New Controller desi...
 
Notifications
Clear all

New Controller design

131 Posts
26 Users
0 Reactions
44.8 K Views
Posts: 3650
Topic starter
(@jhinkle)
Famed Member
Joined: 15 years ago

Phil - thanks for the story on how the Renard got started. Now it makes perfect sense as to why it was designed the way it is.

Joe


Reply
dirknerkle
Posts: 10310
(@dirknerkle)
Illustrious Member
Joined: 18 years ago
P. Short wrote: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 as possible in the interests of easy explanation.

So THAT'S why Borders went belly-up -- you ran them out off coffee and customers stopped visiting the store!

😆


Reply
Posts: 1041
(@n1ist)
Noble Member
Joined: 19 years ago

Interestingly, you see the same in the theatrical world. When DMX first came out, they spec'ed a 2-pair cable (hence the 5-pin XLRs) without any real plan for the second pair. Many places started using that second pair for a second universe. Then RDM was proposed. It was supposed to give status monitoring, remote configuration and commissioning, and all that. It never really caught on, possibly due to the infrastructure issues. Artnet and ACN both allow a back channel; we'll see how that catches on.

Unfortunately, when you are trying to deploy into an existing world, you have to fight issues with the installed base.

I would love to be able to get a simple status report with board temperatures, any issues, dead loads, etc.
/mike


Reply
ebrady
Posts: 670
(@ebrady)
Prominent Member
Joined: 17 years ago
JHinkle wrote:One of my main design points requires that I have a bus with two way communication - firmware upgrades.
I am not going to be pulling cpus out of boards and reprogramming them when I have a firmware update.
I will have firmware upgradeable via the bus. That requires two way handshaking as blocks of data are passed and flashed into the controller.
If I have 10 controller deployed in a field setup - I want two way communication with them ... I want ..

#1. Query - who is alive?

#2. Reflash firmware. This is the big reason. Cost no more in hardware than what is normally on a controller.

#3. Perform debug functions ... control controller's behavior outside of Vixen - debug lighting setup and fuctionality, etc

This is nothing new to me - I expect it. Sounds like its new to this community.

Think automotive. A vehicle has many embedded modules in them. Most are designed so that their firmware is reflashable over the CAN bus. CAN is just a version of RS485 with a custom SCI.

The design would not antiquate what people have. I use DIY SSRs. As I stated earlier - yes someone would not hang this controller on an existing bus - but from what I have read on the forum, most are running multiple buses anyway.

So that is way I was asking about full vs half.

Joe

This feature set is very similar to RDM, which is a bi-directional extension of the DMX protocol that was designed with half-duplex operation in mind. I would give that protocol a serious look before making the decision to implement your own, (I currently am looking into it for my own custom boards). since diagnostics, reporting and firmware update are things done infrequently as compared to the actual control commands I think half-duplex is adequate in most cases, full duplex is a nice to have but may not gain you enough to justify the effort for support hardware and additional programming.

This is pretty much the "meat" of my comment, but if you want a little background of why I arrived at this conclusion, read on.

Similar to what you are doing now, I also designed and built my own custom controller boards. For the 2010 season, I made a custom, half-duplex, single direction, 485 protocol. The protocol worked, but about mid-way thru the season I ended up feeling like I was "re-inventing the wheel", A lot more work was required to go this route as I had to write a plugins for Vixen and hack up code on x-lights to support it. Since the protocol was my own creation, it was pretty much up to me to figure out the problems. Support with Light Show Pro was out the window because it did not allow for plugins to handle custom protocols.

For the 2011 season, I designed a new board which corrected a lot of the mistakes I made on the first board and I removed the custom protocol and replaced it with DMX. Looking back, I have to say that had to be one of the best decisions I made as it opened up the door to a larger audience for support and troubleshooting and allowed the boards to integrate better with off-the-shelf hardware and software. Using RJ's DMX dongle, the boards can now be used with Vixen, X-Lights, LSP, and LOR (with DMX support) right out of the box.

From my experience, I can not say I would go down the route of doing my own protocol again as the reward didn't seem to be worth the effort.

I am looking into RDM this year, as there is some level of existing support in the community, and there are likely some hardware/software packages out there for diagnostics and monitoring already.


