DIY Electric Car Forums banner

1 - 20 of 83 Posts

·
Registered
Joined
·
53 Posts
Discussion Starter #1 (Edited)
Homebrew AC Controller Build

Just thought I would post a build progress on my AC controller.
I am currently working on fabricating the chill-plate.
Using a CNC machine I built a while back.
The electronics are almost done.
As you can see in the photo the electronics are mounted on a cardboard backer just for the initial tests and to keep anything from shorting while I move things around.
The electronics are from the J Huebner kit. You can see more information about that kit on another thread in here

Open to any suggestions for improvement. Haven't really thought out the entire build so I am fabricating on the fly. :eek:
 

Attachments

·
Registered
Joined
·
290 Posts
looks cool to me! (is that Donkey Kong in the background?)

Also, what size is the coolant tubing , what type of coolant and how will you be affixing them inside the grooves?
-josh
 

·
Registered
Joined
·
53 Posts
Discussion Starter #3
Yes that is Donkey Kong in the background! That was my "I didn't have anything to do last summer" project. That is a full sized MAME cabinet that is running off an old PC and some scrap furniture grade plywood. Got it loaded with all the classic coin op games! Every garage should have one. With all the old PC's I see put to rest at my local dump it saddens me. Re-use and re-purpose is the name of the game. I used another old PC to build the CNC machine I have that cut this chill plate (that was the project of 3 summers past)

My project for this summer is the AC controller :)

The machined fluid channel is 1/2 by 1/2 inch so I used those brass hose connectors which have an inside diameter of 1/2 inch. Typical 5/8 OD hose should work fine there with some hose clamps. I will probably use glycol fluid in the chill plate. I have access to tons of it at my job. Everyday anti-freeze is just as good. The hose bibs are mounted on the chill plate cap and are mechanically centered over the beginning and end points of the fluid channel when the cap is screwed down with all those 1/4" bolts.
 

·
Registered
Joined
·
53 Posts
Discussion Starter #4 (Edited)
So I made a little more progress:

Verified that pulse train is working

Here are the waveforms from the IGBT driver boards.

Note the voltages that are required to trigger large IGBT's

Also note the required dead-time between the phases. It is difficult to see on the first photo but once you zoom in it is apparent on the second photo. I know I am stating the obvious here but I will throw it in just in case someone is new to this. I was that guy at one point right? Low or no dead-time will lead to catastrophic failure.
 

Attachments

·
Registered
Joined
·
53 Posts
Discussion Starter #6 (Edited)
Got a little further today.
Ordered some very nice copper bus bar that I drilled & installed.
I haven't really decided how I am going to bring my power cables in so I left them longer than necessary for the time being. Remember, I am designing on-the-fly :rolleyes:

So I am not quite courageous enough to go ahead and put massive current across these IGBT's just yet. I still need a massive capacitor network, some snubbers and a mains fuse. But you know I couldn't resist doing a little something. I put a bench supply in the circuit and drove a tiny 3 phase radio controlled helicopter motor. A video snippet of it running can be seen here:

https://drive.google.com/file/d/0B0yvjhlC62G7SnZfeklPbGZTYnMyRWlFZF9LNFRjSEpJelRZ/edit?usp=sharing

There is still much work to do but this is a promising start.
.
.
 

Attachments

·
Registered
Joined
·
1,437 Posts
i guess that a next-generation design would not have the bolt heads and coolant taps protruding out the bottom--makes it difficult to sit flat while assembling on the bench.
 

·
Registered
Joined
·
53 Posts
Discussion Starter #8
I wrestled with the idea of having the coolant outlets machined in the side of the plate but I decided against it for the following:

a) The main plate is only 3/4" thick. If I wanted to put fittings on the edge of the plate I would have to use smaller threaded hose connector. I didn't want the fitting to be a flow restriction. I used 1/2" NPT connectors. The final machined wall thickness on the top and bottom would be too thin for my comfort. I was envisioning a possibility of slight deforming of the thin walls when tapping or screwing the tapered fitting into soft aluminum. The plate cap and chill plate must remain absolutely flat to maintain a positive seal. When the bolts are tightened the chill plate and cap will be absolutely flush with each other, almost appearing as a solid piece. I did not want to see the NPT thread forming a small bulge when driving the fitting in.

b) I could have bought thicker 1" aluminum but the cost rises exponentially as the thickness increases.

