Ideal scenario for power distribution in console buckets

GroupDIY Audio Forum

Help Support GroupDIY Audio Forum:

This site may earn a commission from merchant affiliate links, including eBay, Amazon, and others.

boji

Well-known member
Joined
Jan 6, 2010
Messages
2,375
Location
Maryland, USA
Due to helpful advice by MANY of you over the months, I've got the coms copper bars worked out.

But...what about power?

I have extra 48" x 3/4" x 1/4" copper rods.  Initial thoughts were to run them parallel to the com bus bars, and use them to distribute 16v +/-,  48v,  12v,  tapping off them with ring terminals at points where lengths of wire to each 8ch backplane (4 per bucket plus logic cards) are short as possible, and so on, moving on down to other buckets.  Does this sound reasonable?

If not, what might you recommend for distribution? The only other way I can think of would be to use longer wire runs to everything all tied to 6'x6' plates at the far end of the console, right behind the amphenol power plug.

Any words of wisdom would be highly appreciated, thank you!
-Boji
 
boji said:
Due to helpful advice by MANY of you over the months, I've got the coms copper bars worked out.
somebody actually suggested that? In my experience finesse (engineering) trumps brute force for most such things.
But...what about power?

I have extra 48" x 3/4" x 1/4" copper rods.  Initial thoughts were to run them parallel to the com bus bars, and use them to distribute 16v +/-,  48v,  12v,  tapping off them with ring terminals at points where lengths of wire to each 8ch backplane (4 per bucket plus logic cards) are short as possible, and so on, moving on down to other buckets.  Does this sound reasonable?

If not, what might you recommend for distribution? The only other way I can think of would be to use longer wire runs to everything all tied to a 6'x6' plates at the far end of the console, right behind the amphenol power plug.

Any words of wisdom would be highly appreciated, thank you!
-Boji
Try to visualize the current flow like plumbing and keep the clean water from being corrupted by the sewage.

JR
 
We had a job at a studio I used work one time , was more an act of charity in some senses ,
The clients  ended up secreting miraculous medals in slots in the console frame , there had to  be a complete tear down and inspection  of the console afterwards .

Sounds like some kind of stand offs or mountings for these buzz bars would be required  , indvertantly turning your masterpiece into an arc welder would be a big mess to clean up .
Those bus bars are probably good for hundreds of amps at mains voltages, its probably about  10 times  what it needs to be ,but the big chunky engineering is probably easier and more fool proof than the more delicate solutions .

https://www.ebay.com/itm/LBY-4pcs-Busbar-Insulator-Polyester-Standoff-Insulators-SM-51-with-Screw-Busbar/163790889819?hash=item2622b2a35b:g:t8kAAOSw09ZdOa~w







 
Hi,

I don't know the background to the project being posted about, so am accepting of incoming "missiles" in any replies.

From a general console engineering point of view, a few cents worth of experience......

Copper, or nickel-plated copper, or aluminium busbars that run the length of the console and connect to each module / assembly are good news for ensuring a low-impedance ground path back to the console ground reference. Even miniscule currents present in and on the ground can be amplified by virtual-earth stages (they all use ground as the non-inverting reference, low impedance is important).  Much has been said and written about this by others.

Power distribution is a bit of a different subject.
How many amps is each of your modules drawing?
Two examples of power draw are, from memory, a Neve VR channel module which has something like 60-plus NE553x chips in it draws about half an amp. An SSL4000 channel module draws about 350mA, give or take a bit.

A 12-channel Neve bucket requires 6 amps and an SSL 8-channel bucket around 3 amps.

Neve VR uses 4mm square (about 13 gauge) as the cross-console power distribution, from bucket-to-bucket.
SSL4000 uses 0.75 sq mm, or 24/0.2mm, or something close to 19 gauge for power from a central distribution point (bucket power isolation switches & circuit breakers) to each bucket.

A 3/4" x 1/4" copper bar is capable of carrying hundreds of amps. Good for low-impedance ground, but radical over-specifying for power distribution. For console power distribution, the busbars will look impressive when installed, but really do not serve a scientifically calculated purpose.

Trusting this is of use / interest.


 
If you have the materials to hand and calculated for the space/physical load of that mass of copper ,
We salut you !

Sure, to deliver the given current/volts you only need a certain gauge of cable , but an ultra low z conection across all your audio modules on both +/- and ground should help minimise potential differences and extraneous noises.
 
