News:

Server migration complete, Welcome to version 2.1.1

+-+-

+-User

Welcome, Guest.
Please login or register.
 
 
 
Forgot your password?

+-Site Stats

Members
Total Members: 15511
Latest: courtmija
New This Month: 19
New This Week: 7
New Today: 7
Stats
Total Posts: 127196
Total Topics: 15612
Most Online Today: 85
Most Online Ever: 7634
(January 21, 2020, 02:14:03 AM)
Users Online
Members: 8
Guests: 61
Total: 69

µSC Max Voltage

Started by vttinman, December 19, 2016, 08:57:20 AM

Previous topic - Next topic

0 Members and 1 Guest are viewing this topic.

vttinman

I have been thinking about converting a PixelNet hub to 24V to drive some RobG floods I have and I was wondering how much voltage the µSC can handle.  I think I read somewhere that it was 16V?  Would a simple resistor based voltage divider work to drop the voltage down?

corey872

uSC input voltage is 5-16VDC per the manual.  You could possibly use a resistor divider, though they are generally used for high impedance measurements vs providing actual power.   Some considerations would need to be made given the varying current load of the uSC at different phases of operation - you would want to make sure your resistor divider is properly sized to keep the voltage within the 5-16V window at all times.

My .02, but I think you'd be far ahead in durability, reliability and efficiency to just add a simple volt regulator.  Anything with a "7812" in the name and voltage input of 24V or greater would work fine...

http://www.digikey.com/product-detail/en/on-semiconductor/NCP7812TG/NCP7812TGOS-ND/2401347

While the regulator does run ~43 cents, you'd likely want to use 1-2 watt resistors for the divider - and those seem to run ~20+ cents each, minimum of two needed, so overall cost is about a wash.

The other option would be to run the hub and uSCs off the standard 12V power and just inject the 24V for the floods.
Corey

vttinman

Thanks for your insight, Corey.   These 24V floods have been driving me crazy, the increased voltage combined with the amp draw interferes with the data signal much easier than the 12V stuff.  Cat5 seems to work okay for minimizing the noise but i get voltage drop if I start daisy chaining them, so I was thinking the hub would be a good way to distribute power and the RS485 would probably be more immune to noise.  I was trying to avoid running extra cable for power injection, but I may have to go that route. 

lrhorer

A resistive voltage divider is a really bad idea as a power supply.  First of all, it's not regulated, at all.  More importantly, it's impedance is going to be very high, comparatively speaking, and for a constant voltage supply one needs as low an impedance as one can afford.  One could reduce the impedance somewhat by using very low resistance values, but the impedance still would not be acceptably low, and the current draw would be hideous.  Not only would the setup be doing little else than producing a lot of heat with nearly all of the input power, but it would also cause the voltage drop across the inlet leads to be extremely high, requiring very large conductors to get anywhere at all.

A series pass element (also known as a linear supply) will most certainly drop a DC input voltage (regulated or not) to a lower working voltage with a very low impedance, and do so for a low equipment price.  It is also a great deal more efficient than a resistive voltage divider, but it is still very inefficient.  With a series pass element, dropping from 24V to 12V will result in the power supply using the same amount of power as the equipment it services, or a 50% efficiency, at best.  If you are using the linear supply to deliver 5V from 24V, then its efficiency is under 21%, with the supply itself using more than 3 times as much power as the equipment it feeds.  Again, although it is not nearly as horrendous a situation as the voltage divider, it still will result in a high line loss in the input leads, resulting in a much shorter maximum total lead length and a much smaller maximum number of elements driven.

A switching down converter (power supply) will do a far, far better job than any of the above solutions.  For currents of less than 3A @ 5V or 3A @12V, one can obtain buck converters that handle 24V or more for under $1.00 on e-bay.  Injecting 24V after the hubs / uSCs  as corey872 mention is also a very viable solution, and is possibly the least expensive, depending on the setup.

