Micro Lebowski Controller - DIY 3kw

[riba2233]

Standoffs can be brass, but yeah, copper is still much better.

 

[izeman]

Standoffs can be brass, but yeah, copper is still much better.

afair steel is 10x of copper and brass is 4x of copper. i don't know if this really matters a lot for these very short distances?!

 

[Futterama]

I think a proper solder joint is way better than a mechanical joint with a brass tube and steel bolt regarding long term stability and electrical resistance. But whether it matters, must depend on how much current you are going to pass through the connection.

Also remember, the solder joint will not corrode (nor will the end of the copper that is covered with solder), but if the bolt or brass tube start to corrode, the connection will deteriorate. Watch out for rainy weather then :wink:

 

[Animalector]

all great points. when I first started down this path I was going to use small bullet conectors like this

but never really decided on a solution, figuring (like many of you suggest) just to solder. which I will probably end up doing.

I like investigating the options though

Andy

 

[izeman]

those connectors look massive. good idea. my experience taught me that for 3kW power of the controller and maybe around 100A phase you will need at least 5.5mm connectors. anything smaller and they will produce massive heat. i made the pvc heat shrink melt with my 1500w mac and 4mm bullets.
now i use 10awg and 5.5mm bullets and 2.5kW are no problem at all.
on the other hand: i see no point in using connectors for parts that never need to be taken apart?! if it's the only way to put thing together than of course it's ok.

 

[izeman]

i just checked the pcb's layout and realized that it doesn't have any mounting holes at all?! why? did you try to save space?
i guess this would be a nice addition to v2 (if there is any planned). glueing the pcb to the case with epoxy or silicone is not my preferred way. or how did you plan to fix the pcbs from flying around?

 

[Animalector]

There's six mounting holes through the FETs

 

[izeman]

There's six mounting holes through the FETs

yes of course. i was speaking of the other two boards. my mistake. but as i rethink it: i guess you planned to stack all boards above of each other. so if the FETs are screwed to the box the other boards will be fixed as well. correct?
have you thought of how to screw down the FETs in the middle. it wil be ok to attach an open end wrench to the outer ones and then turn the screwdriver outside to fix it. but can you reach down to the middle one (which you need to fix first obviously)?

 

[izeman]

afaik there is no shunt to connect the CA to, correct? is there still something to feed to the CA, or will i need an external shunt?

and just to be double sure: if i look at the picture of the pcb: how are the FETs soldered in so they will have direct contact to the heat sink or case? if this is the top view and the FETs are on the bottom side: how can they be bolted to the heat sink?

 

[Animalector]

Have a look back on page 3 at the CAD render.. the fets all mount flat facing down, so the bottom surface becomes the heatsink.. so unlike an Infineon that have the FETs bolted to the side of the case, these will be screwed to the base or top....

should work right???

Andy

 

[izeman]

Have a look back on page 3 at the CAD render.. the fets all mount flat facing down, so the bottom surface becomes the heatsink.. so unlike an Infineon that have the FETs bolted to the side of the case, these will be screwed to the base or top....
should work right???

stupid me. those are to-247 cases. their heat spreader and form factor is different from the to-220 i'm used to work with.
and if you insert the screw from the outside of the box: you put a screw through the box, the FET, and to the board. and the nut sits on the board. correct?
so if you want to install it in a small box you will need to fix the nuts to the board and then you can easily screw the FETs down to the board from the outside. mission accomplished.

 

[Animalector]

Yep. Maybe even solder the nut in place. I'll have to think about that one with regards to isolation. I'll look at the PCB tomorrow to see if I isolated the mounting hole. Probably not. But it would be one track to cut..

I'll let you know
Andy

 

[izeman]

aren't there any of these white spacers as i have them for to-220?
i would just use super glue to fix the nut permanently.

 

[riba2233]

To247 doesn't need those white insulators.

 

[Animalector]

Yep. The center is isolated. Only the pad on the back is electrically connected

 

[Animalector]

Well the Mounting hole PAD is isolated from all other electrical connections.. so we can solder a nut to it without causing a short between anything important. That'll help assembly..

Drill the base of the case.. a 1:1 PDF template is already uploaded on the first post of this thread.. then just insert m4 screw. tighten. Simple.

looking forward to receiving my motor to test with. hopefully I can get to testing phase tomorrow afternoon. I could already do some PWM testing by monitoring the outputs of the Drivers... might try that Thursday afternoon see how things go..

Andy

 

[izeman]

... Drill the base of the case.. a 1:1 PDF template is already uploaded on the first post of this thread.. then just insert m4 screw. tighten. Simple. ...

you're sure it's m4? the hole of a TO247 is 3.56mm which is the right diameter for an m3. a thru hole for m4 is between 4.3-4.8mm.
btw: i can't find the pdf template?! edit: i see there are v0.2 files of the layer pdfs. i'll print that and use it.

 

[Animalector]

sure thing. I didn't measure just guestimated... M3 it is..

the important thing is that we use a nut...

Those PDF's should be 1:1..

Andy

 

[izeman]

sure thing. I didn't measure just guestimated... M3 it is..
the important thing is that we use a nut...
Those PDF's should be 1:1..

thanks. yes absolutely. soldering the nuts to the pcb will make installing as lot easier, if not even possible at all (when you install it in a case).

 

[Alan B]

Using short standoffs would be easier than soldering nuts to PCBs. Solder isn't very mechanically strong, tension, thermal expansion and vibration will likely break the bonding.

 

[izeman]

Using short standoffs would be easier than soldering nuts to PCBs. Solder isn't very mechanically strong, tension, thermal expansion and vibration will likely break the bonding.

how would you fix the standoffs to the board? the problem is: you have a case with the FETs touching the inner side of it. so you insert the screw from the outside. it goes through the box, through the FET, through the board. and then you have to put a nut on it. but you can't reach inside. so you need to fix the nuts first on the board before you start installing it. soldering is just to keep them in place. once you screw the FETs down there will be no more pressure on the nut.

 

[Futterama]

Maybe my recent mini project could inspire a solution. I also needed to attach 2 nuts to a PCB because the nuts are unreachable when the bolts needs to be tightened from outside the case.
Since these are M2 bolts and nuts, the tension is low. I could have used epoxy to secure them, but I don't know how well it would adhere to the PCB. So I used a piece of polycarbonate plate and milled a slot where I could press fit the nuts into. When 2 bolts go through the same plate, the tension torque from the bolts is managed by the opposing bolt being in place. The plate itself is mounted using doublesided tape.

I could also just have used a 2mm steel plate and then drill and tap the threads, but it was just easier on my tools to mill the polycarbonate plate and I didn't have to worry about shorting any PCB connections though the bolts and plate.

 

[izeman]

here are some pics of the controller case i plan to use. IF i can fit everything in it could be a perfect case. it is quite thick and has longer cooling fins than the (more rounded) infineon ones. even though i don't expect the case to be hot it won't do any harm to have more cooling capacity than needed. the board with the FETs will be installed in the TOP side as this is the side with the cooling fins. i will need to mill away some 6mm for the screws to fit in. but after that the screw's heads should be flush with the fins giving a good look. if someone is interested i can give you the source in china where i got the case from.

 

[Alan B]

Using short standoffs would be easier than soldering nuts to PCBs. Solder isn't very mechanically strong, tension, thermal expansion and vibration will likely break the bonding.

how would you fix the standoffs to the board? the problem is: you have a case with the FETs touching the inner side of it. so you insert the screw from the outside. it goes through the box, through the FET, through the board. and then you have to put a nut on it. but you can't reach inside. so you need to fix the nuts first on the board before you start installing it. soldering is just to keep them in place. once you screw the FETs down there will be no more pressure on the nut.

Threaded standoffs can have screws in from both sides. No soldering needed.

 

[izeman]

Threaded standoffs can have screws in from both sides. No soldering needed.

if you add stand offs to the pcb the overall height will be even more (+1cm at least). we (i) try to make the controller as small as possible. the three stacked pcbs add up to some height already.