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

Best practice regarding buffering & impedance, ground?
MUFF WIGGLER Forum Index -> Music Tech DIY  
Author Best practice regarding buffering & impedance, ground?
fingerfarbensound
I am wondering what's the "proper" way how inputs & outputs in a Eurorack module should behave.

Should inputs be buffered / be a high impedance input?
Should outputs be a low impedance driver? Is a current limiting resistor a good idea?
Protection diodes on everything?

From what I gather ring == ground on the jacks, should the module "short" all ground pins from the supply and have a common ground for everything including the jacks?

Any other noise rejection methods I should keep in mind if I want to drive high current loads like LEDs or even motors?

hmmm.....
Graham Hinton
fingerfarbensound wrote:
I am wondering what's the "proper" way how inputs & outputs in a Eurorack module should behave.


There is best practise from an audio engineering viewpoint and what Eurorack does. The two are different.

Quote:

Should inputs be buffered / be a high impedance input?


That will make it noisier. 100k is bad enough.

Quote:

Should outputs be a low impedance driver? Is a current limiting resistor a good idea?


Ideally outputs should be zero impedance with some current limiting and capable of driving down to 2k loads. This is not hard to achieve and allows for different input impedances.

Quote:

Protection diodes on everything?


Protection from what?

Quote:

From what I gather ring == ground on the jacks, should the module "short" all ground pins from the supply and have a common ground for everything including the jacks?


There is no ring on mono jacks.
Connecting jack screens to 0V is not best practise. Talking about "ground" like this is vague. Ideally there should be a 0V Common and a Chassis Ground/Screen, both are grounded, but should be kept separate. Most Eurorack modules are not built like this, but if you are building your own you don't have to copy mistakes.

Quote:

Any other noise rejection methods I should keep in mind if I want to drive high current loads like LEDs or even motors?


Doing either directly from module outputs is not a good idea.
Doing either and feeding the return currents into a module's 0V is not a good idea.
fingerfarbensound
Thanks,


Quote:

That will make it noisier. 100k is bad enough.
...
Ideally outputs should be zero impedance with some current limiting and capable of driving down to 2k loads. This is not hard to achieve and allows for different input impedances.

Ok then, what amount of impedance is a good value?
Lets say I have a high impedance input (like an ADC) should I add a resistor to load it?

Is there some ballpark amount of current I should aim to send over the cables?

Quote:

Protection from what?

Overvoltage I guess, but from what I gather you just assume nobody would send something outside the -12V to +12 into a jack.

Quote:

There is no ring on mono jacks.
Connecting jack screens to 0V is not best practise. Talking about "ground" like this is vague. Ideally there should be a 0V Common and a Chassis Ground/Screen, both are grounded, but should be kept separate. Most Eurorack modules are not built like this, but if you are building your own you don't have to copy mistakes.

Yes I meant the sleeve. So what should be connected to the sleeve if anything?

Quote:

Doing either directly from module outputs is not a good idea.
Doing either and feeding the return currents into a module's 0V is not a good idea.

That much I gathered, I am thinking about including a DC/DC converter after the 12V rail to generate a ~6V supply and hope that by the time the noise gets to the bus board the bypass caps attenuated it enough so it's not a problem. I can also make sure that I do not generate any fundamental frequencies in the audible range ie pwm >20khz and ramp the current up/down slower than 50ms.
J3RK
I just dug these up for another thread. They contain some information about this sort of thing:

https://www.muffwiggler.com/forum/viewtopic.php?t=157779&highlight=
https://www.muffwiggler.com/forum/viewtopic.php?p=1129644&highlight=#1 129644
https://www.muffwiggler.com/forum/viewtopic.php?t=81108&highlight=
fingerfarbensound
J3RK wrote:
I just dug these up for another thread. They contain some information about this sort of thing:

https://www.muffwiggler.com/forum/viewtopic.php?t=157779&highlight=
https://www.muffwiggler.com/forum/viewtopic.php?p=1129644&highlight=#1 129644
https://www.muffwiggler.com/forum/viewtopic.php?t=81108&highlight=


Is ringing really a concern? That would be well above the audible range.
I guess adding filtering at some point has other benefits, but is that really done?

regarding my original question
https://learningmodular.com/glossary/impedance/
Quote:
Eurorack modules tend to have an input impedance of 100k (100,000 ohms); output impedances can vary from near zero to 1k.
J3RK
There is a lot more information than that in those threads, including much of what Graham was getting at above.
Graham Hinton
fingerfarbensound wrote:

Ok then, what amount of impedance is a good value?


