I need to lower an audio signal by 20dB. *Schematic added*

[mkiijam]

RL6 shorts the pots together. Also R54 with R55. The outcome is Thevenin-equivalent to a 11k/5k voltage divider fed by the average voltage of left and right channel.

As for the relays, "it will never happen" are famous last words before something happens
I think they could be rearranged to prevent any possibility of a short, and only two would suffice to choose from three inputs.

Right. I should know my own circuit, except it's mostly ingredients I've been grabbing to make my OWN pie!

The choice on 3 discrete relays was, at least in part, to facilitate lighting up LEDs on the input switches... I thought up that stuff about 2 years ago, so I'm kinda not remembering why I went with 3. The idea of not causing a short is intriguing. Let me think on that for a bit before you give it away.

[Alex Nikitin]

Over 20 years ago I've published (in Ukrainian magazine "RadioHobby" #2 2002), a circuit for the constant input impedance logarithmic attenuator. Here is the circuit, from that printed article, set for binary 1-2-4-8-16-32-64dB steps, giving the maximum of 127dB attenuation in 1dB steps, with only seven switchover contacts (preferably good signal relays) . For a 20dB attenuator step and a 10K load the resistors values should be 9K and 1.11K (two 18K in parallel and 1K1 should do nicely).

Cheers

Alex

[mkiijam]

So this could / would be controlled by a microcontroller or encoder or something?

[Alex Nikitin]

So this could / would be controlled by a microcontroller or encoder or something?

Yes, that was the idea. A microcontroller also allows for a timing trick to make the relay switching less noticeable, the worst example is if you go from -64dB to -63 or back from -63 to -64 and you switch all relays at the same time, you might release the lower attenuation bits before you engage the high attenuation bit, and get a very noticeable signal breakthrough up to 0dB, so you have to adjust the timing accordingly. If it is done right the switching is almost imperceptible on an audio signal. On the photo is the two channel stereo (or a single channel balanced) module from 2001 (I've worked as Chief Engineer at Creek Audio at the time), control was arranged in BCD format, 1-2-4-8-10-20-40 dB, 85dB max in seven stages plus a full mute (8 relays in total) , using laser trimmed resistor arrays.

Cheers

Alex

[Zero999]

Or use analogue switches. There are devices available with a resistance of under an Ohm. If it's significant, reduce the lower resistor values slightly.

[Alex Nikitin]

Or use analogue switches. There are devices available with a resistance of under an Ohm. If it's significant, reduce the lower resistor values slightly.

With very few exceptions solid state analogue switches are not good for audio applications, they more or less destroy the sound quality (at least practically all CMOS switches do). Also a limited headroom can be a problem. The volume control module I've posted above could take 50V RMS continuous signal input, try that with a solid state switch  .

The whole point of designing this circuit was to provide a better sound quality than solid state switches/controls and better precision and longer life than a motorized potentiometer, for a reasonable cost with as few relays as possible.

Cheers

Alex

[mkiijam]

Is there an example of the MCU that would control this or would i need to design it myself?

[Zero999]

Is there an example of the MCU that would control this or would i need to design it myself?

Any microcontroller will do. The relays are more important. Reed relays would be ideal.

Something like this would be great, as it can be driven directly from a microcontroller output pin.
https://www.pickeringrelay.com/product/101-1-c-5-3/

So this could / would be controlled by a microcontroller or encoder or something?

Yes, that was the idea. A microcontroller also allows for a timing trick to make the relay switching less noticeable, the worst example is if you go from -64dB to -63 or back from -63 to -64 and you switch all relays at the same time, you might release the lower attenuation bits before you engage the high attenuation bit, and get a very noticeable signal breakthrough up to 0dB, so you have to adjust the timing accordingly. If it is done right the switching is almost imperceptible on an audio signal. On the photo is the two channel stereo (or a single channel balanced) module from 2001 (I've worked as Chief Engineer at Creek Audio at the time), control was arranged in BCD format, 1-2-4-8-10-20-40 dB, 85dB max in seven stages plus a full mute (8 relays in total) , using laser trimmed resistor arrays.

Cheers

Alex

Of course, 126dB of attenuation isn't much good at low voltages.

The some non-standard resistor values in that circuit. Did you round to the nearest E96/E192 or go for two in series or parallel to get the exact value?

[Alex Nikitin]

The some non-standard resistor values in that circuit. Did you round to the nearest E96/E192 or go for two in series or parallel to get the exact value?

The custom resistor array trimmed to better than 0.1% was used in the original design, so no problem with non-standard values and at least 60dB CMRR in a balanced connection. In practice, one could use standard values (i.e. 1K1/81K for 1dB, 2K0/39K for 2dB, using two resistors combinations for higher attenuation), losing some accuracy but keeping the difference between channels to a minimum.

Any microcontroller will do. The relays are more important. Reed relays would be ideal.

