MUFF WIGGLER Forum Index
 FAQ & Terms of UseFAQ & Terms Of Use   Wiggler RadioMW Radio   Muff Wiggler TwitterTwitter   Support the site @ PatreonPatreon 
 SearchSearch   RegisterSign up   Log inLog in 
WIGGLING 'LITE' IN GUEST MODE

Power wiring for multiple bus boards
MUFF WIGGLER Forum Index -> Music Tech DIY  
Author Power wiring for multiple bus boards
jflower
Simple question that I've not been able to find a definitive answer to in the archives:

On cases with multiple bus boards (I will have 6), how are you distributing power from your power supply to the bus boards? I had envisioned using two terminal blocks, one for +12 and -12, and then ground each board to a common point.

A picture from one that you guys have completed would be grand.

thanks!
jf
Rigo
Yep ... star configuration. Every busboard +/-/ground to terminal blocks, and terminal blocks to +/-/ground of PSU. So not just a common point for the ground, it should be the ground of the PSU (being the reference point for the +/-). In my case I use a PowerOne PSU 3.4A
daverj
mikecameron
ok here goes... my plan is this: Large External PSU with +12, -12, +5 and ground. Terminal block in PSU with 3 outputs.

3 cases, each with it's own cable to the supply, feeding internal Terminal Blocks which feed distro boards (not chained)

this might help:



Do you forsee any issues with this scheme?
decaying.sine
We might consider moving this to DIY and making it a sticky. Dave did you make that diagram? It's awesome.

How would you recommend making the ideal connection? A big terminal strip? That's what I was using but it was only for connecting two supplies.

I suppose you could just solder a big mass of wires but there must be a terminal that is more elegant. Any thoughts?
daverj
decaying: yeah, I drew that in Photoshop. The ideal one would have all the wires going to whatever connections are directly on the power supply. Such as a stack of spade lugs screwed down on terminal screws. Or a bunch of wires soldered around posts.

The Almost Ideal one is where you add a terminal strip or some other kind of multi pin connector as close as possible to the power supply terminals. For example a terminal strip mounted an inch away from the power supply with several terminals tied together and jumped over to the supply using fat wire. Then the distros all go to the terminal strip.

Mike: That'll work, but use fat wires and keep the cables to the cases short. It's not the ideal, but it's similar to the common version.

Personally I prefer to have separate power supplies in each case. Distributing AC power to the cases has less worry about losses than distributing DC from a central power supply.
50quidsoundboy
+1 for making this sticky, that's a really useful diagram...
Graham Hinton
mikecameron wrote:

Do you forsee any issues with this scheme?


There are issues with all power supply schemes because although the diagram is the same the scale and details vary.

If you run a large system in several cabinets from an external PSU where are you going to define as the correct regulation voltage point? Wherever you make it there can only be one place and everywhere else will differ by the voltage drops along the distribution.

We use 4 pin XLRs for outlets to external devices like skiffs and controllers, I wouldn't use them as a inlet for a whole cabinet. Four conductor cable that fits in an XLR will not be suitable for carrying several amps, we use the largest wire that fits the pins and pliosil sleeve it.

If you are powering one cabinet from an external PSU it is better to use remote sensing and that requires a lot more pins as you need four wires for each rail plus a ground reference. In that case 0Vs and Ground are tied together at the destination. If you do have one large PSU you have no redundancy and if there is a PSU fault the best that can happen is that you have no working system at all and the worst is that it takes all your modules out. Large PSUs usually need fan cooling for full rating which means putting them in another room with even longer cable runs, like large mixing consoles. It makes more sense to power up to 2A sections from separate PSUs with short distribution distances.

When you have amps flowing every milliohm of cable or busbar has a mV drop per amp. 18 gauge or 24/0.2 cable is about 20 milliohms per metre so if it is carrying 1A that's about 6mV per foot. 20 gauge or 16/0.2 is 33 milliohms/m and ribbon cable is over 200 milliohms/m. Doubling up cable halves the resistance.

