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

Passive balanced out for eurorack case
MUFF WIGGLER Forum Index -> Music Tech DIY  
Author Passive balanced out for eurorack case
uniquepersonno2
Hi there!

I'm very very new to electronics/DIY so I apologize for the long, probably basic question.
Basically, I want to install a passive stereo balanced output into my eurorack case. Optimally I'd have a small 2 or 3 hp module with two inputs, which would carry the output of the synth to two TRS jacks on the back of my case.

I've been looking at how to best go about this and found this article about building a passive DI, but I have no idea if that would work for what I want to do. The kit they make also comes in a bulky box that doesn't really suit my case as I want it mounted internally with the jacks/switches mounted on the euro panel and on the back of my case.

Could I simply adapt the circuit they demonstrate to fit in my case? My worry here is length of cables, optimally I'd have the input panel in front and the outputs in the back, which is about 24".
What do I need to worry about in terms of power supply interference?
Can I simply mount the circuit in my case or do I need to put it in something to shield it?
How do I ground it? I'm using a Trogotronic power supply which I don't think grounds to the case at all.

Thanks in advance for any advice!
slow_riot
This is a good question, it's something that hasn't really been properly addressed by the eurorack market yet but which a DIY solution can achieve very good performance.

With a balanced transmission line, the receiver is doing the work to subtract the difference between the hot and cold sends. What you need to do is to ensure that both these lines see the same amount of noise. This is done by cable symmetry and matched impedances. There are different techniques on the construction of a balanced output from an electronic point of view, for simplicity a so called "impedance balanced" technique will be shown, I will leave the why.

You could fit this device into a single unbalanced to balanced adapter cable with some careful soldering. Say minijack to 1/4" TRS or XLRm using a 3 wire cable with shield, hot and cold. The best 3 wire cable uses foil screening, perhaps combined with braiding, or a drain wire. The foil is the important part, giving best coverage from external high frequency interference.

The output of your module will be protected by a resistor, likely 1K ohms, you can either read this value, consult the manual or the manufacturer, or measure it yourself with a multimeter, or via a formula (see below). This will be your 'hot' output, connected directly to tip at the minijack, and then directly to pin 2 of your balanced connector if using an XLR (or tip of a TRS).

When you know the output impedance of the driver, find a resistor with the same value. This is to be connected somewhere between Shield of your minjack and pin 2 or Ring on your balanced connector. It is essential for noise rejection that you mount this resistor as close as possible to the unbalanced end. If you can't wire as shown, I would make up a very short TS to XLRm adapter, 5 or 6 inches only, with the resistor at the balanced end, then follow with a regular XLRf to XLRm cable.

The connection for shield (pin1, or sleeve) is contentious for an unshielded signal source. Some practitioners advocate connecting one end only (such as Neil Muncy and Rane who I revised their image), some advocate connecting this directly to 0V. Both may introduce their own problems and I have made this connection via a capacitor, so that the shield will allow high frequency AC signals to flow, but minimises potential problems from common impedance coupling. I suggest a value of around 1nF. You could also try the other 2 techniques.



Formula to measure output impedance Zo. Measure the open circuit output, Vo. Measure the output with a known load Z, Vl. The drop is due to the output impedance in series with the load, so

Zo = (Vo-Vl)/Vl x Z
uniquepersonno2
Thank you for the extremely detailed response and explanation!
Since it is based off of the impedance of whatever module is outputting to it, would that mean it would have to be designed to fit one module only? I was hoping for a solution that I could plug any source in my rack into, which is why I was looking at transformer based solutions.
If the above is not an issue, I'm really looking for something to install into my rack permanently, instead of an adapter cable. Could this be converted to use input jacks as opposed to output jacks as you've pictured?

slow_riot wrote:
This is a good question, it's something that hasn't really been properly addressed by the eurorack market yet but which a DIY solution can achieve very good performance.