One possibility (which I myself employ on my starburst displays) is to deliver power to the displays in a hub and spoke topology.  This will work whether one is driving the displays with 24V, 12V, or 5V.  Place the hub in a central location as close to equidistant from all the display groups, and run a separate power lead pair to each group.  This not only slashes the length of the power supply runs, it also reduces the current drastically in each run.  For example, each of my starburst displays has 8 spokes with 15 pixels per spoke.  The light spacing on the spokes is 2.5cm, but the lead length between each pixel is around 6cm, for a total run length of over 9.5 meters.  The lead size is 20AWG, which has a resistivity of .0333 ohms/meter.  If I were hooking up the power serially to all 8 spokes, and assuming a 60mA per pixel current draw, then the voltage drop end-to-end would be .87V, assuming the best case of a constant current per pixel.  In reality, the pixels may pull more current when the voltage is lower, so the voltage drop may exceed 1V.  That's probably OK if one is running 12V pixels, but if the pixels are 5V, then a 1V drop may be problematical.  If, OTOH, one wires the spokes in parallel, then the lead length is only 90cm, and rather than carrying a 7.2A tapered load, each string only carries .9A across the first lead, with a total end-to-end drop of less than 2mV.  That is a huge difference.

I really don't know what you mean when you say, "...the increased voltage combined with the amp draw interferes with the data signal much easier than the 12V stuff."  For a given power output, the 24V gear will pull at most 1/2 the current of a 12V element, and quite likely less.  Again, for a given power output, the 24V system is going to tend to be considerably less susceptible to power-induced noise than a 12V system.  If they are inducing more noise on the data line, then it is either because they are drawing much more power, there is some sort of issue with their supply isolation, or else because you have some sort of problem with your wiring.  If it is one of the first two, then a moderately sized capacitor across the power input at each device may help.  So might some uAmp buffers.

vttinman

Quote from: lrhorer on December 30, 2016, 01:24:39 AM
A resistive voltage divider is a really bad idea as a power supply.  First of all, it's not regulated, at all.  More importantly, it's impedance is going to be very high, comparatively speaking, and for a constant voltage supply one needs as low an impedance as one can afford.  One could reduce the impedance somewhat by using very low resistance values, but the impedance still would not be acceptably low, and the current draw would be hideous.  Not only would the setup be doing little else than producing a lot of heat with nearly all of the input power, but it would also cause the voltage drop across the inlet leads to be extremely high, requiring very large conductors to get anywhere at all.

A series pass element (also known as a linear supply) will most certainly drop a DC input voltage (regulated or not) to a lower working voltage with a very low impedance, and do so for a low equipment price.  It is also a great deal more efficient than a resistive voltage divider, but it is still very inefficient.  With a series pass element, dropping from 24V to 12V will result in the power supply using the same amount of power as the equipment it services, or a 50% efficiency, at best.  If you are using the linear supply to deliver 5V from 24V, then its efficiency is under 21%, with the supply itself using more than 3 times as much power as the equipment it feeds.  Again, although it is not nearly as horrendous a situation as the voltage divider, it still will result in a high line loss in the input leads, resulting in a much shorter maximum total lead length and a much smaller maximum number of elements driven.

A switching down converter (power supply) will do a far, far better job than any of the above solutions.  For currents of less than 3A @ 5V or 3A @12V, one can obtain buck converters that handle 24V or more for under $1.00 on e-bay.  Injecting 24V after the hubs / uSCs  as corey872 mention is also a very viable solution, and is possibly the least expensive, depending on the setup.

I am just looking to power the uSC board itself and not any pixels or floods.  The uSC has an onboard power supply that would regulate and I believe it only draws a handful of mA.  I think a linear supply would be the best choice based on your and Corey's feedback, but I had some resistors laying around and thought I would ask about using them.

