic.php?f=30&t=35387&start=105#p540239

I have some driver circuits already designed with buffers that supply up to 8A of gate drive based off the IRF4110 chips that could switch in the sub 100nS range if so desired (but might cause ringing or under shoot or some other side effects with that kind of speed). I just need to send the boards off to be made. I'm going to look at the stock setup first. I'm thinking the low voltage (less than 50v) and my high R115 resistor is causing some of these issues limiting current to the totem pole drivers. I'll run it again tomorrow the same way with my scope on the 15V rail to watch the ripple. I'm guessing I'm going to be seeing quite a bit. I'll tweak from there if I do see ripple. This is what experimentation is about While it still may suck in the end compared to a much better driver, it's still worth playing with and many people run the 24 FET version of this board with great success.

Swapping in one of my driver boards after intercepting the gate drive signal would make for a cheap an easy improvement with sub 1uS on/off times, but I want to experiment a little first before I decide to Frankenstein a Xie Chang controller :lol:

Seeing this makes me realize I need to get my driver boards done and check out my water cooled FET power stages based on the 12 FET EB3xx boards which I'm using to feed my IRF4110 drivers with. I also have your chip to play with too. But I need more time to play around.

 

[zombiess]

At 46V and using a 510 ohm resistor the on board power supply can only manage 0.09A of current draw. I'm sure once I up the voltage or drop the resistor value it's going to change the switching times and wave forms, but before I do I'm going to check the 15V regulator for how much ripple it's producing. That should tell a lot more of the story. I've been told 330 ohms is a good R115 value for 126V usage and my other EB318 controllers use this resistor as well. This allows a max current of 0.38A to be pulled by the regulators. The schematics only call for a 1W resistor but I'm using 2W resistors so I should be good.

Now I'm excited to test tomorrow and see what kind of results I get. I want to wrap this project up and put it in it's case and move onto the next project; my throttle interface so I can enjoy riding my bike more with less worry of power wheelies when I don't want them and also my water cooled power stages.

 

[Lebowski]

can you post a schematic or somehting 'cause I got no clue wich resistor you're talking about ?
would also be nice to see the schematic of the gate driver

 

[zombiess]

can you post a schematic or somehting 'cause I got no clue wich resistor you're talking about ?
would also be nice to see the schematic of the gate driver

Sorry, I don't have them. Some models of the Xie Change boards use an on board SMPS to supply voltage to the regulators. They produce about 20V input to the 15V and 5V linear regulators used to power the low voltage side of the board. From the previous gate driver schematics I have seen posted it is a totem pole gate driver (at least I think that's what it is, sorry I'm not expert). I think there is a 6FET schematic which shows a similar driver floating around on here somewhere.

I always knew these controllers didn't have good gate drivers from what others have posted, but it's pretty hard to argue they don't work well even though there is most likely tons of room for improvement. These cheap controllers power many high powered Ebikes on this forum, including mine with great success and usually a long life, even when run at 100% of their absolute max ratings.

 

[liveforphysics]

That does look very ugly.

If I remember right, I made a thread showing the gate drive components on the 6fet, 12fet, 18fet, and 24fet were either all the same, or the 24fet had twice what the 6fet had or something like that. I remember we werent super impressed. lol

Fortunately, as far as mods go, adding a real gate driver that floats on an isolated DC/DC for high side and low side is a pretty trivial thing to setup. About $30-40 in parts to do it right though.

It would be interesting to compare the high side and low side gste drive signals, and scooe the boot strap supppy voltage and low side supply voltage. Its possible adding some decent low ESR caps to those supplies could clean things up quite a bit if they happened to be undersized.

 

[Lebowski]

I have some driver circuits already designed with buffers that supply up to 8A of gate drive based off the IRF4110 chips that could switch in the sub 100nS range if so desired (but might cause ringing or under shoot or some other side effects with that kind of speed).

That's what snubbers and gate resistors are for.

Maybe the slow waveforms you see are the reason why the controller you have doesn't
have snubbers or gate resistors, it just doesn't need it...? The British off-the-road dude
puts 4 kW through a 6 FET, maybe with proper gate drive signals a 6 FET can do the
same power as a crappy gate driven 12 or 18 FET... A huge amount of power is dissipated
in the FET's during the switch on / off transients so faster is better.

