Paul and Sabrina's Cheap 3 Phase Inverter (AC Controller) with Field Oriented Control

[nlc]

On my "small" 100A FOC controller, the STM32 works @ 3.3V, thus I have a 5V cheap linear regulator to supply throttle and brake pedal (with 4.99K/10K divisor before enter in ADC input of the stm32), and some amp op, but the rest is @ 3.3V, even the two isolated current sensor. I use ACS758 (the +/-100A model but there is a +/-150 and +/-200A model) which accept 3.3V and have a ratiometric output (alway VCC/2 @ 0A current) and sensitivity proportionnal to supply voltage, thus it's very well because if the 3.3V is not very precise it's not a problem, the digital result value of the adc will alway represent the same current value even if 3.3V varies over time (effect of temperature for example).

Regards

[sawickm]

Why couldn’t the Micro Controller be placed on a separate daughter card e.g. “Stamp” or “Arduino”. Then it could be upgraded as needed?

A standard interface PCB would be used with the different Micro Controller daughter cards to makup the complete controller.

That concept would also allow different Microcontrollers to be used, e.g. dsPIC, STM32, ARM, etc…..

Just a thought that I have been kicking around for awhile ???

-Mark

[CollinK]

Originally Posted by MPaulHolmes

The main problem seems to be the lack of availability in the future of the dsPIC30F. I think all the current sensors out there also have 2.5v as zero amps, and then +/- x volts for positive and negative current. Everything would have to be in 0-3.3v. I guess it's not the end of the world though.

When they say "not recommended for new designs", how serious are they about that? They are going to stop making them in a few years?

Well, the LEM sensors can be adjusted for a different 0A voltage. Their fourth pin can be used to drive the bias voltage to something other than 2.5V. For a 3.3v system it could be driven to like 1.7V. Also, the dsPIC33 supports 12 bit ADC so there would be a 4x increase in resolution just because of that. So, actually, the current sensors would be more accurate.

The dsPIC30 is still listed as "in production" so currently there is no reason to think that they won't be available for at least a few years. But, commercial developers have moved on for newer designs. Like it or not I'm sort of a commercial developer who is contributing to a hobbyist project. So my goals are to maintain a good commercial outlook for the future and thus I'm leaning toward using the newer, better chips.

[CollinK]

Originally Posted by sawickm

Why couldn’t the Micro Controller be placed on a separate daughter card e.g. “Stamp” or “Adrino”. Then it could be upgraded as needed?

A standard interface PCB would be used with the different Micro Controller daughter cards to makup the complete controller.

That concept would also allow different Microcontrollers to be used, e.g. dsPIC, STM32, ARM, etc…..

Just a thought that I have been kicking around for awhile ???

-Mark

That is a possibility. The dsPIC30 and dsPIC33 chips are almost completely pin compatible. So making daughterboards to support them would be pretty easy. Just maybe a couple of small changes. The daughterboard could be made as a DIP40 or something or maybe some sort of custom pin configuration. However, the pin layouts for STM32 and ARM chips would be likely very different and the daughterboards would have really weird trace patterns. It might be able to be done though. Also, another problem, though not insurmountable, is that the oscillator for the processor has to be really close and tightly coupled to the processor. So the daughterboards would likely need the crystal and caps right on the board.

It is an interesting idea. I don't think I'd really be interested in doing it. It complicates things which I seem to be doing enough of as it is.

[nlc]

Also, there is the possibility to propose a very small pcb which transform the QFP package in DIP package, with only the dsPic33 on the pcb and some decoupling capacitors.

[edit] oupsss, messages crossed ^^

[CollinK]

Originally Posted by nlc

Also, there is the possibility to propose a very small pcb which transform the QFP package in DIP package, with only the dsPic33 on the pcb and some decoupling capacitors.

[edit] oupsss, messages crossed ^^

Yes, it might be a good idea to make a small board which converts the QFN chip to DIP40 (and just leave four pins out) so that the whole thing is once again through hole like people like. Then we all can have our cake and eat it too. Still, I do think it best to have the oscillator and it's caps plus some decoupling caps right on the dip40 board.

[nlc]

Originally Posted by CollinK

Still, I do think it best to have the oscillator and it's caps plus some decoupling caps right on the dip40 board.

I agree for the VCC/gnd decoupling cap, but for the oscillator I am not sure, it will not be very very far

[sawickm]

Originally Posted by CollinK

Yes, it might be a good idea to make a small board which converts the QFN chip to DIP40 (and just leave four pins out) so that the whole thing is once again through hole like people like. Then we all can have our cake and eat it too. Still, I do think it best to have the oscillator and it's caps plus some decoupling caps right on the dip40 board.

FYI: dspic-mcu-daughter PCB

-Mark

[CollinK]

Originally Posted by sawickm

FYI: dspic-mcu-daughter PCB

-Mark

I figured that someone had already made adapter cards to do what we want but I hadn't looked yet. Thanks for finding that! They even sell the cards without processors so we could put whichever chip we want on there (so long as it is 64 or 80 pin). Good stuff! This makes it possible to use dsPIC33 chips and still maintain through hole construction.

[rayjay]

Hope everyone had a great (BURP) turkey day. Was looking around to see if anything new was posted on the switched reluctance project. Guess it is "on hold" status?

Not pushing just curious. This is a direction that I am considering. Was hoping to get started soon, but I will be very, very patient. hehehe


RJ Crawley