PDA

View Full Version : Understanding Design of Horning Dimmer



Steven010
10-06-2008, 09:35 PM
Sorry, I have briefly looked through the posting, do not see the details that concern me with design of Horning Dimmer. Hope someone can help…

I am reviewing a schematic of the Horning Dimmer on Feb 22, 2006 for Olsen 595 with dimming- I am thinking of a re-design. I need to understand a few things before I proceed…. If I understand this correctly, this is 8 channels of 8 bit dimming per channel. It appears the Q11 and Q10 allow for daisy chaining (signal repeating) of controllers over RJ45. I have also installed Vixen SW and attached the logic analyzer so I can better understand what is happening. If possible, can someone answer a few questions on the design and driver for me?

1. It appears this requires 64 bits sent over D0 and Strobe to clock in 8 dimming channels, If I built a 8 boards, this would be 64 channels, and this would take 4096 bits to be clocked in per “frame”. It also appears from my logic analyzer that all channels need to be re-loaded per frame, so in this case only 10 could be loaded in 1 second…maybe less. This would take > 100mS to send. If this is true, dimming would be jerky or non-existant since it takes too long to pulse in each “frame”. I use frame for lack of better term…
2. How many channels should I able to control using dimming and not have bad behavior? I need at least 128 well-behaved channels to make my current planned display…
3. how many steps of dimming could I expect in a transition in a song, realistically?
4. Is D0 channel the only one supported in Vixen driver, or can D1-D7 be used in parallel to speed up each frame load time to channels? Or do you simply rely on using more than one printer port?

Thanks for the help.

Steve VG
Steve.vangent@palm.com

scorpia
10-06-2008, 10:28 PM
not knowing much about the horning dimmer i probably cant help much with your technical questions except to say if you want dimming you might want to check out some of the other designs on the board here. the renard series is alot easier way to get dimming i would think.

if you want remote ssr's then check out the renard 64. otherwise search for the renard 16 and renard 24. they both have onboard ssr's

Peter

RavingLunatic
10-06-2008, 10:38 PM
I'm not sure if anybody is currently using the Horning Dimmer.

Your best bet on getting info on the Horning dimmer is to PM ErnieHorning directly. I'm sure he'll see this thread eventually but if you want to speed up the process send him a PM.

Steven010
10-07-2008, 12:19 AM
SInce the Grinch is a derivitive of the 595, and the Horning dimmer is also 595, perhaps you can tell me how many Grinches or 595's you can tie together using printer port before you can see delays in lights, and also how far you can run the RJ45 jack between them, also in Vixen, does the driver support more than just the D0 and Strobe?

ErnieHorning
10-07-2008, 06:47 PM
Wow, has it really been two years? I came up with this circuit back when even 32 channels was a lot.

Once ‘595 gets loaded with a value, it’s output gets latched and stays that way until you need to change any output. The 40103 continuously reloads the value that it see’s on it’s input on every zero cross. The speed that a 74595 output can be changed is totally based on the speed of the computer controlling it. The maximum speed of the 74595 is limit by the pulse width of the clock pin which is 1 uS. Which means that fastest that 4096 channels could possibly be updated in, is about 8.2mS.

The computer that I was using at the time took a total of 4uS for one clock; 1uS for on and 3 uS for off. So 4096 channels would have taken 16.3ms. I don’t think that anyone could see this delay. With that machine, I should be able to support 194 dimmable channels or 12416 ‘595 digital channels, if Windows didn’t get in the way. In practice it’s going to be less.

I checked this with my current computer awhile back. I believe it was somewhere around 160 dimmable outputs.

Steven010
10-07-2008, 09:04 PM
Wonderful. Thanks for the reply! I am going to hook up logic analyzer tonight and test. I will post results and plots (if I can figure out how).

We are going to try using one PIC controller to control 8 SSR's, as in your design with 8 595's and 840103's. we think the zero crossing can drive an interupt, as can the clock edge from printer strobe, and strobe from Printer linefeed. That should bring cost and size way down, just SSR's and H11AA1 (unless I use BJT). 41k timer will be internal to PIC.

