Why do I need a buffered multiple?

Cwejman, Livewire, TipTop Audio, Doepfer etc... Get your euro on!

Moderators: Kent, luketeaford, Joe.

Post Reply
Epignosis567
Wiggling with Experience
Posts: 499
Joined: Wed Jul 02, 2014 10:28 am

Why do I need a buffered multiple?

Post by Epignosis567 » Wed May 13, 2015 5:55 pm

Why do I need a buffered multiple? What is bad about sending two oscillators into a multiple and out to one source? What exactly happens?

User avatar
dude
fuck yeah!
Posts: 10001
Joined: Sun Jul 05, 2009 2:08 pm
Location: northeastohio

Post by dude » Wed May 13, 2015 6:01 pm

you are fine in many cases to use unpowered mults. the one where a buff mult is supposed to be superior is sending precisions 1v/o type info. i however have had better results using a reg unpowered mult than the one buff mult i ever had. the buff mult got weird (maybe added a little voltage?). anyway, i would try reg mults if or until you found you needed the buff.

that is my anecdotal opinion and based only on my limited buff mult experience. i am sure there are some good cases for the buff mult. i just never found them.

User avatar
dude
fuck yeah!
Posts: 10001
Joined: Sun Jul 05, 2009 2:08 pm
Location: northeastohio

Post by dude » Wed May 13, 2015 6:03 pm

oh and you are talking about trying to use a mult as a mixer.

short answer is just to use a mixer.

i believe you can cause damage trying to use a mult as a mixer.

mixers are very undervalued publicly. i have the belief that they are just as important if not moreso than vcas in the modular concept.

User avatar
ilya.n
Common Wiggler
Posts: 142
Joined: Wed Oct 01, 2014 7:57 pm
Location: Sydney, Australia

Post by ilya.n » Wed May 13, 2015 6:14 pm

dude wrote:you are fine in many cases to use unpowered mults. the one where a buff mult is supposed to be superior is sending precisions 1v/o type info. i however have had better results using a reg unpowered mult than the one buff mult i ever had. the buff mult got weird (maybe added a little voltage?). anyway, i would try reg mults if or until you found you needed the buff.

that is my anecdotal opinion and based only on my limited buff mult experience. i am sure there are some good cases for the buff mult. i just never found them.
It probably just needed calibration.

User avatar
dude
fuck yeah!
Posts: 10001
Joined: Sun Jul 05, 2009 2:08 pm
Location: northeastohio

Post by dude » Wed May 13, 2015 7:14 pm

there wasn't anything to calibrate on the back. no pots or anything adjustable i don't think. anyway your comment had no bearing on the question posed and i haven't had said euro module for about 5 years now. offer him/her some advice, not me.

Epignosis567
Wiggling with Experience
Posts: 499
Joined: Wed Jul 02, 2014 10:28 am

Post by Epignosis567 » Wed May 13, 2015 7:41 pm

Wait, so I can or cannot put two oscillators out into a passive multiple and then out to my DAw or whatever? I was always told it was bad...

User avatar
dude
fuck yeah!
Posts: 10001
Joined: Sun Jul 05, 2009 2:08 pm
Location: northeastohio

Post by dude » Wed May 13, 2015 7:44 pm

yup it is bad to use a mult as a mixer.

see dave's reply in the linked thread

here

User avatar
Sinamsis
Super Deluxe Wiggler
Posts: 4809
Joined: Tue Jul 22, 2014 9:36 am
Location: Richmond, VA

Post by Sinamsis » Wed May 13, 2015 7:48 pm

I agree with Dude. Never owned a buffered mult. Often use them to multiplying 1v/oct signal. They track fine for my mom discerning ears. Granted I may use them over 2 or 3 octaves tops. Also if you read the Make Noise Mult manual (haha yeah I read the manual for a multiple, I'm a nerd), they mention most of their modules do not require a buffered signal. The jacks are balanced or something is the explanation but it's vague in the description. I assume inputs requiring 1v/oct are manufactured so as to not require a buffered signal. That being said I still don't get how that makes up for the voltage drop experiencing by patching a 1v/oct signal into an unbalanced mult.

That's pretty jumbled stream of conscious shit, but the bottom line is it should be negligible. Try an unbuffered mult. They're cheap and you can always use for envelopes and other shit. Use the money you saved to buy a mixer. They both come in handy. I have to argue that mixers in certain situations are not as valuables. It seems like a lot of modules have multiple inputs for CV signals along mixers not as essential. Korgasmatron is a great example. It has inputs out the wazoo.

