Connecting the Negative (Ground) on two PSUs - wire size & fuses ?

[plasmadrive]

Running power supplies in parallel (that aren't designed for parallel operation) is just asking for trouble. Just keep the positive wires separate and have one less thing to cause a problem in your layout.

Actually David, running them in parallel is not exactly what this scenario is doing.. Kind of.... you are putting resistance and load in between the two.... Technically they are still in parallel, but not how most people think of wiring them next to each other in parallel with almost no resistance... (splitting hairs)

See my next post..

 

[David_AVD]

OK, I've started my opinion and I'll leave it at that.

 

[plasmadrive]

For *most* switching power supplies, even a very small voltage difference between two different Power Supplies is problematic. Because The way the regulation works, if the voltage is "over" the PSU's set point it will stop putting energy into its output circuits. This is how they regulate.. the first thing will happen is that the psu with the highest voltage output will end up taking all of the load. ( the last thing that will happen will be the magic smoke will come out of the box, and then everything will stop. )


If you really do want to run PSU's in parallel you can, but you need PSU's that are designed to do this. and generally this is only the bigger ones ( 1000+ W )..


If you are using one PSU to power multiple segments of the same strip, then there is no need to break the + between the sections, but there are some reasons why you might want to as well. ( think what happens if a feeder wire breaks, current will flow on another path, and you might just end up melting copper ).


If you are using more than one PSU that is not intended for parallel operation, you *must* keep the sections separate.


If you are unsure, draw a picture, post it here, and someone will help you out!!

OK.. I ran the testing as I said I would try and here is what I found. First off< note that I did not say to do this without cutting the positive lead, I said "it was brought to my attention..." That made me curious because I would never do this in my own display it but was an intriguing thought.

With one full 5 meter strip RGB 30 LEDs per meter and two switching power supplies each feeding one end, here is what I found.
When unloaded one power supply would be the dominant supply as expected. The other would whine and complain a bit, but no secret smoke. There was a little fighting going on between them according to meters but not as much as I expected actually. However, meters update very slowly so I was only looking at RMS values.

When I started to load them down they started sharing load as soon as the voltage drop across the strip became about the same as the differential that I set the two supplies at, and they both seemed to be doing ok. As soon as I removed the load the dominant supply took over again but this time the other supply whined and complained a lot more for a short time. Again, nothing damaged.

I never pulled out my scope to look at what was happening because at this point I figured I didn't want to burn a strip just to test out a point that was not really relevant because this is not a good practice no matter how you cut it.. It was however something that was fun to actually play with. It did work pretty well up to the max current of the strip but that was a fraction of what the supplies could actually do and either of them could have burned that strip to a crisp had the other shorted out and I not fused them both.

So to recap. When loaded and the voltage differential is exceeded by the drop of the strip, they worked fine which is what was thought would happen. When unloaded the lower set point supply didn't like it much but did not fail. Fun to play with.. but let me repeat. This is not a good practice no matter how you cut it..

For those of you that are just figuring this out, this was nothing more than an experiment because it was brought up to me that it might work... It is not something that you should do in the real world on your display.. as I am typing this I am realizing that it was a bad choice on my part to bring this up in a thread where someone is asking about something new to them.... Sorry about that.. sometimes I am doing too many things at once and just not paying attention.. :-[

 

[fasteddy]

You are correct that the cable 'acts' as a resistive load, be it a low resistive one, but I have deliberately not discussed this as an known option in the past because it runs a risk especially with the cheap low quality power supplies that are used within the hobby, so the only way I support is keeping them separate. In the end what do you gain by adding the risk of not keeping the +V separate.
Suggesting this as an option will cause much more grief than giving any real benefit

 

[plasmadrive]

You are correct that the cable 'acts' as a resistive load, be it a low resistive one, but I have deliberately not discussed this as an known option in the past because it runs a risk especially with the cheap low quality power supplies that are used within the hobby, so the only way I support is keeping them separate. In the end what do you gain by adding the risk of not keeping the +V separate.
Suggesting this as an option will cause much more grief than giving any real benefit

YUP! it is a good intellectual game.. but sadly misplace by me in this thread.. I would never do it in real life and I warned against it when it was brought up to me.. but it was interesting the more I thought about it.. Had I not been doing too many things at one time when I read (skimmed) this thread I would have had the smarts to not bring it up on this particular thread.. one Ah Sh$t for Plasma... I take will take my beating... :-\

 

[fasteddy]

