P.Short originally wrote
half-duplex versus full-duplex only has meaning when there are just two devices communicating
I usually consider that statement when talking about RS232.
I intend to have 10+ of these controllers on a single bus. Normal mode of function - one transmitter - multiple listeners.
Some communication requires two way conversation as in reflashing firmware, diagnostics, etc.
When two way communication is required - next question is two wire or 4 wire (half or full). If CAT5 cable is used to wire the bus - full is cheaper and easier.
I was asking - those that utilize a lighting system that cared about two way communication - if they used half - why.
I take from the replies so far, no one in the DIY community uses two way - so its a don't care so far from the responses.
That is what I was inquiring about.
About 5 years ago there as a similar discussion over on computerchristmas.com, where people were debating various lists of desired features for DIY controllers and ways of implementing them (protocols and devices). At that the only controllers that people in that group were using for light control were completely hardware based, using either 74H595 or 74HC374 devices, and afaik nobody was aware of DMX512. With the exception of one design by Ernie Horning using HC40103 counters, everything was on-off only, no dimming.
There was a lot of discussion about protocols (mostly bi-directional, if I recall right) and features lists, but nobody was actually designing any controllers. In addition, this was in the very early days of Vixen (or maybe even before the first release), so the software running on the PC was a big question. The most accepted controller design at that time was a chain of 74HC595 shift registers. In an attempt to get something going as far implementations go, I went to Borders, bought lots of coffee, and wrote my first program for the PIC, a quick, throw-away (or so I thought) design that was supposed to be as close as possible to the 595 design in the interests of easy explanation. Remember, this was the first time that a PIC-based design was presented on that forum, so I wanted to bring in as few new ideas as possible (just dimming and PICs). My hope was that this would motivate others to create the more complex designs under discussion, although this didn't happen at the time.
So that is one reason why the Renard designs lack the features that you mentioned above. Another, of course, was the early state of the PC software (i.e. Vixen) at the time.
DHinkle - it's early in the day...see who else responds to this topic.
For what we do today, just turning lights on and off, I don’t see a need for two-way communications. For reprogramming, I can see during the development stage that you might need to change the operation of the code or fix a bug. Typically though, controllers are designed in the off season, so operational issues are worked out before mass use.
That said, I know that a few of the guys that do Halloween displays, have asked about input capability of controllers. This can give feedback on mechanical props that have movement. This would allow you to change the operation based on a particular mechanical position and allow Vixen to have control timing and maybe make it variable based on audio or other props. Some have asked for just a switch input that could be used to run a particular sequence.
Another possible use is feedback to Vixen of failures. Maybe a string of lights stopped working or a CFI tripped. Maybe the controller lost power and doesn’t respond to a request.
Currently I can’t imagine that the incoming data stream would need very much bandwidth. Send a packet and get an acknowledgement. Reading an input would just be a blip. But who knows, build it and we’ll figure out some way to use it in a way that nobody had thought about before.
JHinkle wrote:I've been doing embedded for 30 years and have formed a bias towards micros - much like one forms towards politics and religion. I don't care for PICs. I like Motorola/ Freescale products. So feet first - means do it my way instead of using one of the many fine controllers already on this site.I have no expectations of a coop buy, etc. If people like how I think and what I will do different with this controller - then it will be available to the community - hardware and software. (It will include a bootloader - so anyone wanting the software and does not have the development tools to flash it themselves - it will be upgradeable using a PC via a COM port).
I currently designed the controller to use full duplex communication. I am running CAT5 as the RS485 link so there are plenty of twisted wire pairs available so half is not required because there is not a limitation of only one pair.
Also - with full duplex - I don't have to add the additional bias circuitry usually required for half.
My opinion is that half is used commercially because the bus only needs a single twisted pair. Wire is expensive. Since CAT5 has 4 pairs - wire is not the issue - so why half?
Hi Joe,
I too haver a long history of embedded development. I also have a bunch of Motorola M16C development boards in my bottom drawer which would make excellent lighting controllers. I figure that the light show could be downloaded entirely into the M16 flash, eliminating the need for the permanent PC connection. A back-channel also allows the show to get input from viewers (e.g. a PIR to restart the show when someone shows up, buttons to allow viewers to select songs, etc).
Like you, I see no reason to use half-duplex... it makes sense only when comms channels (wire or radio frequency) is relatively expensive... i.e. over long range.
Frank