If you run power on exposed buses, what happens when someone drops a brass dope-pipe inside?
A highly-shocking, shockingly-high possibility.  However those longing for contact high will have to find a creeper, and get at the underside cage to expose the heating element.
somebody actually suggested [heavy metal]?
For the record, JR never recommended that genre. Others did show examples of headbangers engineering. Likely a marketing urge overkill.
Sounds like some kind of stand offs or mountings for these buzz bars would be required
Fershizzy my dizzy.
big chunky engineering is probably easier and more fool proof than the more delicate solutions .
Emphasis on 'fool'.
How many amps is each of your modules drawing?
(~300ma per ch. x  8  ) x 3.5 buckets.  Less draw per 500 slot, but x 48 if stuffed to gills, all relays/logic bricks pull quite a bit;  let's say ~4 amps total.
Led's....um lets not talk about leds, b/c someone in this very thread went so far as to ms-paint how to CCS each channel and I went direct resistors anyway.  :-X  :-[

busbars will look impressive when installed, but really do not serve a scientifically calculated purpose.
In the name of science I may order appropriate sizes, but in the name of fiscal responsibility...what do they say? Smoke'em, if you got em'? Never make expensive repairs to a car if you plan on selling it?  I'm not selling this money/time pit, so... I dunno we'll see.

Neve VR 13ga / SSL4k  19ga / Trusting this is of use / interest.
Exceedingly!!!  I thought I was going to hear how oversized supply lines would make electrons play telephone with each other in a bad way.

We salute you !
No, TT. You got it 1000% backwards.  Thank you everyone.

 




 
boji said:
For the record, JR never recommended that genre. Others did show examples of headbangers engineering. Likely a marketing urge overkill.
Thank you...  to expand upon my dislike for brute force overkill the work of engineering is to craft effective solutions, cost (size and weight) is always a factor.

Even a 100A bus bar is not 0 Ohms, so will still support voltage potentials from any current flow.

By "finesse" I refer to use of differential signal handling to ignore (actually subtract out) voltage potential errors between different 0v reference nodes.

Likewise PS are not an exercise in brute force over design... enough (maybe plus a little) is just right.

JR

 
One solution is to distribute ±24V and locally regulate this down to whatever you need inside of each bucket. By using RC isolators on this ±24V rail, you can remove the need to use heavy conductors or worry about impedance - low impedance is created by the regulators after each bucket's RC isolators.

A 50-100Ω series resistance and 500-1000 µf capacitance will be good for dozens of mA of amplifier standing current. Scale the series R for a couple volt drop at 2x the standing current, and then size the shunt capacitor to have an RC constant around a couple Hz or lower. In this way, no audio band transient current is sourced from the power supply, no two buckets talk to each other via the power rails, transient current is sourced from the bucket's local ground, and each bucket's supply is optimized for exactly the circuitry that is contained within.

The days of one huge  supply providing low impedance to an entire desk should disappear - it was never a good idea in the first place, and successful consoles like the old Neoteks started to move away from that idea long ago. I had an old Series II that used 100Ω series and 100µF shunt to each quad op amp, with no regulators, but it provided isolation from each chip to the power rails, at least from the lower midrange frequencies and up.

I'm using a variant of this idea in some circuitry I'm designing and it works well. Each circuit gets ±24V through a pair of 100Ω series Rs into a pair of 470µF shunt caps. This isolated power then goes to a pair of regulators to make ±17V or whatever is needed. Using 4 layer PCBs, these regulators provide low noise, low impedance power to the amplifiers with only the minimum required regulator bypass caps, and the ±24V rails can be powered by pretty much any supply with exactly the same performance - all of the AC and ground properties are defined inside of the circuit, not by the ±24V feed or its impedance. It's not clear to me that this more expensive than one magic low Z supply, trick cabling and a magic bypass cap arrangement, and it works repeatably.
 
I have explored several variants of this... In the last "big" console I did (>100 stems to L/R bus) I put cheap 3 terminal regulators on every strip... I did discover an issue with noise from crappy 3 terminal regs getting into the 2 mix.  :-[

===

Another old factoid...  I figured out back in the 80s that you could mitigate the rising source impedance of cheap 3 terminal regulators (based on low bandwidth IC processes of the time) by placing a 1,000uf electrolytic cap across the output. The falling impedance of the electrolytic cap nicely compensated for the rising impedance of the 3-term regulator.