Quote from: lrhorer on December 30, 2016, 01:24:39 AMOne possibility (which I myself employ on my starburst displays) is to deliver power to the displays in a hub and spoke topology.  This will work whether one is driving the displays with 24V, 12V, or 5V.  Place the hub in a central location as close to equidistant from all the display groups, and run a separate power lead pair to each group.  This not only slashes the length of the power supply runs, it also reduces the current drastically in each run.  For example, each of my starburst displays has 8 spokes with 15 pixels per spoke.  The light spacing on the spokes is 2.5cm, but the lead length between each pixel is around 6cm, for a total run length of over 9.5 meters.  The lead size is 20AWG, which has a resistivity of .0333 ohms/meter.  If I were hooking up the power serially to all 8 spokes, and assuming a 60mA per pixel current draw, then the voltage drop end-to-end would be .87V, assuming the best case of a constant current per pixel.  In reality, the pixels may pull more current when the voltage is lower, so the voltage drop may exceed 1V.  That's probably OK if one is running 12V pixels, but if the pixels are 5V, then a 1V drop may be problematical.  If, OTOH, one wires the spokes in parallel, then the lead length is only 90cm, and rather than carrying a 7.2A tapered load, each string only carries .9A across the first lead, with a total end-to-end drop of less than 2mV.  That is a huge difference.

This is exactly why I was considering using a modified DLA hub with uSC's to power my floods.  It's hub and spoke by design and would keep me with 1 cat5 for both power and data.

Quote from: lrhorer on December 30, 2016, 01:24:39 AMI really don't know what you mean when you say, "...the increased voltage combined with the amp draw interferes with the data signal much easier than the 12V stuff."  For a given power output, the 24V gear will pull at most 1/2 the current of a 12V element, and quite likely less.  Again, for a given power output, the 24V system is going to tend to be considerably less susceptible to power-induced noise than a 12V system.  If they are inducing more noise on the data line, then it is either because they are drawing much more power, there is some sort of issue with their supply isolation, or else because you have some sort of problem with your wiring.  If it is one of the first two, then a moderately sized capacitor across the power input at each device may help.  So might some uAmp buffers.


The floods draw a good amount of power, I haven't measured it but I would suspect it's north of an amp on white for each one.  The problem came when I tried to use a 3-core connector for power and data, I would get random flickering that I interpreted as interference.  I was using a uAMP to ensure a clean data signal and switching to cat5 without a connector stopped the flickering.  I haven't tried it yet, but I'm hoping switching to RS485 with a 4 core connector will work.

lrhorer

#5
Quote from: vttinman on January 02, 2017, 10:16:17 PMI am just looking to power the uSC board itself and not any pixels or floods.  The uSC has an onboard power supply that would regulate and I believe it only draws a handful of mA.  I think a linear supply would be the best choice based on your and Corey's feedback, but I had some resistors laying around and thought I would ask about using them.

That is correct, and yes, if you are only powering the uSC, then you can probably get away with the linear supplies.

Quote from: vttinman on January 02, 2017, 10:16:17 PMThe floods draw a good amount of power, I haven't measured it but I would suspect it's north of an amp on white for each one.
24V @ 1A = 24 Watts.  What does their spec say?

Quote from: vttinman on January 02, 2017, 10:16:17 PMThe problem came when I tried to use a 3-core connector for power and data, I would get random flickering that I interpreted as interference.
It's possible, I suppose, but it would not be my first suspect, at least not what is usually thought to be "interference" by most people.  Since you are using a single ground for both power and data, my first guess would be noise on the ground lead due to the switching load on the floods, known as common mode noise.  Remember, they are PWM, and a 1A square wave current could be potentially doing some nasty things to the ground potential at the uSC.  I definitely do not recommend using a common ground lead for power and data, and I also highly recommend a differential input for the data input, not single ended.  Indeed, I'm not even sure how you are getting away with a single data lead on the uSC at all.  It's true the uAmp can clean up a lot of dirt on the signal, but a differential input isolated from power is going to do a lot more to increase the S/N than anything else.  Common mode rejection can be quite high on a differential pair.  There is almost none at all on an unshielded, unbalanced line, but as I mentioned before, a $0.10 capacitor across the uSC power input might help significantly.

Quote from: vttinman on January 02, 2017, 10:16:17 PMI was using a uAMP to ensure a clean data signal and switching to cat5 without a connector stopped the flickering.  I haven't tried it yet, but I'm hoping switching to RS485 with a 4 core connector will work.
That would be my suggestion, yes.  Do you have an  oscilloscope?  If so, take a look at the signal using a balanced pair vs. an unshielded, unbalanced input.  You might be shocked .

Powered by EzPortal
Powered by SMFPacks Menu Editor Mod