My company does high-end system integration

Archive
1

We are a small company that primarily does high-end ($1 million / each) system integration for data acquisition and functional testing equipment.

We also provide general maintenance support for our high-end customers.

Working with big companies has it challenges. You pretty much complete the up-front design work before they will cut you a P.O. for a custom project. Our customers generally need to be convinced there is very little risk the project will be successfully completed on-time even if they waste months having their lawyers argue terms and conditions.

Occasionally, we get a small project request that is a rush job and bypasses the normal channels. This recently happened. We were asked to build, essentially, a fancy thermostat controller.

We were provided the schematic of an existing system with several op-amps, dozens of pots and various obsolete parts. The provided BOM was incomplete and generally didn’t match the schematic. It was going to take weeks just to clean up the documentation.

I suggested we could easily do this with a microcontroller like the ATmega328. Surprisingly, the customer agreed.

I had on hand an Arduinio (actually, it was my son’s). I did a breadboard checkout that worked as expected. I wrote the code over the weekend. The etched panel and wired circuit board are due in a couple of days. I fully expect the project will be completed by the end of the week.

The point of this story is to explain why I am searching for a professional grade embedded solution. I want to be better prepared next time.

We rarely build more than one of anything. And if we do, it has never been more than ten units spread over several years. We have a do-whatever-it-takes attitude. Be it .NET, Java, GCC, FORTRAN, machine code, etc.

Most of our customers are comfortable with Windows, ergo my interest in Micro Framework.

We also often have a need for two Ethernet ports (one for internal use and one for external use). And, NO we aren’t allowed to use WiFi for the external Ethernet.

My options so far…

1 - FEZ Domino

Pros - looks like it is ready now. For $110, I can dip my toe in the Micro Framework waters with a system that has an LCD display and five buttons. I can also expand to try other things. I would expect to get it up and running in a couple of hours.

Cons - Mostly customer perception (“It’s a toy”). There is also a legitimate concern of having only 10 bit A-to-Ds and limited memory.

2 - iPac 9302

Pros - Definitely, not a toy. 12 bit A-to-Ds. 500 ma capable digital channels. Available now for less than $100.

Cons - Unsure of Emac’s commitment to Micro Framework. Would expect difficulty in getting it to work with doubtful technical support.

3 - FEZ Cobra / EMX Module

Pros - Less of a toy, but 10 bit A-to-Ds may be considered inadequate. Available soon.

Cons - FEZ Cobra board layout would likely be unacceptable for end customer. EMX Module requires Surface Mount soldering.

4 - GHI ChipworkX

Pros - Pluggable. Professional. Nice Development System.

Cons - $400. Does anyone know of a DIMM200 through-hole socket or a breakout board?

Please excuse the lengthy post, but I would welcome critiques and suggestions.

David Petticord
Complete Networks, Inc.

EDIT - corrected iPac-9302 to read 12 bit.

2

What GHI offers is ideal for what you do. You said you only make one of your device and it has to be done as fast as possible…what is better than FEZ to do all this.

Where did the “it is a toy” come from? FEZ is certainly not a toy but it is so easy that users can use it as a toy. GHI created this separate website for hobbyists so no one would get the idea that GHI is making hobby toys. Looks like you have already been on GHI main website.

About the ADC comment, yes those built in ADCs are not made for accurate data acquisition. I think the best option for your application will be EMX module + external ADC/data acquisition circuitry.

I can go on and on talking about FEZ, not because I work for GHI but because I really love it myself.
What I suggest you do is give GHI a call and talk to one of the engineers. GHI policy is to give the customer honest answer not false promises, so if we know our boards are not a good fit for your application then we will tell you that :wink:
Today is a holiday for us so call tomorrow please.

3

The whole idea that the FEZ is a “toy” came from the whole Arduino thing. The Arduino is labeled a toy because anyone, regardless of skillset, can pick it up and use it. Unfortunately, since the Domino uses the same sort of pinout, it is going to be grouped in this blanket statement.

4

I guess FEZ Domino can be considered a toy because it is just so easy to use and has the “Arduino-pinout”, but then USBizi chipset (the core) has been used by professionals everywhere.

In the embedded world, everything is so damn difficult to use so if we see something easy then we automatically assume it is no good or it is a toy…but this is where FEZ came in to change that :wink:

5

First a correction, the IPac 9302 has 12 bit ADC converters.

I agree that if our customers want 16 bit A-to-Ds then we should go with an external chipset.

As for “It’s a toy”, don’t get me wrong. I listed the choices in order of my personal preferences.

Right now, I am leaning towards FEZ Domino as my first choice.

For rapid prototyping, there are lots of Arduino shields out there.

If we are going to have to deal with SMT no matter what. I can design my own custom board around USBizi (ala Eagle CAD) and outsource the soldering.

Yes, this is a holiday. And only truly sick individuals like us would spend it drooling over the latest technical gadgets.

I’m just trying to get my ducks in order so I can justify spending $110 on a “toy”. First to myself and then, tomorrow, to my business partners.

David Petticord
Complete Networks, Inc.

6

Another thing to point out, GHI offers consulting services and getting a prototype out in days is a GHI specialty…just give us a call.

7

Good to know. Right now we don’t have a specific order. I’m looking at something to play with as a background task.

As I indicated, we often have need for dual ethernet capabilities.

If I could come up with a $250 prototype suggestion, I could probably sell that to my partners.

I’m looking at the FEZ Domino, and two Lantronix xPort RJ45 modules ($51 / each).

http://www.newark.com/lantronix/xp1001000-03r/serial-to-ethernet-module-40-85degc/dp/24M9286

I like the idea of using the xPort because we would have the choice of either doing web hosting in .NET on the FEZ Domino or in Java on the xPort.

8

[quote]I’m looking at the FEZ Domino, and two Lantronix xPort RJ45 modules ($51 / each).
[/quote]
That can be an option.

For the high-end million dollar project you are talking about, does the prototype cost make any difference?

9

Our situation is complicated.

We are a small company. In order to work with large companies we pretty much have to itemize our expenses and markups in excruciating detail.

While we could probably get away with inappropriately charging our customers a couple extra hours for are own work, that is not how we operate.

If I were to make this an official project, we would be talking thousands of dollars for which we could purchase any one of dozens, if not hundreds, of off-the-shelf solutions.

Ergo, I need to do this on our own nickel and pretty much on my own time (like July 4th weekend).

After we build it and then show it to our customers and they find it acceptable (10% chance), then we can charge them for it.

I, personally, want to try it. It makes sense to me. I see it as a good gamble that could pay off big.

One of my partners is very conservative. He doesn’t like taking gambles. He and I balence each other very well. I suspect he will see the wisdom of spending a little money and time on this.

Thank you for being a sounding board.