NB is it time to expand out to other platforms ?

[seulater]

Let me first say that i Love NB and it products!
I the years that i have used the software i have never had one problem, that was not my own fault !

However as times are getting tough, and things are getting smaller and smaller i would like to break away from the larger NB modules like the MOD5270 and go with a single chip or 2 chip solutions from other providers like Luminary's LM3S9B92 (single chip solution) or the new LPC1768 which just needs and external PHY. Using these parts would greatly reduce the cost of projects and size.

A good feature would be to use something like the SST25VF016B 8-pin spi Serial Flash part to store the customers web site in to help save the processor flash area for users code.

So my plea to you guys is that you would seriously consider using another manufacture other than Mot. and try something different. With your expertise in code and things i think you will be pleasantly surprised at how many people will come your way.


Rabbit, well... they just released code to support firmware updates for their network products,after i bitched in a conference call with management & engineering at how silly it was to have a networked enabled product and NOT have the ability to flash it via the network. I mean common Rabbit, get your head out of the sand. Not to mention they have a goofy RTOS like implementation, if i can even call it that. So needless to say NB has no great threat there IMHO, other than they have a great size module at a good cost for the RCM5700 . Again, if you don't mind not being able to remotely flash the board that is. I personally cannot live with out it. Development time with rabbit is crazy taking 15-25 seconds a crack to re-flash the board serially. With NB its just a couple seconds!!!!

Microchips new PIC32MX Ethernet Starter Kit, BLA!
Let me save some of you guys some time if your even thinking about it.
I have 2 of these kits, the speed is great i will say that, but remember no RTOS, and the free TCP/IP stack, well.... you get what you pay for. Documentation is a JOKE to say the least. they have no idea how to think outside the box and explain things to new customers. Even this new part has nothing to explain the TCP/IP stack and how to start a fresh project. There are only 2 TCP/IP examples and one of which is not even for this kit! So with bad or non existent documentation and 1 example i wish you well in that avenue. One more thing, they have a feature where you can store your web site in external flash, there is a whole paragraph about this in the manual. when you select this option and compile you get compiler errors because you were also supposed to define things. yet there is no documentation on it. when i asked support, they told me (the customer) that i need to put my nose to the ground and push though it. That pissed me off beyond belief, dont tell me that when its your job to provide me with the information i need to use your product! Lastly, they have whats called a PMP port which is the address and data bus, ~CS,~RD,~WR lines, which is a nice feature, but they also have this pin sharing feature. as it turns out if you need the PMP port some of its pins are shared with the Ethernet PHY pins so its one or the other. I needed to use the PMP for my 480x272 LCD, only to find out that i cant because of this feature.

OK, enough bitching. there was a point believe it or not, it was to show NB that though there are other guys out in the networking world they dont hold a candle to what NB has release on day #1. NB has their act together and this is why i feel if they would support another uP product like NXP, Luminary or Atmel i believe they would really grab the major share hold of the people out there that are out there with these chips but no quality RTOS and Stack at an affordable price.

[tod]

I would just like to say I concur. I often have projects with lower cost and more importantly lower power consumption requirements and hate having to leave the NB family to pursue them. (The most recent was also for a Luminary product, the Stellaris LM3S3748.) This is all NB's fault, they've made it so easy to develop software they've spoiled us!

[seulater]

This is all NB's fault, they've made it so easy to develop software they've spoiled us!

Could not have said it better

[pbreed]

We are now releasing a dev kit for the SBl2E, it currently does not have a debugging solution, but its a low cost part.
Sales is also discussing selling the SBL2E chips as pre programmed parts to customers that are interested.

I've ported the core RTOS and some of the loader code from the SBl2E platform to a ti/luminary part,
but we have still not decided if we want to make this a product.

It basically doubles our support as we now have two rtos, two compilers, two bsp driver branches etc...

The chip costs for similar freescale parts and similar luminary parts are very similar, the luminary single chip parts seem to have more memory,
the freescale parts seem to be a tad bit faster.

We have been working off all the rough edges for a number of years so our tools and support are really well polished.
Starting that over on a new platform to the same quality level is a daunting task.

[seulater]

Starting that over on a new platform to the same quality level is a daunting task.

No pain, no gain

seriously though, it might be a good investment to sort of split the company off if you will, by hiring a couple more guys to port what you have for software over to and ARM platform and then make a kit. I would be more than willing to design a kit for you guys for free to help accomplish you goal.

I know what i am asking is no trivial thing. However i really believe it would most likely wind up being the major source of your income. At the very least do a web poll on the subject.

[pbreed]

When you say low cost what are you really expecting?

So how low is low enough?

The SBL2E is 19.99 in quantity.

Is that low enough?

We were thinking about selling bare pre programed chips for about a $5.00 adder for the software?

Is that low enough?

Eventually hardware will be free (IE sub $1 for 32 bit parts) how do we make a viable business in this realm?
So that means we have to start charging for software,presently we make nothing on our dev kits, so any software
based revenue model would by necessity be different then our current one.

Would a really low cost non-network solution help?

We've actually contemplated releasing the mod5213 tools to the community,as a marketing effort,
we just aren't sure how to do this where the result does not kill us on support or tarnish our name,
or enable someone else to create a network product that competes against our core business.

The present 40 pin dip for the MOD5213 is a costly assembly.
Would a $15.00 in any quantity pre programmed TQFP version of the MOD5213 be a big seller?

Paul

[seulater]

Maybe i was not explaining myself clearly with all my talk of low cost.

The problem with the SBL2e-100IR, SBL2e-200IR & SBL2E, is that are so striped down that all you basically can do is serial. So this is not what i would be referring to using.

What people need is the functionality of say a MOD5270 but in a smaller and lower cost solution. We still need the ability for web server, TCP, UDP and so on, along with the GPIO I/O, SPI, UART, I2C and even yes the address and data bus. Using the ARM7/9 parts will give you all this, if of course the code can fit into the flash area is the only real concern. I would have to imagine that if using a NXP176x part with 512k of flash you could get quite allot of it in there.

[pbreed]

The basic message is a lower cost smaller MOD5270.

What features would you be willing to give up?
Smaller is hard as the connectors pretty much set the size limit,
what pins would you vote off the island?


Paul

[Chris Ruff]

2 mm connectors, anyone?

-Chris

[pbreed]

Maybe times have changed, but when we surveyed customers before designing the module format
the number one request was connectors that don't break. (about 6 years ago)

2mm connectors are 20% smaller so thats a bit of improvement.
Any smaller than that and the connectors get real expensive.

I believe that per pin the 0.1 connectors are still the lowest cost.
We could go to three row connectors, but that pushes up the layer count for the
board it plugs into.

Just a note to seulater the SBl2E has

GPIO, I2C A/D uarts just no address and databus.