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

[Njay]

Which 2 slopes?

On the turn on it starts with a slow slope then a second faster slope.

That's caused by the total power loop inductance (the loop formed by the DC link caps and FETs). As the FET opens, current starts to circulate and it's the caps supplying it; but the inductance in the loop doesn't like a sudden current change, so it counteracts by "subtracting" voltage from the supply- The time length from the start of the 1st slow slope to the start of the 2nd fast one is the time it takes for the current to switch between 0 and max. See the image below.

 

[Arlo1]

Finally got the solar install done I just have 20 min of adding a breakout box and its done.

 

[SjwNz]

1.21 Giga Watts ? looks good.

 

[HighHopes]

1.21 Giga Watts ?

nice :wink:

 

[Arlo1]

1.21 Giga Watts ?

nice :wink:

We are aiming for 10kw in the long term.

Progress coming on controller just sorting out what and how I want to do it then it will come and pictures will be posted.

 

[Lebowski]

Which 2 slopes?

On the turn on it starts with a slow slope then a second faster slope. Blue trace as you asked for on the left side where is goes down. When it goes up its a off event that I need to address.

Isn't the dual slope due to reverse recovery ?

 

[Lebowski]

1.21 Giga Watts ?

nice :wink:

We are aiming for 10kw in the long term.

Progress coming on controller just sorting out what and how I want to do it then it will come and pictures will be posted.

Do they have sun in Canada ?

 

[Arlo1]

Which 2 slopes?

On the turn on it starts with a slow slope then a second faster slope. Blue trace as you asked for on the left side where is goes down. When it goes up its a off event that I need to address.

Isn't the dual slope due to reverse recovery ?

Dual slope is on turn on not turn off.

 

[Arlo1]

Do they have sun in Canada ?

More then Germany http://www.currentresults.com/Weather/Germany/annual-hours-of-sunshine.php
We are having 2 record years in a row. although this doesn't reflect that. http://www.currentresults.com/Weather/Canada/British-Columbia/sunshine-annual-average.php

And I believe Germany is now at a record 70% solar powered! http://www.renewablesinternational.net/germanys-new-solar-record-25-or-27-gw/150/452/86973/

My panels had direct sunlight on them at 8:00 am yesterday morning and the first picture I posted is at 6.24 pm.

 

[Lebowski]

On the turn on it starts with a slow slope then a second faster slope. Blue trace as you asked for on the left side where is goes down. When it goes up its a off event that I need to address.

Isn't the dual slope due to reverse recovery ?

Dual slope is on turn on not turn off.

Yes so when you turn on low side the reverse recovery of the high side makes that the output signal only drops a little, until reverse recovery is over and the output is pulled quickly down.

During this effect the saturation detection may get confused and think both FETs (igbts) are on and trigger a fault.

To reduce the effect the turning on must be slowed down.

 

[Arlo1]

Yes so when you turn on low side the reverse recovery of the high side makes that the output signal only drops a little, until reverse recovery is over and the output is pulled quickly down.

During this effect the saturation detection may get confused and think both FETs (igbts) are on and trigger a fault.

To reduce the effect the turning on must be slowed down.

You mean the desat must be slowed down? Thanks for the answer. I have never seen that before but I have also never switched 680+ amps before

But its been working so far I think I have the desat timing under control.

 

[Lebowski]

No the turn on of the FET must be slowed down.

Reverse recovery is when you try to stop the current through a conducting diode. The higher the rate of current change (amps per nsec) the longer the reverse recovery will take and the higher the peak current reached. During reverse recovery it is as if both FETs are on, and this is what can trip your desat detection.

All the way at the bottom of this post
https://endless-sphere.com/forums/viewtopic.php?f=30&t=55641&p=828491&hilit=+reverse+recovery#p828491

https://endless-sphere.com/forums/viewtopic.php?f=30&t=55641&p=875398&hilit=recovery#p875398

 

[Arlo1]

Thanks yes. I get that. But I am saying the double pulse tester is working so it is proving the desat is not an issue now that I have it set up correctly.

 

[HighHopes]

U r right lebowski. This is part of the reason desat has a blanking time. So it can ignore all the potential false triggers during turn on.

 

[zombiess]

Non laminated bus bar layout = problems at high power and sometimes low power

 

[Njay]

Which 2 slopes?

On the turn on it starts with a slow slope then a second faster slope. Blue trace as you asked for on the left side where is goes down. When it goes up its a off event that I need to address.

Isn't the dual slope due to reverse recovery ?

Isn't reverse recovery starting only when current through the diode reaches zero? That happens only at the 2nd fast Vds drop (when Id of the switching FET reaches the phase current). My understanding is that phase current is "changing circuits" and exercising the new circuit's parasitic inductances, therefore dropping a "constant" voltage which is the 1st step on Vds drop.

