New DMX dongle

[RJ]

Ok a few assuptions I'm making for this to be a benifit:

Some users already have the cheap usb to serial convertors that can handle the 115K speed (most all can do this its the faster ones that cost)

For those that dont these cheap units cost very little $9 - $20

for some of our members the current offerings of DMX dongles at $60 - $150 is a healty expense.

That the majority of the members are not running lights shows of over 192 channels.

Ok, if this is true then here are some facts :

It is easy to write pic code to send dmx data.
Pic's are cheap
It is easy to write code to recieve 115K serial data
streaming serial data out to the usb to serial convertors is easy for vixen

With this in mind what about this for a cheap homebuilt DMX solution of our own.

Program a pic to recieve a 115k stream of data from vixen just like you do for the entec pro dongles and then to send out the dmx stream from the pic.

So you would have a little sub dongle that plugs into you serial convertor and then you plug you output cable to it and you have DMX. The pic would keep up the dmx stream not the computer just like in the Entec Pro.

The first thought is 115k is not fast enough from the port to feed a dmx stream. This is true for a full 512 channel stream but what about a trunucated or "Short frame" of 192 channels. hopefully you agreed that most of use do not even do 192 ch shows. The data stream would be legitimate dmx and should work other dmx equipment.

The end would be a product for those not wanting to speed the money of the more expensive solutions and not needing 512 channels on one usb port.

I believe this would be a very inexpensive way to do this.

Let me say right up front I am not purposing this for myself I will gain nothing from it I am simply trying to come up with a dirt cheap alternative for our members to use that would like to move to dmx. I am in hope that in time the equipment the DIY'ers are using can move to one protocol which multiple different devices can use instead of continuing to create devices that all use different protocols which require each different type of device to be hooked to a different port and another feed from our computers. I know the value my opinon carrys here but I would hope good ideas are universial. I have learned from the other indurstries I work and play in that without a common standard to work from progress slows. If the computer industry had not given up all the different companies independant standards and settled on solid standards across the industry you would not be reading this right now. (so you can blame them). :lol:

[KC]

If I understand correctly...

Your PIC-based dongle would be maintaining the DMX stream, as you stated. The data from the PC would only be used to update the state of the channels in the DMX stream periodically.

Doing the math...@115,200 you get 14,400 bytes/sec...to update 512 channels at 25 ms (because that number has been thrown around) with 1 byte/channel = 20,480 bytes. So even at 115k, you could still get the full 512 channels of a single universe by taking your timing down to roughly 40 ms (or greater). This doesn't account for any overhead in the packet structure, but still, at 50 ms you could easily fit the full 512 channels. At 25 ms, you're limited to around 360 channels (no overhead accounted for), which is still pretty decent.

This, assuming that I'm on the same page as you...? It sounds like a less expensive version of the Pro unit.

[RJ]

KC,

You are dead on with my thinking. I was using conservative numbers to make sure we can hit what we are saying. Would rather promise 256 channels and be able to deliver 360 than the other way around. And we all no theres overhead that we take a bite out of the perfect world numbers.

[DynamoBen]

I've had a similar thought for a while. My "dream" would be to have a common serial protocol (Vixen Serial Protocol = VSP). This protocol would allow anyone to create any kind of conversion dongle they want. So it could be VSP to DMX, or VSP to Renard. All vixen would do is send out intensity up dates. That gets everything onto one common plug-in.

As far as refresh and configurability that would be left up to the dongle. So the developer could make a little programming app that could change things like number of channels transmitted and what not.

I think its a worth while project, but the first thing is to come up with a protocol. Since I've been toying with this idea for some time I have a skeleton of a protocol. Thoughts?

[andyhough]

I work for an integration software company. We have integration software that allows our 130 or so application adapters to plug into a bus similar to the PCI bus on a PC but for software. There are several protocols already available, like Web Services over TCP or Java Mesage Services. Many of them are shareware.