EDIT: Also wanted to point out that modular levels are very hot. You want to attenuate before hitting your audio interface or whatever you're using. I do it's through my VCA. Does fine though it may impact tone. Does not effect me however. There's also a super cheap Blue Lantern module that's just 7 passive attenuators. I snagged one for my Z8000 but I may use it on my outgoing audio signal as well. A good audio interface module is also a very valuable tool in my opinion. It attenuated outgoing signal and boosts incoming signal. I like the Intellijel AI. The LEDs provide visual feedback. There are some other great ones out there as well.

Final aside would be that there are some VCAs that serve as mixers as well. The Blue Lantern HexVCA is what I use, though I don't use the mix function at all. Intellijel has one and I'm sure there are many others out there. Just food for thought.

User avatar
PM33AUD
Destroyer of BSP
Posts: 987
Joined: Sat Apr 23, 2011 7:28 pm
Location: USA aka dumpster fire

Post by PM33AUD » Wed May 13, 2015 9:06 pm

You may need buffered multiple for these reasons:

1) many modules are designed with out of the loop protection resistors which, when plugged into any module with nominal input impedance, you get a non-neglibile accuracy error (to your ear, a tuning error on something like a 1V/oct tuning input). There are relatively simple ways to fix this but many module makers either do not know about them or choose to continue doing it incorrectly out of legacy or lack of ability to stabilise and compensate an op amp for the common (or worst case) loads it will see.

2) there is no standardised specification for input impedances in euro.

3) there is no standardised specification for output impedance in euro.

4) there is no standardisation of what the acceptable amount of error should be.

These are just my observations but they explain why you need a buffered multiple. Your intuition on questioning 'why' is fully warranted. They should never be needed if things were done right from the start.

On a further note, some buffered multiples can add much more error than they 'correct' in some scenarios. This is due to the use of ash-n-trash (non-precision) op amps with drifty tempcos or high offset voltages, or if there is resistive feedback, poor resistor matching. Investigate (or ask the manufacturers that do not provide specifications directly) what the gain error is for their mults and see if that is within the specification you are after.

Make Noise jacks are not 'balanced' and neither are any others in euro to my knowledge - outside some special requests you may make with a Hinton product or something. Balanced IO has nothing to do with DC accuracy anyways.

At the end of the day, if you cannot notice the errors from loading issues, don't worry about it!

User avatar
Sinamsis
Super Deluxe Wiggler
Posts: 4809
Joined: Tue Jul 22, 2014 9:36 am
Location: Richmond, VA

Post by Sinamsis » Wed May 13, 2015 9:13 pm

PM33AUD wrote:You may need buffered multiple for these reasons:

1) many modules are designed with out of the loop protection resistors which, when plugged into any module with nominal input impedance, you get a non-neglibile accuracy error (to your ear, a tuning error on something like a 1V/oct tuning input). There are relatively simple ways to fix this but many module makers either do not know about them or choose to continue doing it incorrectly out of legacy or lack of ability to stabilise and compensate an op amp for the common (or worst case) loads it will see.

2) there is no standardised specification for input impedances in euro.

3) there is no standardised specification for output impedance in euro.

4) there is no standardisation of what the acceptable amount of error should be.

These are just my observations but they explain why you need a buffered multiple. Your intuition on questioning 'why' is fully warranted. They should never be needed if things were done right from the start.

On a further note, some buffered multiples can add much more error than they 'correct' in some scenarios. This is due to the use of ash-n-trash (non-precision) op amps with drifty tempcos or high offset voltages, or if there is resistive feedback, poor resistor matching. Investigate (or ask the manufacturers that do not provide specifications directly) what the gain error is for their mults and see if that is within the specification you are after.

Make Noise jacks are not 'balanced' and neither are any others in euro to my knowledge - outside some special requests you may make with a Hinton product or something. Balanced IO has nothing to do with DC accuracy anyways.

At the end of the day, if you cannot notice the errors from loading issues, don't worry about it!
Thanks for chiming in. I still don't understand half of that, but I get the spirit of what you're saying. Also balanced was probably a poor choice of words. I actually meant to say buffered, but I'm sure even that is not the right word. This is their exact wording:

"the Make Noise system does not require Buffered Multiples since all critical control signals are already buffered in such a way as to provide a large fan-out capability."

So I guess the output signal is what determines the lack of a need for buffered multiples in this case, though again it's a very vague description.

Epignosis567
Wiggling with Experience
Posts: 499
Joined: Wed Jul 02, 2014 10:28 am

Post by Epignosis567 » Wed May 13, 2015 9:42 pm

Thanks alot guys. So to what module would the damage be done in my scenario, the multiple or the oscillators or the destination?

User avatar
mskala
Super Deluxe Wiggler
Posts: 2750
Joined: Thu Mar 27, 2014 7:33 am
Location: Toronto
Contact:

Post by mskala » Thu May 14, 2015 2:25 am

It's worth emphasizing the most basic concept here: a multiple, with or without buffering, is for sending ONE output to several inputs. As soon as you say "two outputs" you want to be talking about some kind of mixer, not a multiple.Multiples split signals. Multiples are not for combining signals.

