Arlo's power stage Leaf controller runs and drives page 103

[Lebowski]

4) I can't comment on Lubowski's algorithm on turning on the lower FETs of the H bridge. This is key to charging the upper bootstrap caps. I always have my routine do this to precharge bootstrap caps before we use them in the PWM routine.

This is why I added 'low side pulsing' as an option going into version 1.01 .. I had the same issue with the
high side bootstrap not charging properly which caused the chip to stay stuck in drive_0. What I added is
an option for low side pulsing, when turned on it will turn on the low side FETs at a (user) specified
frequency and time (I use 20 Hz and 20 usec myself). When pulsed the low side FET's will pull all three
motor outputs to ground, providing a nice clean path for the high side bootstrap caps to charge. When
properly charge the bootstrap charging current will become 0 and the motor voltages inputs of the
chip (pins 5, 7 and 8 ) will be at 2.5 V (half the 5V supply).

 

[Arlo1]

Ok I got it. The hi side driver fets we in the wrong spot. I think thats the last major thing I had wrong and it should turn a motor but I fried lebowskies chip in the process. So.... Gordo?? You there lurking?

 

[bearing]

What I added is an option for low side pulsing, when turned on it will turn on the low side FETs at a (user) specified
frequency and time (I use 20 Hz and 20 usec myself). When pulsed the low side FET's will pull all three
motor outputs to ground, providing a nice clean path for the high side bootstrap caps to charge.

Is the pulse frequency audible from the motor windings?

I'm just curious, because I think it would be a good thing if an E-vehicle had some sort of "idle" noise to make the user aware whether it is "running" or not. And It's better, IMO, if the noise is a bi effect of a technical function, rather than having to make the noise with a buzzer.

 

[Arlo1]

What I added is an option for low side pulsing, when turned on it will turn on the low side FETs at a (user) specified
frequency and time (I use 20 Hz and 20 usec myself). When pulsed the low side FET's will pull all three
motor outputs to ground, providing a nice clean path for the high side bootstrap caps to charge.

Is the pulse frequency audible from the motor windings?

I'm just curious, because I think it would be a good thing if an E-vehicle had some sort of "idle" noise to make the user aware whether it is "running" or not. And It's better, IMO, if the noise is a bi effect of a technical function, rather than having to make the noise with a buzzer.

It is audible in some settings but not loud. You would never hear it in trafic! And I only think it pulses in drive 0

 

[Lebowski]

What I added is an option for low side pulsing, when turned on it will turn on the low side FETs at a (user) specified
frequency and time (I use 20 Hz and 20 usec myself). When pulsed the low side FET's will pull all three
motor outputs to ground, providing a nice clean path for the high side bootstrap caps to charge.

Is the pulse frequency audible from the motor windings?

I'm just curious, because I think it would be a good thing if an E-vehicle had some sort of "idle" noise to make the user aware whether it is "running" or not. And It's better, IMO, if the noise is a bi effect of a technical function, rather than having to make the noise with a buzzer.

just an idea ... my girlfriends e-bike has a squeaky Brooks saddle

It only pulses in drive_0 which is the startup and safety mode (it waits in drive_0 till the motor is at nearly standstill AND the
throttles are closed, all FETs are off in this mode). The normal running modes (drive 1 to 3) all have full-time PWM pulses going
to the output stage so no extra 'low side pulsing' is necessary to keep the bootstrap capacitors charged.

 

[Arlo1]

OK so I did a digikey order and could not find any diodes like Bigmoose used in stock. So I searched some fast recovery diodes and found these STTH3R02 http://www.st.com/internet/com/TECHNICAL_RESOURCES/TECHNICAL_LITERATURE/DATASHEET/CD00110333.pdf

200v
3A
16ns recovery

What do you guys think?

 

[bearing]

<500ns recovery

What do you guys think?

I don't know, that sounds like a pretty long time. You are making a sturdy gate driver, which probably will be able to switch in less than 100ns, then you need diodes with <100ns recovery.

Considering you can get 600V 50A diodes with 50ns recovery time, then 500ns recovery on a 200V 3A diode seems pretty slow. A standard 1n4148 has 4ns recovery. You should at least be able to find <50ns @ 200V easily.

Hm, OK, I looked at the datasheet, and the ones you ordered has 16ns, so it's not a problem.

 

[Arlo1]

<500ns recovery

What do you guys think?

I don't know, that sounds like a pretty long time. You are making a sturdy gate driver, which probably will be able to switch in less than 100ns, then you need diodes with <100ns recovery.

Considering you can get 600V 50A diodes with 50ns recovery time, then 500ns recovery on a 200V 3A diode seems pretty slow. A standard 1n4148 has 4ns recovery. You should at least be able to find <50ns @ 200V easily.

Hm, OK, I looked at the datasheet, and the ones you ordered has 16ns, so it's not a problem.

Yeh its just the way they list it. I guess I should edit my post.

 

[bigmoose]

Arlo, PM me your address and tell me how many you want of the RGP 15D and Ill send them to you. They are leaded, not surface mount however. I have more than I'll ever use. Just saw that you have to buy 8000 of them at Digikey! The diode you spec'd will work. A little faster than it needs to be, hence a little pricier.