My thought is that finding a standard that already exists, and building an the VSP adapter to plug right into it, would offload a lot of development, and allow any number of adapters to be added later. If you would like to chat more, this would be a nice project for me to get my teeth into for next years, "nifty use of our technology award"

[DynamoBen]

I'm with you in spirit. I'm of the mindset that if I we were to follow any standard it should be one adopted by the lighting industry. I guess the issue is that there isn't really anything in the lighting industry in the way of a standard serial protocol (except for maybe midi). That being said recently the Ethernet standard was finished (ACN) which should help some with future Ethernet based devices.

As far as serial I feel like it should be pretty simple to implement, expandable, and as usual free. The skeleton protocol was as simple as a header, data type identifier, data, and footer. Initially controllers and dongles will use the protocol but over time it will be used to remote control vixen, hence it needing to be expandable.

I would love to see what you come up with, as you find info post for the group and we can discuss it. It’s always hard to standardize but I think it’s worth trying to do.

[RJ]

I think it's time we raised this idea from the dead!

I dropped it as there was very little interest in DMX at the time but it has gained some support so I worked out a couple of design for this. It looks like on the DMX Pro Dongle The fastest that the port settings in Vixen will go is 115K so unless it does not matter and vixen can run faster than that on the usb even with it set that way in the plugin ( help us out on this KC) I have tested 512 channels at 25ms on my setup and it seems to update find. I only had 216 channels in use but I tested it all with vixen setup with 512 channels and had the extra channels doing random stuff. with that said the usb chip is the only problem it is surface mount so we would Have problems with most users building it this way. I could solder it and some other here could but we have to plan for the majority. (It's a .8mm pin spacing so its not super small as they go.) So we could either go with a all in one design and I have asked JEC if I did this would he help get me setup with the people who build his stuff to have them built for us. ( parts would run about $15 minus pcb and assembly.) this would allow full usb speed if vixen can do it, to update the DMX stream. Or we could revert back to my first Idea on this thread and use a cheap usb to serial port adapter and then I use the other design which would be basicly a big pic with enough ram for a full DMX universe and room for working space and a Euart to communicate with the serial port. It would create and stream the DMX stream. Both methods would function like a DMX Pro where the pic does the stream to maintain good timming and the computer and vixen just spits out changes. I have my dongle so I'm set but I am simply trying to help out the community. I believe there is no reason I can not do this. I have enough time playing with DMX firmware that I have become very comfortable with it and I know the pics I work with can communicate easily at 115k. as far as serial protocol I would stay with the DMX pro's protcol so it works with the current plugin.

If we do this I see no negative to DMX for our users and so I am willing to do the work if others will fund the proto's just like with the Freestyle. It would be about $14 a piece for the proto boards for 4 people. I will buy the parts for my proto work.

Lets get a dialog going on this let me hear your thoughts on it.

I would start right after we order the proto boards for the Freestyle if that works out, if not as soon as I come to the conclusion the freestyle isn't going to pan out.

[DynamoBen]

Something important to keep in mind. Even though the baud rate is set to 115K the USB interface actually runs at close to 3M. The Pro uses the FTDI USB to Parallel interface. In the documentation it states that a baud rate must be set but the interface will communicate at a much higher speed in this case up to 3M.

While my math isn't great I do believe that you will run into a bottle neck if all 512 values are changed simultaneously at 115K. So if you go the serial route you may need to be a bit "crafty" in your approach.

I'm all for it if you can get it to work. :)

[rrowan]

Hi RJ

I really love this idea. I read this post when I first join up here and was hoping it would come back from the dead.

While I am just starting out and will be using a Grinch board for 2008. I can see going to DMX after that. The cost of the Pro box was a major factor and then there is a cable conversion from the 5 pin (bnc?) to I guess rj-45.

So yes I would like to help get this going.

Thanks

Rick R.

[snething]

Great work RJ! Count me in to help fund or test this project!

Thanks

[DynamoBen]

RJ if you want some design inspiration, this is a project I've been watching for a few years.

http://www.usbdmx.com/usb_dmx_interface.html

Its a great reference.