With a balanced transmission line, the receiver is doing the work to subtract the difference between the hot and cold sends. What you need to do is to ensure that both these lines see the same amount of noise. This is done by cable symmetry and matched impedances. There are different techniques on the construction of a balanced output from an electronic point of view, for simplicity a so called "impedance balanced" technique will be shown, I will leave the why.

You could fit this device into a single unbalanced to balanced adapter cable with some careful soldering. Say minijack to 1/4" TRS or XLRm using a 3 wire cable with shield, hot and cold. The best 3 wire cable uses foil screening, perhaps combined with braiding, or a drain wire. The foil is the important part, giving best coverage from external high frequency interference.

The output of your module will be protected by a resistor, likely 1K ohms, you can either read this value, consult the manual or the manufacturer, or measure it yourself with a multimeter, or via a formula (see below). This will be your 'hot' output, connected directly to tip at the minijack, and then directly to pin 2 of your balanced connector if using an XLR (or tip of a TRS).

When you know the output impedance of the driver, find a resistor with the same value. This is to be connected somewhere between Shield of your minjack and pin 2 or Ring on your balanced connector. It is essential for noise rejection that you mount this resistor as close as possible to the unbalanced end. If you can't wire as shown, I would make up a very short TS to XLRm adapter, 5 or 6 inches only, with the resistor at the balanced end, then follow with a regular XLRf to XLRm cable.

The connection for shield (pin1, or sleeve) is contentious for an unshielded signal source. Some practitioners advocate connecting one end only (such as Neil Muncy and Rane who I revised their image), some advocate connecting this directly to 0V. Both may introduce their own problems and I have made this connection via a capacitor, so that the shield will allow high frequency AC signals to flow, but minimises potential problems from common impedance coupling. I suggest a value of around 1nF. You could also try the other 2 techniques.

Formula to measure output impedance Zo. Measure the open circuit output, Vo. Measure the output with a known load Z, Vl. The drop is due to the output impedance in series with the load, so

Zo = (Vo-Vl)/Vl x Z
auxren
I made a module that is based of AS007 here: http://www.jensen-transformers.com/schematics/

Sounds incredible and highly recommended.
slow_riot
uniquepersonno2 wrote:
Thank you for the extremely detailed response and explanation!
Since it is based off of the impedance of whatever module is outputting to it, would that mean it would have to be designed to fit one module only? I was hoping for a solution that I could plug any source in my rack into, which is why I was looking at transformer based solutions.
If the above is not an issue, I'm really looking for something to install into my rack permanently, instead of an adapter cable. Could this be converted to use input jacks as opposed to output jacks as you've pictured?


1k output impedance is more or less standard in euorack.

You could certainly adapt this solution to fit behind a panel, just apply the same principles. The performance would be more limited than running as balanced straight from the module output (explanation why is given above).
uniquepersonno2
auxren wrote:
I made a module that is based of AS007 here: http://www.jensen-transformers.com/schematics/

Sounds incredible and highly recommended.


This looks perfect! What transformer did you use?

slow_riot wrote:
1k output impedance is more or less standard in euorack.

You could certainly adapt this solution to fit behind a panel, just apply the same principles. The performance would be more limited than running as balanced straight from the module output (explanation why is given above).


Good to know! I'll definitely consider this option going forward, the cost seems minimal which is quite tempting.
555x555
Thinking through this problem is reminding me of what a mess Euro can be. The problem with the stated cable design is that the "correct" use of the wire shields in an unbalanced system is to connect them to chassis ground, not to signal ground. The signal then travels in the cable, and returns via the bus. The screen exists to shunt noise away from the cable to the chassis, which may or may not exist in Euro, and eventually to the mains earth, which may or may not be connected. So even with a passive balanced connection, you probably want some sort of bus connection to use to tie the ring resistor to signal ground. While my guess is that most modules connect the screen directly to signal ground, if you end up with the odd module which doesn't do this, you might be totally SOL with the cable design. Add that to the nonstandard impedance problem...

