Can an LED be 20' from it's power source?

[Mbira]

I'm planning on having banks of 8 leds up to about 20' away from their power supply.  They'll be connected with a db9 type cable.  Will this be an issue?

Thanks!
Joel

 

[djn111]

Hi depends how you organise it, how is the current limiting organized ? (current source, resitor etc.)

DJN

 

[abbey road d enfer]

Since LED's are current-fed, leads resistance is essentially harmless. Assuming they are DC-fed (not multiplexed) the capacitance of the wires would not matter at all. In industrial applications, LED's are used on so-called "4-20mA loops" on distances up to several miles.

 

[Mbira]

Actually, I am using multiplexing...
I've got 32 LEDs controlled with a max7221....

 

[Gus]

http://www.maxim-ic.com/quick_view2.cfm/qv_pk/1339

You left out a lot in the first post

The spec has slew rate limited drivers to reduce EMI:However when you hang 20 feet of cable with the added wire length capacitance who knows what will happen.  What I would do is try it and look at the IC drive pins with a scope to see how they change with short wire and 20 feet of wire.  Also you are making a 20 feet long antenna with a 500 to 1300 Hz scan rate frequency.

DC drive should be no problem as abbey road d enfer posted
Sanned drive you might have to try it and see if the amp(s) picks up and of the scan frequency or if the drive circuit of the chip can handle the 20 Feet of cable.  Shielded cable non shielded cable?

 

[Martin B. Kantola]

Why not have the MAX7221 at the other end, with the LEDs?

Martin

 

[JohnRoberts]

Th problem is not likely the driver, but the interference you will be throwing into your room from the 20' antenna...

JR

 

[sahib]

First hand experience.

Long time ago I built a counter using LEDs as a special effect for a film. The unit was battery powered and about the same distance away from the batteries. It went haywire like hell. A 1uf fast cap shunted on the power lead on the device end solved the problem.

 

[jdbakker]

The spec has slew rate limited drivers to reduce EMI:However when you hang 20 feet of cable with the added wire length capacitance who knows what will happen.

+1.

On top of that, 20' of wire has a non-negligible inductance (depending on return wire distance etc...), and slew rate limiting or not I would consider a diode to catch the resulting turn-off voltage spike, much like when you're switching a relay with a transistor.

What everyone else said: put the driver chip near the LEDs. Only if it were me I'd use a $1 AVR or PIC rather than a $10 Maxim chip (and there are several microcontroller eval boars that are not much more expensive than that single-sourced LED driver). Hook these together with RS-232 (either full swing or TTL), RS-422, current loop, IR, cheap ISM transceivers or even bleedin' Ethernet, whatever floats your boat.

JDB.
[Maxim chips: gotta love their specs, but only design-in parts which are in stock at at least one major distributor]

 

[Mbira]

I can't put the chip on the end, as there are four db9s running from the board to different parts of the stage-each db9 will go to a band of eight LEDs.

So....
How do I make it work then?  ;D

I already have the max chips (samples).

So in short:

I'm using the maxim chip
The chip is in the rack
The rack has 4 db15 cables going out to the stage.
Each cable will range from 10'-20' in length
I was planning on using shielded cable
If this makes a difference: The db15 cable will also be carrying response signals from piezo elements as they get whacked to activate the LEDs.  Not sure if that needs to be taken care of in therms of interference or whatever.

Keep in mind, this isn't a hospital situation...all that is happening is
1) I whack the piezo trigger
2) It gets fed to the megadrum controller and converts that to midi
3) this goes into ableton
4) ableton sends a feedback signal via midi of what clip is playing
5) that signal goes through a midi->serial converter
6) That goes into the arduino
7) feeds the max7221
8) goes out to light the respective LED to show it is being triggered

whew!
1-7 are done...now on 8.

 

[Mbira]

Last night I had a thought about using fiberoptic cable and keeping the LEDs in the rack. ;D

 

[JohnRoberts]

There are some pretty simple static LED drivers that don't need to be multiplexed. I am using a 16 line driver in my new design. Look at something like TIs TLC5925. These can be addressed by 3 wire SPI and daisy chained in series for as many bits as you want.  4 of these in series would give you 64 lines.


JR

PS: I'm actually muxing mine for  3x12 using some high side switches to drive the 3 banks of 12, so do what I say not what I do.  :-[

 

[jdbakker]

Not all's clear yet -- how fast are the LED blink rates? Is on/off sufficient or do they need to be dimmed?

I can't put the chip on the end, as there are four db9s running from the board to different parts of the stage-each db9 will go to a band of eight LEDs.

So put a chip with each of the LED clusters.

If this makes a difference: The db15 cable will also be carrying response signals from piezo elements as they get whacked to activate the LEDs.  Not sure if that needs to be taken care of in therms of interference or whatever.

Possibly/probably. Depends on whether the piezo signals are pre-amplified before you send them down the cable, and whether the run from the piezo elements is balanced and/or shares a shield with the LEDs.

There are some pretty simple static LED drivers that don't need to be multiplexed. I am using a 16 line driver in my new design. Look at something like TIs TLC5925. These can be addressed by 3 wire SPI and daisy chained in series for as many bits as you want.  4 of these in series would give you 64 lines.

+1. Other interesting chips include the TPIC6595/6A595/6B595/6C595 and TPIC6596/6A596/6B596/6C596 for serial drive, and the TPIC6273 for parallel loading. All are available in DIP from the usual suspects and are dead easy to interface to the Arduino.

