Author Topic: SoR annoucement: microcontroller product (Read 84253 times)

paulstreats · « **Reply #150 on:** May 12, 2008, 12:31:23 PM »

Its strange because im currently designing myself a board with external memory installed. I dont think the internal memory of the mcu is that important, but maybe you should look into building an optional extra board that fits snugly on top with through pins for the ports that the memory board doesnt use. Im thinking of doing it like this for my own unit..... The 250k memory isnt going to let you hold images and pattern files for processing but external memory will

superchiku · « **Reply #151 on:** May 12, 2008, 01:05:14 PM »

why add external memory and go through all the trouble when u are getting 250 k premade in a microcontroller

paulstreats · « **Reply #152 on:** May 12, 2008, 01:10:36 PM »

how many images can you hold in 250k of memory? (consider that the data will be fully uncompressed)

Gertlex · « **Reply #153 on:** May 12, 2008, 01:29:24 PM »

Boo hiss. It looks like I'll be lucky to have this thing in my hand 8 weeks into my summer. A summer which I had hoped to spend continuously playing with robotics stuff. At least I'm not in high school any more; that would leave me with 10 weeks of summer total. I have more now as a college student...

And this doesn't sound like a particularly small change. I will be unsurprised and further disappointed if you do yet another revision afterwards.

Admin · « **Reply #154 on:** May 12, 2008, 01:52:02 PM »

paulstreats, is your argument that if you need a lot of memory, you should be using external memory anyway?

Gertlex, I agree, this has taken too long. I'm going to take time off my job to get this done faster.

This isn't a major change actually. The 640 is pin for pin with the 2560. Prototyping just takes time with shipping and testing . . .

Unless I hear a compelling reason otherwise, I'm moving to the 640. I can still do custom 2560 Axons for extra $ for the 1% of users that might need more memory . . .

JesseWelling · « **Reply #155 on:** May 12, 2008, 05:22:30 PM »

My 'demo' FreeRTOS program compiles to 9210 bytes, so I think you may be spot on in using the 640 version. From what I can tell it's completely code compatible with my Atmega128 except it has more Output Compare Registers for more PWM outputs (which I've discovered doesn't matter as much as I thought) and more ADC channels, which is always a welcome thing.

Good cost reduction. You have my vote.

superchiku · « **Reply #156 on:** May 12, 2008, 10:00:40 PM »

personally i think 64k is better as it reduces cost of manufacture and buying besides if u want to store images , microcontrollers ar not good options go for ur laptop or pc instead

airman00 · « **Reply #157 on:** May 14, 2008, 05:45:33 PM »

Quote from: JesseWelling on May 12, 2008, 05:22:30 PM

more Output Compare Registers for more PWM outputs (which I've discovered doesn't matter as much as I thought) and more ADC channels,

sounds good

Admin · « **Reply #158 on:** May 14, 2008, 07:43:16 PM »

Thought I'd give another quick update on progress.

I figured out and tested my distribution system. Advertising is set up too.

After quite a lot of thought and changing my mind several times, I decided to go with the 640. The default software right now for the Axon takes up ~17kb, and I can't imagine people using up more than another 20kb, leaving still more memory. Its exactly the same as the 2560 except the memory. Maybe I'll move to the 2560 when more people require that amount of memory, and the cost of the chip goes down. If you still need the 2560, I can do a custom job for a few extra bucks.

I finished the latest design, but it hasn't been prototyped yet. I made only minor changes so Im not worried, this is the last prototype for sure. I got my ATmega640 in the mail today so Im going to try it out on an older prototype board tomorrow - basically it'll take just a day to verify it and modify code for it. I'll release the documentation as soon as I verify the 640 works as planned - this Sunday the latest?

I'm currently in negotiation with the manufacturer about prices and other options I'd like, so this is the bottleneck right now. I may have to make short-term concessions to speed production up . . .

