HELP with designing a headphones distributor please

[Hubbub]

At the risk of over-simplifying, do you even need stereo, if it's just a speech feed?

If mono will do, it's now totally simple:

View attachment 117570

You just need a DPDT switch and a single-gang pot, and drive both headphone channels from one transformer.

(A 4k7 pot will probably do just fine. With a pot driven from a low-impedance source, the maximum output impedance is one-quarter the total impedance, which occurs at 50% track resistance).

Your solution is 99% correct. My only quibble is 4k7 is a fixed resistor value, which is not common for potentiometers unless you pay for custom tooling and buy maybe 10,000 or more. Fortunately a value of 5k is common and readily available.

Even a linear taper pot would work decently as loading it reasonably heavy bows the transfer curve down approximating a "linear in dB" type log-taper pot. FWIW I've done this, plotted out via spreadsheet for proof, just recently for an oddball DI box I'm building.

 

[Voyager10]

Also: I've deliberately shown a dual-pole switch switching both sides of the transformer between the A and B channels.

I'd avoid the temptation to use a common ground / -ve for both channels, because the wiring resistance is going to be noticeable here, and this will lead to crosstalk between A and B. (e.g. 1k resistance between B+ and B-, 10 ohms ground resistance on a shared A- and B- line, gives -40dB crosstalk between A and B. That's going to be annoying if they're totally separate audio streams).

As @MidnightArrakis says, paralleling each channel over two RJ45 pairs will lower the wiring loss, and give you a degree of resilience if the cables/connector get dodgy.

 

[Hubbub]

[10w-20w amp into 8 Ohms -> 1 Box Bus -> volume pot -> transformer -> headphones] -- Equals a 5K load.

[10w-20w amp into 8 Ohms -> 25 Box Bus -> volume pot -> transformer -> headphones] -- Equals a 200 Ohm load.

>> Does this load difference make any difference to any kind of apparent signal-level shift?

[Even a linear taper pot would work decently as loading it reasonably heavy bows the transfer curve down approximating a "linear in dB" type log-taper pot] -- I interpret this as meaning the -- taper -- of all of the pots will vary according to the number of boxes plugged into the system. Does this matter?

[If mono will do, it's now totally simple] -- Should a mono signal and system be used, would there be any benefit to paralleling the RJ-45 conductors to become "2-sets of 4-conductors"? Wouldn't doing so lower the total wiring loss? Just wondering.....

/

The load difference between 5k and 200 makes no difference. The amplifier has an output impedance of less than 1 ohm, think of it as a voltage divider: 5000/5001 is almost the same as 200/201, at most they differ by about 0.5% or 0.043 dB. That is why the A-B toggle switches, regardless of position, will have no perceptible effect on volume.

The taper of one pot depends on the load to that pot alone. This presumes the pot is fed from a load-independent supply which the above sentence indicates is a fair assumption here.

Yes, paralleling conductors for mono will lower the wiring loss and would be beneficial, albeit to a rather small degree. Just don't parallel conductors in the same twisted pair, but rather parallel each wire from pair "a" with a wire from pair "b". This keeps the twisted pair nature of the signal & return (now 2x) so as to maximize immunity to induced noise.

 

[Hubbub]

Just had a bizarre thought: Have independent control of left and right headphone channels, each side gets its own A/B switch and volume control. Now one has the option to have the native language in one ear and the translation in the other. This could be beneficial (or distracting?) to someone who is conversant in both languages. You could offer this as a specialty "C" box, building only one or two just for the experience.

If there is a lot of give and take between speakers of different languages this could be helpful as a user could hear all languages, including theirs, in one ear, while always hearing the translation in the other ear. I'm picturing: understand left ear, my guy speaking; understand right ear, their guy speaking (via translator).

 

[sonolink]

Just had a bizarre thought: Have independent control of left and right headphone channels, each side gets its own A/B switch and volume control. Now one has the option to have the native language in one ear and the translation in the other. This could be beneficial (or distracting?) to someone who is conversant in both languages. You could offer this as a specialty "C" box, building only one or two just for the experience.

If there is a lot of give and take between speakers of different languages this could be helpful as a user could hear all languages, including theirs, in one ear, while always hearing the translation in the other ear. I'm picturing: understand left ear, my guy speaking; understand right ear, their guy speaking (via translator).

I'll ask my friend but I think that won't be very useful to him. Thanks for the thought though

I've found these amps. Would they be suitable? Any preference?

https://es.aliexpress.com/item/1005...id=2023112507263515577450821769550007496923_5

https://es.aliexpress.com/item/1005...4354e4b6c!12000017039207603!rec!ES!784183733!

https://es.aliexpress.com/item/1005...5607e4b6c!12000032735053321!rec!ES!784183733!

https://es.aliexpress.com/item/1005...id=2023112507263515577450821769550007496923_1

Thanks a lot for your help
Cheers
Sono

 

[sonolink]

I just bought these:



40 transformers = 12,41€ shipping included!!!

