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

New Product Announcement: Low Impedance Bus Board
MUFF WIGGLER Forum Index -> Eurorack Modules Goto page Previous  1, 2, 3 ... 19, 20, 21 [all]
Author New Product Announcement: Low Impedance Bus Board
GenusModu
John/Mungo:

Mungo wrote:
You made a clearly falsifiable claim that the power supply would not affect module crosstalk,

Incorrect, I never said that.

Mungo you are interpreting things very poorly and arguing against statements I have not made. That is not productive.

Mungo wrote:
…while you promote power supply impedance so heavily it is disingenuous to ignore the power supplies contribution to this or completely misrepresent it.

If I ignore the power supply impedance contribution it is because it is something that the consumer can do nothing about, since that information is not available as you agreed to below.

Mungo wrote:
The power supply is one of the limiting factors in crosstalk by power rail impedance, along with distribution impedance from the wiring and bus boards. With your excellent bus boards and thicker wiring the power supply becomes the dominant factor in the audible range.

It might surprise you that I almost agree with this statement, but I would change “power supply becomes…” to “power supply could become the dominant factor.” But just to be clear, I would apply that to a LIBB/heavy wire gauge system, not to some of the common euro power distribution cables and 1 ounce boards with ~20 milliohms of ground resistance and ~70 milliohms of power line resistance.

Mungo wrote:
To see a difference across the audible range when changing over to lower impedance bus board such as the LIBB there would need to be already a very low impedance distribution system and power supply. The LIBB can't hide problems there. All your capacitance is only significant above the audio range 10kHz and up. This sort of message while accurate is misleading:
GenusModu wrote:
First and foremost the goal was to reduce audio frequency coupled noise through use of heavy copper ground and power planes. Once I had that, I realized that I could effectively build a distributed capacitance scheme that had ultra low resistance from any power connector on the board, and filter into the MegaHertz range with MLCC caps placed throughout the board.

You sure can filter into the Megahertz range, but its yet to be explained how that is important when you are not greatly affecting the audible range below 10kHz/5kHz/wherever you want to put the cutoff. Once the distribution impedance is reduced the power supply is the dominant factor across the audible range.

This has been discussed before several times, look up PN Junction Rectification, Video Synthesis and the examples in the white paper. The goal of high frequency filtering was developed with feedback from Lars Larsen of LZX systems to support improved video synthesis results to 6 MHz and better.

Mungo wrote:
GenusModu wrote:
Power supply impedance is not a parameter that can be controlled or determined by the user. I cannot recall every seeing an output impedance spec for a power supply. The information is not available to the consumer. So no one cannot compare Linear Power Supply A to Linear Power Supply B without $2K+ active probes, differential amplifiers or $8K+ VNAs and measuring the impedance, as you have done, or doing a fuller noise comparison.

Can you identify the "good" linear power supply (blue and green lines) and "bad" linear PSU (orange and purple lines) used for your tests? That would be useful to know. Are you saying these are power supplies from the Genus Modu recommended list?


There are very few power supplies that specify their output impedance, I have a few on hand for lab use (one with a full frequency range typical specification, more common with limits at a few specific frequencies) but they are not really appropriate or cost effective for a modular rack. If you want to stand behind your simulated impedance results then you'll need to point to the power supplies that can produce that sort of performance, and if there aren't any with the specifications measure it yourself.

It is a very slippery slope for me to create my own specifications for other commercially released products. The designs can change at any time, and there can be previous versions that I know nothing about, so the parameters I test or publish could easily be different for the same models. What I was curious about was whether you saw those variations for recommended power supplies. It seems that the answer is no and it was a theoretical argument.

Mungo wrote:
Trying to make simple binary separations between linear and switching supplies is fundamentally marketing at work. You can find linear supplies with poor output impedance control and noise, and switching supplies with excellent output impedance and noise. If you just pick random supplies without testing them then the situation is likely to be reversed, but clinging to linear supplies as inherently superior is simply false.

One of the two points you are trying to argue against is one I have never made. I have never said that linear power supplies have better impedance control than switching supplies. Impedance control depends on every individual PSU design, not linear or switching design, in that sense we agree.

I have generalized in terms of noise output because most linear PSUs have less noise than switching PSUs. I have one make of switching PSU on the list, so they are not completely ignored. I will list more if I find any. Just about any linear PSU with specs will make the list.