Oh and I decided to test every single board and every single pin for flaws. SoR seal of quality

Today I designed an external circuit to quickly do this for me

airman00 · « **Reply #159 on:** May 14, 2008, 07:48:21 PM »

Quote from: Admin on May 14, 2008, 07:43:16 PM

Oh and I decided to test every single board and every single pin for flaws. SoR seal of quality
Today I designed an external circuit to quickly do this for me

now thats what I like to hear!!!!

if only every store was as great as you , lol

alessio136 · « **Reply #160 on:** May 15, 2008, 02:21:22 AM »

Quote from: Admin on January 11, 2008, 08:41:42 AM

As many of you have realized, I have not really been happy with any of the microcontrollers available on the market. They are all missing something . . .

So I am developing a microcontroller for SoR to sell (called the Axon), with revenue going to further develop/grow the SoR website. I've done a manufacturing run and it mostly works, but I'm working out the final bugs and documentation. I've been working hard on this for the last ~6 months, and it should be ready within a month or two from now for shipment.

For those who have used the Arduino, it will basically be the Arduino x 3 (in features, support, and cost). Yet it will be smaller and lighter, too.

A few teaser features:
16 ADC
built in USB
3 additional UART (for camera, wireless, hyperterminal, etc)
no programmer required
control up to 40 servos (building a hexapod/biped, anyone?)
has a power bus
designed for the intermediate/expert builder, simple enough for a beginner
-actually, much much easier than the $50 Robot
2.5"x2.5"x.7" profile
full support, easy to use software/code
RoHS compliant (environmentally friendly, no lead, etc)

Price will be between $100 and $150. Send me an IM if you want to reserve one (no money down) so I can better judge how many to initially make. Those who reserve for the first produced limited batch will get ~20% off.

did you really make the microcontroller integrated circuit or "just" the demo-board using an existing microcontroller?

Realizing IC costs hundereds thousends dollar due to the silicon masks, doesn't it? moreover microcontroller are very complicated devices and would require maaany masks... how could you afford that invesment?!?

have you ever tried Texas Istruments MSP430? they are very powerfull microprocessor and would be very good for robots...

TrickyNekro · « **Reply #161 on:** May 15, 2008, 02:46:33 AM »

Quote from: alessio136 on May 15, 2008, 02:21:22 AM

Quote from: Admin on January 11, 2008, 08:41:42 AM
As many of you have realized, I have not really been happy with any of the microcontrollers available on the market. They are all missing something . . .

So I am developing a microcontroller for SoR to sell (called the Axon), with revenue going to further develop/grow the SoR website. I've done a manufacturing run and it mostly works, but I'm working out the final bugs and documentation. I've been working hard on this for the last ~6 months, and it should be ready within a month or two from now for shipment.

For those who have used the Arduino, it will basically be the Arduino x 3 (in features, support, and cost). Yet it will be smaller and lighter, too.

A few teaser features:
16 ADC
built in USB
3 additional UART (for camera, wireless, hyperterminal, etc)
no programmer required
control up to 40 servos (building a hexapod/biped, anyone?)
has a power bus
designed for the intermediate/expert builder, simple enough for a beginner
-actually, much much easier than the $50 Robot
2.5"x2.5"x.7" profile
full support, easy to use software/code
RoHS compliant (environmentally friendly, no lead, etc)

Price will be between $100 and $150. Send me an IM if you want to reserve one (no money down) so I can better judge how many to initially make. Those who reserve for the first produced limited batch will get ~20% off.

did you really make the microcontroller integrated circuit or "just" the demo-board using an existing microcontroller?

Realizing IC costs hundereds thousends dollar due to the silicon masks, doesn't it? moreover microcontroller are very complicated devices and would require maaany masks... how could you afford that invesment?!?

have you ever tried Texas Istruments MSP430? they are very powerfull microprocessor and would be very good for robots...

