DIY Switcher w/ Volume Control & Effects Loop

[saturnaal]

Really appreciating all the feedback, advice, and discussion everyone. I've got no real experience with analog electronics, my expertise is quite a few layers of abstraction away from this.

This could mean variable FR with volume position.

Oof, exactly the kind of thing I want to avoid. I read through the thread linked below by @Wombat and I'm still working on understanding these interactions.

Look what happens when the attenuator is all the way to 0.
As it is switched now you actually SHORT input 1 as wel as loop input 1

It could be solved by adding a 10k resistor between wiper and signal wiring.

Thanks for bringing my attention to this. If I understand correctly, this issue should be corrected in the latest revision.

KISS is the go. Not even bridged.

T-Pad

I'll dig into this a bit when I have some time and see if I can wrap my mind around it.



Some updates...

First, a revised overview that should illustrate a little more clearly than my clumsy original attempt.

In my case, the tube amp has active outputs. As @solderdude mentioned, I don't want to clip the inputs on anything connected to the switchers outputs when the loop toggle is on. Fortunately the tube amp has a volume control that I'll be able to use to set the level so I don't anticipate an issue.

None of the other amplifiers (electrostatic for the Koss, solid state headphone amp, or speaker amp) have pre-outs in my case. Only the electrostatic headphone amp and the solid state headphone amp have volume controls, but I want to "set-and-forget" those and rely on the switchers volume control.



Next, I took some a couple of the tips here into consideration and re-worked my diagram to have inputs on the left, and changed the pot wiring to avoid the issues mentioned by multiple users. When I have some more time I'm going to try and produce a version using a T-Pad attenuator.

What would be the use case or reasons for using a bridged T-pad vs. a non bridged T-pad?

 

[solderdude]

This schematic too has a flaw.

When you follow the signal when there is no loop via the tube amp then it is obvious the signal goes into the output of the tube amplifier.
Assuming is has a low output resistance the RCA input signal is 'pulled down' by the low output R of the tube amp.

Solution 1: using the same switch is to connect the loop-out of the tube amp directly to the input selector (so always in parallel) and switching the hot end of the volume control between the loop in and input selector.
I suspect it is perfectly safe to leave the tube amp always connected, even when powered off I don't expect this to be a problem.

Solution 2: the loop switch is changed to 4PCO (so 4 chance-over contacts) where the volume control input is switched between the tube amp output and input switch.

 

[saturnaal]

This schematic too has a flaw.

When you follow the signal when there is no loop via the tube amp then it is obvious the signal goes into the output of the tube amplifier.
Assuming is has a low output resistance the RCA input signal is 'pulled down' by the low output R of the tube amp.

Interesting, I realized when I was laying this out that with the loop switch in the "off/bypass" position the signal would be connected to the amp outs, but I wasn't sure if this would be an issue.

Solution 2: the loop switch is changed to 4PCO (so 4 chance-over contacts) where the volume control input is switched between the tube amp output and input switch.

I like this solution, seems clean to me. Unfortunately, there doesn't appear to be a 4PCO switch symbol in the KiCad / Eeschema symbol library. I'll have to find an add-on or spend some time figuring out how the symbol editor works to add it myself.

 

[solderdude]

Interesting, I realized when I was laying this out that with the loop switch in the "off/bypass" position the signal would be connected to the amp outs, but I wasn't sure if this would be an issue.

The input connected isn't an issue (in most cases) but the output is usually low resistance so would 'load' the sources.

 

[BDWoody]

Really appreciating all the feedback, advice, and discussion everyone. I've got no real experience with analog electronics, my expertise is quite a few layers of abstraction away from this.

I'm enjoying reading this... Can't help but think you are probably already ahead of JK...

Oops...too soon....

Cheers...I am learning along with you.

 

[MRC01]

Many years ago I built a passive preamp based on Ron Welborne's Reveille kit. It had 4 selectable inputs, a single output, a tape loop, and a pair of mono 24-position Goldpoint stepped attenuators that I filled with Dale 1% metal film resistors selected so the volume clicks were 2 dB apart and each position provided a 10 kOhm load to the selected source device. It's similar to what you're doing here so if you can find the plans for this kit online, it might give you some ideas. I changed a couple of things in the original design, both related to grounding.