Mungo wrote:
What explains much of the improvements in noise floor is not power rail impedance but reduced 0V fluctuations through reduced 0V distribution impedance. Again the distribution wiring is the majority of the contribution and a well planned 0V block/rail along with some thicker 0V wires is usually the place to start, making more of a difference than the choice of bus board.

I do not agree the wires make more difference than the bus board in the most common cases which are 1 ounce bus boards.

Here are the three bus board plus wire ground resistance examples from the Eurorack Power Distribution and Noise White Paper http://www.genusmodu.com/products/libb-wp-01.html.

1 ounce bus board plus 10 inches of 3x16 AWG ground = 17.04 mohm BB + 1.02 mohm wire
4 ounce bus board plus 12 inches of 12 AWG ground = 2.02 mohm BB + 1.45 mohm wire
LIBB plus 10 inches of 10 inches of 2x12 AWG ground = 0.69 mohm BB + 0.61 mohm wire

LIBB and the 4 ounce bus board are nearly balanced between bus board and wire resistances, while the 1 ounce board is unbalanced with most of the resistance on the bus board. Replacing the 1 ounce bus board, not the wire, is the priority for noise reduction.

Now if you had the 4 ounce board with single 16 or 18 AWG ground wires, then sure the priority is reversed and replace the wires first. That is a less common scenario but it can happen.

If you have the 4 ounce bus board with 12 AWG wire, then the power supply can be the next improvement point. That is a specific case, not a general case.

Mungo wrote:
There are power supplies and bus boards with such low impedances that the wiring will always dominate the result. You've spent a lot of time and effort to end up with a product in that category, engineered right to the edge of practical performance and limited in the audible range by everything around it. Sell it on the actual value, not all the misleading marketing puff you continue to push out.

That was sounding great until the last bit…

John Loffink
Genus Modu
meatbeatz
GenusModu wrote:
I believe you are addressing semantics.

The two aspects are not mutually exclusive. LIBB can do both. I started with the low resistance board and bulk caps on the first LIBB prototype. The MLCC caps to address switching noise, both from switching PSUs and switching regulators on modules, were added in the second revision


That's right, semantics. Just like the "Quietest Eurorack bus board you can buy" forgetting a whole class of busboard as if that was no big thing and when it was pointed out your business partner threw a 20 page wobbly. This is where it all started and you both continue to make out as if this wasn't a case of puffy marketing and that you've both been transparent throughout. This is why you are being scrutinised. Paul can't understand why Make Noise isn't scrutinised to this degree. Let me spell it out for you. Because they aren't talking shit, and making conveniently vague claims while lecturing us on how great they are and how the rest of us know nothing. I will add that using passive distro on such large systems is not "how it's done", it's an arguably clunky way it can be done necessitating ridiculously large cable runs and busbars when you could've used an active solution. I'll check back in another 6 months to see if you are still persisting in making out as if you've not been sneaky with your wording and if so I'll repeat it over for you.
Mungo
GenusModu wrote:

Some clarification here, inter module noise is not improved by linear power supplies. That is a function of the power distribution, primarily wires and bus boards.

Linear PSUs are known for lower generated noise and ripple levels, normally under 5mVpp, whereas common switching PSUs including most euro PSUs will have 50-150 mVpp of noise, much of that in high and ultra audio frequencies.

Here is the blurring of lines between generalisations and absolute statements. You can't put those sentences right next to each other and pretend like you haven't made a statement on power supply impedance. Like output noise different supplies have a range of difference characteristics.

GenusModu wrote:

LIBB provides 2-4 locations for 12 AWG wire. Many passive bus boards can only support 16-18 AWG distribution wire, limiting their effectiveness.

GenusModu wrote:

1 ounce bus board plus 10 inches of 3x16 AWG ground = 17.04 mohm BB + 1.02 mohm wire
4 ounce bus board plus 12 inches of 12 AWG ground = 2.02 mohm BB + 1.45 mohm wire
LIBB plus 10 inches of 10 inches of 2x12 AWG ground = 0.69 mohm BB + 0.61 mohm wire

LIBB and the 4 ounce bus board are nearly balanced between bus board and wire resistances, while the 1 ounce board is unbalanced with most of the resistance on the bus board. Replacing the 1 ounce bus board, not the wire, is the priority for noise reduction.