For god shake man... If admin was producing a micro then he wouldn't sell it like this...
It's just a development board based on a ATmegaXXXX with a build in bootloader...

Lefteris

Admin · « **Reply #162 on:** May 15, 2008, 01:27:11 PM »

Actually a fab would probably cost $500 million . . .

Although we all say 'mcu', uC', and 'microcontroller', sometimes what we really mean is 'augmented microcontroller'. Basically a misnomer, its just much easier to not type out the word 'augmented'.

The Axon (not spelled Axion

) is an augmented microcontroller.

Anyway, another news update . . .

I wired up the 640 and wrote the new software and new bootloader. Everything works

Next step, order the new prototype (when manufacturer gets back to me hopefully tonight) and try it out on my ERP for a few days to make sure no bugs pop up.

Admin · « **Reply #163 on:** May 17, 2008, 07:00:21 PM »

I am now releasing the documentation. It is still in rough draft form, but most of it is finished.

http://www.societyofrobots.com/axon/

There are a few things I'll be adding in the coming weeks:
A video demonstration
Updated code, many are dead links right now
A much better 'Getting Started' section that goes into detail for software
Current draw tolerances may change
Updated images of the newest version

If you have any comments, I'm open to consider them.

Now an update for manufacturing . . . I'm still waiting on getting quotes from the manufacturers. I should have them by Monday hopefully . . .

Gertlex · « **Reply #164 on:** May 17, 2008, 09:34:15 PM »

Just to be sure... are orders through Google Checkout being taken *now*?

Admin · « **Reply #165 on:** May 17, 2008, 10:15:42 PM »

Quote

Just to be sure... are orders through Google Checkout being taken *now*?

Well, not really. But if you put in an order, you can reserve an Axon. It will get you at the top of the list, but I don't play to run out.

Those who reserved during those first two weeks for a special 20% discount (no longer valid for new reservations), I'll have a different place for you to pay so don't worry.

benji · « **Reply #166 on:** May 18, 2008, 03:26:10 AM »

Quote

control up to 40 servos (building a hexapod/biped, anyone?)

what do u exactly mean by that? ,,, 40 I/O s

?
by the way , in the http://www.societyofrobots.com/axon/
its mentioned 4 PWM CHANNELS,, but actually if its based on the ATmega128 then it can provide 7 PWM channels

timer0 outs ocr0
timer1 outs ocr1a,ocr1b,ocr1c
timer2 outs ocr2
timer3 outs ocr3a,ocr3b,ocr3c

so why 7 not 8 ? its actually because the ocr2 and ocr1c are the same pin

Admin · « **Reply #167 on:** May 18, 2008, 07:49:49 AM »

Quote

Quote
control up to 40 servos (building a hexapod/biped, anyone?)

what do u exactly mean by that? ,,, 40 I/O s ?

Where do you see that written at? Its actually only 29 on the board, but if you make an external power bus more can be added.

Quote

its mentioned 4 PWM CHANNELS,, but actually if its based on the ATmega128 then it can provide 7 PWM channels

Ehhhh no its based on the ATmega640, which can do 16 PWM. But I only pinned out 9 of them - not enough room!

benji · « **Reply #168 on:** May 18, 2008, 09:10:48 AM »

Quote

Where do you see that written at?

your very first post

Admin · « **Reply #169 on:** May 19, 2008, 07:14:58 PM »

Oh that was like four versions ago . . . the current specs are here:
http://www.societyofrobots.com/axon/
(about halfway down)

I might still tweak some of the specs, so don't quote me on anything until I've manufactured the final version . . . I'll be choosing a manufacturer tomorrow . . . quotes are coming in slowly . . .

Admin · « **Reply #170 on:** May 21, 2008, 01:45:18 PM »

Turns out I'm more inexperienced than I thought when it comes to manufacturing.

