Evolution

Evolution.jpg (1641374Bytes)

Just had a rummage through one of my boxes and found *most* of the different revisions of the board that I use as my main controller, just thought some of you might be interested in how things evolve as needs change.

On the far left was my very first PCB, with a PIC16F88, a dual h-bridge, RS232 interface, ADC, Current Monitor and Tri-colour LEDs,

Next was a very 'analog' version, this time with a PIC24EP256, two dual h-bridges, real time clock, sd card and 10 channels and USB.

Thrid is another attempt, this time with protection against short circuits etc, added bluetooth short range, dual can, dual lin, dual RS232.  Note the two much bigger 30A hbridges now :)

4 and 5 are the wrong way around, as you can see the strappng on '5', when I discovered that my adc's were backfeeding through the hbridge power supplys and causing noise.  The solution was to add some diodes on the hbridge inputs and buffer the adc inputs.  Changed to a dsPIC33EP256

Six, a bottom up redesign of the board, now with RELAYS too :) added programmable voltage outputs, more capacitance for the h-bridges, changed to a 100meter bluetooth module, added some port expanders, onboard temp senders, quad encoder inputs - 2 PICS now with a high speed data bus between them, forgot to connect some of the wires (expensive mistake) so more strapping.

Se7en, this is currently the current state of play, all of the above goodies, some more protection, better filters, more I/O and dual 16-bit dsPIC33EP512s running at 70MIPS each (280MHz PLL).

Obviously this is not designed to be a robot controller, but it really does a great job of it.

I think the point of this post is that with some effort and determination cool things can happen.  I would also like to point out that I was a complete noob at electronics when I started this project (I used to manufacture windows and doors), so any can learn this stuff too. :)

Hi bdk6,Thanks for your

Hi bdk6,

Thanks for your comments.  This is 3 years developement from first board to seventh board, with the first board taking about 2 months, then about 10months for the next one, and then pretty rapid development between the later stages.  I have been wanting to move over to an arm based board for some time now, but none of the suppliers seem to want to talk to me.  I just build 3000 of these things, thats 6000 top-of-the-range 16bit dsPICS, so you can imagine the bill on that.  I said to a couple of the big suppliers at the start of the year - send me an arm dev kit, and i’ll move to arm and buy my processors from you?  Nobody was interested, and microchip made another 6000 unit sales as a result.  (Thats $43000 someone lost out on)  :(

EDIT: Just out of curiosity, what would you recommend processor wise that would give the same or better performance in a single chip, and still come in cheaper than the pair of dsPICs?

When I saw evolution as the title

I wasn’t quite expecting this but thanks for sharing this. Apart from your skill at board layout design you must have good business skills as well. To sell 3000 boards and have demand for another 3000 is not too shabby. What is their main purpose?
I’m surprised in a way about your arm experience. I’m not sure what kind of dev kit you asked them for but with so many cheap dev boards available to investigate their possibilities why don’t you just buy one. Unless I misunderstand your requirements. Is it the crippled free IDE’s that come with them your main concern?

 

I will look into the M3 and
I will look into the M3 and M4s, they sound promising. I would love to try out the ARM thing, but its catch 22, i cant convince my boss man to buy things unless i understand them, and i cant understand them without trying them :frowning:

I would not like to say too much about the intended purpose of the board, suffice it to say that if you look at the list of comms interfaces above you can probably figure it out. These boards are in use all around the world now, in use by an international company that everyone knows.

It disappointed me when I contacted the large component houses, and none of them would give me an arm dev kit. I refuse to buy one, and by my thinking it would be good business sense to give me a £100 dev kit and get $40k worth of sales from me. Some of my other projects use the 64 pin version of the dspics, again in quantities of 1000s. I know they are overpowered and overpriced for what i need, but i know them inside and out, and dont need to refer to a datasheet every 5 minutes.

I have to say though, it has been a fun, if stressful at times process learning how to do electronics and embedded programming, and i learn so much with each revision. Making the change from free pcb software supplied by a pcb fab shop to using eagle, then electronics workbench, and ultimately altium. Each time thinking, i could have finished this project months ago, if only i had these features.

Do you guys get many of your project off the breadboard onto pcb? Also do you surface mount or stick to through hole?