 

[bigmoose]

Nice traces, easy conclusion: the stock gate drive is incorrectly designed for any amount of upgrade and likely marginal for a stock configuration. When the drive is so marginal, doubling the capacitance it has to drive will slow it proportionately. I like the transitions around 1 uS.

If you have it post a schematic of the stock gate drive and we'll help dissect it.

 

[zombiess]

Nice traces, easy conclusion: the stock gate drive is incorrectly designed for any amount of upgrade and likely marginal for a stock configuration. When the drive is so marginal, doubling the capacitance it has to drive will slow it proportionately. I like the transitions around 1 uS.

If you have it post a schematic of the stock gate drive and we'll help dissect it.

Just imagine that this came for use with some 60 or 75V FETs. I know the stock FETs aren't very good and can't even find a spec sheet for them but I'm wondering what the Qg is on them.

Here is a gate drive schematic of a 6 FET which I believe is the similar but I'd need to really trace it out to make sure. Gate driver is on page 3, this is from EB206-A-2 version. My 36 FET is a EB236-AS-1 version. All the EB2xx boards I've seen use very similar gate driver setups. This one has individual gate resistors but they are all zero ohms.

That does look very ugly.

If I remember right, I made a thread showing the gate drive components on the 6fet, 12fet, 18fet, and 24fet were either all the same, or the 24fet had twice what the 6fet had or something like that. I remember we werent super impressed. lol

Fortunately, as far as mods go, adding a real gate driver that floats on an isolated DC/DC for high side and low side is a pretty trivial thing to setup. About $30-40 in parts to do it right though.

It would be interesting to compare the high side and low side gste drive signals, and scooe the boot strap supppy voltage and low side supply voltage. Its possible adding some decent low ESR caps to those supplies could clean things up quite a bit if they happened to be undersized.

I can't find your thread Luke, if you know where it is I'd love to see it. I can add three IR4110 gate drivers with current buffers, cut off and intercept the signal from the processor to the board pretty easily. This should provide a big advantage even if I just get the on/off times down under 1uS. I already have all the parts I need except the PCB's which I'm getting 3 sets of 3 drivers made up for about $55. I can power them from the on board SMPS power supply, but I might swap out the 15V surface mount regulator for a TO220 version. If you can recommend a cheap DC/DC power supply with a decent voltage operating range I'd like to take a look at it if the on board one can't supply enough current for the drivers.

I'm planning to experiment with the on board SMPS tonight to see if the current limiting resistor to it changes the times/wave forms due to lack of proper current. I keep bringing up the lack of current to the on board SMPS based on how long it takes the signal to reach the full 15V. To me that seems like a current limiting issue due to a weak power supply. Only hooking it up and testing again will tell if that's really part of the issue here, 0.09A fed into the SMPS is pretty low. When fed with 46V the other side of the resistor which is directly connected to the 46V B+ was sitting around 36V. I can also measure the current the SMPS is drawing to figure out how many watts it's drawing, but if the R115 resistor I have in there now for 126V usage is limiting it to 0.09A at 46V in and I'm only seeing 36V (measured somewhere around this voltage last time I checked) on the other side of the resistor, that's a whopping 3.24W being fed into the SMPS, not nearly enough to power both 15V and 5V linear regulators which are rated for 300ma each I think.

Someone correct me if I'm wrong here, but the on board SMPS produces 20V (measured it) so to power both the linear regulators to their maximum current I should probably want to let the SMPS pull at least 15W at what ever voltage I'm feeding it.

 

[Arlo1]

This might be the link luke is thinking about. http://www.endless-sphere.com/forums/viewtopic.php?f=16&t=21502

 

[zombiess]

I found Luke posted some info in this thread http://endless-sphere.com/forums/viewtopic.php?f=2&t=19642&p=286373&hilit=driver#p286373

These guys have a 20ohm resistor, and switch ~10v (i dont remember exactly). ****EDIT I was totally 100% wrong here, somehow confused the RC controller FET design with the infinion design...Ooops! The infinion boards tie all the gates common, and use an up-stream gate drive resistor***

It's half an amp of load per FET for the first nano-second of switching, then tapers to zero amps. (then bounces current back into the gate drive rail, oscillates a bit a few times etc etc, as you know of course).