The process of getting and negotiating price quotes is like a 2-3 week process, when I thought it was a 2-3 day process. Just when I'm ready, another company comes in with a better offer. At least I managed to save a few thousand by comparing ~12 different companies and negotiating with them

So yea, I'm going to spend a huge sum of money in the next day or two. From that point on, its going to be about a ~15 day process. Then I'll spend a few more days putting on the final touches (testing, programming, etc).

But then again, my inexperience has made me really poor at estimating time . . .

pomprocker · « **Reply #171 on:** May 21, 2008, 02:26:54 PM »

I don't know if anyones already said this...

maybe you can write a tutorial when your done!

haha

Nyx · « **Reply #172 on:** May 21, 2008, 03:27:42 PM »

It would be nice if you could provide a firmware version that allows the Axon to be used as an I/O board for those of us who use PCs/Laptops/etc, and controlled through a serial/USB port. That could possibly get you more clients. If not, maybe that could be another product. There aren't that many competitive solutions out there, besides phidgets, which in my opinion, are fairly limited (few I/Os, each board does very little) and highly priced. I would be especially happy if you provided some standardized/published reference of the communication protocol used, and the source code for the firmware.

Admin · « **Reply #173 on:** May 21, 2008, 03:42:47 PM »

Nyx, I'm way ahead of you!

Thats one of the first optional software packages I plan to write. It will turn the Axon into a USB servo controller and USB I/O board. You send it a serial command however you like from a PC, and it will do stuff for you. No programming required.

If you have more feature requests that you think others will like, I'll write the software for a firmware upgrade.

Most importantly I'm going to encourage open source as much as I can.

Nyx · « **Reply #174 on:** May 21, 2008, 03:55:13 PM »

The only other feature I would like is an expansion board with lots more I/Os, which could also be used with a PC-interface firmware. This would be more economical (and more practical) than having to get multiple boards. It would also make your solution significantly more competitive. I'm no expert, but perhaps this could be done for cheap by using weaker microcontrollers just for their I/O lines (I recently got 5 atmega8-16s for $8 shipped...).

On the software side... It would be great to have linux software support. Linux tends to be more practical to develop on, and more stable over time. If you program windows vista drivers/software, who's to know how hard it will be to make it all run on the next windows version, or the one after... But if you make a linux solution, chances are it won't be so hard to keep it working 15 years from now.

Admin · « **Reply #175 on:** May 21, 2008, 04:06:50 PM »

I want the Axon to be multi-purpose, basically multiple products in one. Phidgets are designed so that you have to keep buying to add capabilities (which is how they make money off of you). The Axon is designed so that capabilities can be added by software.

I could make an I2C expansion board in the future. But not in the next 6 months probably.

Quote

It would be great to have linux software support.

The USB will work with both Linux and Mac, too. You will need to find a compiler however that works for your distro of Linux. I'm sure you can find it with a quick google search . . .

Nyx · « **Reply #176 on:** May 21, 2008, 04:12:33 PM »

Quote

The USB will work with both Linux and Mac, too. You will need to find a compiler however that works for your distro of Linux. I'm sure you can find it with a quick google search . . .

gcc can probably target the microcontroller you're using, and should be usable on windows too.

If you get this business working right, I think there's a big potential for profit. Most amateur robotics platforms seem to be so poorly done, they'll be easy to beat

gamefreak · « **Reply #177 on:** May 22, 2008, 09:12:48 PM »

and yet it hsant been done.......

Nyx · « **Reply #178 on:** May 22, 2008, 09:35:21 PM »

Quote from: gamefreak on May 22, 2008, 09:12:48 PM

and yet it hsant been done.......

What hasn't been done?

airman00 · « **Reply #179 on:** May 22, 2008, 09:37:50 PM »

umm I think he means a good robotics platform

Admin , im so excited for this board!!!

I have so much stuff I wanna connect to it , lol

Can you arrange some business deal with the guys who make Blackfin camera?

News:

Author Topic: SoR annoucement: microcontroller product (Read 84253 times)

paulstreats

paulstreats