What is the furthest run you have done between each controller on RJ45? Should I repeat the printer signals?

AM I RE-INVENTING THE WHEEL?

ErnieHorning
10-07-2008, 10:53 PM
I originally designed this to be simple to understand; and it was. It didn’t cost me anything for parts, because I used parts that were being thrown away from previous projects at work. It turned out, that it was pretty expensive for the number of usable channels that were available to use. The biggest fallacy was that it still used the parallel port, though it seemed like a good idea at the time. When a lot more people started experimenting with the cheaper 595 only board, we found out how non-standard the parallel port was. Some ports tended to function very differently at very high speeds, which function just fine at the much slower printer speeds.

I had thought about using a microcontroller to replace nearly all of the added dimming circuitry. About this same time, Phil Short informed me that he was working on an external micro that would control a standard 595 board. It would use a serial port for data and he was pretty sure that he could control it accurately enough to cause a TRIAC to function as a dimmer. He did succeed, its call the REN-C and the result is here. http://www.doityourselfchristmas.com/wiki/index.php?title=Ren-C Now with a 64 channel string of 595’s, you’re able to have 64 channels of dimming outputs. With the serial port input, the 595 is now very reliable. You also have the ability to use a USB to RS232 converter on new PC’s that don’t come with parallel ports. You can also extend the serial cable to over a thousand feet by converting it to RS485 and back.

The REN-C was also a stepping stone. A lower cost method is to eliminate the 595, and control the SSR directly using just a microcontroller. This is known as Renard, which now has several variations depending on how many channels you want on a single board and whether want the SSR’s on board or remotely located.

Are you re-inventing the wheel? Maybe.

P. Short
10-07-2008, 11:07 PM
Basically everything here is re-inventing the wheel, one way or another. I suspect that almost everything that has been presented on this forum is a re-invention...I certainly do not claim any great originality in anything that I've done here. Nothing here is truly rocket science, but go for it anyway. That is the only way to learn, and therefore is definitely not a waste of time.

Steven010
10-13-2008, 06:43 PM
See attached GIFs, total time for 164 chanrels on printer port was 18mS (approx), I set up 8 channels for partial dim. Notice it is loading a decimal FBh into each of the 8 channels. The rest were 00h.

It does appear the printer port is fast enough for this. The only question would be how far between each 8 channel drop run could be done, and if the speed of data transfer could be maintained.

ErnieHorning
10-13-2008, 11:30 PM
The original SSP mode was speced at 2 meter or 7 feet. I don’t know how far I could have gotten but I had about 40 feet of ribbon cable between two PC’s running communications software without any problems. There may have been error correction involved but I was able to transfer harddrive data at full speed.

Bi-directional was speced at 10 meter or 33 feet, which uses the same hardware as the EPP that you should be using. I found that one guy on the internet had a printer working at 75 feet with low capacitance cable. This would have likely used hardware handshaking so it probably wasn’t at full speed.

I don’t know how far you’re looking to go, but distance will depend on the cable that you use and the environment that it’s placed in. People here that use this type of interface keep the hardware close to the PC and run SSR control cables or the power cords the longer distance.

scorpia
10-14-2008, 03:29 AM
steven,

may i ask why you are trying to push this design to what seems to be its limits and beyond?

from what i can see there are easier ways to get 160+ dimming channels over a long distance.

Peter

Steven010
10-14-2008, 02:18 PM
Great question. I have asked myself this recently.

I have a huge lot, and in the past my christmas displays consist of litterly 100 long extension cords, and hundreds of short 3-outlet cords. I HATE extension cords. I want to get to 1 cord per 8-16 AC outlets per location. like a "smart power strip".