Anyway, it's aggravating that there really isn't enough of a standard with these things to solve problems that ought to be pretty simple. Like, we should have a sane and standard output impedance that doesn't try to be a terrible mixer. We should have screens connecting only to a chassis built from panels that aren't anodized on the back side/made of papier-mâché/etc. Balanced connections should be either standard everywhere, or at least standard for connections between standardly smallish cases. Etc.—others know all this stuff better than I.

What's especially annoying is that this lack of good practice actually makes the module design practice more difficult. This thread is a case in point. We should be able to offer a simple DIY design for case egress, rather than saying "it depends..." and all that.

There's not really a design practice to implement to get out of this problem in the future, either. Like, does one choose a 1k output and screens to ground so that people building weird cables like this will be happy? Or should one choose a low resistance output for maximum voltage transfer in order that people using lots of stackables will be happy? Or does one tie screens only to chassis ground so that people with noise problems can build a proper dedicated balanced egress? There's no good answer; everything won't be interoperable with some common problem/practice. That's not good for a modular synth. Semi-pro gear got the balanced/unbalanced 1/4" jack and it is usable with everything and usable properly balanced with properly balanced things. Can there be an equivalent design paradigm for Euro?[/endrant][/sorryboutthat]
auxren
uniquepersonno2 wrote:


This looks perfect! What transformer did you use?



I used the transformer in that app note, but, you can use most transformers with similar properties. Here are some other recommended transformers: Cinemag CM-DBX, Lundahl 1935, Jensen JT-DB, Sowter 4243

Just keep in mind that the transformers aren't cheap. $50+, but, worth.
slow_riot
555x555 wrote:
Thinking through this problem is reminding me of what a mess Euro can be. The problem with the stated cable design is that the "correct" use of the wire shields in an unbalanced system is to connect them to chassis ground, not to signal ground. The signal then travels in the cable, and returns via the bus. The screen exists to shunt noise away from the cable to the chassis, which may or may not exist in Euro, and eventually to the mains earth, which may or may not be connected. So even with a passive balanced connection, you probably want some sort of bus connection to use to tie the ring resistor to signal ground. While my guess is that most modules connect the screen directly to signal ground, if you end up with the odd module which doesn't do this, you might be totally SOL with the cable design. Add that to the nonstandard impedance problem...


The cable is not going to solve the lack of defined shield in eurorack gear, but how could it? The hybrid bonding via a capacitor is better than the one end only rule and tying 0V to shield (which modules marketed as high end do), except where there is bonding between the 0Vs of the 2 interconnected systems, which is likely out of the scope of this request.

The simple adaptor will outperform a transformer DI.

There are modules on the market with high performance balanced IO with correctly defined shield.
555x555
slow_riot wrote:
The cable is not going to solve the lack of defined shield in eurorack gear, but how could it?


Of course. I’m not trying to critique your design, I’m just pointing out that in the Euro world there’s some unavoidable problems with any design. Actually, the problem is that to me it’s clear that the passive extra resistor to ground and shield to screen is the simplest, best way of driving a balanced connection from an unbalanced system, but I also know that, because of problems with the Euro format, (a) this connection is still not up to spec, and (b) if a module “correctly” ties the screen to chassis instead of signal ground, you’ll have to implement this as a module rather than a cable. Given all the currently unsolvable problems, I’d basically use this design, but in a module, capacitor between output screen and chassis ground, resistor to ground via a power header. Although I suppose once you’re adding a power connection, you might as well also add a buffer so you can set the output impedance directly.

What I’m more usure of after this discussion is whether in a new module design I should tie screen to 0V and use a 1k output impedance, or follow best practices and use a <100 ohm out and tie screen to chassis.
Graham Hinton
555x555 wrote:

What I’m more usure of after this discussion is whether in a new module design I should tie screen to 0V and use a 1k output impedance, or follow best practices and use a <100 ohm out and tie screen to chassis.