[RJ]

Ben can you confirm that vixen is able to spit the data out at the faster rate even with the DLL set to 115k? It would just simplify the diy part if the we could start with a cheap usb to serial adapter. I do have a usb chip coming but the suface mount part is going to be a problem for everyone. I know the usb port can do it but I was curious if it does under inpout.dll control since it was designed for standard serial ports. If not I can drop that idea and simply work on the direct usb version.

[DynamoBen]

From the DMX pro API manual:

"Install the VCOM FT245BM device driver on the PC. Since the driver is a virtual COM port, the baudrate setting used to open this COM port is a dummy value, and does not control the USB communication speed."

They are using the FTDI FT245BM chipset. If you want to get up and going w/o surface mount try these:
http://www.dlpdesign.com/usb/usb245.shtml
http://www.dlpdesign.com/usb/245pb.shtml (has onboard pic 16F)
http://www.dlpdesign.com/usb/245pl.shtml (has onboard pic 18F)

BTW You will notice that the usbdmx link I posted earlier uses a similar USB to parallel interface. It is all about speed. Personally I think you should have enough speed to be able to do two DMX universes.

BTW I will help in any way that I can. I do have access to a real DMX tester but need some lead time to get a hold of it.

[RJ]

I looked at these the cost was the only thing as the parts to do it with just using all dip except the one surface mount and the part cost was only $18 and that alone is $25 plus the other parts. We made need to go that way but i was hoping to keep this thing in the $25 - $30 range total.

[DynamoBen]

Even if the cost went to $50 it would be worth it if the dongle had multiple DMX universes.

[JEC]

I looked at these the cost was the only thing as the parts to do it with just using all dip except the one surface mount and the part cost was only $18 and that alone is $25 plus the other parts. We made need to go that way but i was hoping to keep this thing in the $25 - $30 range total.

What package does your SMT chip come in?

As I've learned this week, a $15 syringe of solder paste and an ex-kitchen pancake griddle make short work of all things surface mount.

It's surprisingly easy to do... and nearly magical to watch 30 different components simultaneously solder and perfectly align themselves, just from the heat of the grill.

[RJ]

Yes I have thought that way many times myself. I can even hand solder most packages with the drag and braid clean method but I just don't feel too many here are ready for that. It is a matter of time before we all must accept the surface mount parts and learn to work with them but I think it would just scare too many at the moment.

I wish it were not that way as I am working on layout for the Freestyle and if I could use the SMT version of the parts on it we could have a very small very neat, clean and state of the art little controller doing 128 channels. But I fought the urge and just made it a bigger board to fit every thing in. One of these days I will do it just to have it. LOL

[DynamoBen]

As I've learned this week, a $15 syringe of solder paste and an ex-kitchen pancake griddle make short work of all things surface mount.

It's surprisingly easy to do... and nearly magical to watch 30 different components simultaneously solder and perfectly align themselves, just from the heat of the grill.

Ahhh yes the reflow skillet, look here for more details.
http://www.sparkfun.com/commerce/present.php?p=Reflow%20Skillet

I too have been nervous about moving to SMT...I may have to bite the bullet too. I just need some nudging I guess. :oops: Maybe I will use the SMT version of the Propeller on my project instead of the huge DIP version. ;)

[P. Short]

SMT is not a big deal. The only reason that I've been avoiding it on the coop projects is because of the perceptions of other people, not because I agree with them.

--

Phil

[grages]

RJ,

I would be willing to help fund a prototype board.

Shawn

[JEC]

SMT is not a big deal. The only reason that I've been avoiding it on the coop projects is because of the perceptions of other people, not because I agree with them.

--

Phil

Indeed.

[RJ]

SMT is not a big deal. The only reason that I've been avoiding it on the coop projects is because of the perceptions of other people, not because I agree with them.

