DIY Switcher w/ Volume Control & Effects Loop

[MRC01]

With all this discussion a summary is worthwhile. That is, what are the limitations, or guidelines for when a passive switch-box with a stepped attenuator should out-perform an active preamp. Put differently, what can go wrong with a passive attenuator:

1. Don't add attenuation unless you need it. If a switch-box does the job, your design will be simpler and cheaper, and your impedance matching is simplified: just make sure downstream device input impedances are at least 10x higher than upstream device output impedances. Note: most typical solid state devices will have a ratio of more than 100:1, like a CD player with a 100 ohm output impedance to a headphone amp with a 30k input impedance = 300:1.

2. If you have attenuation, use a ladder / voltage divider with metal film resistors having the lowest impedance you can: 10k is typical but if all your upstream sources are < 100 ohms then 5k will be better.

3. Upstream source devices have low output impedance, ideally < 500 ohms or at least 10x lower than the attenuator input impedance.

4. Downstream devices (amps) have high input impedance, ideally > 25 kOhm or at least 10x higher than the attenuator output impedance.

5. Use short low-capacitance cables for all connections -- especially to downstream devices (amps).

6. Ensure your design individually grounds de-selected upstream devices, to prevent de-selected devices from driving each other or leaking into the selected source.

Anything else?

 

[saturnaal]

Wow, lots of discussion here. Where to start....

With respect to @restorer-john 's insights... Sure, I could probably find a commercially produced pre-amp that more or less does what I'm looking for, with reasonable performance. There's a few good reasons I'm not interested in this approach. I wrote out a long winded response, but ultimately it boils down to these three points.

1. There's no fun in it for me.
2. I wouldn't learn anything.
3. There are things I do not want to compromise on.

I think these pretty well exemplify a few of the core motivations for DIY projects everywhere. At the end of the day, some people prefer to eat out, some people prefer to cook.

Getting back on topic, love the summary of technical considerations there @MRC01 . All this input / output impedance discussion has demonstrated to me that I have some analog fundamentals to catch up on.

You've mentioned your own experience with DIY "preamps" a few times now, do you have any pictures or schematics from any of your efforts? Always nice to have some other points of reference.

With regards to other products, like the Nobsound / JDS / Schiit audio switchers & volume controls, how is it that they address the issue of FR roll-off across the volume range? If I put one of these into my signal path and measured, would I find the same problems discussed here? Why have I not seen this discussed anywhere else where these products are talked about?

 

[restorer-john]

but ultimately it boils down to these three points.

1. There's no fun in it for me.
2. I wouldn't learn anything.
3. There are things I do not want to compromise on.

Fair enough. I totally agree and respect that position. I too was there once and built everything from scratch because I didn't want to compromise.

I reckon there's a good opportunity to knock up a very good preamplifier with customisable numbers of inputs, outputs, loops and switchable gain/passive etc as a fun project. Someone here is probably super good with PCB layout (I'm not).

 

[MRC01]

... You've mentioned your own experience with DIY "preamps" a few times now, do you have any pictures or schematics from any of your efforts? Always nice to have some other points of reference.

Here's the one I built years ago: http://mclements.net/Mike/audio/mrc-PassivePre.html
I used it for about 10 years then sold it when I got rid of all my vinyl & turntable, and replaced it with a DAC+preamp. I think I packaged the circuit diagram with it when I sold it. But knowing it was a Welborne Labs Reveille Passive, you may be able to find that online. I can send you the spreadsheet I used to calculate the resistor values for the attenuators. This site won't let me post it as an attachment, so PM me and I can email it to you.