Reed relays (as much as I like Pickering, use them a lot) have one particular problem in audio applications - a direct pickup from the coil (a closed contact makes it into a small transformer). A "normal" sealed signal relay with gold contacts is easier to use.

Cheers

Alex

[mkiijam]

Of course, 126dB of attenuation isn't much good at low voltages.

I've never understood this. My feed to the output amps / line drivers is from the output of the volume control which can be at ground? How is this anything but infinite (in theory) attenuation?

[Zero999]

Of course, 126dB of attenuation isn't much good at low voltages.

I've never understood this. My feed to the output amps / line drivers is from the output of the volume control which can be at ground? How is this anything but infinite (in theory) attenuation?

I was talking about Alex's circuit which has an atenuation range of 0 to 127dB in 1dB steps. 127dB is over 2.2 million, which is pretty pointless at low level signals. One might as well just disconnect the signal, using a switch. At releatively high voltages, such high levels of attenuation make more sense.

[magic]

The point of 127dB is that it costs only two resistors and a relay more than the obvious alternative of 63dB, which OTOH may be not enough

[mkiijam]

RL6 shorts the pots together. Also R54 with R55. The outcome is Thevenin-equivalent to a 11k/5k voltage divider fed by the average voltage of left and right channel.

It seems like I read someplace that the two signals combined will be louder, so this is why there are larger resistors?

[magic]

Two signals mixed by means of equal resistors become simply the average of the two signals.
For instance, if the signals are equal, the output is also equal to each of them.

The output of such passive "mixer" is X = (A + B)/2 = A/2 + B/2.
So each signal is attenuated by half and then they are summed together.

[mkiijam]

What I meant to say was: https://www.goldpt.com/mode.html

I just used this for MONO.

Is that wrong in my case? Better to use an active buffer for the mono mode?

[magic]

I believe the issue was that switching to mono increases output resistance of the mono/stereo module and causes further attenuation?

Several options to deal with it.

1. Simply live with it. An equal signal on L and R will be become quieter in mono mode, even though nothing really changes. You may want to ensure that the attenuator/volume module has constant input impedance so that the attenuator still makes exactly 20dB difference in mono mode too.

2. Insert a buffer between the mono/stereo module and the attenuator/volume module. It may be switched into the signal path for mono only or simply always connected.

3. Maybe you could find a way to modify mono/stereo to always have 22kΩ output impedance per channel, then the same attenuation will happen in stereo too. Compensate for it by adding gain or reducing attenuation somewhere else. Watch out for noise/clipping/dynamic range issues.

4. Reduce mono mixdown resistor values to reduce output impedance and the amount of attenuation in mono mode. This increases loading on the preceding stage when L and R aren't equal.

[Bltzbrg]

Hello,

I couldn't find any help with google or chatgpt and this forum is the closest to a possible solution.

I am designing a PCB basically to merge the signals from two 3.5mm audio jacks to a headphone and into a recorder. I want to enable an about -15db attenuation with a simple 2 position switch. I tried to understand which resistors I need but I couldn't figure it out. All I know is that the source that is too loud is usually connected to headphones with 64 ohm and a driver sensitivity at 1kHz/1mW (db) - 88db.
The signal is splitted. It should go unchanged into the headphones and with -15db into the audio output jack to my audio recorder.

I couldn't understand why some people calculate with 10kOhm.

Any help would be much appreciated

Thank you!

[magic]

10kΩ is a common input impedance of line level inputs (i.e. each channel is loaded to ground by 10kΩ), although certainly not guaranteed - it may be higher or sometimes lower.

A simple passive mixer (two equal resistors between the two inputs to be mixed) has high output impedance and won't drive headphones directly, so this device will need some active electronics: a buffer/amplifier and a power supply for it.

[Bltzbrg]

I don't know which numbers I should look for.

I want to put the signal of the Rode mic on the left audio channel in my gopro and the signal of my mp3 player on the right channel. The headphones of the mp3 player have got following specs:

Driver sensitivity at 1kHz/1mW (db) - 88db
Impedance - 64 Ohm
Dynamic Frequency range - 20-20.000 Hz


The microphone's technicals specs (Rode Smartlav+) are:

Acoustic Principle: Permanently Polarised Condenser
Active Electronics JFET
Polar Pattern Omni-directional
Frequency Range 20Hz - 20kHz
Output Impedance 3k Ω Typical
Signal To Noise Ratio 67 dB
Equivalent Noise 27 dB Typical
Maximum SPL 110 dB
Sensitivity -35dB (17.8mV @94dB SPL)
Dynamic Range 83dB Typical
Power Requirements Powered from iPhone TRRS
socket (2.7V)

Unfortunately I cannot find any technicals specs to the mp3 player.

Hearing the microphone on the left channel and the mp3 player on my gopro footag already works. I just want to make the mp3 player more quiet (but I need the Volume on the headphones). Its just too much for the gopro.

I want to lower the volume of the mp3 player just for my gopro by about -15db. I don't know with which impedance I should calculate and how many Watts the resistors should be able to handle. Also is Pi or T attenuator better?