If you look at pro-audio circuitry an input "bridging load" is generally between 10k and 20k and you could have several paralleled. A 10k input resistor is 20 dB quieter than a 100k one.

Quote:

Lets say I have a high impedance input (like an ADC) should I add a resistor to load it?


You should avoid connecting ICs directly to the outside world. ADC usually require some form of buffering, range changing and offset.

Quote:

Is there some ballpark amount of current I should aim to send over the cables?


No. You aim to drive the correct voltage on the cable and to provide enough current into a reasonable load. 50R or 600R is not a reasonable load and requires specialised drivers.

Signals have >80dB range so the currents involved do too.

Quote:

Overvoltage I guess, but from what I gather you just assume nobody would send something outside the -12V to +12 into a jack.


Even if you interface to +/-15V systems you wouldn't get greater than +/-12V on an IC pin, there would be a resistor in the path.

You might want to consider the situation of connecting or mis-patching to a mixer mic/line input when phantom power is on. Most modules don't though.

Quote:

Yes I meant the sleeve. So what should be connected to the sleeve if anything?


Chassis Ground if you have one.

Quote:

I am thinking about including a DC/DC converter after the 12V rail to generate a ~6V supply and hope that by the time the noise gets to the bus board the bypass caps attenuated it enough so it's not a problem. I can also make sure that I do not generate any fundamental frequencies in the audible range ie pwm >20khz and ramp the current up/down slower than 50ms.


It isn't as simple as this, you need to consider where currents go. e.g. if you turn an LED on and off and the current goes to 0V you will be raising the 0V on the module by a voltage drop created in the 0V cable and pcb traces. This may cause audio breakthrough and is nothing to do with filtering the supply voltage. Modules with seven segment LED displays are particularly prone to this.

Quote:
Quote:
Eurorack modules tend to have an input impedance of 100k (100,000 ohms); output impedances can vary from near zero to 1k.


That is an observation, not something to aspire to.
Mungo
fingerfarbensound wrote:
I am wondering what's the "proper" way how inputs & outputs in a Eurorack module should behave.

Should inputs be buffered / be a high impedance input?
Should outputs be a low impedance driver? Is a current limiting resistor a good idea?
Protection diodes on everything?

From what I gather ring == ground on the jacks, should the module "short" all ground pins from the supply and have a common ground for everything including the jacks?

Any other noise rejection methods I should keep in mind if I want to drive high current loads like LEDs or even motors?

hmmm.....
The existing practices of modules already out there put some limits on what you can do while maintaining compatibility. Elby (Laurie) put together a vision for how they would like to ideally do things while maintaining as much compatibility as possible:
https://www.elby-designs.com/webtek/documents/eurosynth-specification. pdf
Unfortunately the best decisions for any one specific module interact with the other modules in the system, so many specific "rules" will only apply under ideal or contrived circumstances. Eurorack isn't really the place for modules drawing 10W of power each since the supplies are widely variable and mostly poorly specified. If you need a lot of power you'll either want to specify a known good set of power systems, or supply an auxiliary power source for the high draw sections.
fingerfarbensound
@Mungo thanks alot, that document is what I was looking for. thumbs up
555x555
100k (input)/1k (output) is the "standard" for Eurorack, but it is better to understand the tradeoffs and make intelligent decisions than to use that. That being said, if you don't understand the tradeoffs, putting 100k/1k everywhere will almost definitely work and not cause any major problems.

