How many PIC chips would it take?

[pmuldoon]

A customer insisted that I ask Microchip the following question. I'm still waiting for a reply, and afraid that my request may be taped to the refrigerator door of their break room for laughs. But, I've read enough messages here to know that there are some amazingly talented people out here. Maybe you'll tell me it CAN be done. javascript:emoticon('')

Email to Microchip 1-18-10
Our customer is insisting on using a PIC chip for his design and that I get your opinion of it's suitability for the application.
Can you give your opinion as to if it is a suitable choice, or if we should go to something with more performance?

The application will be a proprietary Laptop-like device that must support:
1280x1024 32-bit color graphic LCD Display
Webcam video/audio capture and display
Stylus handwriting and recognition
WIFI
USB thumbdrive support
Instant Messaging between devices
Headphone/Microphone
Touchscreen

[bungee-]

Well I believe that you could do this with PIC32 as heart of the system. They have USB OTG peripheral, DMA's, Ethernet MAC ....

How to integrate and connect everything together, well that's another question

[Guest]

Depending on quantities involved it's probably way cheaper to take a $200 'notebook' or similar device and cut appropriate code. R&Ding this project from PIC to working prototype will comsume HUGE $$ in not only hardware design as well as software.let alone the 'can you add this' requests.
I'm not saying it can't be done, but 'run the numbers' on alternatives.
Heck, I just found a 4channel DVR WITH nice software,SATA drive,ethernet for under $300 ! So invest more time in getting the hard facts/numbers first.

Jay

[mbradley]

Just in my opinion;
Suitable: no
Possible: yes

Sounds like the criteria is not an embed controller, but rather a proprietary PC. Cheaper, easier to just write your own OS with a mini laptop.

Ok, just my take on the project, it would take several pics, or one large pic, and several smart components.

With smart components, I am refing to not doing your own video, but purchasing an embed video chipset. Smart wifi daughter board.

Again, my 2 cents, not feasible, but possible.
_________________
Michael Bradley
www.mculabs.com
Open Drivers and Projects

[pmuldoon]

Thanks for the thoughts, guys.
My thinking is that if I could get it to work, it would be at the marginally acceptable level and dependent on making some hardware and software optimizations that would be very application specific. But what the customer sees is something that looks like a laptop, acts like a laptop, so he will think I should be able to make it do anything a laptop could do.

The biggest, most obvious hurdle to me is how fast a PIC can move data.
A frame of 1024x1280x32 = 5.2MB. A USB2 (external) runs at about 480Mbps which is 60MB/s. I think an 80MHz PIC would have trouble moving that much external data around that quickly. Probably need some sort of external DMA (I looked at the PIC32 datasheet, and I believe it's DMA is for internal use only.) At that point, I thought it just made much more sense to look at something that supports 100's of MB's of fast DRAM and DMA to a video chip.

Thanks for the comments, guys. It's very helpful.

[bkamen]

Well, I've seen slower do LCD's because that's what an LCD controller with buffer space is for..

but doing something like live mpeg decoded video on a PIC, probably not without some external help.

This is what the 500MHz+ microcontrollers are for.. and when you look at them, they start looking more like small general purpose microprocessors (that need all sorts of external support) than embedded micro controllers.

My Palm Pilot T|X will do 480x320 video.. has some of what you mention (WiFi, bluetooth, SD card, touch screen) and runs a 312MHz PXA255.

so scale accordingly.

-Ben
_________________
Dazed and confused? I don't think so. Just "plain lost" will do. :D

[Ttelmah]

First thing, USB video cameras, usually send compressed video, not 'raw'. Few do 1280*1024. Most are lucky to send 640*480*30fps....
Second, the USB peripheral, is 'internal', so is of course supported by the DMA controller.
The programming for this project is enormous. The hardware could be 'made to work', but you would need to write a USB stack, USB driver software for each peripheral, decoder code for the camera data, video drivers, a basic OS, etc. etc.. Without looking too hard, I'd estimate something in the order of perhaps 5-10 man-years of software development involved, even using off the shelf modules.
I'd really say several orders of magnitude simpler, to use one of the embedded Linux distributions, and hardware to go with this...