Caveat circa 1980's electrolytic caps, and maybe even 3 terminal regulators are different animals from today so YMMV.

JR
 
JohnRoberts said:
I have explored several variants of this... In the last "big" console I did (>100 stems to L/R bus) I put cheap 3 terminal regulators on every strip... I did discover an issue with noise from crappy 3 terminal regs getting into the 2 mix.  :-[
I'm taking the pricey approach here and using the Analog ADP7142 and ADP7182 regulators, and it is working well. The circuits I'm using are nowhere near that large and have modern high PSRR op amps, but still, previous attempts using modern LM317 and LM337 worked just about as well. The only reason I'm using these fancy regulators is that the 317/337 regulators can get cranky with output bypass caps, and I'm trying to avoid using tantalum or large electrolytics on their output.

Another old factoid...  I figured out back in the 80s that you could mitigate the rising source impedance of cheap 3 terminal regulators (based on low bandwidth IC processes of the time) by placing a 1,000uf electrolytic cap across the output. The falling impedance of the electrolytic cap nicely compensated for the rising impedance of the 3-term regulator.

Caveat circa 1980's electrolytic caps, and maybe even 3 terminal regulators are different animals from today so YMMV.
The modern Analog regulators each require 1.5µF of output capacitance, which I get using three 1µF X7R 1206 SMD caps to accommodate the 'squish' of X7R with applied voltage. Using a 100Ω probe driving the regulator output, it behaves like a 25mΩ resistor up to about 3kHz with a single pole rise. I guess these caps are too small to flatten out the regulator, but the impedance seems low enough for the op amps, and relatively little signal induced ripple appears on the rails. For circuits that have little or no PSRR, the brute force 1000µF output cap might be a better idea, but again, modern op amps are not so finicky. Plus, the three 1206 packages are a lot smaller, thinner, and long lived.
 
A 50-100Ω series resistance and 500-1000 µf capacitance will be good for dozens of mA of amplifier standing current. Scale the series R for a couple volt drop at 2x the standing current, and then size the shunt capacitor to have an RC constant around a couple Hz or lower. In this way, no audio band transient current is sourced from the power supply, no two buckets talk to each other via the power rails, transient current is sourced from the bucket's local ground, and each bucket's supply is optimized for exactly the circuitry that is contained within.
!!!
Wow thank you for that thoughtful solution! Could be isolatedly reworked for a sub-bucket of x51  24v 500 series too.  I'll have to test this out and give it real consideration.  I do have the old school PSUs at the ready but let me look for 24v alternatives. Thanks MM!!!
 
Monte McGuire said:
The only reason I'm using these fancy regulators is that the 317/337 regulators can get cranky with output bypass caps, and I'm trying to avoid using tantalum or large electrolytics on their output.

The modern Analog regulators each require 1.5µF of output capacitance, which I get using three 1µF X7R 1206 SMD caps to accommodate the 'squish' of X7R with applied voltage. Using a 100Ω probe driving the regulator output, it behaves like a 25mΩ resistor up to about 3kHz with a single pole rise. I guess these caps are too small to flatten out the regulator, but the impedance seems low enough for the op amps, and relatively little signal induced ripple appears on the rails. For circuits that have little or no PSRR, the brute force 1000µF output cap might be a better idea, but again, modern op amps are not so finicky. Plus, the three 1206 packages are a lot smaller, thinner, and long lived.
I targeted low flat PS impedance in the mOhms region out past 20 kHz, but this was arguably overkill and most modern circuitry also provides good PSRR (power supply rejection ratio).

When over analyzing this, like I was wont to do back in my days on the design bench, you will notice that PSRR also tends to degrade at HF, where the regulators are also rising impedance.... This is why HF capacitors are your friend also. Look at the whole  picture working together, not just over-engineering one single aspect of the design (like using 100A bus bar).

JR
 
The long busbar has significant inductance. Making it fatter does not reduce the inductance much.

Optimizing Op Amp Performance by Graeme, Jerald G. has a chapter on this. While most of the book avoids practical values, this chapter uses some older assumptions and figures that 1 Ohm at the opamp supply terminals should be good. But ceramic caps with very low ESR are liable to complex resonances which can *degrade* supply rejection, typically in the MHz where "hot" audio amps still have gain. He shows how electrolytics' part-Ohm ESR can damp this out.
 