I will go double check the 6fet, 12fet, 18fet and 24fet boards I've got sitting here, but I think they all used the exact same FET driver, which is a 3amp unit if I'm remember reading the specs correctly, which unless I'm seeing something incorrectly on the scope and from the numbers, means no noticeable difference in switching performance until you go do a 36fet controller or so, but perhaps I'm seeing something wrong?

***Edit*** Corrected, the infinions all use the same SMT zenner clamping diode called "T4"

The infinions are way different than the RC controller FET stages... lol This is what they use.
The 24fet model uses an external SMT gate drive mosfet for each channel labeled "Y6" for the highside FETs, and one labeled "Y1" for the low side fets.
The 18fet model uses an external SMT gate drive mosfet for each channel labeled "1HD" for the top and bottom.
The 12fet model uses an external SMt gate drive mosfet for each channel labeled "Y2" for highside FETs, and one labeled "Y1" for the low side fets.
I don't have a 6fet in an easy to disassemble way handy right now to check it.

 

[bigmoose]

Were the scope traces of the upper FET of the bridge or the lower?

You can speed up the lower turn on time by decreasing R6 and the turn off by decreasing R8. I wouldn't go too low in R8 to drive it deep into saturation, perhaps 330 or 220 ohms. On the top side decrease R4 to speed up turn on, and decrease R3 to speed up turn off. Don't take R3 much below 220 ohms.

Just some quick thoughts.

 

[zombiess]

Were the scope traces of the upper FET of the bridge or the lower?

You can speed up the lower turn on time by decreasing R6 and the turn off by decreasing R8. I wouldn't go too low in R8 to drive it deep into saturation, perhaps 330 or 220 ohms. On the top side decrease R4 to speed up turn on, and decrease R3 to speed up turn off. Don't take R3 much below 220 ohms.

Just some quick thoughts.

scope was high side directly on the fet gate leg with a micro hook less than 1mm from the body. Ground was the phase output / bootstrap cap ground.

 

[zombiess]

Alright did some more playing around tonight. Looks like reducing the R115 resistor value from 510 ohms to 305 ohms helps a little bit. I was able to shorten up the pulses by 1.3uS on the 0V to 7V transition from 5.6uS to 4.3uS so at higher voltages it the resistor value I'm using will produce the same results,. I even tried just a 50 ohm resistor and also completely bypassing the resistor there was no further noticeable improvement. The on board SMPS was drawing a whopping 36-60mA from the 46V battery when in use depending on the R115 value. Higher values caused a voltage drop which in turn caused a higher current draw as expected.

Here is a picture of the low side I scoped tonight to see what it was doing, it has no PWM at all. About 1.4uS to go from 0V to the 5V plateau and about 3.8uS to get to go from 0V to 7V. The on board driver is not good at all. This is the same driver used on the 24 FET versions as well, with the only difference being the 3 of the 4 gates on both high side and low side use 1 ohm resistors and it has one FET in each bank with no resistor going to the gates of their respective banks both high and low side (that's pretty odd and I wonder why the 24 FET is designed this way). Only real differences that I spotted at all so my guess is it too suffers from the same slow switching speeds.

So we all know people run stupid amounts of power through these controllers. Someone I know with a EB224-AS-1 controller just like the board I checked using IRFB4110 FETs was able to run something crazy power through it. 250A battery and 420A phase and it works fine and has for quite a while. An EB218 board with IRFB4110 FETs was run at 150A battery and 360A phase. How much is really left to gain in performance out of these controllers? I know having switching times around 800nS or faster is desirable, but these silly slow drivers that have on/off times that can almost be measured with a stop watch seem to work OK which just blows my mind after reading lots and lots of info about FET switching and drivers.

Is it because they switch at about 16khz? For the high side if we go with the figures of full turn on voltage of 7V Vgs and say it takes 4uS to get there and then 2.5 uS to turn off that gives a total time spent transitioning through the linear region of 6.5uS Since the PWM period is a 63uS period the controller spends 10.3% of it time generating heat in the linear region, that's a really long time.