If you do the former then you perpetuate the problems. If you do the latter then you need to get a separate 0V and Chassis Ground to the module. If you want to retain power connector compatibility use a MOTM header like this:

Pin1: +ve
Pin2: 0V
Pin3: Chassis Ground
Pin4: -ve

If this is connected to an old distribution Pins 2 & 3 will both be connected to 0V, but at least the cable screen currents will not be going through the circuitry.

A similar thing can be done with a Euro header by allocating three 0V pins on one side and three Chassis pins the other.

The Chassis Ground may be taken around the edge of the pcb to the jacks either as a thick track or a discrete wire.

If you are not interested in backwards compatibility then you can choose a better connector.
slow_riot
I've not done the testing myself but Neil Muncy has. The hard part of proper shielding is the custom metalwork, but CAD is better now than it ever has been. I use FreeCAD. Finding someone with a press brake who can accurately fold who has time is not so easy.

Yes the cable is a hack but it still beats a DI. It may be possible to add a switch for alternative impedances. No good for variable impedance output like Turing Machine.
555x555
Graham Hinton wrote:
555x555 wrote:

What I’m more usure of after this discussion is whether in a new module design I should tie screen to 0V and use a 1k output impedance, or follow best practices and use a <100 ohm out and tie screen to chassis.


If you do the former then you perpetuate the problems. If you do the latter then you need to get a separate 0V and Chassis Ground to the module.


The chassis ground can’t be the faceplate? I know Front Panel Express offers an anodized front, chromated rear option. I’m not really sure what chromated is, but I read elsewhere it was conductive. I wouldn’t be surprised to find out that isn’t true—so many diyers who just barely dip into the reading. I suppose you’d have to use a jointing compound with the rails to get a good electrical connection anyway. Either way, how careful do you have to be about this? What are the effects of more or less resistive chassis grounds?

In general the idea is to figure out how not to perpetuate problems, but also to keep compatibility. I suppose the output impedance compatibility issue is minor enough that you can safely lower it. The 0V vs Chassis screen seems a bigger compatibility issue, as people are used to randomly cross patching. I guess that’s the good thing about putting chassis ground on the power connector: there is likely little overlap between the people who would freely cross patch unbalanced and those who would use a distribution that differentiates between signal and chassis grounds.
Graham Hinton
555x555 wrote:
The chassis ground can’t be the faceplate?


Not in Eurorack, the rails are usually anodised and so are not a reliable connection.

Quote:

I know Front Panel Express offers an anodized front, chromated rear option. I’m not really sure what chromated is, but I read elsewhere it was conductive.


https://en.wikipedia.org/wiki/Chromate_conversion_coating

There are other finishes now due to hexavalent chromium being toxic and banned by RoHS. They tend to be inferior and react with sweat to make unsightly handmarks (typical unthought out EU legislation).

The best way to make contact with a finished metal panel is with a buried stud.

Quote:

What are the effects of more or less resistive chassis grounds?


A chassis ground is an equipotential and should not be conducting large currents. There may be tiny RF currents, but there is also the safety aspect to consider that it should be capable of conducting 25A (briefly).

Quote:

I suppose the output impedance compatibility issue is minor enough that you can safely lower it.


1kohm outputs are a mistake. People copying something they didn't understand. There is no good reason for having it.

Audio outputs should be 50 ohms or less, pitch CVs zero ohms.

Quote:

The 0V vs Chassis screen seems a bigger compatibility issue, as people are used to randomly cross patching.


Cable screens should not be a signal return between cases. If you want to crosspatch between different unbalanced cases there should be a 0V connection between them at least an order of magnitude lower resistance than cable screens. For fixed systems I use large busbars, for gigging systems I am now using Dinse connectors and 25mm2 cable.
555x555
Quote:
[quote="Graham Hinton"]
555x555 wrote:
The chassis ground can’t be the faceplate?


Not in Eurorack, the rails are usually anodised and so are not a reliable connection.


But, if it were common practice to use a little abrasion + jointing compound to cut through the anodization when mounting a new module? (or maybe this is still insufficient). Just checking that there isn't some other reason that it would be preferable to send chassis ground through the power header.