Be careful of ratings for cable and pcb trace sizes, when it says that it can carry so many amps that means for a certain temperature rise, often 10 degrees C which you don't really want. The heat comes from the (I^2 x R) losses which are power and becomes very significant with larger currents. The same thing is happening everywhere, but with milliamp currents the voltage drops are in microvolts and so tend to be ignored.
jflower
Thanks guys,

I understand that it's best to solder everything directly to the pins on the PSU, I just wanted to see how everyone was doing it and making it clean and pretty. I can't imagine soldering 6 wires to one post on the PSU and making it not look like a volcano. Perhaps soldered on the side, but not into the center of the post with shrink wrap over it.
Graham Hinton
As an addendum to this question I thought this would illustrate the problem well:

This is one of our FullPower PSUs delivering +/-15V @ 1.5A into a pair of 10 ohm resistors on the heatsink (which gets too hot to touch). The meter is connected either end of the red 24/0.2 wire which is two feet long and is showing a 120mV drop.

This is why you should not put the total output through one cable or pcb trace. We don't normally connect this way, the PSU is split on a 15A faston commoning block to multiple distribution cards, but it demonstrates clearly why you shouldn't. Actually I would have at least 16/0.2 for every module and it would be better and cheaper than using ribbon cable.

You can thicken the cables and reduce the lengths to reduce the resistance, but there will always be some voltage drop. Remember that this is "only" 1.5A and people talk of running large systems from an external 7A PSU.
mikecameron
Obviously I'm not an EE, this probably goes without saying. But my idea was to have overkill with the size of the power supply, thinking it would compensate for any drops from cabling from the external supply.

In my original plan above... would I be better off running the external supply at +/- 15V and regulating it down to +/- 12V and 5V internally inside each case? or per-distro board?

I'm not necessarily looking for the absolute ultimate perfect scenario, but I want to avoid pitfalls if possible.
Graham Hinton
mikecameron wrote:
Obviously I'm not an EE, this probably goes without saying. But my idea was to have overkill with the size of the power supply, thinking it would compensate for any drops from cabling from the external supply.


I'm afraid it does not work like that. The same current in the same wire will always make the same voltage drop. It doesn't matter where it comes from or goes to, it's just Ohm's Law: the voltage drop = the current x the resistance.

Quote:

In my original plan above... would I be better off running the external supply at +/- 15V and regulating it down to +/- 12V and 5V internally inside each case? or per-distro board?


Yes, but that is wasteful. Essentially you are making a PSU without the mains transformer and small mains transformers are cheaper than big PSUs. You would be better off with three smaller PSUs. It is far more efficient to transfer power at higher voltage and lower current and use a transformer to change it to low voltage/high current where it is needed. That's what the electricity companies do on a larger scale.

Quote:

I'm not necessarily looking for the absolute ultimate perfect scenario, but I want to avoid pitfalls if possible.


If you have a point away from the PSU separated by a small cable resistance and which all the modules are drawing current you have a mechanism for interaction between modules without patchcords. In my demonstration above I have a passive load which always draws the same current (ignoring heating & cooling effects), but a real synthesizer system has dynamically changing loads and as the currents change so do the voltage drops and you are introducing ripples on the power rails of other modules.

If you use any PSU larger than, say, 1A externally it really needs to be connected with sense lines to compensate for the drop in the cable length, but that only gets the voltage right at the point of entry and you still get drops in the internal wiring. It's better to have multiple smaller PSUs.

You can achieve a lot by beefing the distribution up in a low cost, low tech way. Simply by replacing the red faston cables with blue fastons and larger cable gauge, or two doubled up, will reduce all the drops. So will reducing any unnecessary length.
daverj
mikecameron wrote:
Obviously I'm not an EE, this probably goes without saying. But my idea was to have overkill with the size of the power supply, thinking it would compensate for any drops from cabling from the external supply.

In my original plan above... would I be better off running the external supply at +/- 15V and regulating it down to +/- 12V and 5V internally inside each case? or per-distro board?

I'm not necessarily looking for the absolute ultimate perfect scenario, but I want to avoid pitfalls if possible.