--

Phil

Posted @ 3:28 am

Man Phil,

Do you ever sleep? LOL

Exactly but these are small parts also ssop 28 etc so I think we should go with the little $25 dip module at his point anyway. I think we could still get the cost in at around $40 would that be cheap enough for most you think? I know the serial adapters run less but I just can't get the cost less unless we do one of two things.

1 - Go SMT and make the users work with a .64mm pitch ssop 28 which is 10 mm long x 5mm wide.

2 - Design it up and talk JEC into getting us setup with a china assembly plant to just build them for us. I understand from him it cost hundreds of dollars just to get setup with them.

What do you guys think?

RJ

[DynamoBen]

I think we should get some units working first then decide what the final board layout will include. So let’s go with the fastest and most obtainable parts, which sound like DIP.

JEC and Phil either of you want to make or know of a tutorial to help folks like me get over the SMT fear? Part of the reason I've avoided it is because I tend to make boards for prototypes at home. With that being said not having to drill tons of holes would be a welcome change.

[JEC]

I think we should get some units working first then decide what the final board layout will include. So let’s go with the fastest and most obtainable parts, which sound like DIP.

JEC and Phil either of you want to make or know of a tutorial to help folks like me get over the SMT fear? Part of the reason I've avoided it is because I tend to make boards for prototypes at home. With that being said not having to drill tons of holes would be a welcome change.

The folks over at sparkfun have done some great work already.

http://www.sparkfun.com/commerce/hdr.php?p=tutorials

And scroll about halfway down the page for lots of information.

[JEC]

2 - Design it up and talk JEC into getting us setup with a china assembly plant to just build them for us. I understand from him it cost hundreds of dollars just to get setup with them.

RJ

If there's enough interest, I'm happy to ask the folks overseas to quote on this.

However, once tooling, setup, and shipping & customs fees are considered, outsourcing probably doesn't make sense for less than 50-75 units. 100 would be even better.

That being said, a multi universe dongle sounds very intriguing.

@ Ben - maybe this needs a propeller. Getting two discrete streams out of a PIC is tricky unless you use a bigbig chip with more than one UART. Would hate to try bit-banging in software so quickly.

But a nearly idle prop could drive, what, seven universes?

Definitely design it to support RDM, even if it's not implemented immediately.

Once could even add a footprint for one of those XPort ethernet gateways. The bones would be there for a drop-dead inexpensive ArtNet bridge.

There are stacks of alternatives are out there, but priced north of $1K.

Mostly because it's a (1) relatively tiny niche product running (2) decently complex firmware which (3) can't fail. Ever.

For the guys doing this sort of thing for real, stability, consistency and bombproofiness trump price every single time.

[DynamoBen]

NOTE: This topic has been split

I split the discussion about at mutli-universe dongle into its own thread so that RJ and others can continue discussions about a DMX Pro replacement here.

Multi-universe is now here:
http://www.christmasinshirley.com/forum/viewtopic.php?t=2216

[wbuehler]

Thanks for the links on the Re-Flow skillet.
I belive that I would like to try some SMT now. I just need to go buy a skillet.

Bill

[mrpackethead]

Using both the skillet and the reflow oven are great ways of going things.. I'll never go back to through hole by choice.. for prototyping its great.. You can swap components out multiple times without damaging the copper if you are carefull..

I'd highly recommend if you can using the oven over using the skillet.. its a bit more work, and you need a controller ( the sparkfun one is great, i bought one ).. but its a lot more reliable than using the skillet.. I got a oven that also had a hotplate on the top for about $80..

I use my oven to 'dry' out components prior to loading onto the board.. ( 70C for 6 hours ).. Moisture thats in/on components that rapidlly heated can cause cracking of components if you are not lucky.. This is only a problem for components that are loose, or coming in tubes.. If you can get components in cut tape ( or full tape if you need a lot ), they are sealed and dont have the same issue.. Just dont' take ages from the time you open the tape, till you've soldered them. Also make sure you keep your solder past in the fridge it goes off otherwise., it will last years past its expiry if you do this.. Only other tip is keep everything clean.. make sure your boards are free from any contaimination ( including oils/grease from fingers! ). Clean with detergent and water, dry well, and handle by the edges seems to work well. Careful use of solder flux is also a really good idea.. You'll love it the first time you do, its almost magic watching it melt and all teh components align themselves..