I know how tempting it is to press on with an almost-working solution, but I'd strongly recommend ditching the Maxim chip. A stage can be more EMI-sensitive than a hospital room, especially if you run these lines at floor level vs in the rigging with all the other lighting stuff. Besides, direct drive will be brighter than multiplexed, which is probably what you'd want.

Good luck,

JD 'DMX-512? What's that then?' B.

 

[Mbira]

Not all's clear yet -- how fast are the LED blink rates? Is on/off sufficient or do they need to be dimmed?

On/off is all that's happening.  I'm reading it refreshes around 800 times a second.

So put a chip with each of the LED clusters.

possible.  I'd then need to send the +5v, Load (CS), Digital in, and Clk to each of the chips from the arduino down those 20' lines.

they get whacked to activate the LEDs.  Not sure if that needs to be taken care of in therms of interference or whatever.[/quote]

Possibly/probably. Depends on whether the piezo signals are pre-amplified before you send them down the cable, and whether the run from the piezo elements is balanced and/or shares a shield with the LEDs.

unbalanced, and un preamped.  These are just feeding a midi controller.

I hear what you guys are saying about the better way.  I literally have everything sitting here and built and ready to go...I'm just fixing a last bit of code to properly read the info sent to the max.  If it doesn't work out at those length of lines, then I can just as easily start with a new driver chip and make new code.

;D

 

[JohnRoberts]

You do not want to send the serial data over those 20' lines. Noise on a slow edge rate clock could cause unreliable data transfer. And a fast edge clock is another source of EMI while admittedly lower current than the m'pexd LEDs.

Whatever you send out needs to be robust.

If you are just making a one-off and have the parts, just do it and see what happens..  It's only time and money.

JR

 

[jdbakker]

You do not want to send the serial data over those 20' lines. Noise on a slow edge rate clock could cause unreliable data transfer. And a fast edge clock is another source of EMI while admittedly lower current than the m'pexd LEDs.

Whatever you send out needs to be robust.

Differential transceivers such as those used for RS-422 should do the trick. Assuming that the cable isn't too crappy (twisted pair helps) it's very robust against interference, and generates little EMI by itself. They're used a lot in industrial environments, and standard chip solutions can be had for under a buck, with many modern chips offering adjustable slew rate. You do need two wires per bit/line, but on a DB-9 it looks like you have enough for CS, CLK, DATA and power/gnd. Still eight dedicated lines for the LEDs plus one for common is simpler, sure.

Not all's clear yet -- how fast are the LED blink rates? Is on/off sufficient or do they need to be dimmed?

On/off is all that's happening.  I'm reading it refreshes around 800 times a second.

That bit I got from the chip's data sheet, yes. But how fast do you need the LEDs to blink, and do you require the dimming function that's in the MAX chip?

Possibly/probably. Depends on whether the piezo signals are pre-amplified before you send them down the cable, and whether the run from the piezo elements is balanced and/or shares a shield with the LEDs.

unbalanced, and un preamped.  These are just feeding a midi controller.

...so you have a naked (un-amped) piezo element directly driving a 20' cable which runs in parallel with the LED cable?

If it doesn't work out at those length of lines, then I can just as easily start with a new driver chip and make new code.

If you are just making a one-off and have the parts, just do it and see what happens..  It's only time and money.

Sure, if it's not much work you can always hook up a cable and see what happens. Do try to emulate your 'real' environment as much as is feasible, and if you find you get EMI it's probably best to go straight to direct drive rather than trying to get the MAX quiet.

JDB.
[how many free digital I/Os does the Arduino have once you've hooked up MIDI?]

 

[JohnRoberts]

You do not want to send the serial data over those 20' lines. Noise on a slow edge rate clock could cause unreliable data transfer. And a fast edge clock is another source of EMI while admittedly lower current than the m'pexd LEDs.

Whatever you send out needs to be robust.

Differential transceivers such as those used for RS-422 should do the trick. Assuming that the cable isn't too crappy (twisted pair helps) it's very robust against interference, and generates little EMI by itself. They're used a lot in industrial environments, and standard chip solutions can be had for under a buck, with many modern chips offering adjustable slew rate. You do need two wires per bit/line, but on a DB-9 it looks like you have enough for CS, CLK, DATA and power/gnd. Still eight dedicated lines for the LEDs plus one for common is simpler, sure.


JDB.
[]

I was talking about the SPI which I don't trust over 20'. I suspect he needs to KISS.... The LED driver chips I am familiar with don't have differential receivers.

JR

 

[Mbira]

Ugh, after a few days reluctantly learning more about code than I want to, I just learned that the max7221 "isn't really set up to blink LEDs". The struggles I've had with that and your guy's warnings are making me look at these direct drive chips. So, my question is: how the hell do you use these with an arduino? :-D Am I still controlling these with the digital pins or what?

 

[Mbira]

To add clarity: I'm feeding the arduino midi signals that have been converted to serial. So the arduino is receiving three bytes of information: midi note number, on/off command, and velocity.

Depending on what I get, the led needs to be on, off, or blinking.

 

[JohnRoberts]

Sorry I don't know what an arduino is.

The inexpensive chips I talked about take in a simple 16b serial word and set or clear individual output lines based on the individual bits.

If you are driving alphanumeric displays you need to map the bits to the segments in code somewhere.

JR