For the low side it spends 3.8 uS to go from 0 to 7V and 2.5uS to turn off it's spending 6.3uS switching so another 10% of it's time spent generating heat in the linear region.

***** I'm going to take a SWAG at the power dissipated per phase with these slow switching times assuming a 125V battery and a 250A phase current limit*****

I have no idea if this makes sense to anyone else or if I'm even in the ball park. Experts, please feel free to chime in any time.

Some power dissipation numbers I've run based on resistance measured at 4v Vgs with these FETs are listed here and how many uS they spend in that region. I'm unable to measure the mOhms range at 5V since my meter doesn't read that low so I'm going to guesstimate numbers with full on at 7V. I'm also going to use the faster switching times I saw by lowering the resistor value.

At 4.0V Vgs I measured the RDSon of my FETs between 40 ohms and 150 ohms for these 36 FETs (all similar ones were matched to the same banks). At 3.0 Vgs all the FETs were in the 30K to 60K area. I'll average to simplify this a little. I'll assume I'm limiting the phase amps to 250A

1. At 3.0 Vgs the RDSon is an average of 45k Ohms / 6 for 7.5kohms which puts the power dissipation at 2W and it takes 1.2 uS to get from 0 to 3V
2. At 4.0 Vgs the RDSon is an average of 95 Ohms / 6 for 15.8 ohms which puts the power dissipation at 989W and it takes 0.4uS to go from 3V to 4V
3. At 5.0 Vgs the RDSon is well under 1 ohm / 6 and my meter reads 0 ohms so I'll just guess and say 50 mOhms / 6 for 8.3 mOhm which is above the 250A limit putting power dissipation at 519W and it takes about 1.2uS to go from 4.0V to 5.0 to the end of the plateau.
4. At 6.0 Vgs we should be near full on so I'll say it's sitting at 20mOhms / 6 for 3.3 mOhms is above the 250A limit for a power dissipation of 206W and it takes about 2uS to go from 5.0V to 6.0V
6. at 7.0 Vgs the RDSon should be 10mOhm / 6 for 1.67 mOhms and is above the 250A limit for a power dissipation of 104W for .8uS
7. from 7.0V to 15.0V the RDSon should be the same 10mOhm / 6 for 1.67 mOhm above the 250A limit so power dissipation is 104W for 55.4uS

Using the above estimated data points that gives me an average dissipated power from 0 to 7V of 311W for 5.6uS and an average 0 PWM resistive dissipation 100W for the entire 6 FET bank on for 61uS (allowing 2uS to turn off) and 200W per phase. Since there are 3 phases that average 200W each and commutation causes a 1/3rd duty cycle so 200W / 3 = 66W average power dissipated for each bank during a full commutation cycle with 0 PWM.

It's late and I hope I got my math and thinking are correct. These numbers looked at in the time domain make more sense to me as to why the controllers work with such slow switching times without producing massive amounts of heat. This controller will have to shed 200W of heat at WOT, part throttle is a whole different story and I'd need to use a spread sheet to figure it out. Right now I'm only interested in 0 PWM numbers since they are the easiest to calculate. My guess is things could get ugly pretty fast with PWM in some of the part throttle regions. I'll have to see if I can build that spread sheet to figure it out. If I'm even close to correct in my thinking this controller with some forced air run between the FET high and low sides could do some serious power at WOT and even in PWM because with a 400W soldering iron it's hard to get an FET hot even if I held the tip on the middle leg that's attached to the heat sink for 2 mins straight. Heat sink got warm, FET got pretty warm, solder was workable, soldering iron tip and element started to glow a dull red, but after just a few seconds I could touch the FET so the massive heat sinking caused by all the solder/ 12 gauge and 8 gauge copper added on the traces + the 8 gauge FET bank buss bars, extra 8 gauge wire connecting the buss bars and also better thermally conductive electrical insulators must really be doing a heck of a good job at pulling the heat away.

***** End of SWAG *****

I guess the upside to such slow times is you don't need to worry about the gate ringing :lol:

Now the question is do I bother playing around with changing the gate drivers to something different. Seems like a no brainer, but realistically these controllers can produce some high power even with pathetic switching times. I know it would run cooler with a faster driver but very few of us high power guys can be on the power for long enough to even produce lots of heat and my board has lots of extra heat sinking. My experience with my modded 18 FET shows me that it sheds heat really fast as well and I don't even have it ventilated. With ventilation or even forced air cooling from a fan the temps could be kept well in check.