With input impedance, the problems are: noise, which is higher with a higher impedance; frequency response, which is generally not an issue, but in the rare case you run into an AC coupled output this will be better at a higher impedance; and level drop, which can be a problem with passive mults from other modules with high output impedance. You have to decide: do I need to accept ac-coupled signals? In modular, probably no, not really. Then, do I need to be precise about V/oct signals? Next, what is my acceptable voltage drop? For output impedance, there are fewer tradeoffs with lowering this value; a lower value provides lower noise, better frequency response into an ac-coupled input, and less signal drop. The disadvantages are: that it is potentially higher current, which can be a concern for shorted wires and *ahem* creatively plugged (output->output) signals; and, although this is not really about the output impedance itself, some amps in some configurations will need to see a resistive load before the capacitive load of the wire in order to maintain stability. There are other ways to design an output with stability and a reasonable current limit, but the easiest is to just tack on a resistor. 150 ohms or so is usually sufficient to solve most of these problems, and isn't a big deal unless V/oct accuracy is important. Note that typical input/output impedances of pro audio equipment are in the 10k-30k/20-100 ohm range, so 100k/1k is quite high, and disproportionately higher in the output impedance.

Lastly, having a higher input impedance in most cases makes you a good citizen—a greater number of modules that you plug it into won't have any problems—but this is true only up to a point, since you also don't want to be adding excess noise to the system. 30k is probably the lower limit you should be shooting for in a system which potentially has 1k output impedance. The opposite is the case for output impedance: the lower it is, the better citizen you are. And if everyone used lower output impedances, everyone could then lower their input impedance without worry.

As far as your other questions, buffers are only necessary if they're necessary for some reason, i.e. you want to make the input impedance higher/lower than it otherwise would be, or you need some other kind of isolation, e.g. from an input with a limited max voltage range. Generally, if you've got the impedance you want, no point in adding a buffer.

You don't really need protection diodes in most cases, as you have to be able to gracefully accept a pretty high signal to begin with. But of course this could be a thing depending on an individual case. Basically, you don't need them unless you have a specific reason that you need them.

And the sleeve either goes to chassis ground or signal ground. Chassis ground is better if: (1) you actually route some of the ground pins to chassis ground—this is not common, (2) you have only a single case or multiple cases that are connected via good solid ground connections, and (3) if you additionally take your input and output through proper output modules or somehow otherwise make sure that the signal and the signal return (or both parts of a balanced signal) make their way to your non-Euro gear in some manner other than through the chassis ground. Otherwise, sleeve should be signal ground, and signal ground just connects to all 6 pins shorted.

Note that a lot of Euro cables connect tip and sleeve to two different wires internally, and so sleeve is not always actually a copper shield surrounding the wire anyway.
fingerfarbensound
Thanks for the detailed responses!

Quote:
With input impedance, the problems are: noise, which is higher with a higher impedance

Noise from where exactly? Sorry if that feels pedantic.

What I've gathered so far is it's ok that if I have a high impedance input already I do not need to artificially lower it's impedance?

But lets come up with a practical example:
If I have this ADC: https://octopart.com/ltc2357clx-16%23pbf-analog+devices-88942561
Which is a 4 channel device which can handle +/- 10V directly into it's inputs.
(Lets for a minute ignore the fact that I can use a cheaper device and external offsetting & attenuation)
The inputs are of course buffered and high impedance.

I could just connect the jack input directly to the ADC pins, right?

Or I add 100K "pull down" resistors to tie the inputs to 0V.
(Which also would have the advantage that the ADC inputs wouldn't float if there is no cable in the jacks)
EATyourGUITAR
you are shorting the tip of the output to the sleeve of a jack when you are patching. this is why you need a resistor on your outputs. it only needs to be big enough to prevent damage when shorted to ground. input impedances of 100K with 3 modules in parallel is equal to a 33K load (5V/33K = 0.36mA). this is all the outputs need to drive most of the time. but sometimes they will see a 0v short where the total max output current becomes 5v/Rout = 100mA where Rout = 50 Ohms. we do not need linearity at 100mA while the jack is being inserted. we do need to prevent permanent damage that may occur when an opamp sees this kind of a load. it may also depend on how long the output is shorted. there are switches in eurorack that short things like this. if the input impedances were 1K it would lose brightness. or it would depend on the opamps used. high impedance presents less loading on purpose to get the voltage without changing ideal conditions. If I don't know what op amp is patched to my module's inputs, I will put a good op amp in my module and design it so that it does not load the cheap module. that is why 100K is standard eurorack and other 10vpp systems. in consumer gear, 10K is more common where the signal is 1vpp.
555x555
fingerfarbensound wrote:
Thanks for the detailed responses!
Noise from where exactly? Sorry if that feels pedantic.