 

[Arlo1]

Ok I made some major progress. At 319v battery I was getting a overshoot of 360-380v I worked on many things incl experimenting with laminated buss which did not help at all! Then I added a jumper between the actual emitter and the emitter screw for the driver board and bam that 1 thing reduced the over shoot by 46% I did try this before but I tried so many things I was not sure of anything I then spent a much time as I could tonight making small progress until I think its as good as it can be at 673 amps and 345v on the battery. I will test 470v and more current asap.
I must have 400 scope shots saved on my computer for just the double pulse tester with 10 days of testing and anywhere from 1 hour to 9 hours testing.

I will now try to speed it back up a bit I tried a 250nf cap parallel to the gate VS the 150nf and it made no difference. I will try 100nf and see if its any different and speed things up as much as I can without adding much if any ringing. High Hopes has been helping with PM but I wanted to keep it off the thread so I could keep it a bit cleaner while I learned what I could to show what helped. I still have lots of testing and reading to do. But it is looking better.

I will never try to build anything like this without my double pulse tester again!
I will keep the laminated buss bars to help keep noise out of the rest of the system.

Here is a before picture followed by a few after pictures. In the first two pics The blue is the DC- and Phase with a probe at 200:1 the yellow is a very zoomed in current sensor the third pic is the gate. And the fourth is both the off and on at 673 amps and 345 volts battery (the rail sags to about 280v)

 

[HighHopes]

This looks acceptable to me. Double check your peak voltage spike us still 20% less it more than max voltage limit of Igbt DC link cap and snubber cap.

Why your gate OFF does not show 2-step?

 

[Arlo1]

This looks acceptable to me. Double check your peak voltage spike us still 20% less it more than max voltage limit of Igbt DC link cap and snubber cap.

You confuse me here? Do you mean make sure the peak of the voltage spike is no more then 20% above the max rating of the DC link Cap and snubber cap?
Right now its at 480v with a 345v battery which is 20% below. But the plan is for a 470 fully charged battery which will put it close to 740v which will be right at the 20% above mark....

I will look at finding a better snubber....

Why your gate OFF does not show 2-step?

It does its just very hard to see.

Here is a better pic from yesterday and
AN OLD SCREEN SHOT key word OLD from testing a couple weeks ago.

 

[zombiess]

Nice job Arlo1!

Looks like your car just got closer to running! It's amazing we have such a knowledgeable person like High Hopes to help us work out the bugs.

Just imaging how much learning you would of missed out on if it had worked correctly on the first go. Now you have much more experience to pull from Failing forward with electronics is pretty much mandatory when learning electronics.

 

[HighHopes]

Do you mean make sure the peak of the voltage spike is no more then 20% above the max rating of the DC link Cap and snubber cap?

i mean the peak voltage should be 20% below. IGBT is extremely sensitive to overvoltage, don't go near it.

It's amazing we have such a knowledgeable person like High Hopes to help us work out the bugs

as you know, we're just getting started. this fall we're going to kick off a full up EV showing all the design beginning to end. now if you would just respond to my email...

 

[HighHopes]

I will look at finding a better snubber...

i don't remember which cap you chose but i remember recommending a couple. so long as its a proper snubber cap AND it is physically located shortest possible between upper IGBT collector and lower IGBT emitter, then that's all you can do. after this, its about finding the right value for your system. its impossible to know before hand other than experience says the right value will be somewhere between 0.1uF and 2uf.

remember, just one snubber per phase leg

 

[Arlo1]

Do you mean make sure the peak of the voltage spike is no more then 20% above the max rating of the DC link Cap and snubber cap?

i mean the peak voltage should be 20% below. IGBT is extremely sensitive to overvoltage, don't go near it.

Ok I measured at the cap and it looked like this. If I want to use a snubber its only going to work if I can physically put it closer.

Cap is blue and current measure is yellow.

 

[Arlo1]

BAM I manned up and added another 100v. Its at 440v once the pack is charged it will be 470 max. But upping the voltage didn't seem to effect the overshoot and I also cranked the current as high at the Arduino is programed which is 930us and BAM I was able to see 820 amps with 420v at the cap while under sag. So yes I put 344,400 watts through the coil with success. My goal is to have peaks of 750-800 amps while cold as a max that I can hopefully achieve with a fully charged battery at just shy of 470v and a goal to break 300whp even if its for a few seconds. Things are stacking up nicely. Lebowski is working on field weakening and that will help get the HP up as well. These igbts are rated for 1200 amps when cold and 800 at 80degC....

I think it looks ok.

Before all of this I changed the cap on the gate from 250nf to 100nf and it brought just the slightest ring in but super small amount. I will order some 1.5 ohm resistors so I can try 4.5 ohms on and off in stead of 6 to speed it up just a touch.

I'm not sure why but the second off has a smaller over shoot.
Blue is Diff probe set at 200:1 and yellow is current. The diff probe was on DC- to Phase for the first 2 pics then the last one was on the gate.

 

[Arlo1]

Having a really great visit from someone really awesome and great for the progression of EVs Proud to call him a friend!