 

[Hubbub]

Looking at the 4 amplifier choices the first 3 are scary - high-efficiency digital but without the necessary filtering to drive a distant speaker. These would radiate interference like crazy! The fourth one (https://www.aliexpress.us/item/3256805679473958.html) is class AB and would be my choice of the 4 to try.

There is also this: https://www.aliexpress.us/item/3256801429284041.html but it definitely needs a heatsink. If you can power your amp from a regulated supply, this last one will work up to 18V if I read the amplifier chip datasheet correctly.

Words of warning: When using amps that expect a bridge-tied load (BTL) do NOT ever make any connection between one speaker channel and the other. BTL amps have outputs that must remain un-grounded and independent. (Once you get through the 1300:8 transformer you can have a common connection, but the XFMR outputs cannot be wired in parallel.)

 

[sonolink]

Words of warning: When using amps that expect a bridge-tied load (BTL) do NOT ever make any connection between one speaker channel and the other. BTL amps have outputs that must remain un-grounded and independent. (Once you get through the 1300:8 transformer you can have a common connection, but the XFMR outputs cannot be wired in parallel.)

Thanks for the tip

Would this B Box schem be correct?

 

[sonolink]

A) Is SW2.2-B1 supposed to be GROUNDED?

No, the GND is the shield of the RJ45 connector (case)

B) Is there a reason why one pot is 5K and the other one is 5K1?

Typo, sorry...

C) Where do you plan on placing your little power-amplifier? In the "A" box?

Correct

D) Why are the connections going to the switches "A3" pins going to different RJ45 pins (1 & 2), but the connections to the switches "B3" pins go to the -- SAME -- RJ45 pin (4)?

Typo. Corrected

E) Why is RJ45 Pin-5 left unconnected?

Same as above

Cheers
Sono

 

[sonolink]

Is this better and clearer?

 

[ccaudle]

No, the GND is the shield of the RJ45 connector

Standard schematic practice uses a distinct symbol for chassis connections.
See the picture on the Wiki page: Chassis Ground page at Wikipediahttps://en.wikipedia.org/wiki/Chassis_ground
Using the proper symbol will reduce confusion regarding the design intention.

The pin selection in your schematic of post #91 is decidedly non-optimal from the standpoint of both interference rejection and inducing crosstalk.
To take advantage of the twisted pairs in the cable the entire current loop should be contained in one pair, so A1/B1 should be on one twisted pair, A3/B3 should be on another twisted pair.
If you want to double up conductors to reduce resistance, then you should pick a second twisted pair for each and wire in parallel, so A1 could go to pin 1 and 3, and B1 could go to pin 2 and 6; A3 could go to 4 and 7 and B3 could go to pins 5 and 8.

 

[sonolink]

Notice how in your schematic that the connections going to SW1.1 are in a linear "step-like" pattern, but the connections going to SW2.2 are in an "up and down" pattern. In addition, the connections to SW1.1 also follow a nice linear right-to-left 1, 3, 5, 7 descending order, but the connections to SW2.2 are "jumbled" as 2, 6, 4, 8!!! Is this how things are supposed to be? Just wondering.....

View attachment 117686

I followed the RJ45 Cat6 pinout. Twisted pairs are 1-2, 3-6, 4-5 and 7-8.

[the GND is the shield of the RJ45 connector (case)] -- OK.....fine. But.....where is GND coming from? Is there a separate GND/Shieldconductor within the RJ45 cable not shown?


/

Sorry, I should've said that my idea was to connect the cat6 shield to each box chassis and make a final connection at box A (since box A will always be used) at the power supply ground.

The pin selection in your schematic of post #91 is decidedly non-optimal from the standpoint of both interference rejection and inducing crosstalk.
To take advantage of the twisted pairs in the cable the entire current loop should be contained in one pair, so A1/B1 should be on one twisted pair, A3/B3 should be on another twisted pair.

Like so?

If you want to double up conductors to reduce resistance, then you should pick a second twisted pair for each and wire in parallel, so A1 could go to pin 1 and 3, and B1 could go to pin 2 and 6; A3 could go to 4 and 7 and B3 could go to pins 5 and 8.

Is this what you mean?





Thanks a lot for your time and help

Cheers
Sono

 

[ccaudle]

Like so? [showing twisted pair used per channel]

yes
Incidentally that is the same thing which Hubub pointed out in post #82 on Friday. His phrasing was "Just don't parallel conductors in the same twisted pair, but rather parallel each wire from pair "a" with a wire from pair "b". This keeps the twisted pair nature of the signal & return (now 2x) so as to maximize immunity to induced noise."
Don't want anyone to think I am claiming credit for the idea, it is pretty obvious once you realize why wires are twisted in pairs to begin with.

