#$%@$#@ <--- (insert favorite swearword here), IT WORKS !!!!

[salty9]

20661 bytes filled using assembly

.

And how many megabytes if you had coded in C?

[Alan B]

And how much work to port it to the ARM CPU?

[Lebowski]

And how much work to port it to the ARM CPU?

I would actually like to port it to ARM, I would like to learn ARM assembly.
Problem is that I have a soldering station from the 1960's and cannot
deal with SMD's or very small chips. Maybe an SOIC would work
in combination with a breakout board but that's the absolute limit.
I should look for a new soldering station maybe.

A 32 bit ARM would be nice (as some stuff is calculated with 32 bit precision)
especially as ARM has lot's of conditional instructions... At the
moment the 30F is capable of making ADC measurements, doing 32 bit computations
and writing fresh output to the PWM at a rate of 100kHz. Can you imagine how
fast an ARM would be

First lets see what the beta testers have to say

[Alan B]

Lots of small boards with ARMs aready mounted that will piggyback onto your board for the rest of the logic. Techniques are available to solder fine pitch SMTs with 1960's soldering equipment. Adequate soldering equipment is also available for less than the cost of a moderate Lipo battery.

[ctirad]

I would actually like to port it to ARM, I would like to learn ARM assembly.

You should consider to write in C. It will make the code much more transparent and portable. The speed loss is minimal and you allways can inline assebly routines for better speed when needed.

BTW, I'm not sure, the ARM is generally better option than dsPIC or another MCU/DSP dedicated for motor controll tasks. However, If you want to try ARM, look at STM32 line of chips.

Problem is that I have a soldering station from the 1960's and cannot
deal with SMD's or very small chips.

Don't worry, SMD is easy to handle even with a plain micro soldering iron. I commonly use mine even for FPGAs, bigger MCUs and so. The real problem is BGA.

[Arlo1]

I would actually like to port it to ARM, I would like to learn ARM assembly.

You should consider to write in C. It will make the code much more transparent and portable. The speed loss is minimal and you allways can inline assebly routines for better speed when needed.

BTW, I'm not sure, the ARM is generally better option than dsPIC or another MCU/DSP dedicated for motor controll tasks. However, If you want to try ARM, look at STM32 line of chips.

Problem is that I have a soldering station from the 1960's and cannot
deal with SMD's or very small chips.

Don't worry, SMD is easy to handle even with a plain micro soldering iron. I commonly use mine even for FPGAs, bigger MCUs and so. The real problem is BGA.

We have been bugging lebowski to write in C for months and he is to stubborn lol. Im learning it its not easy but it does seem better in the long run!

[Kingfish]

We have been bugging lebowski to write in C for months and he is to stubborn lol. Im learning it its not easy but it does seem better in the long run!

I wished it was written in .NET - then I could be of some actual use... instead of useless
Great work, KF

[Alan B]

Lebowski has chosen assembler, it is too late for a change here.

But for other realtime projects C is the most supported portable embedded language that gets to the performance of the hardware.

Dot Net is for another application space. Software engineering requires more than one language. Some are more suited than others for a purpose. Dot Net is well suited for Microsoft's purpose.

I've used many languages over the years. C was a mainstay many years ago, but for most purposes we have moved on. But, for getting close to the hardware and coaxing the maximum performance from a chip, C still reigns supreme if you can manipulate the force. Many old languages we would never use again, but C is the one that keeps on keepin on. Some of the bolted on plusplus features awake the behemoth and warp into bloatspace, but the efficient functional machine is still inside there and it will do amazing things in small time and code space if you avoid waking the wrong objects.

[Lebowski]

My preference for assembly comes from it's simplicity, there's only a handful of simple instructions and that's it.
I dislike C because I still remember, back in the day I tried to write programs for the Commodore Amiga
90% of the program lines were procedure and function definitions and only 10% of the program did something useful.

I don't know about C being so portable... Let's say I want to port the whole thing to a 33F device (+30 % speed, jay ),
the main problem I see is figuring out how to operate all the peripherals. Rewriting the algorithm is easy and not
that much work, but figuring out how to access the 33F's ADC's and PWM outputs, that's where I see the main task...
And this would not be any different for C.

And yes, it's too late to change now I just programmed the devices to send around, going in the mail tomorrow.

[Alan B]

You are right about hardware related software not being portable, but that is language independent.

Some of the new processors have hundreds of assembler instructions to learn, and very complicated behavior. Especially the reduced instruction set architectures with delayed branching and overlapping execution and multiple execution units.

PIC assembler has a very low instruction count. AVR and ARM are much larger.

It takes some time to learn C. But when you move to the next CPU you still know C. You have to learn its hardware. If you use ASM you also have to relearn that. If you move from PDP8 to 8080 to 6800 to 8086 to PDP11 to Sparc to PowerPC to PIC to AVR to ARM you will not want to use ASM anymore.

[pelle242]

The portabilty of Lebowskis code i mostly academic for now. What are your intentions for it? Keep it your personal baby or open source it and let others contribute? Just curious...

[Lebowski]

The portabilty of Lebowskis code i mostly academic for now. What are your intentions for it? Keep it your personal baby or open source it and let others contribute? Just curious...

Well, it runs a unique sensorless algorithm with
automatic timing advance which I invented
myself. To get to where it is now took me about
500 hours of designing and programming.

So, I don't really just want to give all this work
away, I'm hoping to start selling controller chips.

[Lebowski]

I posted the schematic and setup manual in a new thread:

viewtopic.php?f=30&t=36602

The setup manual gives an idea of the options my controller IC has...

[deVries]

And yes, it's too late to change now I just programmed the devices to send around, going in the mail tomorrow.

About how many beta testers do you have now? How many have got it working so far? Is there a thread for beta testers & feedback?

Ahhh, I see you linked the beta test thread in the post just above.

Thanks!