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

Fixed Gain factor for Euro -> Line Level?
MUFF WIGGLER Forum Index -> Music Tech DIY  
Author Fixed Gain factor for Euro -> Line Level?
Tristana
Heya- Is there a fixed gain factor that's good to use for attenuating Eurorack to line level?

Is it 5/12th, or do most audio signals in Eurorack not reach that voltage? Even if not, will this allow a fair bit of headroom without making the level too low?



Working on building an out module that directly connects to out jacks on the side of my case.
RISC
Eurorack input resistance is usually 100kOhm so you may want to go a bit higher than 24k if you intend to connect it to arbitrary modules.

I (too) can't find any reference level for eurorack, but Wikipedia says audio is "typically" max. 10Vpp and from your ratio I would guess you want to go down to +4dBu so:
1.736V * 100kOhm / 5V = Rfeedback ≈ 34.7kOhm

Now you can either pick the lower resistor value from e.g. E24 or put some in series or use a trim pot if you want to be exact.

Some things to consider:
- If you are at +4dBu you may also want a differential signal, this can be achieved with the two op amps you already have.
- If you're not going differential, do you really need to keep the phase (by re-inverting) with electronically generated signals? Your ears don't care and you would have a spare op-amp for e.g. another channel.
- You may want DC-decoupling on output and / or input
Graham Hinton
Tristana wrote:
Heya- Is there a fixed gain factor that's good to use for attenuating Eurorack to line level?


No. What people mean by line level varies. Still people believe that +4dBu is a maximum signal level, it isn't, it is a reference level for calibrating metering. Audio equipment that runs on +/-12V or +/-15V has the same signal range as synthesizers and often has input level trims. You only need to attenuate to connect to pedals running on 9V or cheap soundcards running on +5V.

What you really need to know is what the maximum input signal level is and that should be in the product's spec. If it is over +20dBu you don't need to do anything. Beware of cheap mixers that combine the mic and line inputs, they can clip at +7dBu.

The other more important consideration is what length of cable needs to be driven. Most synthesizer outputs, even with 1k vestigial tails, can drive 20m, but if you have a longer line you need a line driver with low impedance output capable of driving the capacitance. Most op amps won't be happy driving a 1k load as you have drawn, you need specialised audio op amps for that.
You also need to consider the grounding because if you have a balanced output on TRS or XLR the cable screen should not be connected to a module's 0V.
guest
if you dont want to put an attenuation knob on, then i would go with a factor of 1/4. this will allow you to go into tape decks and home stereos without adjusting much. but, as graham points out, the ideal signal level will vary depending upon what youre sending it to.

you can also just put a resistor divider in front of an opamp buffer, and save one opamp.
Tristana
This is specifically being designed around use with an Ultralite Mk4:



Not entirely sure what 'dBu' translates to in terms of voltage (decibels have always confused me when referring to physical values...). But going by what you're saying, Graham, I should be fine to just set the attenuation on the MOTU interface itself (even though that's controlled by software rather than a pot)?

I've also decided (potentially incorrectly) that amplifying the 8 outputs is in order-


Since it's 4.7v is a gain factor of 12/5 good enough for scaling CVs to encompass the full range needed? or is only a gain factor of 2 needed to encompass the ±10v pitch CV signals?
cackland
Without knob control, as guest mentions, a 1/4 of the input signal (5vpp) would give you 1.241v, which is pretty close to +4dbu (1.228v).

Non inverting op amp buffer voltage divider, 1 op amp and two resistors.

Graham Hinton
Tristana wrote:



Not entirely sure what 'dBu' translates to in terms of voltage (decibels have always confused me when referring to physical values...). But going by what you're saying, Graham, I should be fine to just set the attenuation on the MOTU interface itself (even though that's controlled by software rather than a pot)?


It's a lot easier if you start thinking in dB levels and dB gains. You don't see faders on mixers calibrated in volts do you?

dBs are an absolute voltage if they have a letter after them which indicates its reference, "u" means relative to the voltage across a 600 ohm resistor dissipating 1mW (sinewave assumed). dBu are used for voltage levels. A +/-5V typical VCO waveform is about +14dBu, it varies with waveforms because there are different harmonics.
if you are reading dBs on a computer screen it is dBFS (full scale) so you have to know what 0dBFS corresponds to as a voltage on your interface to convert it to dBu. If your maximum output is +22 dBu add that to the reading. (Oh dear, the noise floor is worse than you thought!)


dB without another letter is gain and so is relative and may be negative. dBs are logarithmic unit so you can add gains in a chain instead of multiplying them. That is why they are so popular in audio. If you remember only two values you can easily work out more by combining gains and adding:
x2 = 6dB = 20.log10(2), so x-0.5 = -6dB, x4 = 12dB, x8 = 18dB, etc.
x10 = 20dB = 20.log10(10), so x100 = 40dB, x100 = 60dB, etc.
also
x1 = 0dB = 20.log10(1)
Note these are voltage gains not power gains.