I don't have the time at the moment to mess around with building a new driver and installing it so I think for now I'm just going to let it be as is, I'm just shocked at how slow it is. Makes me wonder what my 36 FET water cooled power stage is really going to be capable of driven by a 12 or 6 FET version of these boards through a tuned IR2110 FET driver setup with a nice 8 amp current buffer and sub 1uS on/off times

I spent quite a few hours playing around with this but the hands on experience is priceless. I now know so much more about these boards.

 

[Lebowski]

funny to see the all-American attitude

What you describe is basically the equivalent of an American V8. Kind of old technology
with the single cam between the cilinders, rockers and push rods, only 2 valves per cilinder,
limited rpm, questionable efficiency etc etc. But if you just make it big enough it
will deliver lots and lots of power.

On the other hand there's the German approach (Merc, BMW, Audi) where everything
is engineered to the smallest detail for weight, efficiency etc etc. Also lots and lots
of power by the way.

Like you said, it works also upto crazy power levels but it's based on using lots and lots
of FETs and a big heatsink. With relatively slow hubmotors all will be fine. But it's not efficient,
not high e-rpm, not capable of providing sine-wave out, not capable of going to 50kHz PWM
for low inductance motors, questionable reliability because of all the excess heat etc etc.

It all depends on what you want. And, looking at the design of the EB236, most important on
everyone's mind is dirt-cheap

 

[zombiess]

funny to see the all-American attitude

What you describe is basically the equivalent of an American V8. Kind of old technology
with the single cam between the cilinders, rockers and push rods, only 2 valves per cilinder,
limited rpm, questionable efficiency etc etc. But if you just make it big enough it
will deliver lots and lots of power.

On the other hand there's the German approach (Merc, BMW, Audi) where everything
is engineered to the smallest detail for weight, efficiency etc etc. Also lots and lots
of power by the way.

Like you said, it works also upto crazy power levels but it's based on using lots and lots
of FETs and a big heatsink. With relatively slow hubmotors all will be fine. But it's not efficient,
not high e-rpm, not capable of providing sine-wave out, not capable of going to 50kHz PWM
for low inductance motors, questionable reliability because of all the excess heat etc etc.

It all depends on what you want. And, looking at the design of the EB236, most important on
everyone's mind is dirt-cheap

Look over my time based wattage average and see if you think I missed anything.

BTW, nothing wrong with modern Chevy single cam push rod V8's. They make tons of power and are quite fuel efficient. The 5.7L in my 2002 Z06 makes around 460 HP (not even close to stock, but still stock heads that only flow 260cfm which is better than most DOHC motors) and gets 27MPG averaging 75MPH on the highway (once again better than most DOHC motors) but only revs to 7000 rpms so not super high. It also doesn't cost a fortune to fix like most "modern" cars :wink:

Sometimes you just gotta go with what works, cheap is a bonus too. I'm also only using this for hub motors. For everything else, their is Lebowski's controller

 

[Lebowski]

Talking about 'Lebowski's' controller, I wonder whether it would be worth the effort to design
a proper output stage. What I've been thinking about is a module based solution.

An output stage can be built using 6 FETs, 3 half-bridges (a half-bridge being a low side and a high side FET
together). What if you build a module which has a NCP5181 driver IC and a low and a high side FET (2 x 1 4115 ?). It would
need some investigation to see how big the FETs can be, maybe instead of 2 x 1 one driver IC can drive 2 x 2 or 2 x 3
FETs. The module would be a small PCB with the 2, 4 or 6 FET's on one side with the 2 battery supplies and the motor
terminal. On the other side would be the 12 to 20 V supply and the 2 digital control signals for turning the FET's
on. The 150 V module would also include a 1uF 250V cap and snubbers around the FETs, maybe even one or more 220uF 160V
caps.

It would take 3 of these modules to build a 150V motor controller, one for each motor terminal. Assuming 2 x 2 FET's
this would give you a 12 FET controller. You can also chose to drive each motor terminal with 2 modules in
parallel, you would need 6 modules and get a 24 FET controller....