 

[Blumlein 88]

One thing to consider is what size steps in a stepped attenuator. I tried 2 db steps and it wasn't as much too large as you might think, but it was too big a step. I tried 1.5 db steps and those are almost close enough or for some people are close enough. Even for me only rarely did I find myself wanting something in between two settings. JND (just noticieable difference) for levels is 1.2 to 1.4 db depending upon frequency a little bit. I built some with 1 db steps I used for quite sometime. As afar as I am concerned 1 db steps are effectively equivalent to continuously variable pots. You don't need anything closer just for controlling music volume. Also having made a few for others if resistor values result in some steps being .9 db and some being more like 1.1 db don't worry about it or try to come up with some weird value resistors to reach perfection in step size. In use you'll never notice the difference.

 

[restorer-john]

In use you'll never notice the difference.

Trouble is, you'll know the steps weren't exactly even...

 

[Blumlein 88]

Trouble is, you'll know the steps weren't exactly even...

Notice I said building them for others. I just told them it was 1 db steps.

I eventually went with the sort of a digital ladder DAC arrangement where I had lots of small steps. I knew the octal-code for volume if you will, and did this mostly for exact volume matching. It was what I was using for those series amplifier tests.

 

[617]

You guys have it easy. I am looking for the same but with 6-8 channels and remote control.

Funnily enough, this and the cables are what prevent me from going to an active set up again. Even 4 channels is a pain in the ass, and 8 is really preferable. I should ask Romy the Cat what he does.

Last time I did this it was a rat's nest of wires and getting 8 conductor cables connected to speakons was not fun.

 

[restorer-john]

Last time I did this it was a rat's nest of wires and getting 8 conductor cables connected to speakons was not fun.

Not what I would call a fun day with cables.

 

[saturnaal]

I like the idea of 1dB steps, provided I can find a part with sufficient steps to give me the range I need. In practice, it seems to me that -40dB is the maximum amount of attenuation I can see myself using. Currently I have my E90X set so that with my DAC at -0.0dB, the volume is nearly intolerable for any amount of time, any sound. I usually listen in the -16dB to -28dB range, but I've never turned it down beyond -35dB.

I've also been playing with the idea of using a rotary encoder driving a bunch of relays to provide volume control with a sufficient number of steps and precision. I don't know if this would actually be any better, and it adds the necessity of a power supply and control circuitry. But, I'd be willing to accept some extra complexity in the design if it aligns better with my goals.

EDIT: I'll add, the goal here for it being "passive" isn't necessarily to keep any and all active components out of the design. It's just to keep the signal path untouched except for volume control (and whatever the device connected between the loop terminals might do when the loop is enabled).

 

[MRC01]

One thing to consider is what size steps in a stepped attenuator. I tried 2 db steps and it wasn't as much too large as you might think, but it was too big a step. I tried 1.5 db steps and those are almost close enough or for some people are close enough. ...

If I were building another one, I'd make them non-equal. Put the top few steps 1 dB apart, then gradually increase the step spacing toward the lower steps. The bottom step can be -36 dB since I don't use more than that. This could be done with Goldpoint V24C rotary knobs which aren't too pricey. You can buy them pre-loaded but then you lose the fun of picking your own resistors and soldering them into place.

I can hear about a 0.5 dB difference in overall level, but I don't need the volume steps to be that close together. My current DAC-Preamp's volume knob changes the gain-feedback resistors in 64 positions with 0.3 dB to 0.5 dB steps but I don't need them that close together.

... db don't worry about it or try to come up with some weird value resistors to reach perfection in step size. In use you'll never notice the difference.

True that. Your resistors will only come in certain values so you have to play around with what pair combinations give the attenuation you want while providing a consistent 5k or 10k impedance to the upstream source. You won't be able to get perfect steps, but you can get them within about 0.25 dB of your target step size which is plenty close enough.

 

[restorer-john]

I've also been playing with the idea of using a rotary encoder driving a bunch of relays to provide volume control with a sufficient number of steps and precision. I don't know if this would actually be any better, and it adds the necessity of a power supply and control circuitry. But, I'd be willing to accept some extra complexity in the design if it aligns better with my goals.

Ah, it gets more complicated doesn't it?

The relay deal can give you 64 different levels with 6 relays (R2R).

