Seeking backers for IO board project on kickstarter

If it is truly rated for industrial conditions, then $100 is a steal, have you done any testing on how well it integrates with Rockwell PLCs or Siemens, etc?

Thank you for the feedback.

We were wondering too how much sense it made to have an add-on card to cost more than the PI. Here is some of the rationale.

- First, $100 is what we ask the backers. This is to pay for the development which involves a lot of work: protyping, firmware development for the on-board MCU, software development on the Linux side, ... We figured that the money we raise would only pay for that cost, and the cost to build and ship the product in low volume. After that, the card would eventually be made in high volume and we had a target resale price of around $75 for the basic version and $150 for the version with the Inertial Naviation Unit.

- Even assuming $100, we figured that the resulting system (PI + RIO) would be an extremely capable and inepensive solution for the money. No PLC or other industrial computer would come even close in terms of price/performance.

- The I/O card processes the input signals using its on-board MCU and configurable firmware. The best example is the ping ultrasound sensor that we show in the demo. It take a fair amount of processing to generate the pulse trigger and then meaure the echo pulse, and it must be repeated every few miliseconds. If the card had no I/O processor, it would be a lot harder and time consuming to interface such a device, and the PI would be wasting cycles.

- We also figured that the DCDC converter would be a valuable feature that would save the cost of an external adapter for user looking at operating the PI from batteries.

- We have not talked about this feature in our product presentation but we intend to make the MCU on the card programmable by the user using a Basic-like language. This way, the user can process the I/O locally and keep the RI for the supervisory function. This onboard programming language is also what would be used to make the CAN and RS232 communication work with other commercial devices (sensors, PLCs, ...)

Cosma

You are very coy about the real world specifcations of this device - the RPi is not an industrial product and the RPi people have never claimed that it is.

You need to firm up on the spec  a bit before anyone could use this in a real industrial application:

What temperature range will your board and the Pi operate over.

How are your inputs and outputs protected.

What EMC specs will you meet.

Will your software comply with any recognised standards for quality and safety.

MK

Good points. The goal is not to go after Siemens or Allen Bradley, although chips that are made for mobile phone are very enduring and it is quite likely that the PI will prove very robust in most applications. Again, our prime target is mobile robotics, and in particular autonomous or semi-autonomous applications where the 700MHz ARM will make the difference and where the alternative solution would cost many time more, be much larger and consume more.

We obviously believe that our proposed card opens the PI to a world of new applications. But at the end of the day, it is for the community to decide if this card would be a useful addition to the PI ecosystem. This is why we created the kickstarter project instead of making a direct investment in this idea.

I agree with Michael, while the Rpi is an interesting board as a low cost Linux powered controller, it is not really a product designed and built for an industrial control environment. About PLCs you can get a CLICK PLC for $69, already proven and certified for control applications.

Still the project sounds interesting and a good approach to extend the capabilities of the Rpi. I understand the costs associated with developing such a product (that is what I do for a living) and I backed several kickstarter projects, my impression was that compared with other projects of higher complexity like Parallela that raised almost $900K, $100 feels too high to my pocket, even at the target $75 sale price feels too high, on top of that $50 to ship outside the US ?

What MCU are you planning to use on the board ?

Wish you best of luck on your campaign.

-J

The MCU is an STM32 ARM Cortex M3, In the version with the AHRS navigation module includes its own STM32, so the board contains 2 processors in that configuration.

As for price, $75 is a max that we are targetting and it is for single quantity. If the demande is very high, then manufacturing in volume would allow making it cheaper.

But fundamentaly, price is a matter of value. If someone needs to do something and the board allows to do that job faster and/or cheaper - or for that matter, to do it at all - then it will be worth the price. Also, in applications that could use this board (or other IO boards), there are other system costs: wires, sensors, motors, joystick, batteries, etc, which can add up to considerably more than the PI+RIO cost.

Thank you for your encouragement.

The MCU is an STM32 ARM Cortex M3, In the version with the AHRS navigation module includes its own STM32, so the board contains 2 processors in that configuration.

As for price, $75 is a max that we are targetting and it is for single quantity. If the demande is very high, then manufacturing in volume would allow making it cheaper.

But fundamentaly, price is a matter of value. If someone needs to do something and the board allows to do that job faster and/or cheaper - or for that matter, to do it at all - then it will be worth the price. Also, in applications that could use this board (or other IO boards), there are other system costs: wires, sensors, motors, joystick, batteries, etc, which can add up to considerably more than the PI+RIO cost.

Thank you for your encouragement.

Hello Cosma,

I think you are using  an excessively weedy processor - the cost of the connectors and conditioning on an IO board will swamp the cost of the processor - use an STMF4 - same footprint (almost) - but you can get 192k of ram, floating point and DSP on chip and 168MHz. The NXP M4s are even nicer but much more work to change.

MK