Now if you had the 4 ounce board with single 16 or 18 AWG ground wires, then sure the priority is reversed and replace the wires first. That is a less common scenario but it can happen.

Three 16 AWG cables in parallel is not typical, look though all the examples posted on this site or the cases shipped from manufacturers. A single 18 AWG cable is the norm and matches well to typical bus boards. When the majority of the noise improvement is from better control of the 0V connections (not the much discussed here power supply impedances) it is comparing 6-10mOhm of 18 AWG wire and connections with a similar 10 mOhm or less of busboard 0V resistance.

12" lengths only apply to the ideally placed bus boards or small cases, once the cases exceed 3 rows and/or 120HP they are typically going to be longer, or using multiple power supplies and the inherent isolation that provides.

GenusModu wrote:

If you have the 4 ounce bus board with 12 AWG wire, then the power supply can be the next improvement point. That is a specific case, not a general case.

That benchmark would also include anything lower impedance such as your boards with their recommended wiring. Thats the point, systems are approaching and passing to the level where they are limited by the power supply impedance and changes to it will make the most difference. Low impedance systems are the specific case under discussion, not some arbitrary straw man to create an argument.

GenusModu wrote:
What I was curious about was whether you saw those variations for recommended power supplies. It seems that the answer is no and it was a theoretical argument.

Once agin the deceptive language. I have measured a range of power supplies and know the impedance characteristics of them, there is nothing theoretical about that. If you aren't presenting the output impedances of the power supplies you recommend (they don't have to become specifications as I have shown) then you are the one who remains theoretical, relying on models/assumptions/simulations.

The gains from improving the power supply impedance are almost always minor compared to the improvements of the 0V distribution. Your product and recommendations help make that happen but you focus all the marketing on your products unique aspects (lower supply impedance and capacitance) which are either minor effects or happening at frequencies outside the audible range. Its understandable and straight from a marketing playbook. There is a lot of good information and advice common to all power systems, the LIBB makes it easier and is a leading example of a bus board. But being such low impedance the wiring and power supply are critical to getting the most out of it.
Leverkusen
Mungo wrote:
Having a mixture of bus boards won't introduce new problems. If you started with all Doepfer boards and swapped out some improved boards such as LIBBs one at a time, it would only improve with each one.


Thank you @Mungo for confirming this - it’s great to know that a step by step transition is possible. Also by improving the wiring. This will be the way to go for me.

Mungo wrote:

The higher impedance bus boards introduce more crosstalk between modules, so the general thinking is to put the most sensitive modules in the lowest impedance place, and then put modules which will introduce noise somewhere else with as few common paths as possible.

It strongly depends how everything is wired together, using a bus type connection going from one board to the next, or using a star type connection all coming from a single point (creating more isolation). The result also changes if you are trying to reduce noise at a single module instead of across all modules.

On a bus type connection you get lowest noise overall when the noise creating modules are closest to the power supply (at the lower impedance position!) this sort of non intuitive result is very confusing and the opposite of what most people would think is the correct answer. The module with the lowest noise is closest to the power supply so the logical answer is to put the most noise critical module there, but that will increase the noise in all the other modules!

There was a great thread looking at the detail of where the wiring connects in the DIY forum:
https://www.muffwiggler.com/forum/viewtopic.php?t=211316&highlight=
That sets up a case with each board having its own path back to a star 0V point and star power paths for improved crosstalk reduction. With a star arrangement noise creating modules mostly affect the modules on the same bus board, so you can group all the noise sensitive modules on their own bus board to keep them quiet. With a mixture of different bus board impedances (and/or different distribution impedances from different wire lengths) you have the same choice to reduce noise overall (noise producers connected to lowest impedances), or prioritise reducing noise on some specific modules (sensitive modules at lowest impedance).

As always, the answer is complicated and changes depending on your specific priorities and choices.