Is this what you mean? [doubling up pairs]

Yes, but I thought that was what you were effectively doing before (in the first diagram). Since I did not see an updated Box A design (might have missed it, there was a holiday in the US) I thought that the A3 pins at switch 1 and switch 2 would connect together back at the A box. User 1 and User 2 will both have access to the same two channels, correct? What was the plan shown in the first diagram, four amplifier channels to drive the two audio signals?
It would be a little easier to follow with signal names on the pins. In the diagram with switch 1 A3 connected to pin 1, and switch 2 A3 connected to pin 2, what signals were going to be on pins 1 and 2?

 

[sonolink]

Ok.

First of all, please accept my apologies for the inaccuracies I'm providing. I'm just sketching up, trying to go from "coarse to fine" ok? I also thank you all for pointing out those errors

Based upon your schematic, I get the impression that the -- IN GND-CHASSIS -- and the -- OUT GND -- * ARE NOT * connected together. Is this correct? Otherwise, they would have the same -- Net Name -- shown at each of the Pin-9 locations to be electrically connected together.

They are connected through the GND plane copper pour. I will correct it so that Diptrace will connect both nets.

What are you planning on doing with these grounds and how is it that you wish for them and the entire system to be connected together? I'm confused.....

My idea is to connect each Box B chassis (aluminium) to the Cat6 shield through the RJ45 pin 9, so that all boxes will be shielded. Box A will be the same but will connect said shield to the PSU GND (kinda like a star ground). If this is wrong please point me out in the right direction

Headphones + and - circuits are only connected to the amp in Box A through the transformer, switch and pot.

It would be a little easier to follow with signal names on the pins. In the diagram with switch 1 A3 connected to pin 1, and switch 2 A3 connected to pin 2, what signals were going to be on pins 1 and 2?

Sorry, I will redraw the whole thing to make it clearer so that you have a complete vision of what I'm trying to make

Thanks
Cheers
Sono

 

[ccaudle]

They are connected through the GND plane copper pour. I

In the A box? Because the B box has nothing but wires and transformers now, there is nothing that needs a copper pour connected to chassis.

 

[sonolink]

In the A box? Because the B box has nothing but wires and transformers now, there is nothing that needs a copper pour connected to chassis.

I wasn't thinking of going P2P, so I need something to solder the transformers, switches, pots and RJ45s to in each box: a PCB with thick traces and a copper pour. That would make easy connecting the RJ45 female connectors to the bus.
Anyway, let's not get ahead too fast
For now I've drawn a new diagram:



And a new schematic of the Box B with pin names:



Thanks
Cheers
Sono

 

[Hubbub]

If you connect CHA+(1) to CHA+(3), etc. in every "B" box you end up making a slew of unnecessary redundant junctions. The first schematic in post 95 is all you need (provided you clean up the different grounds). The paralleling connections between pairs need only be made in the "A" box. Note that this is still effectively paralleling pairs in the cable for less loss. In one case you get lots of junctions that feed 2 loads per box from 2 pairs, in the other case per box you need no junctions and feed 1 load from 1 pair, just 2 times over. For 20 boxes either approach puts at most 20 loads on one pair, same as 40 loads on 2 pair. I vote for the simpler solution with fewer loops.

One could argue that with the junctions as shown in post 100 you have some redundancy in case you have a failed RJ-45 contact. But you would never know you had a failure that needed repair, which I think is a serious flaw. Ultimately that is a design tradeoff decision that is up to you.

 

[Hubbub]

FYI my experience with shielded RJ-45s is that they usually have two shield pins, so you might need to add a pin 10 to your RJ-45 model.

 

[sonolink]

If you connect CHA+(1) to CHA+(3), etc. in every "B" box you end up making a slew of unnecessary redundant junctions.

Of course, you're right. I think I need some sleep
I'll stick to this schematic for Box B then.

Note that this is still effectively paralleling pairs in the cable for less loss. In one case you get lots of junctions that feed 2 loads per box from 2 pairs, in the other case per box you need no junctions and feed 1 load from 1 pair, just 2 times over. For 20 boxes either approach puts at most 20 loads on one pair, same as 40 loads on 2 pair. I vote for the simpler solution with fewer loops.

Basically I understand that the best idea is to load the bus directly from Box A and avoid redundancies further down the line. So, in Box A I'd have to feed from the amp output, ChA+ to pins 3 and 4, ChA- to pins 5 and 6, ChB+ to pins 1 and 7 and ChB- to pins 2 and 8. Right?

Thank you very much for your help
Cheers
Sono

 

[Hubbub]

Of course, you're right. I think I need some sleep
I'll stick to this schematic for Box B then.

View attachment 117706



Basically I understand that the best idea is to load the bus directly from Box A and avoid redundancies further down the line. So, in Box A I'd have to feed from the amp output, ChA+ to pins 3 and 4, ChA- to pins 5 and 6, ChB+ to pins 1 and 7 and ChB- to pins 2 and 8. Right?

Thank you very much for your help
Cheers
Sono

I think it is right, just make sure your switches are oriented correctly so both operate with A (similarly B) in the same direction for both. Get some sleep!