Quote:

A chassis ground is an equipotential and should not be conducting large currents. There may be tiny RF currents, but there is also the safety aspect to consider that it should be capable of conducting 25A (briefly).


Let me check if I understand. With a chassis ground, low resistance is not a huge concern, but rather coverage. This is true as long as the resistance from any point to earth is low enough that the I2R to briefly shunt 25A (a) doesn't blow up the chassis, and (b) pulls enough current away from a human conductor. Not melting under 25A seems like it would be a problem with the typical tiny Euro power wires, correct? Setting that aside, if one does put chassis ground on some of the ground pins of the Euro header, wouldn't it be better to use only one or two, thus lowering the resistance of signal ground as much as possible?
Graham Hinton
555x555 wrote:
But, if it were common practice to use a little abrasion + jointing compound to cut through the anodization when mounting a new module?


Jointing compound is intended to pierce natural aluminium oxidisation which is only a molecule thick, not an anodised layer which is much thicker.

Quote:

Just checking that there isn't some other reason that it would be preferable to send chassis ground through the power header.


You need to ensure that it gets there, not maybe. A tab or screw terminal could be used, but that would make installing modules unnecessarily more complicated. It would also mean that one omitted would mess up the whole scheme.

Quote:

With a chassis ground, low resistance is not a huge concern, but rather coverage.


Qualified by how high frequency it is intended to work.

Quote:

Not melting under 25A seems like it would be a problem with the typical tiny Euro power wires, correct?


Don't guess, look up the fusing current for that wire size.

Quote:

Setting that aside, if one does put chassis ground on some of the ground pins of the Euro header, wouldn't it be better to use only one or two, thus lowering the resistance of signal ground as much as possible?


If you are that worried don't use a Euro header. I'm only pointing out that you can design modules that work better and may still can be used in existing systems. In the latter case the 0V and Chassis are joined together and there is no improvement.

Once you realise that a power cable doesn't need to have the same connector each end and every module does not need to have the same connector then you are free to make improvements.

Getting back on topic, if you have a balanced output module the output jack screens should not be connected to the signal 0V, but as most systems only have a signal 0V the question is where should the cable screens be connected? There is no right answer in a system that has no Ground. If you see XLR or TRS jack connectors mounted on front panels they are probably incorrectly wired.
555x555
Quote:
Don't guess, look up the fusing current for that wire size.


OK, so 28AWG wire, which you can get bigger but is common for IDC connections, has 14.5A fusing current. You'd need at least two wires.

Quote:
I'm only pointing out that you can design modules that work better and may still can be used in existing systems.


Yeah, that's what I like. This seems like a pretty viable upgrade path for Euro: standardize with two ground connectors designated for chassis ground, connected to screens. Then you can use unbalanced/balanced style 3.5mm jacks a la semi-pro equipment. Everything interoperates, but if/when you're so fortunate as to have only modules and a distribution system built this way, it will have a decent design.

I know you're going one better than this with your stuff by using the superior bantam jacks. I'm curious, though, why you aren't (AFAIK) dedicating a power pin to chassis ground. Is it just that you're using superior rails/faceplates that create a reliable, connected chassis? Where are you connecting screen to?
555x555
Returning to the topic, this doesn't limit itself to passive, but it's a very good, short discussion of the relevant concerns and performance for interconnections:

http://www.douglas-self.com/ampins/balanced/balanced.htm
Graham Hinton
555x555 wrote:

OK, so 28AWG wire, which you can get bigger but is common for IDC connections, has 14.5A fusing current. You'd need at least two wires.


You have two or three, but why use ribbon cable at all when you can use discrete 20awg?



Quote:

This seems like a pretty viable upgrade path for Euro: standardize with two ground connectors designated for chassis ground, connected to screens.


It's not an upgrade path, it's a way of using modules intended for a better system in existing systems. Actually it's a downgrade path.