So after debating it in my mind I decided I was going to do one of two things. I would put a number of vibration dampening spacers under the unit to afford me space in my final mounting solution (as shown in the attached photos). I was also thinking of just tapping some bolts into the edge of the chill plate and mounting some beefy aluminum L shaped stock.

Anyway you are correct, it is not ideal placement for the hoses but I do not have an infinite budget for this project so it seemed like a good compromise and it in no way effects performance. In the long run it may turn out to be far cooler this way. For the time being I can live with it not sitting flat on my workbench. I am more concerned with function over form. Sometimes that is how it goes when designing on-the-fly :eek:
 

Attachments

·
Registered
Joined
·
3,219 Posts
...I still need a massive capacitor network, some snubbers and a mains fuse. But you know I couldn't resist doing a little something. I put a bench supply in the circuit and drove a tiny 3 phase radio controlled helicopter motor....
Be very careful running even miniscule voltage and current levels through the IGBTs without any decoupling (aka "snubber") capacitors mounted on the bus bar. I managed to blow up a 1200V/600A IGBT module while bench testing a gate driver design at a mere 24V because I was too lazy to mount a snubber cap right on the module... That was a $200 mistake.

However, if you do manage to destroy one or more of those CM400DU modules during testing, I just happen to have several boxes of them which I purchased from the Azure Dynamics bankruptcy auction... :D
 

·
Registered
Joined
·
53 Posts
Discussion Starter #10 (Edited)
Be very careful running even miniscule voltage and current levels through the IGBTs without any decoupling (aka "snubber") capacitors mounted on the bus bar. I managed to blow up a 1200V/600A IGBT module while bench testing a gate driver design at a mere 24V because I was too lazy to mount a snubber cap right on the module... That was a $200 mistake.
Wow..thanks for the heads-ups. Amazing that these things can crush 400 amps but could get toasted at 24V with little current.
 

·
Registered
Joined
·
1,466 Posts
Got a little further today.
Ordered some very nice copper bus bar that I drilled & installed.
I haven't really decided how I am going to bring my power cables in so I left them longer than necessary for the time being. Remember, I am designing on-the-fly :rolleyes:

So I am not quite courageous enough to go ahead and put massive current across these IGBT's just yet. I still need a massive capacitor network, some snubbers and a mains fuse. But you know I couldn't resist doing a little something. I put a bench supply in the circuit and drove a tiny 3 phase radio controlled helicopter motor. A video snippet of it running can be seen here:

https://drive.google.com/file/d/0B0yvjhlC62G7SnZfeklPbGZTYnMyRWlFZF9LNFRjSEpJelRZ/edit?usp=sharing

There is still much work to do but this is a promising start.
.
.
In the commercial units I've seen they laminate the bus bars for induction reduction. Nice looking work.
 

·
Registered
Joined
·
53 Posts
Discussion Starter #12
In the commercial units I've seen they laminate the bus bars for induction reduction. Nice looking work.
Thanks for the comments :)

When you say "laminate" do you mean they encapsulate it in a resin or something?

I was going to put large black and red shrink tube on the exposed bus sections just as a safety thing but I don't know if that would achieve the same effect as what you are describing. Waiting on my snubber caps and main caps before I do anything else with the rails.
 

·
Registered
Joined
·
1,466 Posts
If we make 2 large sheets of conductors ,each sheet covers the of + and -
then say starting with the + side . The 3 holes are drilled to fit. Then the - side of this bar are oversized for insulators and a conductor ring that will be insulated or isolated from the + bar and high enough to reach the bottom of the - bar . After the + bar is on a insulation layer is bonded to it. Then the same is done with the - bar except the - holes are drilled to fit and the plus are oversized . I haven't seen any of these in over a year so the details may be off. The other thing I remember is less mass in the bars
means less inductance . And there may be some advantage to more thin layers then a single thick one ( inductively). As an example I'm taking apart a Chevy Volt . It uses 1" wide by less then 1/5" "bar" made of laminations . If I remember right battery cables have a lot of inductance that will have negative effects on the igbt's. So this laminated bar helps reduce the need for caps.
 