The wiring for passive mixers and passive multiples is similar enough that one can be used to makeshift for the other, but you should still be talking about it as a makeshift mixer. It's not really a multiple anymore as soon as you plug more than one output into it; it has become a (probably bad) mixer.
North Coast Synthesis Ltd.
Twitch stream (Mondays, 3pm Eastern)

If you tell me that your goal is systemic change toward radical acceptance, and I see that you treat those you perceive as lesser-than with the same kind of scorn and derision that pushed me toward this insular little subculture where I feel comfortable [. . .] then you’ve successfully convinced me that your acceptance is not radical and the change you want not systemic. - "When Nerds Collide"

User avatar
TechnoPrisoner
Common Wiggler
Posts: 152
Joined: Tue Oct 01, 2013 6:14 am

Post by TechnoPrisoner » Thu May 14, 2015 6:32 am

what about the Intellijel Hub ?

its the same thing with a mult only its not a module,right ?

so it has all the negative of regular mult ?

for example,last night I tried to mult the gate out from a Dark Time with A-182-2 1 in -> 3 outs ,when I patched the last cable on the 3rd out the module stopped sending signal completely to the other 2 outs above.I removed the patch from the 3rd out and it started working again.

User avatar
Sinamsis
Super Deluxe Wiggler
Posts: 4809
Joined: Tue Jul 22, 2014 9:36 am
Location: Richmond, VA

Post by Sinamsis » Thu May 14, 2015 6:50 am

It's passive. Again I'm guessing, but sounds like the voltage dropped to the point that it was not high enough to trigger whatever you were using it for.

My question would be is the voltage drop equal across outputs or does it increase the further from the input? In other words is the outgoing voltage of all outputs the same? I would assume so, but that with each additional output patched you have a further voltage drop in all outputs. Don't know if this is correct though.

O-Ma
Common Wiggler
Posts: 66
Joined: Wed Dec 03, 2014 11:17 am
Location: Tokyo, Japan

Post by O-Ma » Thu May 14, 2015 9:14 am

Very interesting about the gates ducking out. Thank you for sharing that info.

Does this mean it is better to use a buff mult for clock signals?

Is there a risk of missing clock steps if I use passive mults? And does this mean the more u split the gate/trigger signals the more likely it is to duck out, (stop responding)?

Epignosis567
Wiggling with Experience
Posts: 499
Joined: Wed Jul 02, 2014 10:28 am

Post by Epignosis567 » Sun May 24, 2015 9:35 pm

So what makes it unfeasible to develop shared standards in terms of what levels inputs and outputs function at?

User avatar
brandonlogic
Super Deluxe Wiggler
Posts: 3233
Joined: Tue Jan 27, 2015 9:17 am
Location: Milwaukee WI

Post by brandonlogic » Tue May 26, 2015 8:16 am

O-Ma wrote:Very interesting about the gates ducking out. Thank you for sharing that info.

Does this mean it is better to use a buff mult for clock signals?

Is there a risk of missing clock steps if I use passive mults? And does this mean the more u split the gate/trigger signals the more likely it is to duck out, (stop responding)?
I am wondering the same thing. I had a clock signal split 5 ways this weekend with a passive multiple and it seems that after splitting it wasn't consistently triggering my envelopes. wondering if a buffered mult would take care of this?

User avatar
PM33AUD
Destroyer of BSP
Posts: 987
Joined: Sat Apr 23, 2011 7:28 pm
Location: USA aka dumpster fire

Post by PM33AUD » Tue May 26, 2015 9:08 am

Epignosis567 wrote:So what makes it unfeasible to develop shared standards in terms of what levels inputs and outputs function at?
ego and ignorance

but slightly more seriously, there are standards organisations for audio like AES who do publish standards and the professional audio community uses them. here, there is no market pull to do so and the user base, by an overwhelming majority, doesn't notice many of these sorts of issues. or maybe they do and just don't care.

:despair:

JohnLRice

Post by JohnLRice » Tue May 26, 2015 9:40 am

Related but only for one specific 5U brand. For the heck of it . . .
[video][/video]

User avatar
PM33AUD
Destroyer of BSP
Posts: 987
Joined: Sat Apr 23, 2011 7:28 pm
Location: USA aka dumpster fire

Post by PM33AUD » Tue May 26, 2015 9:54 am

Hopefully I can settle the 'what's going on' questions or issues by explaining some of what is actually going on.

:omg:

A passive mult is like having a bunch of patch cables all soldered together putting them in parallel.. This is just wire - there is no 'in' or 'out' - that's up to you.