A bigger supply can't compensate for what happens in the wiring after the supply.

There are supplies with "sense" connections that can sense and compensate for losses in wiring by running an extra set of wires to the destination that then send the sensed voltage back to the supply to adjust the supply. But this only works with one point of load. Not with 3 loads tapped from one supply.

Theoretically having a higher voltage distributed and then regulated locally is a better solution. But the reality isn't always so good. A power supply like the Power-One is a precision adjustable regulator and has a precise voltage output (if you adjust it), and very little fluctuation with load. If you used regulators similar to what it uses internally, then that would work well. But typically people would use 3-terminal regulators for distributed power and they are not very accurate (5% typical, 2% for the more expensive ones), and their fluctuation with load is comparable to a fairly long length of wire. So it loses the advantage of distributed regulation.

The one thing distributed regulation does do is isolate loads from each other. So a module that adds noise to the power rails can be isolated from other modules by plugging it into a different regulator than others.
infradead
daverj wrote:
mikecameron wrote:
Obviously I'm not an EE, this probably goes without saying. But my idea was to have overkill with the size of the power supply, thinking it would compensate for any drops from cabling from the external supply.

In my original plan above... would I be better off running the external supply at +/- 15V and regulating it down to +/- 12V and 5V internally inside each case? or per-distro board?

I'm not necessarily looking for the absolute ultimate perfect scenario, but I want to avoid pitfalls if possible.


A bigger supply can't compensate for what happens in the wiring after the supply.

There are supplies with "sense" connections that can sense and compensate for losses in wiring by running an extra set of wires to the destination that then send the sensed voltage back to the supply to adjust the supply. But this only works with one point of load. Not with 3 loads tapped from one supply.

Theoretically having a higher voltage distributed and then regulated locally is a better solution. But the reality isn't always so good. A power supply like the Power-One is a precision adjustable regulator and has a precise voltage output (if you adjust it), and very little fluctuation with load. If you used regulators similar to what it uses internally, then that would work well. But typically people would use 3-terminal regulators for distributed power and they are not very accurate (5% typical, 2% for the more expensive ones), and their fluctuation with load is comparable to a fairly long length of wire. So it loses the advantage of distributed regulation.

The one thing distributed regulation does do is isolate loads from each other. So a module that adds noise to the power rails can be isolated from other modules by plugging it into a different regulator than others.


huge plus, i had to deal with this on a smaller scale and it was a PITA.
mikecameron
sad banana

Thanks for all the help guys. Time to rethink my plan.
Graham Hinton
daverj wrote:

But typically people would use 3-terminal regulators for distributed power and they are not very accurate (5% typical, 2% for the more expensive ones), and their fluctuation with load is comparable to a fairly long length of wire.


Only if you are talking about the obsolete 780x/790x series. Many three terminal regulators available now have better performance than the 723 used in Power Ones and the performance of a PSU is dependent on more than the specification of one component within it. E.g. trimmers often are too coarse or work over too wide a range to allow accurate adjustment.

Many PSUs are run dropping out of regulation, due to underrating of compoments, e.g. the transformer is not designed to work on the lowest mains voltage that could occur in a region (-10% or 207Vac in Europe). There is a lot more to it than meets the eye and a lot of problems that people frequently complain about here are PSU and Distribution related.
synthomaniac
Graham is absolutely on the money about volts drops in the wiring. I built a big system of 12 Zeus boards for a client with 7 of those being powered types (maybe overkill but there were several Metasonix modules to consider). The ±12V was not really an issue because of many powered boards each regulating their own busses but there were there were volts drops to the passive boards to consider. So large gauge short links were the way forward. The problem I had not considered (assumptions again d'oh! ) was that the supply to the Zeus DC-DC converters is not isolated from their inputs (so input 0V is connected to ouput 0V) which gave the following problem:-