First, the kit connected the outputs of all de-selected sources together. If they're playing anything, each "drives" the others outputs and even its L and R channels drive each other. That didn't seem right to me. My first solution was to leave de-selected sources "hang" not connected to anything. That seemed simple and clean, but it allowed a very low level of a deselected source to leak into the selected input. That is: you hear the selected input, and very quiet almost in the noise yet audible, you hear any de-selected sources that were playing. How this could happen when the signals were never connected, was a mystery to me. The only explanation I could imagine was that the voltage from de-selected sources that were playing, was radiating inside the chassis like a radio and getting picked up by the selected source. To fix that, my next solution was wiring each de-selected source independently, through its own 500 or 1k resistor to ground. The idea was to select a resistor that was high enough to avoid damaging the source device, but low enough to absorb the energy. Whatever the cause of this leak-through - radio effect or something else - this worked, eliminating leak-through in the output.

Second, the kit treated both + and - on each channel the same, wiring each straight through the attenuator resistors to the output. Sometimes I heard a low-level hum in the output, sounded like a ground loop. I found by experimenting, that connecting the - of one channel (I think it was the left, but it shouldn't matter which as long as you pick just one) to the frame ground (the thick metal chassis) eliminated this hum. I don't know exactly why, but this asymmetric grounding eliminated the hum. Ideally, you could make this optional with a switch labeled something like "ground loop".

 

[Wombat]

You could add the loop to this:

https://www.amazon.com/Nobsound-Bal...b89bdd7464a9f3ea1700ee0ae069ab&language=en_US

 

[restorer-john]

I'd just buy a proper preamplifier. Done. Easy. You know, those things that are specifically designed to buffer incoming sources, have plenty of loops for processing (use the switchable rec-outs to send sources wherever you want independently of others), have switchable outputs, mostly also come with a bonus headphone amplifier onboard, maybe remote control (ooh, how modern) have buffer stages that can drive any length cable/impedance and don't have wild frequency response variations depending on the the position of the volume pot?

Many thousands of models were made from the 1950s to the 1990s. You may have heard of this long forgotten era.

Reinventing the wheel if ever I saw it, except it's like going from this wheel:

To this wheel:

Progress. Bah! Who needs it.

 

[MRC01]

I'd just buy a proper preamplifier. Done. Easy. You know, those things that are specifically designed to buffer incoming sources, have plenty of loops for processing (use the switchable rec-outs to send sources wherever you want independently of others), have switchable outputs, mostly also come with a bonus headphone amplifier onboard, maybe remote control (ooh, how modern) have buffer stages that can drive any length cable/impedance and don't have wild frequency response variations depending on the the position of the volume pot? ...

There is much to this pragmatic advice.

However, no active preamp is as transparent as a stepped attenuator which will have nothing more than a metal film resistor in the signal path. It should introduce no measurable distortion and worst-case noise should be -120 dB or less (thermal noise of metal film resistor). If your upstream sources have low output impedances and your downstream devices have high input impedances (which covers virtually all normal solid state devices) then you won't have frequency response variations with volume.

Regarding active preamps: remember that specs can be misleading. For example SINAD is usually measured at full volume which we never use. As you turn down the volume, the SINAD drops with it. The 50 mV SINAD that Amir measures is a better representation of what you get when actually listening. And the best preamps that measure 120 dB on the first, often drop into the 80s on the second. So in comparison, at actual listening levels the passive gives you a SINAD of > 120 dB where the active gives you around 90, and that's if you get one of the best. The difference should at least be measurable, if not audible.

No doubt, the stepped attenuator switchbox certainly is less pragmatic. It's for those who are willing to do the extra work and find satisfaction in designing, building, and listening to it.

 

[restorer-john]

If your upstream sources have low output impedances and your downstream devices have high input impedances (which covers virtually all normal solid state devices) then you won't have frequency response variations with volume.