Not at all :-). The noise comes from the resistance itself. It's known as "Johnson Noise." I'm not sure I understand it 100% myself, but I know it's there, and if compounded with many connections, or with an amplified low signal, it can become significant. The vague gist is that electrons move around because they have a nonzero temperature, and when it's harder for them to move, this makes more voltage noise. Wikipedia has more, as probably do many other people on this forum.

Quote:

What I've gathered so far is it's ok that if I have a high impedance input already I do not need to artificially lower it's impedance?


You might not *need* to, but you might *want* to. E.g. putting a resistor to ground on the noninverting input of a JFET op amp (like to TL07x/8x) is probably a good idea. It does seem counterintuitive that you can lower noise by adding components in parallel with an existing circuit; I'm not 100% sure about this, but I think it actually *would* lower noise. At the least, though, it's a good idea to throw a lower impedance onto something like that to lower coupled noise, for which you'd otherwise want to lay out a guard on the PCB.

Quote:

If I have this ADC: ...


In so far as it's an IC that can accept that input, that should be OK. There are sometimes special concerns in driving the input of an ADC though, e.g. do you need to do the first/last stage of antialiasing filtering yourself? I don't know that much about it, but this would be something to look into. At any rate, the question is not much different because the input is coming from a cable rather than from some other part of the circuit.
devinw1
Thanks for sharing that elby document mungo!. thumbs up
ljwall83
This is handy --- I'm just starting to design a first Eurorack module of my own (well, connect some standard circuits into a cohesive whole).

Sounds like to interface with other Euro rack modules you're more or less obligated to keep your input impedance up at 100k, but output could go down to whatever you can tolerate getting shorted to ground?

Reading that elby document it talks about 330R as a good general target output impedance, but <= 50R for CV outputs driving a v/octave input.

How do you achieve impedance that low from an output being driven by, say, a TL071/2/4, without it causing problems when shorted to ground as it's being plugged in to an input?
ljwall83
Sorry --- I should have read the links more fully before asking this question above ^^^.

Seem like the answers are all in there, and the TL07X don't actual need current limiting resistors: those resistors are there either to serve other purposes, or because cargo-culting.

Gonna read this all more thoroughly, and crack out the scope, before blinding adding those 1k's that seem to be on all the circuits I'm looking at!
EATyourGUITAR
ESD can be 2000v spike for a very very short time. Having no output resistor exposes the TL072 to damage if ESD happens. We are handling these patch cables with bare hands while it is on without ESD protection.
ljwall83
I admit I had not considered ESD. What's a good inline resistor value to protect against that?

I guess by the same token, an inline resisistor on inputs could be useful? Though I've seen a lot of designs without. Or is the high impedance of, say, an op amp input, protection enough?
devinw1
ESD protection first line is generally done with protection diodes, not a resistor. Check this out:

http://www.ti.com/lit/sg/sszb130c/sszb130c.pdf

and

https://www.analog.com/en/analog-dialogue/articles/esd-diodes-as-volta ge-clamps.html
ljwall83
thanks devinw1. Was reading about TVS diodes in the context of power supplies just the other day --- didn't connect the dots with regard to general input/output protections on other devices..
megaohm
devinw1 wrote:
Thanks for sharing that elby document mungo!. thumbs up


Is it just me or are most of the diagrams in the pdf blurry and unreadable?
Sandrine
I usually use 100 ohm on the drive side of pretty much any type of signal, while on analog input 100K, and digital 10K both with light 1M pulldowns
That said I have used 500 ohms drive, depends on what it is. @1V/oct lower is better because as mentioned above, it's possible you may be driving a 10K load which, at 500 ohms, will end up significantly lower (5% or 3 semitones at 5V)

On any I/O from ADC, DAC, microprocessor, CMOS ins I always buffer always. Also inputs to ADC or uP should normally have a schottky diode pair, especially if audio or hi-freq driving an ADC with low impedance from an op-amp. But sometimes, if a 1K resistor is added, ADC's are internally protected so no need.
MIDI input always isolated totally, and ouput drive limited with min. 200 ohms

Just my 2 cents
neil.johnson
https://www.muffwiggler.com/forum/viewtopic.php?p=1129644&highlight=#1 129644

Neil
MUFF WIGGLER Forum Index -> Music Tech DIY  
Page 1 of 1
Powered by phpBB © phpBB Group