If one powered board is less loaded than another they will have different volts drops in th 0V supplied to them - this means that one 0V is at a different potential to another - maybe just 50mV (just!!). Here's the rub: when you patch betweem modules on one bus to one on another that 50mV forces current through the 0V connections between the two modules i.e. through the ribbon cables and the patch cable shield. You may think so what seriously, i just don't get it, well I tried this: I connected a DMM from one 0V bus to an others 0V bus (the 50mV difference) but I used the DMM in DC current mode and it measured a whopping 200mA. This is what is known as a circulating current or ground loop and is undesireable. very frustrating

I revised my wiring to parallel up on the 0V wiring and reduce lengths even more and eventually got down to a couple of mV difference between an unloaded board and a fully loaded board. Job done and works a treat thumbs up

Lessons: use fat wiring and short runs, don't daisy chain if can avoid it, use a star distribution, don't make assumptions, don't underestimate the importance of a good power system and know that 0V is not necessarily 0V by some form of magic (very low impedance is required).

Hope that helps
Tony Steventon
Synovatron
jflower
Thanks for all the replies here - good info.

Unlike mikecameron, I'm not using an external power supply. More on my setup:

I've got a DIY case with 6 rows of 152hp each. I am building the case expecting to add a second power supply in the future since I have known for a while that I don't want to have a huge PSU. For now I have a single 3.4A Condor PSU with sensing, which I felt was a good compromise between the issues Graham has mentioned and trying to run a 6 separate power supplies in the case.

I was planning on having my distribution blocks literally right next to the PSU, so the high current load will be on wires that are only perhaps 5-7cm long at most. However the wiring from this point to distribution board may be up to 75cm/30" for the highest rows of modules, depending on how i can run the wires. But those top rows of modules will be powered with the next power supply so i'm not crossing that bridge yet.

perhaps I should entertain the notion of using something like the TTA powered zeus boards for the top few rows of modules, which could keep the cable runs short enough to not run into serious voltage drops? I won't have enough depth in the case to mount a big PSU closer to the top rows of modules.
Graham Hinton
jflower wrote:
For now I have a single 3.4A Condor PSU with sensing, which I felt was a good compromise between the issues Graham has mentioned and trying to run a 6 separate power supplies in the case.


It isn't a compromise, there are two different issues here. Sensing only brings one run of cable into the PSU feedback loop, so only the point where the sensing wires are connected is regulated. After that you have the same distribution problems. Separate power supplies is not a bad thing, there are many advantages over one large one.

Quote:

I was planning on having my distribution blocks literally right next to the PSU, so the high current load will be on wires that are only perhaps 5-7cm long at most. However the wiring from this point to distribution board may be up to 75cm/30" for the highest rows of modules, depending on how i can run the wires.


What matters here is the rating of the distribution block and cable. You need to buy the heaviest you can find. Ideally 1" square copper bars should be used, anything lighter is going to be a compromise smile
Think big, look at how cookers, heaters, car batteries are wired.

Quote:
I won't have enough depth in the case to mount a big PSU closer to the top rows of modules.


Another thing to be aware of with large OEM PSUs is that they all use rectangular transformers. At high loads these will buzz and the stray magnetic fields will cause any steel panel nearby to vibrate which can be incredibly annoying when you are trying to work with sound.

There is no point in spending money on a custom case and then trying to shoehorn a PSU in afterwards. My standard advice is have 19" rackbars on the rear, this keeps all your options open. Also consider the heat and ventilation. If you multiply the maximum output current by 2 by the rail voltage and then add 25%, i.e. for a 3A Euro system 3 x 2 x 12 x 1.25 (or current x 30) you get the total power used in the modules + PSU, in this case 90W and nearly all of that will be heat. Compare with a filament light bulb to get an idea of how hot that is.

I've always advised people building studios to allow 30% of the equipment cost for cabling. They rarely do and get all the problems they deserve. Large synthesizer systems are no different. If you spend thousands on modules you ought to spend hundreds on the infrastructure that makes it all work together. PSUs and patch cables are not an area that you should be trying to cut corners on because you will be plagued with problems that prevent the whole thing working as it should.
consumed
moved to DIY. damn lovely resource too.
MUFF WIGGLER Forum Index -> Music Tech DIY  
Page 1 of 1
Powered by phpBB © phpBB Group