I think that wraps it up for me and puts me in the position to try what works best in my case while knowing what I am doing and can expect from the changes. Thanks a lot!
Also for the link to the other thread which I missed until now. I am pretty sure that I have read every single power thread on the forum but this one makes it clear by showing the diagram evolve with every learning step.
GenusModu
I am adding Acopian Gold Box Linear Single Output power supplies to the recommended list. If you buy the A Series you will get noise and ripple specification of less than 0.25 mVrms across input voltage range and temperature. That is the best ripple spec I have seen for any euro capable power supply. Current output ranges from 4 to 21 Amps for the A Series, and down to 0.5 Amps for the B Series with 1.5 mVrms ripple spec. As these are single output versions, you will need to buy two for +/-12V, wiring one as the negative voltage.

Acopian Gold Box Linear Single Output

resynthesize
does anyone have a mouser part # for appropriate faston tabs for wiring a LIBB to an L-1 PSU?
GenusModu
resynthesize wrote:
does anyone have a mouser part # for appropriate faston tabs for wiring a LIBB to an L-1 PSU?


Tabs for soldering onto the L-1 PSU are TE Connectivity part number 1217861-1, or Digi-key part number A100452CT-ND or Mouser part number 571-1217861-1.

resynthesize
GenusModu wrote:
resynthesize wrote:
does anyone have a mouser part # for appropriate faston tabs for wiring a LIBB to an L-1 PSU?


Tabs for soldering onto the L-1 PSU are TE Connectivity part number 1217861-1, or Digi-key part number A100452CT-ND or Mouser part number 571-1217861-1.



Thank you! I lol'd at minimum order qty of 10,000 from mouser. Luckily arrow has them in stock too: https://www.arrow.com/en/products/1217861-1/te-connectivity
boubak
Enter in the land of modular synths
    --> Start planning an eurorack modular
      --> Start thinking about a case
        --> Start reading about power supplies
          --> Start reading topics on MW
            --> WTF HOLY FLAME WAR TECHNICAL ARGUING SH*T STORM screaming goo yo
              --> Scary first introduction to this community, but learned several stuffs on the way.
                --> Ok found possible contenders for power supplies, distribution system, wires and etc. smile







Strange exhausting and interesting experience grin
jroman
The information I could find from the LIBB Wiring Guide suggests using 12 AWG wire from the power supply to a barrier strip, but I can't find a specific suggestion for what to use from the barrier strip to the LIBB. My instinct says you can't go wrong with 12, but maybe you would suggest otherwise?

Also, is there any difference for using solid vs stranded wire? Stranded seems more convenient for its flexibility but I figure it is worth asking.
resynthesize
Here's my installation. I'm running two DIY L-1 PSUs, each powering two LIBBs. For each PSU, I ran two wires for +12, -12, and ground to a terminal block, and then ran +12, -12, and two grounds to each LIBB. The maximum cable length is a little over 16 inches. All wire is 12 gauge. The PSUs are both wired to a single IEC inlet, but are otherwise not connected at all. Voltages are measuring +-12.00 at the IDC headers.

Before I fill my case back up with modules, any suggestions for improvement? Would it be worthwhile to run ground cables between the two terminal blocks?








[/img]
L-1
resynthesize

Don't forget about ventilation because PSU generates some heat. Make some holes to have air movement inside the case.
GenusModu
jroman wrote:
The information I could find from the LIBB Wiring Guide suggests using 12 AWG wire from the power supply to a barrier strip, but I can't find a specific suggestion for what to use from the barrier strip to the LIBB. My instinct says you can't go wrong with 12, but maybe you would suggest otherwise?

Also, is there any difference for using solid vs stranded wire? Stranded seems more convenient for its flexibility but I figure it is worth asking.


Use 12 AWG stranded wire for the the entire installation. There is no benefit for solid wire.

John
jroman
GenusModu wrote:
Use 12 AWG stranded wire for the the entire installation. There is no benefit for solid wire.

John
Great, thanks!

I was wondering if you could help me sort out the wiring topology for the eurorack power supply system I am building. The system is meant to potentially run 12U x 84HP of LZX style video modules, so I needed a high-power low-noise solution and Acopian supplies with the LIBB seemed to be the solution. I'm not quite sure what the appropriate way to hook everything up is, though.