When I first looked at the Horning Dimmer circuit, I felt it was the answer. It had 8 dimming channels, and the ability to pass through to next "power strip". I can envison my yard with a few extension cords, and 1 four conductor phone cord tieing all power strips together to 1 PC in the house. I then thought "gee, I can replace 16 ICs with 1 pic". THis boiled design down to 1 pic, 1 crystal, 2 phone jacks, 8 optos and 8 Triacs, and 1 5V power supply brick. I could house this in 1 -four gang outlet box with 8 individual outlets, and 1 feeder cord for incoming AC.

However, the more and more I look at the REN-C, the more I think I can use it instead. I will need to have more than one serial port, and the 64 channels will still need to be broken into 4-16 channels or 8 - 8 channels.

The other way still seems more "modular", the jury is still out IMHO...

scorpia
10-14-2008, 05:33 PM
Well if your looking at trying to limit the amount of cables running around the yard, have you thought of the ren-16. 16 dimming channels with the ssr's on board and you can do 1 cable in then 1 cable to the next ren-16.

If you dont allready have the 595 boards, but you wanted a central board with remote ssr boards then replace the ren-c + 585 idea with a ren-64 board.

i think there is a 4 or 8 channel SSR with dmx logic on the board available over on the DLA site that RJ runs.

the big advantage of these designs for you as far as i can see is that they all get the dimming that you want. and they are rs232/rs485 based. which means that once converted to RS485 you can run 1000+ feet inbetween controllers without a huge problem.

And the parrallel port that has caused so many people issues is no longer being used.

The biggest problem i can see with the 595/grinch + horning dimmer is that you are working with a complex design that would only give you say 8 dimming channels per board. and trying to join multiple boards you will run into issues with the parrallel port in both interfacing to the pc and also joining the boards. as i wouldnt trust the parrallel port for running more than a few feet reliably.

So i guess the more information about what equipment you allready have and what type of layout/design you want to make that yiou give the peopl here the better advice we can give you .

Also do you want to make your own boards or do you want to just build a allready designed board.

anyway which ever way you decide to go please let us know and keep us informed as there are lots of people here who would like to know how others go.

Peter

Steven010
10-30-2008, 03:27 AM
Ernie, Peter,

After reviewing Renard and Horning, I decided to build Renard-8. first 8 channels are hand built, work great. I got sampled on 16F688 Microcontroller, Triacs(BTA08-600), Max232, and zero cross. I had some FR4 proto board, only parts I had to buy were MOC3023. Got enought to build 100 channels. Have 100 spare Triacs. Also have spare 595, 40102.

Ended up using dry-wall-blue box and duplex outlets, very cheap. 1 15amp cord supplies each 8 channels. I did not use the external crystal, so I am running at 19,200 instead of 57,600.

THanks for recommendation. Very clever, having PIC responsible for zero-cross and refresh each of 8 triacs.

What is the max channels I can run on 19,200 at 100mS resolution?

scorpia
10-30-2008, 08:16 AM
good to hear you had some sucess.

im not sure about the maximum channels you can get at 19200 but there is a table in the wiki with the worst case numbers somewhere (double byte characters on every channel0. and i think with the new renard pluggin does away with those so basicly you could double the channel numbers they are talking about.

Also if your interested have a look at the ren-16 or ren 24, both are prebuilt boards based on the ren-8. think both boards can be gotten pretty quickly if you required some.

anyway, allways good to hear people doing something and getting a working outcome.

great work

Peter

Ronp
10-30-2008, 03:03 PM
Number of Channels (*) Refresh Interval
Baud Rate 100 ms 50 ms 25 ms
......115200 575 287 143
......57600 287 143 71
......38400 191 95 47
......19200 95 47 23

ErnieHorning
10-30-2008, 05:25 PM
The note in the wiki is incorrect. These are the original numbers when the two byte code was being used. If you are using the Modified Renard Plug-in included in Vixen 2.X you can double those numbers. Guaranteed, I’m running 50 mS using 64 channels at 19.2 K baud with no external oscillator on the board.

rstehle
10-30-2008, 06:08 PM
Yes, those numbers in the Wiki can be doubled using the new Renard Plugin. I successfully tested 192 channels of Renard at 50ms on one com port earlier this year.