You will have frequency response variations. Most normal solid state commercial amplifiers have 10K-50K input impedance and that impedance varies with frequency. Many of the 'new' Class D and pared down basic amps-in-a-box have very low input impedances. The passive pot you are likely to us will be 10K-50K in itself, and most of the time, the volume pot will be towards the lower end with a consequent source impedance more than enough to throw some wild fluctuations at the top end.

I suggest you grab your favourite passive front end and put it between your source and your power amp. Run some 20-20K FR plots at various volume pot settings and varying levels and report back with your findings. Funny, I was just doing that yesterday chasing a design flaw which coincidentally involved a direct (pot) vs preamplifier stage in a vintage, but TOTL amplifier.

Regarding active preamps: remember that specs can be misleading. For example SINAD is usually measured at full volume which we never use. As you turn down the volume, the SINAD drops with it.

And to your noise comment. Amplifiers are measured for S/N and residual noise (uV) with a shorted input or very low impedance, not with a high impedance source via a pot. S/N is severely compromised in most amplifiers with a high impedance source. Not only that, many preamplifiers maintain their S/N as the volume is reduced. Do not tar all preamplifiers with the same brush. Many techniques (multiple ganged pots in two stages of the preamplifer etc) for doing so were employed when people actually knew how to design proper preamplifiers; ones that happily coped with all manners of input impedance, signal levels and a plethora of multiple connected devices.

 

[Blumlein 88]

I messed with lots of passives. Nothing wrong with DIY rather than off the rack. My final version used switches similar to how ladder DACs were arranged. It let me have many, many small steps, keep impedances in line, use expensive Vishay resistors and only need a few of them. Basically a 7 bit arrangement with 128 steps.

The worst case is going to be - 6db. If you do your own pairs or even switched multiples you can keep everything working pretty well though it gets a bit more complex. Using a simple L-pad style, at -6 db, you could have a 5 k ohm input impedance, and a 1250 ohm output impedance. Depends upon your source and your amp if that works or not. Every other position could have higher input impedance and lower output impedance.

Now days if I were trying this, I'd maybe build a simple switch for the HT pass through. Everything else would feed a quality ADC, and the digital result would go to a nice DAC. Control volume in software. I believe you'd get excellent results this way, and while perhaps a painstakingly designed passive made specifically to optimize the gear you have would match or slightly exceed it, it is a pain in the rear eventually. To get really clean passive performance I always had to optimize resistor values for the actual bits of gear in use.

 

[restorer-john]

Everything else would feed a quality ADC, and the digital result would go to a nice DAC. Control volume in software. I believe you'd get excellent results this way, and while perhaps a painstakingly designed passive made specifically to optimize the gear you have would match or slightly exceed it, it is a pain in the rear eventually.

Do you really think there would be an improvement over existing analog preamplification where every single input is individually buffered (unity gain), sent to low value pot and buffer stage and then either sent straight to the second pot gang or via the tone stages and final output buffer? I've got preamplifiers that do that in direct and have rated residuals <1.5uV and THD <0.001% S/N >106dB. I can't near that figure with my test gear.

Surely the A/D is going to let the whole A/D-digital volume-D/A deal down?

 

[Blumlein 88]

Do you really think there would be an improvement over existing analog preamplification where every single input is individually buffered (unity gain), sent to low value pot and buffer stage and then either sent straight to the second pot gang or via the tone stages and final output buffer? I've got preamplifiers that do that in direct and have rated residuals <1.5uV and THD <0.001% S/N >106dB. I can't near that figure with my test gear.

Surely the A/D is going to let the whole A/D-digital volume-D/A deal down?

I'm not sure either way. It could be. I'm thinking in regards to the passive volume control. Plus the routing the OP wishes to achieve.

I might suggest a recording interface in place of the DAC in his diagrams. Then all his sources are digital anyway except for one. If the OP didn't want the option of using his tube headphone unit as a preamp for the other two then he'd not have much difficulty.

Maybe he just needs a plain switch for that purpose. In one position the headphone unit is the preamp (with digital volume control via the DAC) or in position two he bypasses the tube headphone unit and still has digital volume control via the DAC.