I've created a diagram showing the components that I have. As you can see, there are 2x Acopian VA12MT400 supplies for +12V and -12V and one B5TN100 for +5V. I added in hookups from the power inlet to each of the power supplies (distributed from a WAGO 221 wire nut). But I'm unsure how to connect the Acopian supplies to the barrier strip / LIBB. In the wiring guide, the COM ports of the power supply are wired to the GROUND connectors on the LIBB. But with dual (or, in my case, triple) supplies there is no COM connector. The Acopian usage guide warns against linking the output of different supplies together:
Quote:
Do not attempt to directly parallel the outputs of two power supplies. This would result in current flowing from the higher-set output into the lower-set output, and probable damage to both circuits. Outputs may be connected in series to obtain a higher voltage provided that a reverse-biased diode, having PIV and current ratings exceeding the combined output, is used across each output; however, keep in mind that the output current to be drawn cannot exceed the output current rating of the lowest rated supply used.

And if the total current is limited by the smallest supply anyway that would kind of defeat the purpose of having two A12 supplies in the first place.

Finally, is there any way to power a fan off of these rails without generating lots of electrical noise?

[img][/img]
GenusModu
jroman wrote:
GenusModu wrote:
Use 12 AWG stranded wire for the the entire installation. There is no benefit for solid wire.

John
Great, thanks!

I was wondering if you could help me sort out the wiring topology for the eurorack power supply system I am building. The system is meant to potentially run 12U x 84HP of LZX style video modules, so I needed a high-power low-noise solution and Acopian supplies with the LIBB seemed to be the solution. I'm not quite sure what the appropriate way to hook everything up is, though.

I've created a diagram showing the components that I have. As you can see, there are 2x Acopian VA12MT400 supplies for +12V and -12V and one B5TN100 for +5V. I added in hookups from the power inlet to each of the power supplies (distributed from a WAGO 221 wire nut). But I'm unsure how to connect the Acopian supplies to the barrier strip / LIBB. In the wiring guide, the COM ports of the power supply are wired to the GROUND connectors on the LIBB. But with dual (or, in my case, triple) supplies there is no COM connector. The Acopian usage guide warns against linking the output of different supplies together:
Quote:
Do not attempt to directly parallel the outputs of two power supplies. This would result in current flowing from the higher-set output into the lower-set output, and probable damage to both circuits. Outputs may be connected in series to obtain a higher voltage provided that a reverse-biased diode, having PIV and current ratings exceeding the combined output, is used across each output; however, keep in mind that the output current to be drawn cannot exceed the output current rating of the lowest rated supply used.

And if the total current is limited by the smallest supply anyway that would kind of defeat the purpose of having two A12 supplies in the first place.

Finally, is there any way to power a fan off of these rails without generating lots of electrical noise?

[img][/img]


I will respond in more detail in the email you sent separate from this thread.

The LIBB System Wiring Guide shows a bipolar power supply with +, COM and - connections. When using floating single voltage power supplies, the terminals are labeled + and -. Wire ground to -, and the + terminal provides a positive voltage. Wire ground to +, and the - terminal provides -12V. So the two positive voltages, +12V and +5V, wire the - to the LIBB GND terminals, while the -12V supply wires + to the GND terminals. LIBB can take 4 GND wires, so that will work. When wired this way, the +/-12V supplies are effectively wired in series as in the Acopian quotes.

If you only have two bus boards you probably don't need the barrier strips.

John Loffink
Genus Modu
gruvsyco
I finally got mine all put together. 9u x 104hp. uZeus with the 3000mA Adapter. I'm marginally satisfied with the wiring I did. It might have been cleaner with the 38770-0305 instead of the 38770-0105.

Other than that, I am really pleased. The whole modular powered up with the standard PS, I put the beefier PS on anyway. My uO-C which would not power up with just under 6u x 104hp on the same PS with flying busboards fires right up. It feels good to have an almost legit modular case now. I might upgrade to a linear supply eventually.
seank
anyone here have a recommendation for housing external linear PSU's? i ordered two LIBB's to replace the tps80w boards i have in my two submodular 6u 104hp cases, but i don't have enough room for the PSU's inside the case.

i had planned to mount two L-1 PSU's in a custom enclosure and then connect them to the cases with a heavy duty 4-pin cable/connector but i'm not sure what the best option is.
MUFF WIGGLER Forum Index -> Eurorack Modules Goto page Previous  1, 2, 3 ... 19, 20, 21 [all]
Page 21 of 21
Powered by phpBB © phpBB Group