Reply
Posts: 3650
Topic starter
(@jhinkle)
Famed Member
Joined: 15 years ago

I was looking for a USB to RS485 or Ethernet to 485 device to use as the head on my bus.

To my dismay - commercial products all assume RS485 is two-wire communication (multi-point - half duplex) where multiple xmitters can negotiate for transmit rights.

The same devices supported RS422 (4 wire - multi-drop - full duplex) where there is only one transmitter and multiple receivers.

No one supports RS485 - 4 wire - full duplex.

They do support 4 wire RS422. Most consider it differential RS232 - which now relates to statements Phil made earlier.

What I am actual doing is what the industry calls a "Quasi" RS422 bus. I am using RS485 transceivers with transmitter enable pins so I can control who has the transmit rights to the "Receive" side of the 4 wire bus. The head unit (USB or Ethernet) to RS244 connects to my 4 wire bus. All controller have their transmitters disabled until one of them is directed to transmit (which was my plan from the beginning - based on protocol) - since no controller can transmit unless directed to do so by the master (head unit) --- it is referred to as a quasi RS422.

I can get USB to RS422 for $20 and Ethernet to RS422 for $60.

What has others found to work well in their system - USB or Ethernet. I am leaning towards Ethernet because the head unit does not have to be directly attached to the computer which would be the case with USB. My only concern with these Ethernet to RS422 converters is how they build the UDP frame. One byte per message or do they wait for a specific time period and then frame multiple bytes. If you have some background in this process - please share.

Originally posted by ebrady
This feature set is very similar to RDM, which is a bi-directional extension of the DMX protocol that was designed with half-duplex operation in mind. I would give that protocol a serious look before making the decision to implement your own, (I currently am looking into it for my own custom boards). since diagnostics, reporting and firmware update are things done infrequently as compared to the actual control commands I think half-duplex is adequate in most cases, full duplex is a nice to have but may not gain you enough to justify the effort for support hardware and additional programming.

This is pretty much the "meat" of my comment, but if you want a little background of why I arrived at this conclusion, read on.

Similar to what you are doing now, I also designed and built my own custom controller boards. For the 2010 season, I made a custom, half-duplex, single direction, 485 protocol. The protocol worked, but about mid-way thru the season I ended up feeling like I was "re-inventing the wheel", A lot more work was required to go this route as I had to write a plugins for Vixen and hack up code on x-lights to support it. Since the protocol was my own creation, it was pretty much up to me to figure out the problems. Support with Light Show Pro was out the window because it did not allow for plugins to handle custom protocols.

For the 2011 season, I designed a new board which corrected a lot of the mistakes I made on the first board and I removed the custom protocol and replaced it with DMX. Looking back, I have to say that had to be one of the best decisions I made as it opened up the door to a larger audience for support and troubleshooting and allowed the boards to integrate better with off-the-shelf hardware and software. Using RJ's DMX dongle, the boards can now be used with Vixen, X-Lights, LSP, and LOR (with DMX support) right out of the box.

From my experience, I can not say I would go down the route of doing my own protocol again as the reward didn't seem to be worth the effort.

Thanks for your comments. My experience does not align with your experience. Mind you - I have no experience with Vixen 2 - but it took me an hour to get my controller module working in Vixen 3.

I am really excited about Vixen because I can do anything I want and very easy.

My use of the term protocol relates to what data is packaged and sent to a controller. Vixen 3 allows me to do that very quickly.

I can use one protocol to send simple 255 dimming values - one per channel.

I can use another protocol to send 10 bit dimming values.

I can use another protocol to not send dimming values but timing event that would allow me to go to 12 bits of dimming or higher if there was any benefit to doing that.

I can use a protocol to reflash software, query the controller for status data, etc.

By having the flexibility to easily package the message frame (Vixen 3 provides that) and firmware that understands the message frame - I can do any thing I can think of - and I'm not bounded by what others have lock down - to me - that is the real essence in DIY.

To me - the implementation of protocols is one of the easiest aspects of a custom controller.

I'm sorry to hear you had issues.

Joe


Reply
Page 15 / 27
Share: