Power PCB for Lebowski Controller to be combined with Bobc's processor board

Is this the next gen chip or the same but perf board variant of bobc brain?
 
This might become the next generation... depends on whether I get sensorless full torque at standstill working...

The power pcb makes for a nice development platform, only need to build a new processor board and can use the power pcbs output stage and current sensors.
 
Well, I wish you LOTS of fun, and at the end of it, a Nobel prize or something! :lol:
But seriously, I hope you will be able to sell your design one day.
(But don't give them the exclusivity, to the detriment of ES builders) 8)
 
progress from yesterday

don't do it like this. using scrap extrusion from a upvc window
IMG_20180412_202702web.jpg
here is were my screws went in (white arrow) to be attached from the outside
IMG_20180412_202710web.jpg
A lot of tedious filing and grinding went into this and it is useless.
after all this work I realised I couldn't solder the fets and then take out the heatsink to place the insulating pads in. I had a long hard think about where my other scrap aluminium was and after about an hour of tearing the house apart I found it.


do it more like this (still needs insulating pads) as Lebowski originally demonstrated
IMG_20180413_100035web.jpg
IMG_20180413_100045web.jpg

so now I'm on to the other power board bits which I shall try to create a video for.
 
There is an issue with the Zip on the first page.

It contains a file called component list, it is missing some of the cheaper more generic components on the board

here are the missing parts.
View attachment Power_PCB missing from original component list.ods

components missing.png
I only realised when attempting to build my powerboard so I could start testing.

This will set me (and others in the group buy) back till I can buy in the bits I'm missing.

How this happened is I didn't open the project first in KICAD and produce the BOM for myself but thought that every thing was in the file called component_list.

:roll:
 
Make sure you get the correct gate resistors for the Fet you are using.

Also, leave some room between Fets and PCB, this make it easier to replace a broken FET and for fattening uo the traces.
 
Make sure you get the correct gate resistors for the Fet you are using?

Which are the correct resistors for the 4468 fet, Nobody is using anything else?

Ive just put in a digikey order with the components in the picture before. Am I missing something?

This just added £40. I feel like I'm gaining nothing but experience at this point. and frustration.

components missing2.png

this is what I ordered (10 sets of to get free shipping) and some dip8 sockets
I think these are right for the 4468 fets as you posted on the first post of this thread Bas.
 
IMG_20180413_192550_800x410.jpg
This is as far as I could get today. Hopefully Monday I'll have more parts and can finish it.
 
Just be meticulous and building the board should pose no problems.

I would however add copper wire to the high current traces. Solder basically has about 10 times the resistance of the equivalent area of copper. So you need 10mm^2 of solder (a 2 mm thick layer on 5 mm wide traces) to get the equivalent resistance of only 1 mm^2 of copper..
 
I twisted some 1mm copper ground cable to a coil and soldered it to resemble your beefing up of the power traces.

Also I've left a decent gap from the board to the bottom of the heatsink.

Just got to wait for parts now
 
Digikey parts arrived. so should be able to finish my power board and update the other 3 kits from the group buy. I bought 6 more assemblies worth to make up the shipping so It was free and will sort those and sell them with the spare boards I bought, that way in the long run it wont matter about this mistake.

Anyway I'm getting quite a bit of interest now in the brain boards so it will be great when I can test the boards before bothering bas about them.

:D

Sorry to all those who are patiently waiting.
 
coleasterling said:
If you come up with a heatsink design I'll gladly machine it and send them out to the other beta guys.

That's a very cool offer. I'll look into it.
 
Tonights progress
IMG_20180416_215918_800x449.jpg

So I didn't do any in the brain boards tonight. I was up at 6am and I didn't feel awake enough to solder the fine pitch stuff. So tomorrow I should have something to test with. bas before I power up this bad boy are the 4r7 resistors in r18-r23 correct for the 4468 fet.

IMG_20180416_220718_480x854.jpg

So this is the beefy trace on the underside. You cant see much of the on top under the fets.they are similar though.
 
Yep, R18-23 for 4468's should be 4.7 Ohm, R12-17 should be 15 Ohm.

R1, I dont know the value of this one yet, it is a filter resistor for the processor board supply. Ive been using 100 Ohm so far, but shorting it out should be ok too. Plan is to drop around 5V across this resistor, assuming the power dissipated wont be too high.
 
Lebowski said:
TI has it (and is the only one I think) but as far as I know has not published how it works... plus figuring out for myself is more fun :D

Do you have any theories as to how instaspin differs from other FOC controllers? I know they aren't using HFI by default, the only notable difference I have been able to find is that they use the phase voltage sense combined with hardware filters to do something, the firmware has to know the value of these filters, either omitting the phase voltage capacitors or having the wrong value in the firmware causes the entire thing to fail. I think they are using them for some other purpose than just tracking rotor position when its coasting, the filters appear to be there to reject PWM which indicates they use them while the motor is being driven.
 
Tja, I don't really know what TI does with Instaspin. If you have a schematic somewhere I'd be interested to have a look...

What I have seen somewhere (but this is not FOC or rotor position detection at standstill) is that they integrate the phase voltages to get motor flux (as voltage is the derivative of motor flux). And then commutation is based on motor flux.

I don't do this as I drive all three motor voltages constantly... there's no point in measuring motor voltages if you supply them yourself by means of the output stage... I also try to stay motor parameter independent. One of the nice things that I have in my sensorless running is that it does not need to know motor parameters for the control loops to be stable. This means that for instance the Nissan Leaf motor from Arlo1's car runs with the same control loop parameters as Izemans MAC motor...
 
Lebowski said:
Tja, I don't really know what TI does with Instaspin. If you have a schematic somewhere I'd be interested to have a look...

What I have seen somewhere (but this is not FOC or rotor position detection at standstill) is that they integrate the phase voltages to get motor flux (as voltage is the derivative of motor flux). And then commutation is based on motor flux.

I don't do this as I drive all three motor voltages constantly... there's no point in measuring motor voltages if you supply them yourself by means of the output stage... I also try to stay motor parameter independent. One of the nice things that I have in my sensorless running is that it does not need to know motor parameters for the control loops to be stable. This means that for instance the Nissan Leaf motor from Arlo1's car runs with the same control loop parameters as Izemans MAC motor...
It's fairly mundane, note the 0.1uf caps after the voltage dividers on the phase voltage sense. I stumbled on to the impact it has by accident, I built a power stage that allowed me to switch between a VESC compatible MCU and Ti instaspin launchpads. The VESC has a nice GUI that allows you to generate graphs of the various analogue inputs, normally when running in FOC mode you get no useful data from phase v sense as you said, when I was running the VESC on the power stage with filtered vsense I discovered you could suddenly perfectly observer the sinusoidal BEMF voltages while running FOC. I'm not entirely sure how it works or what instaspin uses the data for, I think the capacitor basically holds the BEMF voltage sensed during the PWM deadtime until the ADC is triggered in the centre of the next PWM cycle, presumably this introduces a delay so the frequency of the filter must be known so as to apply compensation.

How do you deal with the impact of motor inductance and resistance for your sensorless controller? Doesn't the stator flux have a delay as ERPM increases which requires motor details to be able to compensate for?
 

Attachments

  • voltagefilter.png
    voltagefilter.png
    9.5 KB · Views: 3,343
basically the circuit is a lowpass filter. With the VESC it allows you to filter out the PWM and see the sinusoidal content of the output.... though it should look more trapezoidal than sinusoidal.

You need the motor parameters to calculate the correct timing advance for FOC. But mine does not need to know the motor parameters for stable control loop operation (and this is unusual).
 
Back
Top