Others that are fine MUR120 and US1D. Looking at prices they are all about the same.

 

[Arlo1]

Got it running. I had it spining the 12kw colossus on a dummy program lebowski give me then I used gordos chip and got it just there but I have the current at what it should be for a max for 6 fets and I had issues with the fets when I installed them I cracked the plastic because they are to close to the bottom on the buss bar and there is a bend of around .002-.005" at the bottom so My next plan is to use a file and relive the edges of the fets just a bit so the metal will conduct better.
Thanks Everybody.

 

[Arlo1]

Edit: lol I had hit the hold button on my scope and... LOL then the other problem was the powersupply I was powering the powerstage with was crapping out under any load! So now to get lebowskie's chip sorted out! [youtube]rGkxre_IOwg[/youtube]

 

[Lebowski]

you have to post the settings dude it's not supposed to jump between drive modes like that when you give constant throttle !

whoops I saw you posted all in another thread, will post a reply there

 

[Arlo1]

So.... IM hoping to be dyno testing soon first on 6 mosfets then 48 Im using these ATM http://ixapps.ixys.com/DataSheet/DS1001 ... -FX230N20T).pdf

How do I determin the right PWM frequency. I am at 40 khz on the bench but lebowski's controller is so much better I may not need it that high? If I set the PWM then test the outputs in the menu and scope them then look at how fast current rises with the excell spead sheet bigmoose give me in my motor and select a pwm that will keep the current inside the safe operating area in fig 12. is this right?

Here is the last video I made check out the cool sound at the 40sec mark
[youtube]rr6dhJEZ5Yo[/youtube]

 

[Lebowski]

thats what happens when you've set the max e-rpm unrealistically high Its like you said, it thinks its
still running the motor. I've had this happen too and noticed it always gives a increasing frequency. so what
it has is a 20% limit above the max e-rpm. When it reaches this it just stops and jumps back to drive_0. But
with realistic settings it should not do this type of thing....

 

[Arlo1]

Yeh but is sounds cool. I will set it to realistic settings later. I think 150k erpm will be where i set it to test at 40s



Now does anyone know how to make sure i use optimal PWM frequency?

 

[Arlo1]

Ok so I collected some data. I spun colossus at 20v 45v 90vloaded then 160v before load. I had a pass though event. Phase C blew the source legs off both hi and low side fets. I now have both the hi and low gate resistors (1/4w 22 ohm) blown as well. Is this what caused it or a symtom of haveing the pass though event? I have not set up lebowskie's controller 100% yet so once its running I will finish setting it up. Then work up to 40s again.
Im currently at 40khz pwm and 78khz sample. I might need to up it to 50 or 60 khz pwm for 170v
This was exciting because this is the first time Colossus has spun to 10500-11500 rpm ever. I was trying to get a scope capture when the thermal event happend lol.
The other good news is the Graphite paper is working AWESOME for conducting heat and power to the buss bar!
[youtube]Q71tqVDyFWc[/youtube]

 

[Arlo1]

A few more pics.

 

[amberwolf]

Almost certainly the gate resistors blew after the passthru, as current thru them increased total power beyond their ability to dissipate the heat.

 

[Arlo1]

Almost certainly the gate resistors blew after the passthru, as current thru them increased total power beyond their ability to dissipate the heat.

Thanks man. I will set it up more at 8s later this week. I still have to set the dead time I just guessed at 599nS. I have a couple other settings to work on as well. Then I will work up to 40s again.

 

[Arlo1]

I was thinking about it and the failure was at ~11000 rpm (max speed). So its more likly a voltage spike that caused it. I have to make the positve and negative bussbars yet which will help keep the spikes down. I Also have some plans to add some of methods favorite caps down the center of the positive and negative rails for the small realy spiky nasty spikes! Edit: I also forgot the 10k pull down resistors on the gates!

 

[Lebowski]

are you sure there's no diode between the lipo pack and the rest of the circuit ? Did you calibrate the current sensors ? Have you thought about snubbers ?

cool stuff by the way

 

[Arlo1]

Havent calibrated the sensors lol sent you a pm about that. And no snubbers and no diode. Im just using jumper wires its time to finish it with bus bars!
And yes VERY cool stuff IM the first one in the World to spin that motor that fast!

 

[Lebowski]

Also, the no-load (?) current of 20A at 90V seems very high, 1800W to spin the motor ? 11000 rpm is high but I cannot
imagine it needing 1800W.... I would definately make sure you have no shoot-through current (check with
PWM test function, battery current should remain low) and maybe change deadtime accordingly.

 

[Arlo1]

Also, the no-load (?) current of 20A at 90V seems very high, 1800W to spin the motor ? 11000 rpm is high but I cannot
imagine it needing 1800W.... I would definately make sure you have no shoot-through current (check with
PWM test function, battery current should remain low) and maybe change deadtime accordingly.

Thanks man that was my next step.
And this 19-20a at 90v is what I had with another controller So it does not suprise me but yes I will see if I can make it better.

 

[Lebowski]

How many e-rpm is 11000 rpm for this motor ?