Now most op amps used in synths like the TL07x series clip about 2.5V less than the power rails, so the maximum signal level you can guarantee with +/-12V power rails is +/-9.5V or +19dBu. You have less than 6dB headroom, but you can't clip the input to your Ultralite at +24dB.

Quote:

I've also decided (potentially incorrectly) that amplifying the 8 outputs is in order-


Yes incorrectly, you definitely won't need any amplification.
Be careful how you read that chart, it doesn't say, but it looks like those are peak voltages. You already know that the maximum output is +20dBu which is slightly too hot, so you could turn your interface setting down a notch. Match your analogue and digital clipping points will give the best range.

The information that is lacking is that the outputs are balanced so at maximum each side will have half the signal, that's where the 4.7V pk comes from. Some balanced outputs will double the other side if you short one, eg a TS plug in a TRS socket, other types definitely don't like that at all and you have to leave the ring unconnected and make do with half (-6dB) of the signal. The best way is to always go into a differential input amplfier and then you will always get the right level.

If you use a real meter to measure a signal in dBs it is using the rms values, but when you are talking about clipping and headroom you need the peak values. It should be obvious by context, but sometimes you have to read between the lines a bit.
Tristana
dB as describing gain I'm familiar with, having done some plug-in design.

Graham Hinton wrote:
Yes incorrectly, you definitely won't need any amplification.
Be careful how you read that chart, it doesn't say, but it looks like those are peak voltages. You already know that the maximum output is +20dBu which is slightly too hot, so you could turn your interface setting down a notch. Match your analogue and digital clipping points will give the best range.

The information that is lacking is that the outputs are balanced so at maximum each side will have half the signal, that's where the 4.7V pk comes from. Some balanced outputs will double the other side if you short one, eg a TS plug in a TRS socket, other types definitely don't like that at all and you have to leave the ring unconnected and make do with half (-6dB) of the signal. The best way is to always go into a differential input amplfier and then you will always get the right level.

If you use a real meter to measure a signal in dBs it is using the rms values, but when you are talking about clipping and headroom you need the peak values. It should be obvious by context, but sometimes you have to read between the lines a bit.


The current plan was to connect the Ultralite mk4 to my euro case via floating ring cables (1/4" TRS to 1/8" TS) as recommended by MOTU in that chart. The ring signal will thus be floating / an open circuit. Is this going to halve the voltage output?

There is no intention for long cable runs (10 ft absolute tops, likely a fraction of that) so I'm uncertain if I instead maintaining a TRS to TRS connection and having noise rejection by a differential amp for sending signals Ultralite<->Eurorack is worth the cost, design-time and pcb real estate. BUT, if implemented, I think I'm now understanding that this will give me the full 20dBu output of the mk4 in addition to the bonus noise cancellation?

A conversion calculator I found online says that 20 dBu is 7.5v RMS, won't this mean that the full ±10v range of pitch CVs is not covered? (edit: does say that the Peak-to-Peak voltage is ~22v, so perhaps it *is* covered? ...or would be covered with diff amps built in to prevent the signal halving)

- - - -

Getting back to the signals going from Euro into the Ultralite mk4; is there not something to be said for attenuating them anyway, so that the Euro outputs can play nicely with other gear that isn't as forgiving on its maximum input dBu? Are there any additional challenges a buffered voltage division circuit like what cackland showed above would cause, beyond it being unnecessary for use with my Ultralite mk4?
Tristana
To TL;DR my conclusions, it now seems that it's best to:

1. Have differential amps for each of the 8 inputs from Ultralite mk4 -> Eurorack, to make full use of its 20dBu outputs which encompass all possible eurorack CVs (and getting some bonus noise rejection)

2. Buffer the outputs from Eurorack -> Ultralite mk4, optionally attenuating them with a voltage divider to play nicer with line level equipment


With that all in mind- any suggestions on 1/8" TRS jacks that mount nicely into euro style PCBs (vertically?)? Is there a TRS equivalent to the beloved Thonkiconn / Qing Pu's?

For the diff amps, something like the LT1990 or OPA1632 preferable to standard opamp chips, or perhaps something else?
Graham Hinton
Tristana wrote:

The current plan was to connect the Ultralite mk4 to my euro case via floating ring cables (1/4" TRS to 1/8" TS) as recommended by MOTU in that chart. The ring signal will thus be floating / an open circuit. Is this going to halve the voltage output?


Yes, but try sticking a bare TS jack in and see if the tip voltage doubles. If it doesn't pull it out quickly.

Quote:

There is no intention for long cable runs (10 ft absolute tops, likely a fraction of that) so I'm uncertain if I instead maintaining a TRS to TRS connection and having noise rejection by a differential amp for sending signals Ultralite<->Eurorack is worth the cost, design-time and pcb real estate. BUT, if implemented, I think I'm now understanding that this will give me the full 20dBu output of the mk4 in addition to the bonus noise cancellation?


Interference on that short a cable will be minimal, but a balanced connection will be better for overcoming the inevitable grounding problems.

Quote:

A conversion calculator I found online says that 20 dBu is 7.5v RMS, won't this mean that the full ±10v range of pitch CVs is not covered?


