Multi plexing SSR's
Kaos Kidd
Posts: 614
While in the sticky 'HOW TO SAFELY INTERFACE A 5V SIGNAL TO THE PROPELLER?' thread, I posted a question about using opti coupliers and recieved a handfull of very helpful advice.· The issue·being the·SSR's I was looking at (cheapies) needed 5V for control.· I was looking at a cheap method of supplying the 5V without really complicating the whole project.· My suggestion of Opticoupliers resulted in a lot of reading, and explanation by Phil about how's and whats with opticoupliers.· The guts of the whole project was to multiplex a number of SSR's to control xmass lights for a nice music display.
This thread can be read here:· http://forums.parallax.com/showthread.php?p=585920
I read all that I could find on the subject of Charlieplexing;· Multiplexing both the sink / source for each object.·
Charlieplexing links:· http://www.maxim-ic.com/appnotes.cfm/appnote_number/1880· (Google is your friend!)
After all the reading for opticoupliers, I discovered the real limiting issue here were the SSR's.· With response times of the better ones being => 100us, this project is becomming more like a servo controller when considering the switching times.· Phil suggested the usage of 74HCT595's.· This would reduce the timing to that of the serial speed supported by the '595's,·and·decreases the pin count to the interface from a total of 7 wires for 42 devices (the original wiring using charlieplexing), to·3 (2 for serial communications, 1 for the storage register signal).
Now, here's where my math gets fuzzy because I don't understand the timing charts in the chip docs.
Trying to determin if how fast it will run.· The SSR's, the best max response time is <= 100us.
The '595's latching time is about 100 ns, indicating the 595's slower then the ssr's; we cant switch the 595's faster then the ssr's can handle.
This moves the fastest we can move data into the 595's is at the serial speed, which is listed at "typically 100 Mhz"
Dasiy chaining 8 (8 outputs each chip, total of 64 outputs) = 64 bits need to be shifted in for any single output change.
(Someone, anyone, correct this math if it's in error... or any part that is in error for that matter...)
100,000,000 / 64 = 1562500 updates per second is the max speed, without overhead, correct?
Knowing and understanding the propeller to run at 80 Mhz results in the following:
80,000,000 / 64 = 1250000 updates per second max (excluding overhead, and coding allowances)...
My intentions is to code this as a spin application, with a simple GUI for selecting the time line and which lights are to come and off.
My idea is to code a single cog to·output 64 bits, toggle the storage register line, increment the data index,·wait a constant time period, then loop.
So, I guess what I'm asking is how many updates per second can I expect to get?
Is this a good method, or is there another way that's better?
These number seem to high, but I don't want to pick an arbituary number and say "this is it" without it being true.· At some point after this is done, there was talk of attaching the output of an st013 (mp3 decoder) channel indicator to some math and making that the "wait constant".
All and any crits or comments are welcome on this.
Thanks...
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Propeller + Hardware - extra bits for the bit bucket =· 1 Coffeeless KaosKidd
·
This thread can be read here:· http://forums.parallax.com/showthread.php?p=585920
I read all that I could find on the subject of Charlieplexing;· Multiplexing both the sink / source for each object.·
Charlieplexing links:· http://www.maxim-ic.com/appnotes.cfm/appnote_number/1880· (Google is your friend!)
After all the reading for opticoupliers, I discovered the real limiting issue here were the SSR's.· With response times of the better ones being => 100us, this project is becomming more like a servo controller when considering the switching times.· Phil suggested the usage of 74HCT595's.· This would reduce the timing to that of the serial speed supported by the '595's,·and·decreases the pin count to the interface from a total of 7 wires for 42 devices (the original wiring using charlieplexing), to·3 (2 for serial communications, 1 for the storage register signal).
Now, here's where my math gets fuzzy because I don't understand the timing charts in the chip docs.
Trying to determin if how fast it will run.· The SSR's, the best max response time is <= 100us.
The '595's latching time is about 100 ns, indicating the 595's slower then the ssr's; we cant switch the 595's faster then the ssr's can handle.
This moves the fastest we can move data into the 595's is at the serial speed, which is listed at "typically 100 Mhz"
Dasiy chaining 8 (8 outputs each chip, total of 64 outputs) = 64 bits need to be shifted in for any single output change.
(Someone, anyone, correct this math if it's in error... or any part that is in error for that matter...)
100,000,000 / 64 = 1562500 updates per second is the max speed, without overhead, correct?
Knowing and understanding the propeller to run at 80 Mhz results in the following:
80,000,000 / 64 = 1250000 updates per second max (excluding overhead, and coding allowances)...
My intentions is to code this as a spin application, with a simple GUI for selecting the time line and which lights are to come and off.
My idea is to code a single cog to·output 64 bits, toggle the storage register line, increment the data index,·wait a constant time period, then loop.
So, I guess what I'm asking is how many updates per second can I expect to get?
Is this a good method, or is there another way that's better?
These number seem to high, but I don't want to pick an arbituary number and say "this is it" without it being true.· At some point after this is done, there was talk of attaching the output of an st013 (mp3 decoder) channel indicator to some math and making that the "wait constant".
All and any crits or comments are welcome on this.
Thanks...
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Propeller + Hardware - extra bits for the bit bucket =· 1 Coffeeless KaosKidd
·
Comments
If you're switching an AC load, especially if it's an incandescent lamp, it only makes sense to consider switching times greater than 1/120th second, or 8.33 mSec. Even then, one's persistence of vision won't notice any changes until an individual lamp is switched at a much lower rate. But let's consider the worst case: 8.33 mSec to switch 64 lamps. That's 8.33 / 64, or 130 uSec cycle time max for the shift clock (i.e. 7692Hz.) The '595 and Propeller will both be loafing at that rate!
-Phil
I take it my math is somewhat in line, and by setting a "frame" rate of 30 frames per second (motion pictures use 24 frames per second without special effects), is well in line with the 595, propeller and what spin can do... Sweet...
Thanks Phil...
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Propeller + Hardware - extra bits for the bit bucket =· 1 Coffeeless KaosKidd
·
http://www.bgmicro.com/prodinfo.asp?sid=051732638888888915286133139&prodid=REL1080&page=5&cri=REL&stype=2
dan
I'll check into the specs of thoes... they are chaper then the ones I'v seen so far...
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Propeller + Hardware - extra bits for the bit bucket =· 1 Coffeeless KaosKidd
·