 

[MRC01]

You will have frequency response variations. ...
I suggest you grab your favourite passive front end and put it between your source and your power amp. Run some 20-20K FR plots at various volume pot settings and varying levels and report back with your findings.

I did that with the one I built. FR was flat at all volume settings. For flat FR you want at least a 10:1 ratio to the worst case highest output impedance which is 2500 Ohms for a 10 kOhm stepped attenuator. So make sure the downstream device input impedance is at least 25 kOhm. IME traditional solid state amps are typically 30 - 50 kOhm which should be no problem.

... many preamplifiers maintain their S/N as the volume is reduced. Do not tar all preamplifiers with the same brush. ...

Perhaps, though none of the ones Amir has measured here do. The best seem to have about a 30 dB drop from full output. The highest 50 mV SINAD measured here so far is in the mid 90s which is about 25 dB worse than the highest noise from a 10k stepped attenuator.

BTW I agree an active preamp is more pragmatic. Plug and play. But if he has upstream devices with low output impedances (< 100 Ohm) and downstream devices with high input impedances (> 25 kOhm), which is typical of most solid state devices, the passive attenuator should be more transparent than any active preamp.

 

[Wombat]

Two of these switchers(very low cost) back-to-back with the loop/volume inserted between them.

 

[restorer-john]

Two of these switchers(very low cost) back-to-back with the loop/volume inserted between them.

QC could be a bit hit and miss on those- look at the pushed-in video input RCA on #1, the left RCA on #3 and the bashed-in left on #4...

 

[Wombat]

QC could be a bit hit and miss on those- look at the pushed-in video input RCA on #1, the left RCA on #3 and the bashed-in left on #4...

Let me add, or similar. Lots of versions around. Any self respecting 'audiophile' couldn't get friend cred. with these units, though.

I would use a good preamp but their goes the cred. thing again.

 

[restorer-john]

I would use a good preamp but their goes the cred. thing again.

I'll get a t-shirt printed for us: "Nothing to gain with being passive"

 

[Blumlein 88]

I did that with the one I built. FR was flat at all volume settings. For flat FR you want at least a 10:1 ratio to the worst case highest output impedance which is 2500 Ohms for a 10 kOhm stepped attenuator. So make sure the downstream device input impedance is at least 25 kOhm. IME traditional solid state amps are typically 30 - 50 kOhm which should be no problem.


Perhaps, though none of the ones Amir has measured here do. The best seem to have about a 30 dB drop from full output. The highest 50 mV SINAD measured here so far is in the mid 90s which is about 25 dB worse than the highest noise from a 10k stepped attenuator.

BTW I agree an active preamp is more pragmatic. Plug and play. But if he has upstream devices with low output impedances (< 100 Ohm) and downstream devices with high input impedances (> 25 kOhm), which is typical of most solid state devices, the passive attenuator should be more transparent than any active preamp.

One quibble with your post. The 10 to 1 ratio is a good guideline for some purposes. It isn't about upper frequency response however. Output impedance and parallel capacitance create a 1st order low pass filter at some frequency. So an output impedance of 2500 ohms is going to curtail upper FR no matter the input impedance of the next stage. If you have 2500 ohms output impedance with a short low capacitance cable it won't be a problem. With a long or long high capacitance cable it will roll off the treble audibly. The input impedance of the next stage will have negligible effect making it worse if it follows the 10 to 1 rule, but that won't prevent the roll off.

With a rather low capacitance cable like RG6 you probably will be alright to 3 meters with 2500 ohms. If you used some audiophile cables a meter might be pushing it. Of course that is a worst case. Always good to keep the output side cables very short with passives.

 

[MRC01]

... If you have 2500 ohms output impedance with a short low capacitance cable it won't be a problem. With a long or long high capacitance cable it will roll off the treble audibly. The input impedance of the next stage will have negligible effect making it worse if it follows the 10 to 1 rule, but that won't prevent the roll off.
...

Yes, that's true. I was using Blue Jeans LC-1 cable (12.2 pF / ft) 3' long from the attenuator to the power amp for precisely that reason.