You'll need some good tweezers, or a vacumn pick up, good lighting to work with. I have one of those lamps that has a magnifying glass in teh middle, and lighting around the egde.. its almost impossble to read whats on an 805 sized component..

Good luck, its really worth the effort..

[NogginBoink]

RJ,

I'm coming to this party late, but are you planning on putting the FTDI chip on board?

For $4.95, I'd encourage you to include the FTDI chip in your design and make your board a 'complete' solution, rather than force people to buy a $10-25 cable.

I think we as a community have to come to terms with the fact that we either embrace surface mount parts when we need to, or limit the functionality of our designs. There's just too much stuff out there available as surface-mount only for us to ignore it.

[grages]

Posted @ 3:28 am

Man Phil,

Do you ever sleep? LOL

Exactly but these are small parts also ssop 28 etc so I think we should go with the little $25 dip module at his point anyway. I think we could still get the cost in at around $40 would that be cheap enough for most you think? I know the serial adapters run less but I just can't get the cost less unless we do one of two things.

1 - Go SMT and make the users work with a .64mm pitch ssop 28 which is 10 mm long x 5mm wide.

2 - Design it up and talk JEC into getting us setup with a china assembly plant to just build them for us. I understand from him it cost hundreds of dollars just to get setup with them.

What do you guys think?

RJ

RJ,

Is this dead? Or just on hold waiting on other projects to get out of the way:). I know you have been busy! I choose option 1 and keep the price down.

OR.. a variation on option 1

Maybe use something like a Schmartboard (http://www.schmartboard.com/index.asp?page=company) that the company claims
We can teach a 10 year old, who has never held a soldering iron, to hand solder a .4mm Pitch chip...flawlessly, in 1 minute. and has a pretty good video that explains it. 3 minute flash (http://www.schmartboard.com/index.asp?page=movie)

More Videos soldering here (http://www.schmartboard.com/index.asp?page=resources_howto) with schmartboards.

One idea would be to get a quote on having the boards made using SchmartBoard's "EZ" technology ( Link (http://www.schmartboard.com/index.asp?page=licensing) ) I have no idea how big the boards would be so I didn't take this step, and don't know how expensive it would be to do so. Prior to having boards made I think using one of their EZ prototype boards (http://www.schmartboard.com/index.asp?page=products_qfp&id=66) to breadboard one up would be a good Idea. I would order and supply the board if you were interested. I would fund to whole breadboard if you want, if you have time to work on it.

I understand that this DMX dongle may not support all 512 channels or only limited channels at various timings.

P.S. I realize you may be fully familiar with the schmartboard technology and may have stayed away purposefully from that solution.

Hope you haven't given up on this idea.

Shawn

P.P.S My lynx didn't show up yesterday, guess it was to much to hope for the USPS could do it in one day even though it only has to travel 100 miles. Maybe today.

[RJ]

Currently I'm working on a Dmx dongle but I am asking that no one makes plans based on this. It is just a toy project as Phil and Ben are making real effort at it. I went with an all dip solution using a USB capable PIC chip so that the cost could be really cheap. I think it's possible to do it for in the $20 range including the pcb we will see.

RJ

[daviddth]

I think there are a number of us that dont need DMX yet, but would like to in the future. If your dongle works out, then I will get one. If not, then its no loss as I am not relying on it for the display this year. I am considering getting a few DMX items and sitting them aside so I can beta test any of the DMX dongles that may arrive, but thats down the track as I'd prefer to get the real work of the display out of the road first :)

[DynamoBen]

It is just a toy project as Phil and Ben are making real effort at it.

"Real effort" who says mine isn't a toy. ;)

[wbuehler]

I need a DMX dongle

Bill