Optimizing Op Amp Performance by Graeme, Jerald G

Thanks PRR, I've ordered the book this evening.  Regarding bi-polar coming off the bus terminals, is it a waste of time/money to use PTFE twisted pair, shield to chassis?  The bars make the runs short, so it would not be too expensive to go milspec.  ::)

Edit (removed impish wire picts)

Would copper coated aluminum camper/trailer wire do just as well?
 
I did a bench stint at a company that two of their brand lines were consoles.

On the inexpensive side we had a IDC connector with a ribbon. I forget how many pins each IDC was but it ran the length of the desk and each module had an IDC connection coming off the ribbon.. it was a Pain to make those cables so lucky us, we didn't have to unless it was for repair. In turn that IDC held all the bussing, auxes, and power to each module and to each submitter and master. The the rear of the desk was a PCB that had IDC on one end ad wiring to a large military circular connector used for power from the power supply.

On the higher end line, we had a backplane with connectors one per bucket. The backplane had  connectors for the channel strips and IDC  to connect the different panels such as the I/O panel  to the mother board. We then did IDC to IDC to join buckets together. But for power we did it differently.  We had  a similar circular connector getting power from the PSU, to the desk, the circular connector to a PCB, but on the PCB we had molex power connectors and each bucket had it's own wiring from the power connector to the bucket. It worked really well.
 
I see that we're talking about 500 series buckets here and so I assume that a good number of the modules in the buckets will be using some form of clone of the older "classic" discrete op-amps.
If this is so, then something to keep in mind is that these older op-amps didn't always have the best PSSR.  You often see non active current sources for the diff-amp and resistive loading rather than a current mirror for instance.
Of course, a lot of these older circuits incorporated some sort of brute force approach with power lines coming into the module and feeding a series R and a large cap before going to the op-amp power pins.  This may or may not be adequately implemented with a DIY'ed module that uses 2520's or some such gain back. 
Just something to keep in mind.


I like the approaches presented by John, Monty, Gareth et al. above regarding using an elegant rather than overkill system but, when dealing with older type technology, I could easily be tempted to put decent 3 terminal regs on each channel so's not to worry about them talking to each other or picking up crap from the power lines

 
but on the PCB we had molex power connectors and each bucket had it's own wiring from the power connector to the bucket. It worked really well.
Cool, glad it worked well. That's the plan.

modules in the buckets will be using some form of clone of the older "classic" discrete op-amps.
Yessr, mostly. Some hybrids.
...series R and a large cap before going to the op-amp power pins.
(2)100uf, as close to opa's as possible for each channel/group/aux board.
an elegant rather than overkill system
Still trying to learn how to implement RC iso.  If I understand correctly, the regulator is the iso, and the RC stabilizes? Not much online specific to audio consoles. I'll see if there's something in Self's Small signal design.

So the ADP7142 and ADP7182 are good up to 200ma so that would mean a 2 IC's per channel (A ton of them).  What I don't get is if the PSU is rated 800 watts and 24 channels are drawing roughly 200w, does the crosstalk come from the supply (slew?) or the fact that the bars have a small amount of inductance?

Edit: Ok so the LM317 and 337's can supply roughly six times as much current as the AD's, but they would be working fairly hard per 8 channels. Also I'd need to replace the two PSU's to get the higher voltages required.  If I go with what I got, (Wheatstone TV800, Amek Big SMPS) would large lytics near each point of distribution be at all helpful?

Thanks again for all the advice so far. Wish I had the chops to simply run with the thoughtful suggestions already given.
 
I like the idea John mentioned above where he used a pair of 3 terminal regs per channel on a desk.  I don't see it as a particularly extravagant or OTT approach myself, especially if dealing with gain blocks not having top notch PSRR. 
I've not had the pleasure of designing a complete desk but I've done an eight channel box that I did the same for and it worked well.  Another way to go might be using cap multipliers, you could get away with not much more than a Vbe drop across them so that's sometimes advantageous if you don't have excess surplus V available.