7.5V rms x 1.414 [sqrt2] = 10Vpk for a sinewave.
Most modules have a 10V CV range, not +/-10V, so +/-5V will cover that.

Quote:

(edit: does say that the Peak-to-Peak voltage is ~22v, so perhaps it *is* covered? ...or would be covered with diff amps built in to prevent the signal halving)


Yes, both

Quote:

Getting back to the signals going from Euro into the Ultralite mk4; is there not something to be said for attenuating them anyway, so that the Euro outputs can play nicely with other gear that isn't as forgiving on its maximum input dBu?


Matching the maximum signals is as nicely as you can play. You then use as much headroom as you want working down from there.

Quote:

Are there any additional challenges a buffered voltage division circuit like what cackland showed above would cause, beyond it being unnecessary for use with my Ultralite mk4?


The 1k output is not necessary either. You haven't given the input impedance, but it won't be as high as 100k.
In general fixed gains or attenuations tend to never be right. Fixed gain and an attenuator is wasteful too. Look at mixers: you usually have a trim on line inputs and outputs, like +/-20dB on inputs and +10dB on outputs. Adjustable gain is the universal solution so that you never use more that you actually need and keep the snr better.

Quote:
With that all in mind- any suggestions on 1/8" TRS jacks that mount nicely into euro style PCBs (vertically?)? Is there a TRS equivalent to the beloved Thonkiconn / Qing Pu's?


Don't even think about mixing stereo 3.5mm jacks. You can put a bantam jack on the same panel density for individual inputs or bring 8 balanced over on a DB25 snake.
Tristana
First of all, thanks much for all your help- I've learned a lot.

Quote:
The 1k output is not necessary either. You haven't given the input impedance, but it won't be as high as 100k.

So your recommendation for outputting from Euro->mk4 is to simply have a physical connection, no active circuitry / buffering / etc.?

Graham Hinton wrote:
Don't even think about mixing stereo 3.5mm jacks. You can put a bantam jack on the same panel density for individual inputs or bring 8 balanced over on a DB25 snake.


Any particular reason as to why?

The original design goal was to make use of the same PCB twice for cost saving/efficiency; the one present on the side of the case would be powered & populated with the active circuitry (diff. amp etc.), whereas the one racked would be jumpered to bypass all of that. A ribbon cable would carry signals between the two.

Making use of connectors that aren't 1/8" TRS for both would mean either using two PCBs, or having one PCB with a very bulky, eclectic layout. Instead, couldn't the 1/8" TRS on the euro panel have their ring leg chopped off and insulated? [Edit: actually, without the circuitry populated it'd be an open circuit from the ring regardless; could be soldered for structural purposes]

Graham Hinton
Tristana wrote:

So your recommendation for outputting from Euro->mk4 is to simply have a physical connection, no active circuitry / buffering / etc.?


You already have active circuitry / buffering in the output of every module so you don't need more unless you want to drive long cables.
First try using a 3.5mm TS to 1/4" TRS screened twisted pair cable. The ring should be connected to the TS screen and the TRS screen may or may not need to be connected to screen at the TS end. Look into more complex schemes only if that doesn't work, 99% of the time that is all you need.

Quote:

Any particular reason as to why?


3.5mm stereo jacks are for headphones. Period. It's bad enough that they are being used for MIDI. If you mix mono and stereo jacks on a panel the wrong plugs will be inserted at some point. Are you sure that the circuitry both ends will handle that?

As for a DB25 it's a standard (AES59) way of connecting eight audio lines. You can buy ready made cables.

Quote:

The original design goal was to make use of the same PCB twice for cost saving/efficiency; the one present on the side of the case would be powered & populated with the active circuitry (diff. amp etc.), whereas the one racked would be jumpered to bypass all of that. A ribbon cable would carry signals between the two.

Making use of connectors that aren't 1/8" TRS for both would mean either using two PCBs, or having one PCB with a very bulky, eclectic layout. Instead, couldn't the 1/8" TRS on the euro panel have their ring leg chopped off and insulated?


Lateral thinking: why pcb mount jack connectors at all? Put a 3 pin header on the pcbs and then you can use any jack you like. Jacks get worn and broken so make them easy to replace.
Tristana
Graham Hinton wrote:

Lateral thinking: why pcb mount jack connectors at all? Put a 3 pin header on the pcbs and then you can use any jack you like. Jacks get worn and broken so make them easy to replace.


Original thinking was to mate the PCBs to an aluminum panel via the jacks, as I've seen many modules do. But, standoffs could be used instead, and/or the side-of-case I/O PCB could be mounted into the case's wood.

I'm considering bypassing use of a PCB altogether for the racked passive jack panel, and instead just directly wiring 10 TS jacks to a 12-pin socket header (to avoid making it look like a euro power connection, & to provide a common ground in case that's needed?) that'd be insulated and glued to the aluminum front panel.
MUFF WIGGLER Forum Index -> Music Tech DIY  
Page 1 of 1
Powered by phpBB © phpBB Group