You plug one of the wires of the patch cable cluster into an output - let's say it's the gate output of a MIDI>CV converter. You don't want to plug any other outputs into this cluster of wires. The rest have to go to inputs. This was discussed above.

The gate output will have some sort of output 'resistance' in series with the jack of the module. This resistance is behind the panel and part of the design of the module. On some modules it is very low and on many others it is 1000 ohms. On others it doesn't contribute to the error.

Every time you take any one of the other wires in the patch cable cluster and plug it into an input of a module, that module's input has some 'resistance' to ground - current will now flow through that wire from the gate output. Most input impedances are 100,000 ohms.

So now you take that 1000 ohm output 'resistance' and the 100,000 ohm input 'resistance' and use a voltage divider equation to find a best case DC error:

Vout = Vin * [Rinput / (Routput + Rinput)]

For this example with, let's say, a 5V input signal from a gate module you get:

Vout = 4.95V. You lose 50mV. If this were a tuning CV, you'd easily hear the oscillator was out of tune.

If you start plugging this same output into more inputs with 100,000 ohm input 'resistances' you get even more drop... the equation then becomes:

Vout = Vin * [(Rinput/N) / (Routput + (Rinput/N))]

Where N is how many places you've sent that output.

For 4 total destinations and 5V input, you get:

Vout = 4.81V. You loose 190mV.

This is assuming people are using 1k output resistors (common) and 100k input impedances. This may not be true. Outputs can be designed to auto correct the mismatch errors. It depends who made the module, I suppose.

There is a lot more going on in regards to what sorts of loads an output stage can actually drive and how fast they can drive them. In the above examples, the capacitances of each cable are now in parallel, making a larger capacitive load to the output which can cause an output stage to become unstable. Or at the least, that nice square edge becomes rolled over and not so square any more. And with enough 'C', the square's amplitude wont be 5V anymore . But this is all a bit more advanced for a second digression.

The above is the best case error for this example and the math is easy. You can intuit what happens when you make Rinput closer to infinity and/or take Routput down closer to zero using the above equations.

User avatar
mskala
Super Deluxe Wiggler
Posts: 2750
Joined: Thu Mar 27, 2014 7:33 am
Location: Toronto
Contact:

Post by mskala » Tue May 26, 2015 3:03 pm

One small quibble: even with a 1k output impedance outside the feedback loop, any module putting out a pitch CV should and probably would have been calibrated to drive a 100k input. So in that worst-case example, it'll be putting 5.05V into the top end of the 1k resistor and delivering exactly 5.00V to the single input. You'll only start losing voltage when you add the second input in parallel.
North Coast Synthesis Ltd.
Twitch stream (Mondays, 3pm Eastern)

If you tell me that your goal is systemic change toward radical acceptance, and I see that you treat those you perceive as lesser-than with the same kind of scorn and derision that pushed me toward this insular little subculture where I feel comfortable [. . .] then you’ve successfully convinced me that your acceptance is not radical and the change you want not systemic. - "When Nerds Collide"

User avatar
PM33AUD
Destroyer of BSP
Posts: 987
Joined: Sat Apr 23, 2011 7:28 pm
Location: USA aka dumpster fire

Post by PM33AUD » Tue May 26, 2015 3:19 pm

I use personally use 50k input impedances to reduce the noise on unbalanced input stages (since 100k isn't really a standard). 100k input impedance on an oscillator 1V/oct CV input isn't standard either. Some are higher to account for a variety of source impedances to keep the error small enough.

You can only correct things in the manner you mention if everyone follows a standard. There is no standard so correction for one will be an even greater error for another!

You don't need the 1k OP resistor at all. I think people keep using it for legacy reasons.

I would love if I could use out of the loop series resistors. Compensation for wider variety of loads is easier and there are less parts. But this requires standardised input impedances for all modules. It also requires specifying how much error is tolerable. Someone uses a 5% 300ppm resistor yet I use a 0.05% 25ppm resistor... what's the point if we have to rely on this alone to get the required error?

I'm a precision I&M sort of guy so forgive me if I get a little edgy in regards to this sort of thing! :deadbanana:

Back to the OP, I'd say get a nice buffer if you see errors - if you do, follow my advice and ASK the manufacturer for the error specifications. If you want to be super fancy-pants, you can ask for this error over temperature too :ripbanana:

User avatar
edgek8d
Common Wiggler
Posts: 197
Joined: Sat Dec 27, 2014 11:49 pm
Location: Chicagoland, IL USA

Post by edgek8d » Tue May 26, 2015 4:16 pm

I love when seemingly common knowledge gets discussed, and everyone walks away more knowledgeable. :tu:

Post Reply

Return to “Eurorack Modules”