Quote:

Then you can use unbalanced/balanced style 3.5mm jacks a la semi-pro equipment.


NO. That's exactly the sort of thinking that has made Eurorack such a mess.
There is not the same level of TS/TRS compatibility as there is with 1/4".

Quote:

I know you're going one better than this with your stuff by using the superior bantam jacks.


Then why suggest using one of the worst connectors in the world?

Quote:

I'm curious, though, why you aren't (AFAIK) dedicating a power pin to chassis ground.


What makes you think I'm not?

Quote:

Where are you connecting screen to?


The Technical Earth point of the power supply., i.e. where the mains Earth from the inlet splits.

Quote:
http://www.douglas-self.com/ampins/balanced/balanced.htm


I wouldn't recommend articles like this that are full of mistakes. It doesn't even know which Ground symbol to use where. Most of the similar internet pages are based on AES Journal articles (some looser than others) so it is better to go to the source:

Shields & Grounds AES Journal special excerpt ($15)
AES48-2005: AES standard on interconnections. ($30)
555x555
Graham Hinton wrote:
It's not an upgrade path, it's a way of using modules intended for a better system in existing systems. Actually it's a downgrade path.


Yeah, that's probably a better way of looking at it; a downgrade path from the better future smile.

Quote:
Quote:
Then you can use unbalanced/balanced style 3.5mm jacks a la semi-pro equipment.


NO. That's exactly the sort of thinking that has made Eurorack such a mess.
There is not the same level of TS/TRS compatibility as there is with 1/4".


OK, good to know. In that case, I guess the best way is either to use adapter cables or to just limit the connections outside the case to dedicated modules.

Quote:
Then why suggest using one of the worst connectors in the world?


For compatibility, but it seems that was erroneous. It won't happen again lol.

Quote:
Quote:
I'm curious, though, why you aren't (AFAIK) dedicating a power pin to chassis ground.


What makes you think I'm not?


All the pictures of cables on your website going to only 3 busbars.
Graham Hinton
555x555 wrote:

All the pictures of cables on your website going to only 3 busbars.


Because they are power systems for other manufacturers' modules.
If it were totally for my modules if would be done differently, at the moment that is PinMix, microphone amplifier and custom systems only.
All my current modules have modified MOTM power connectors as detailed above and may be downgraded for use in other systems.
fluxmonkey
uniquepersonno2 wrote:
I've been looking at how to best go about this and found this article about building a passive DI, but I have no idea if that would work for what I want to do. The kit they make also comes in a bulky box that doesn't really suit my case as I want it mounted internally with the jacks/switches mounted on the euro panel and on the back of my case.


this requirement seems like the source of all your (unnecessary) problems. the enclosure is not bulky, it's the size of a small effects pedal. it provides isolation from your power supply and shielding. it saves you some front panel real estate, which in euro format is quite expensive. these are great little DIs for the money, and the best solution IMO.
notmiserlouagain
fluxmonkey wrote:
uniquepersonno2 wrote:
I've been looking at how to best go about this and found this article about building a passive DI, but I have no idea if that would work for what I want to do. The kit they make also comes in a bulky box that doesn't really suit my case as I want it mounted internally with the jacks/switches mounted on the euro panel and on the back of my case.


this requirement seems like the source of all your (unnecessary) problems. the enclosure is not bulky, it's the size of a small effects pedal. it provides isolation from your power supply and shielding. it saves you some front panel real estate, which in euro format is quite expensive. these are great little DIs for the money, and the best solution IMO.


+1 It´s a tool put it in series with your output and close to it, it gives you symmetric signal and galvanic isolation and in the best case a little bit of good sound, needs no power
Put one behind every module output hmmm.....
555x555
notmiserlouagain wrote:
fluxmonkey wrote:
the source of all your (unnecessary) problems.

+1


C'mon guys, dubious necessity is the mother of DIY invention.
MUFF WIGGLER Forum Index -> Music Tech DIY  
Page 1 of 1
Powered by phpBB © phpBB Group