Best Wishes

[pmuldoon]

Well, that would last me until retirement!
Actually, I wasn't completely clear. I was thinking HS-USB which the PIC doesn't support internally. And the webcam would be a popup window of 640x480 or possibly even 320x240. Minor points, but could shave an easy 3 weeks off of your projected sched?

I have pointed out to management that simply having to pose this question should make it pretty obvious that I/we aren't up to the task. (Laughing out loud never seems to go over well.)

Thanks again for all of the responses.
I owe you each a Red Bull!

[mbradley]

I looked into this just a tad bit, as if I was offered the job, how I would takle it, and to be frank, I would just source out a plain vanila laptop and co-brand it, and load on a stripped down linux OS with custom applications that hide the OS GUI.

Total hardware investment would be under $5k with co-braning, then its just a software development issue.

For a CNC shop, I had placed mini pc's with linux on them and custom software so the operators didnt know it was a PC, total cost $400 per workstation (less monitors-touch screen).

Hope that helps.

ps. if the client insist on a PIC, add one to the custom laptop, to do real world interfacing or authentication, problem solved
_________________
Michael Bradley
www.mculabs.com
Open Drivers and Projects

[bkamen]

Just look at the power with the average Netbook these days.

It's pretty incredible.

Yet the transmitter I made for someone last year or the atmospheric probe the year earlier would have been a little overkill having a netbook transmitting simple A/D data or turning on a battery powered transmitter.

:D

-Ben
_________________
Dazed and confused? I don't think so. Just "plain lost" will do. :D

[mbradley]

Would the probe really been overkill, really?

I mean you could have live video feed that people on the internet could watch, you could have set up an IM BOT and people cults around the world could have IM'd questions to the probe, etc...

Sorry, just kidding..... I see so many products with useless features... I even once saw a USB lady's electric leg shaver, USB so you could set it for hi med low, and check battery and recharge it.
_________________
Michael Bradley
www.mculabs.com
Open Drivers and Projects

[bkamen]

[asmboy]

how many? short answer = NONE

its not the hardware its the software that will defeat U.

Write an OS that meets user expectation for a pic, any pic ?

good luck - you have an idiot for a customer who will make a fool out of you if you play along.
EXCEPTION: Are you being paid by the hour to TRY?

[pmuldoon]

LOL
Well, I'm glad I was able to provide a little comic relief to the forum.

My company has a hard time differentiating what we are capable of doing vs what we are capable of learning to do. And I seem to be the only one ever to speak up. (kind of like the 'emperor has no clothes' thing).

You guys seem to have avoided my comment about the PIC not being able to move data sufficiently fast for the HS-USB2 or video frame.
Was I wrong?

if(I was wrong)
{
Is the PIC able to move all that external data fast enough?

Or, am I just that wrong about how much and how fast the data may have to move?
}

Actually, I'm sorry. I'm getting way off track of what the forum is for.

Thanks again for your help.
(I'm hoping to get back to simple 18F projects with misplaced semicolon bugs real soon.)

[bkamen]

The PIC32 does have a USB2.0 full speed capable host.
So the interface is there.

Whether it can stuff 60MB/s is another question all-together.

If the CPU is moving the data while running at 80MIPS, I would say "no".

I haven't looked at the datasheet for the DMA module and the CPU datasheet doesn't go into that level of detail.

So my guess on a large scale is "probably not" in terms of a PIC being able to slam full BW of USB2.0 for long periods of time transferring something like video streams or Hard Disk data.

I believe any data at such rates would be bursty.

-Ben
_________________
Dazed and confused? I don't think so. Just "plain lost" will do. :D