Stamp network ideas??

Turnbull2112

I want to build a network of stamp "nodes" that are all connected via serial or some other type of comms protocol to a master stamp processor. I want each node to have a PID control that is written from the master stamp processor. Is the stamp the right type of device to use in this application? I want each node stamp to locally control a PID setpoint and feedback to the main processor for display. There may be as many as 10 or more stamp "nodes".. What does anyone think of this? Doable?· What type of communication protocol would work best? Serial? I2C? There may be some distance between processors.
Thanks.

▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Rob Turnbull
Audaci Favet Fortuna

Franklin

There is not enough information to make an opinion. What are you controlling? What is the distance? How fast do they need to communicate? How are they communicating to the end device? Things like that.

▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
- Stephen

Kevin Wood

You might be interested in this post on a similar project: http://forums.parallax.com/showthread.php?p=765140.

While I'm sure that you could find an appropriate BS2 for this, I personally wouldn't use the BS2 for for cost reasons. I would go with either SX or Propeller protoboards. You would get more performance and flexibility than with the BS2 series, and save quite a bit money on hardware costs. The learning curve would be a little bit steeper, but considering the type of application that you are suggesting, I don't think you would have any problems.

Well, re-reading you're post, and giving more consideration to what you're proposing, I would suggest looking more closely at using the Propeller. I think that on a test bench you could build a system to do what you described with maybe 2-3 protoboards. In actual application, it would depend on how far apart they would be and the limitations of your communication method - ie, max wire length, etc.

Turnbull2112

Yes, I started looking into the propeller this morning. Very interesting.. I have some serious studying to do to learn the new software. More details of what I am doing- multiple stand alone processors in control panels. Each processor monitors a process datapoint and controls a valve to maintain a setpoint (PID). All of these remote panels are monitored from one main control panel and the PID setpoints are entered from that main panel. The distance between these nodes could be as much as 50 yards. I want each remote panel to operate on this network and if communication is lost, the valve will fail closed and the main panel will indicate and issue with that particular node. I would like to be able to add or remove "nodes" on the fly without affecting the whole network. Basically I am trying to mimic an industrial PLC control system but I want something proprietary and lower cost. If I get a proof of concept model running this thing may take off and I will more than likely farm out the finer programming details. Any feedback is appreciated.

▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Rob Turnbull
Audaci Favet Fortuna

stamptrol

The previous posters all make good points.

If the Stamp boards are doing something they are capable of ( speed, number of i/o, etc) and the update time required back to the master is in the order of a second or so, the BS2 will work fine. The systems I've built have used half-duplex RS-485 commuication out to several thousand feet without an issue.

To make it work, the slaves have to be listening a good part of the time so as not to miss the master request. That may reduce the computing time available to run your PID's and do other tasks. An external communication buffer can really help that situation at the expense of a more complicated board.

Cheers,

▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Tom Sisk

http://www.siskconsult.com
·

Kevin Wood

Just to let you know, the Propeller has a basic compiler available from Terry Hitt. It's very similar to the SX/B compiler, in that it compiles to assembly code. It supports multiple cogs as "tasks", and also supports the LMM (large memory model) kernel. So, in a nutshell, it might be easier to start with than jumping in to Spin or PASM.

Since the Propeller has 8 processing units, I think that you could easily dedicate 1 or 2 for a communications bus, 1 or 2 to manage your PID setpoints, and have something left over for driving a display, etc. I would write the code (create an object) so that it allowed for configuring the controller selectively as a master or slave, and all controllers would run the exact same code. This way you could add nodes or reconfigure your network on the fly.

Just an FYI, I'm not trying to sell you on the Propeller. The chip basically does on a small scale exactly what you're trying to do on a large scale - multiple individual processors with a coordinated communication mechanism. So given that, it seems like it could be a good fit.

Turnbull2112

I think the propeller already sold itself!! I definitely like the unit price much better. I really need to get up to speed here, I will probably buy the developement board today and start screwing around. Are there individuals out there who could be contracted to develop this software? I am more focused on the end result and if I go at this alone it may take a year before I am proficcient enough to produce a viable proof of concept. So far I like what I'm reading about the Propeller.

Thanks!!

▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Rob Turnbull
Audaci Favet Fortuna

Kevin Wood

Just send Parallax support an email asking for a list of Propeller consultants. Also, post your question in the Propeller forum, or ask one of the mods to move this post over there. You'll get plenty of feedback.

Just to let you know, if you decide to use the Propeller's native languages (Spin & PASM), you might find suitable objects in the Propeller object exchange that can get you started. I don't know if there is anything there in PropBASIC; you'll need to check.

Post Edited (Kevin Wood) : 6/19/2010 2:22:45 PM GMT

Turnbull2112

Thanks a lot guys!

▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Rob Turnbull
Audaci Favet Fortuna