·
Registered
Joined
·
53 Posts
Discussion Starter #14
If I remember right battery, cables have a lot of inductance that will have negative effects on the igbt's. So this laminated bar helps reduce the need for caps.
That is awesome. Kinda sounds like those layers of copper form a rudimentary capacitor in itself! Definitely something I will explore.
 

·
Registered
Joined
·
1,466 Posts
It does sound like a capacitor .
I looked over the Volt battery conductors , it looks like about 10 laminations , total thickness 1/8"X less then 1" for about 400 amps.
200 hp and 360 volts.
 

·
Registered
Joined
·
53 Posts
Discussion Starter #16
No real news on my build except that I mounted snubber caps.
Still waiting on main caps.
But now that I have snubbers on there I feel a little better about bringing that tiny 3 phase motor up to some higher speeds. I got it to go pretty fast. I had to push it off with my fingers to get it started. I am sure I have to play with some more parameters to improve that. My bench power supply cannot supply much current either so that may also be a factor.
 

Attachments

·
Registered
Joined
·
1,437 Posts
Looking at your pictures i see two things you might want to consider for your build:

1. Typically the igbt gate drive will be thru a twisted pair of wires instead of two separate wires with a big air gap. This is to reduce EMI noise and the possibility of firing both gates at the same time. That big loop created by your white and green wires acts like an antenna to radiate and receive unintended signals.

2. Flat ribbon cable is notoriously unreliable, especially for critical applications. It may be fine for benchtop prototypes in the lab, but is not a good solution in a vibration environment such as an EV. A sliding electrical contact that relies on weak spring pressure to maintain continuity should be avoided.

3. Focus on the oscilloscope settings and measurement traces so we can tell what is being measured and what it shows about your circuit.
 

·
Registered
Joined
·
53 Posts
Discussion Starter #19
1. Typically the igbt gate drive will be thru a twisted pair of wires instead of two separate wires with a big air gap. This is to reduce EMI noise and the possibility of firing both gates at the same time. That big loop created by your white and green wires acts like an antenna to radiate and receive unintended signals.
I have had others recommend changing that as well. I will definitely address that in the next few days. Thanks. While I think these particular IGBT's would require a significant amount of stray current to produce a false trigger it is better to be safe than sorry.

2. Flat ribbon cable is notoriously unreliable, especially for critical applications. It may be fine for benchtop prototypes in the lab, but is not a good solution in a vibration environment such as an EV. A sliding electrical contact that relies on weak spring pressure to maintain continuity should be avoided.
Those are not my favorite connectors either. I plan on replacing the ribbon feeding the IGBT drivers with twisted pair cable for some noise rejection. In the interim I also plan on relying on old motorbike racing techniques to keep the other ribbon cables latched. I am going to drill tiny holes in the plug/jack latches and tie them together with safety wire making it impossible for the plug to be removed without first cutting the safety wire. If I do make it to the point where I actually install this in a vehicle I will probably hard-wire everything point to point. I am so far away from that reality at the moment.

3. Focus on the oscilloscope settings and measurement traces so we can tell what is being measured and what it shows about your circuit.
Will do.
 

·
Registered
Joined
·
53 Posts
Discussion Starter #20 (Edited)
I got my main capacitors in yesterday.
So I have been tossing around ideas on how to mount them.
Remember I am designing this on the fly :eek: so let's see how it goes.

I decided to try using some more of that white PVC board to make cap holders. Machined a pair of them. It got kinda late here so I could not finish so I will continue tomorrow. Tomorrow I will shave off the fake wood grain and also make the brackets a little smaller by shaving off some of the excess PVC material.

Also I machined the holes for the caps to the exact size. I want a clamping action on them to prevent movement so I will cut the brackets in half, forming a lower and upper half and then use machine screws to hold the two halves together sandwiching the caps in place.

I have a nice aluminum plate that I will mount the caps to. I was thinking of machining an access hole in the plate (as shown by the marker outlines) to pass flexible bus-bars thru and connect direct to the main bus-bars. The plate would be mounted directly above the IGBT's.

Here is the CNC in action. Sorry, videographer I am not.

http://youtu.be/1CcyO-kTokI

I am looking for suggestions or observations or "No-No's" about mounting the caps in this fashion.

Thanks for looking.
 

Attachments

1 - 20 of 83 Posts
Top