You are correct that the cable 'acts' as a resistive load, be it a low resistive one, but I have deliberately not discussed this as an known option in the past because it runs a risk especially with the cheap low quality power supplies that are used within the hobby, so the only way I support is keeping them separate. In the end what do you gain by adding the risk of not keeping the +V separate.
Suggesting this as an option will cause much more grief than giving any real benefit

YUP! it is a good intellectual game.. but sadly misplace by me in this thread.. I would never do it in real life and I warned against it when it was brought up to me.. but it was interesting the more I thought about it.. Had I not been doing too many things at one time when I read (skimmed) this thread I would have had the smarts to not bring it up on this particular thread.. one Ah Sh$t for Plasma... I take will take my beating... :-\

Nothing wrong with a good discussion, that's how we all learn, so don't feel bad about bringing it up, others reading this will now understand the reasoning behind why we keep them separate, even when in practice it may seem to work for the most part.

 

[davrus]

It has been interesting following this "theoretical" discussion, and reading what happens in practice was like being back at school again !!!


That was one experiment that I couldn't do !


Thanks

 

[arw01]

Curious if will you have this set up still, if any way to definitively decide connected grounds or not? Seen EEs on both sides of that issue too.

 

[davrus]

I have one PixLite 16, and will be using 5 power supplies. Two directly connected to the PixLite, and the others used for power injection on strings that only receive their data from the PixLite - and of course, all the common (-ve) will be interconnected as all the signal grounds go back to the PixLite.

 

[fasteddy]

Curious if will you have this set up still, if any way to definitively decide connected grounds or not? Seen EEs on both sides of that issue too.

What do you gain by doing this compared to the risk.

The electrical world is full of these scenarios where there is good practice and bad practice and many of these bad practices may work OK in many scenarios, the benefit of doing these things though does not warrant the risk because most times there is no real gain.

Its like saying a nail or paperclip works fine as a fuse

 

[arw01]

Let's take a look at all the advice shell got on her matrix blaming power and wires when it was flacky firmware.

I believe she was advised to tie the grounds together because of ground loops etc if she didn't. Like I said EEs on both sides of that one.

 

[plasmadrive]

Let's take a look at all the advice shell got on her matrix blaming power and wires when it was flacky firmware.

I believe she was advised to tie the grounds together because of ground loops etc if she didn't. Like I said EEs on both sides of that one.

Actually I was the one that was involved with the ground loop issue. Shelly told me the board was a good working board and that proved not to be the case. Everyone had tried everything else so I started in on the grounding. I had loop problems with the Plasma Icicles that turned out to be ungrounded conduit mounts that acted as antennas because the wiring was tied to them in a linear fashion.

The old phrase "ground ground the world around" works when applied properly. I am of the school that believes you should indeed tie the power supply grounds (commons) together if you feed from a single location. But, with power injection what do you do? I think the ground or common path should be as low of an impedance as possible, everywhere. I don't know if that is practical feeding from two side of a 5m strip or sting.. but it sure would help stop any possible float.. It is surely NOT a safety hazard and I believe it is "best practices" to do so.

I look at the common as the reference. The more stable the reference, the more stable the signal.. experience tells me this... but I have by no means "done it all".

If however you have several props (2 or more) that are not tied together by injection supplies but rather stand alone and you want to deal with the signal only (like DMX or other) the ground loop issue is an entirely different animal and not a power issue.. IMHO that is.

Anyone wanna jump in on this one?

 

[lonewolf]

Had you asked me earlier I would of said yes, tie the grounds together at the supplies just to make sure they're all well connected, but after thinking about different failure scenarios I'm going to have to go with no, only tie them together where logic signals meet. The problem with tying them together at the supplies is what happens if a ground connection goes high resistance or open between the supplies and the lights? If tied you will now have twice (or more, depending on supply size) current flowing on the remaining ground wire(s) and things could get very hot if the wiring isn't sized for this.

 

[plasmadrive]

Had you asked me earlier I would of said yes, tie the grounds together at the supplies just to make sure they're all well connected, but after thinking about different failure scenarios I'm going to have to go with no, only tie them together where logic signals meet. The problem with tying them together at the supplies is what happens if a ground connection goes high resistance or open between the supplies and the lights? If tied you will now have twice (or more, depending on supply size) current flowing on the remaining ground wire(s) and things could get very hot if the wiring isn't sized for this.

And how do you believe allowing the current to flow thru the signal common will change that? Current will find a way.. whether or not it is destructive depends on how well you did what you should have done.

 

[bigstott]

I have run two power supplies in parallel and it did work but I did see some gliching. This was just a day or two of testing but when it is easy to power inject correctly, I see no reason to make this part of your permanent setup. Besides I did see so issues with effects especially in fading.