Many thanks!

[Bltzbrg]

I found some more information about the GoPro Mic-Adapter:
Bias-Voltage: 2,5 V

Input impedance 2,2 kOhm in setting "standard-microphone" and "standard-microphone+"
Input impedance 47 kOhm in setting  "condenser microphone" and "condenser microphone+"
Input impedance of the GoPro AUX-Input 8 kOhm
unbalanced 3.5mm stereo

[magic]

I want to put the signal of the Rode mic on the left audio channel in my gopro and the signal of my mp3 player on the right channel.

Hearing the microphone on the left channel and the mp3 player on my gopro footag already works. I just want to make the mp3 player more quiet (but I need the Volume on the headphones). Its just too much for the gopro.

I want to lower the volume of the mp3 player just for my gopro by about -15db. I don't know with which impedance I should calculate and how many Watts the resistors should be able to handle. Also is Pi or T attenuator better?

So no need for actually mixing mp3 and mic signals onto a single channel and you don't care about hearing mic signal in the headphones?

This sounds fairly straightforward now.
First of all, connect the headphones directly to the mp3 player and use a passive splitter to get a second output to the GoPro.

To attenuate the mp3 signal, use an ordinary resistor divider. You already have 8kΩ resistors to ground on each input pin inside the GoPro (if aux input is the one you are using for mp3), so simply add 36kΩ in series with the cable for 0.18x attenuation and that's it. Power dissipation in these resistors is negligible - output voltage of mp3 players is no more than a volt or two RMS at full volume, which makes fractions of mW when applied to such high resistance.

[Bltzbrg]

I checked the settings of my GoPro and it uses the "standard-microphone" as default. I want to stitch on that to stay end-user friendly.

I will not hear the mic signal in the headphones because it is routed differently.

So 2.2k Ohm it is.

So how exactly should I place the resistors? So the standard 65 mW is enough? 100mW does cost 1 cent or even less more so I could also go for that.

I just mentioned all the other devices and their specs because I read something about T and Pi Attenuator and Input and Outpur Impedance and it confused me. I thought having some context may prevent some errors in the answer.
Do I need a T or Pi Attenuator?

A small drawing would be nice, if it doesn't bother you I guess I just need the part: Ground, Signal wires from the switch -> Resistors -> GoPro TRS

Thank you very much already for your fast replies and willing to help

[magic]

This should work. Power rating really doesn't matter. I would try it with ordinary 0.25W THT resistors, and if it works as intended put the resistors and cables in any random plastic box with panel mount jack connectors for inputs/outputs.

Code: [Select]

       +---------------- R HP out
       | +-------------- L HP out
       | | +------------ HP GND
       | | |
       | | |
R in --+------ 22k --+
         | |         +--- R out
L in ----+---- 22k --+
           |
GND -------+-----+------- GND out
                 |
GND -------------+
Mic in ------------------ L out
This assumes the receiving device has a stereo electret mic input with separate 2k2 loads on each channel - one for the divider (R) and one for the electret mic (L).

It is important that HP ground is connected close to the input connector and MIC ground close to the output connector, as drawn. Don't allow HP return current to flow through the wire between output GND and mic input GND.

[Bltzbrg]

Let's start with the fact that your code drawing looks really good

I don't have a stereo signal. I merged L and R right after the TRS Socket to make things easier. For the headphones I just split the audio signal to L R again. This is the "end recording device" : https://gopro.com/en/us/shop/mounts-accessories/pro-3.5mm-mic-adapter/AAMIC-001.html

I hope it has two seperate loads. I think I have to assume it.
I connect all the GND with a Copper pour. So everything that is not a signal line is ground (classic PCB design). How did you calculate the 22k resistor? The online calculator (https://www.everythingrf.com/rf-calculators/attenuator-calculator) is probably not suitable for me right? So for my case I just need one resistor with 22k Ohm right before the output?

It would be nice if you could explain how you calculated it. Maybe I have to make adjustments to -20db or -10db in the future.

Thank you so much. You already helped me a lot!

[magic]

For mono use two paralleled 22k or simply 10k, doesn't matter for this kind of accuracy.

But how are you "merging" L and R after the TRS socket? You shouldn't simply short them together, the two output amplifiers of the mp3 player will "fight" and one will source lots of current while the other will sink it, both attempting to set a different level on the same line... They are probably protected enough to survive undamaged, but battery life will be impaired if nothing else.

These Π and T attenuators are for RF systems with 50Ω (or similar) input and output impedance everywhere, that's not how line level audio signals are transmitted and you don't need these circuits.

The mono circuit I suggest is a simple voltage divider consisting of 10k and 2k2 (inside your mic adapter) giving (10+2.2)/2.2 = ~5.5 attenuation ratio. The stereo version replaces 10k with two 22k in order to mix two input signals; if both signals are equal then it behaves the same as mono. For other ratio, the usual voltage divider rule applies.