It's pretty easy to do much better than what was done in the old desks so, it'd really be a matter of looking at the cost involved, space available to implement the scheme, effort needed etc.  and picking the solution that works for you.







 
boji said:
Still trying to learn how to implement RC iso.  If I understand correctly, the regulator is the iso, and the RC stabilizes? Not much online specific to audio consoles. I'll see if there's something in Self's Small signal design.
The RC isolator isolates the transient currents away from the power supply, and re-routes those transient currents through the shunt capacitor so that they are sourced directly from the local ground. The price of all of this is that the voltage at the top of the shunt capacitor can bounce along with the transient current drawn by the amplifier from the isolator cap - that cap is now driven by 100Ω and not a low impedance supply, so if you draw current from the shunt cap, it cannot re-charge instantly, since it is powered through the series isolator resistor. However, that's what you want - that's how the transient current is kept away from the power supply.

To counteract this load current induced 'voltage bounce', you use series regulators between the isolators and the amplifier circuits you're powering. This provides a nice low impedance supply to the amplifier circuits, isolated from the actual power supply by the isolators.

One thing to keep in mind is that the regulators driving the circuits need a voltage margin (dropout voltage) to work properly, so the desired output voltage plus the regulator's dropout voltage should be at least a volt or two less than the main supply voltage minus 2x the idle current drop through the isolator resistor. As an example: say you have ±24V and you want ±17V at your circuitry: you have a combined 24-17=7V of drop that you can lose across your series resistor and in your post isolator regulator. Modern regulators require little voltage drop, maybe only a half volt, but they can work better with several volts. The isolators also should not drop too much voltage when the circuit is idling, since the circuit will probably draw more supply current when it supplies signal to a load. So, you should do some thumbnail calculations and then once you get things running, verify that your assumptions made sense with actual operating conditions using actual signals.
So the ADP7142 and ADP7182 are good up to 200ma so that would mean a 2 IC's per channel (A ton of them). 
Yes, the ADP7142 and 7182 are really tiny. In my world, they work well for one amplifier circuit plus a servo and maybe a buffer, or maybe a couple of these if you're lucky. But, they're small, and their output capacitor requirement is very modest, making them work with my world well. In my quest to make current loops among multi-module circuits exceedingly small, using one regulator per "tiny circuit chunk" makes sense.

However, LM317/337 are a lot cheaper and produce much more output, so those can be preferable if you use the regulator to power more than just a tiny circuit. TI makes some nice modern regulators as well - the TPS7A33x and TPS7A47x are similarly modern and high performance as the AD regulators, but can source 1A and have larger packages to dissipate the heat. There are relatively few + and - regulator pairs, but these should suffice for most uses.
What I don't get is if the PSU is rated 800 watts and 24 channels are drawing roughly 200w, does the crosstalk come from the supply (slew?) or the fact that the bars have a small amount of inductance?
The crosstalk isn't gonna be huge, but it will come from the fact that the power supply will have a finite output impedance, the power rail leads will have some inductance and resistance, and the local circuit ground is so far away from the regulator's ground that the two won't be tightly coupled. None of this is huge, but with many lower end, virtual earth mix bus consoles from the 70s and early 80s, a lot of 'gunk' seemed to build up using 16+ channels at a time, making a multitrack mix get pretty murky. It seems clear to me that if you could make the current path of each amplifier in a console much smaller and more local, this 'gunk buildup' would be reduced. The isolators do this by removing transient current from the power supplies and sending it to the local ground. The regulators are there just to make up for the voltage bounce that the isolators cause, and to provide quality, low impedance rails to each chip, which can be only a few millimeters away if you lay it out like that.
Edit: Ok so the LM317 and 337's can supply roughly six times as much current as the AD's, but they would be working fairly hard per 8 channels. Also I'd need to replace the two PSU's to get the higher voltages required.  If I go with what I got, (Wheatstone TV800, Amek Big SMPS) would large lytics near each point of distribution be at all helpful?
No, large post-regulator caps aren't needed in this system since the RC isolators will eliminate the transient load to the supply anyway. Use them if the supply requires them for stability or for low noise, but again, that's the point of the isolators - you don't need the supply to do much more than source the bias current of your amplifiers, along with a heavily lowpassed "extra draw" when power is sourced to a load. The shunt C moves all of the transient current to ground, so it does not have to be sourced from the supply.

Many supplies can be re-worked to provide a higher regulated output voltage by adjusting a pot or a couple of gain resistors, but the drop across the series R might be too much to 'squeeze out' of an existing design. On the other hand, 24V is a somewhat standard supply voltage, so you might be able to ditch the huge old supply and get something smaller, maybe even a switcher, since its output will be heavily filtered, and it's not doing much more than providing 2x the idling current.


 
Back
Top