Schiit do something like that in one of their products.

edit: they use 7 relays to to give 128 steps.

 

[MRC01]

...I eventually went with the sort of a digital ladder DAC arrangement where I had lots of small steps. I knew the octal-code for volume if you will, and did this mostly for exact volume matching.

... The relay deal can give you 64 different levels with 6 relays (R2R).
Schiit do something like that in one of their products. ...they use 7 relays to to give 128 steps.

When you do it like this, then most of the volume positions will be passing the signal through multiple resistors - not just 1. Does that raise the noise level? It wouldn't necessarily...
Consider 2 MF resistors having values of 1 and 2 kOhm. Do both in series have the same noise as a single 3 kOhm MF resistor? It seems they should.

 

[Blumlein 88]

One way I put one together for a friend was using 4 rotary switches. Two per channel. The lowest setting in the first rotary switch simply sent signal to the second one for an additional 23 steps. You could do this and have 0 to -46 db in 1 db steps if you want to do it that way.

 

[Blumlein 88]

When you do it like this, then most of the volume positions will be passing the signal through multiple resistors - not just 1. Does that raise the noise level? It wouldn't necessarily...
Consider 2 MF resistors having values of 1 and 2 kOhm. Do both in series have the same noise as a single 3 kOhm MF resistor? It seems they should.

All that matters is the final combined resistance. So whatever you need for a voltage divider value it doesn't matter if one leg is a single 3k or a 1k+2K they both have the noise of a 3 k resistor. Alternatively paralleling them reduces noise.

 

[MRC01]

One way I put one together for a friend was using 4 rotary switches. Two per channel. The lowest setting in the first rotary switch simply sent signal to the second one for an additional 23 steps. You could do this and have 0 to -46 db if you want to do it that way.

Goldpoint also makes a 47-position switch.

 

[saturnaal]

Benchmark uses this approach in some of their products as well, and a brief search reveals a handful of DIY PCBs and designs floating around for this which would greatly simplify the process. Most of them rely on a fairly standard input voltage like 12v, and it would be pretty easy to find a power supply board to supply this.

Starting to think this might actually be a better, even easier approach than using a stepped attentuator. Even a fancy one like https://www.khozmo.com/ which is 48 steps with -60dB attentuation, combination of 1dB and 2dB steps.

 

[restorer-john]

I'd do it with an A/D converter with binary parallel outputs driving the relay resistor arrays via Tr buffers. One linear pot to control it all.

Something using the TLC-1550x if they are still around.

Simple, fast and could offer an instant mute/resume option too.

 

[Blumlein 88]

Benchmark uses this approach in some of their products as well, and a brief search reveals a handful of DIY PCBs and designs floating around for this which would greatly simplify the process. Most of them rely on a fairly standard input voltage like 12v, and it would be pretty easy to find a power supply board to supply this.

Starting to think this might actually be a better, even easier approach than using a stepped attentuator. Even a fancy one like https://www.khozmo.com/ which is 48 steps with -60dB attentuation, combination of 1dB and 2dB steps.

That looks like a good way to do it.

Funny thing is I worked out my idea of the best and purchased some 36 position NASA surplus switches of fantastic quality (just stumbled across them). All silver multi-contacts, and teflon board already spaced for filling in with normal size resistors. I planned on using all Vishay Metal foils. Then I had the idea of the R2R arrangement which need only 7 resistors for an abundance of small steps and never used those NASA switches.

 

[saturnaal]

I'd do it with an A/D converter with binary parallel outputs driving the relay resistor arrays via Tr buffers. One linear pot to control it all.

An interesting approach, but I'm not quite ready to tackle PCB layout for A/D & D/A chips. If it can't match or beat the performance of the Topping D50s, I don't want to put it between my DAC and transducers. I admit though, I have long term fantasies of building my own AK4499EQ implementation, or whatever the "top of the line" is if I ever get to that point.

Alas, neither my own skills or home electronics laboratory are quite up to the task yet.