Build your own caps shaped the way you want? 
Time for MORE POWER! 1400a 1200v IGBT controller build!
- Thread starter Arlo1
- Start date
Lebowski
10 MW
Arlo1 said:
I am with HH on this one... I think a lot of attention should be paid to how the currents flow and how they change throughout the full switching cycles. Throughout the switching cycle the current should (as much as possible) stay in the same 3D space, minimizing current and voltage spikes and associated extra switching losses.
Lebowski said:
So maybe you guys don't understand the current paths?
In a rectangular controller they flow in awkward paths. And the inductance from the end caps on each side is quite large.
What I have here is the current flowing from battery > to top busbars which link all caps together > to caps > to IGBT + - with another cap parallel link on the + - end of the IGBTs as well.
Time for an Update. Today I Came over to DSS to see Casey who has been doing all kinds of stuff for the EV world including helping the kids with an electric only Drag race high. Casey is helping with some 3d printed buss bars and heat sinks to test assembly process for THE BIG CONTROLLER. Once this is up and running Casey I will both be running high power 3 phase builds at the track!!?
Attachments
-
IMG_20171214_104547335.jpg -
IMG_20171214_111801209.jpg -
IMG_20171214_111817906.jpg -
IMG_20171214_111830503.jpg -
IMG_20171214_121650996.jpg -
IMG_20171214_121648471.jpg -
IMG_20171214_121656070.jpg -
IMG_20171214_122041510.jpg -
IMG_20171214_122547801.jpg -
IMG_20171214_124942516_HDR.jpg -
IMG_20171214_124946760_HDR.jpg -
IMG_20171214_124955423_HDR.jpg -
IMG_20171214_125131521.jpg -
IMG_20171214_125115132.jpg -
IMG_20171214_125022281.jpg -
IMG_20171214_125013326.jpg -
IMG_20171214_125251482.jpg -
IMG_20171214_125301595.jpg -
IMG_20171214_125305113.jpg -
IMG_20171214_125859922.jpg -
IMG_20171214_130442831.jpg -
IMG_20171214_082321340.jpg
Lebowski
10 MW
Man you got some cool stuff there !
marcos
1 kW
- Joined
- Nov 19, 2016
- Messages
- 348
That array of snubber-like dc link could have a tenth of the parasitic inductance if you could use a laminated bus bar.
The question is if that monster cares at all about a few nH of inductance. Depends on your rise times, I guess you have heatsinking room to switch it slower.
Its looking good!
The question is if that monster cares at all about a few nH of inductance. Depends on your rise times, I guess you have heatsinking room to switch it slower.
Its looking good!
I think it will be quite awesome. Because each + and - will have 3 paths from top to bottom and a parallel link on each the top and bottom cutting inductance into 1/3 But I am well aware it should have laminated DC buss... EACH IGBT is laminated inside but the power buss I designed under the caps is not. I know this is a risky move but so far the CRX has worked without any laminated buss what so ever!marcos said:
When I was researching laminated Buss I was under the impression it was for reducing the fields caused by high voltage and the swing of those high voltages. But if the DC bus moves slow and stays fairly smooth it will not be as big of a deal. I think with the bad ass cap arrangement I came up with it should be VERY stiff!
Lebowski
10 MW
Bus bars are about low inductance.... Actually it is all about preventing ringing due to switching. A cap and inductor combination will have undamped forever continuing ringing. To dampen out the ringing you need to add a resistor in there. Typically the caps have some resistance in them.
The resistor, you either have enough or not enough. The lower the inductance, the lower the minimum necessary resistance. So with a good busbar setup the inductance is low and the caps ESR (electrical series resistance) is enough to prevent bad ringing.
No bus bar means more inductance so you will need more resistance, you may even need to add extra (this is what a RC snubber does). Not something you want in a high voltage high power setup.
The resistor, you either have enough or not enough. The lower the inductance, the lower the minimum necessary resistance. So with a good busbar setup the inductance is low and the caps ESR (electrical series resistance) is enough to prevent bad ringing.
No bus bar means more inductance so you will need more resistance, you may even need to add extra (this is what a RC snubber does). Not something you want in a high voltage high power setup.
zombiess
10 MW
One thing to keep in mind if you do not try to cancel these fields is you will become an intentional radiator. It doesn't look like it would take much to include a overlapping bus design while keeping that same style layout. Caps would most likely need to go on a carrier board with mechanical connections to the bus.
At high current even something minute such as rotating a cap's position 90 degrees (so leads form a smaller loop area by following the current path) can have a significant effect on turn off overshoot. It's so much more sensitive than I thought it would be when I started measuring high current layouts.
The philosophy behind the design rules is similar to wearing a helmet + gear. You don't plan to crash, but if something goes wrong, you'll reduce the injury.
Double pulse testing should provide some good insight as to how well it will work.
At high current even something minute such as rotating a cap's position 90 degrees (so leads form a smaller loop area by following the current path) can have a significant effect on turn off overshoot. It's so much more sensitive than I thought it would be when I started measuring high current layouts.
The philosophy behind the design rules is similar to wearing a helmet + gear. You don't plan to crash, but if something goes wrong, you'll reduce the injury.
Double pulse testing should provide some good insight as to how well it will work.
I am quite confident in this design.
When you lay things out on a rectangular layout like most conventional controllers the inductance from each end can be quite large. Meaning the caps don't all share current as well.
With this design the caps have a very short and low inductance path from each cap to ALL IGBTs.
I think this will be awesome. Laminated buss is a consideration in the back of my head. I am not using it in the controller I am running in my CRX and I am sure it might make some differences in things but I tested with and without the laminated buss and nothing showed any difference in the double pulse testing.
The first test will be without laminated buss on the caps each end will be laminated but not the DC bus under the caps.
If I need to add it I will think of something but I am not interested in soldering the caps to another separated buss bar to increase resistance inductance and add another failure point...
When you lay things out on a rectangular layout like most conventional controllers the inductance from each end can be quite large. Meaning the caps don't all share current as well.
With this design the caps have a very short and low inductance path from each cap to ALL IGBTs.
I think this will be awesome. Laminated buss is a consideration in the back of my head. I am not using it in the controller I am running in my CRX and I am sure it might make some differences in things but I tested with and without the laminated buss and nothing showed any difference in the double pulse testing.
The first test will be without laminated buss on the caps each end will be laminated but not the DC bus under the caps.
If I need to add it I will think of something but I am not interested in soldering the caps to another separated buss bar to increase resistance inductance and add another failure point...
So three igbt to control the phases. I’m wondering if I Could apply this to any mcu chip such as kelly 8080i controller mcu because it’s all I have at moment? What would I need to do to make it work draw out the circuit. And go from there?
Sent from my iPhone using Tapatalk
Sent from my iPhone using Tapatalk
