We haven’t done a proper postmortum on the Leannan Sidhe album project yet, partly because of time, and partly because it’s not through mastering. We’re pretty much done, though; principal recording is over, and Alec has done all or nearly all the mixing; I just ran a set of 50 pre-mastered discs for Shanti this weekend, to take to Arkansas for a festival.
Despite that, I have made a few notes and workflow changes. For one thing, I’ve always known regions – which are windows onto recorded tracks, like snippets of tape, but adjustable – had names, could be renamed, and those names were somehow inherited. But it’s only now I’ve actually started using that, naming them by take numbers and such. It’s made compiling composite takes much less confusing.
But hang on, this is a DIY post, promise
But we did have one problem that cost us seven or eight takes during the process.
We do all the recording in isolation headphones. The performer hears the playback in them, as well as what they’re doing, and none of that gets into the mics. Every so often we stop and listen to what we’ve done on the studio playback speakers. (The “monitors.”) Then we’ll go back to recording.
And all this works great… unless, of course, you forget to turn the monitors off, which we did. Several times. In isolation headphones, we can’t hear the speakers, so we don’t necessarily catch it.
Meanwhile, speaker selection has been a little weird – the amp has “A” and “B” buttons, and “B” went to a three-way box, because I had four total speaker sets. And there aren’t any indicator lights.
Or, rather, there weren’t any indicator lights.
My first idea was “can I make something that’ll flash when there’s speaker activity?” I wasn’t planning on changing the speaker setup or any of that; I just wanted to piggyback onto what I already had. I came up with a simple solution – literally just a resistor and a diode, pulling off 0.8% of the output circuit to cause lights to flicker at loud volumes. It’s not everything I’d like but it’s as close as I’m getting at the moment.
Separately, I’d decided I didn’t like the “amplifier on” light I’d jury-rigged up before, with those little green rectangle nightlights on an extension cord, and figured I could put both of those into one box with like four cords running to it. Seemed a bit excessive, but worth doing anyway.
So I went looking for an experimenter’s box to hold all that. I wasn’t really finding a good-sized one, when I stumbled across an old Radio Shack audio-video selector box.
I found a picture of the same model, elsewhere online:
Switches audio left, audio right, and composite video
The left side of the box is just blank space. The box is that wide only so there’s enough room on back for all the plugs. While I couldn’t find a “before” picture of the insides online to show you, there’s just not a whole lot inside, either; it’s four switches on a single circuit board.
And that’s when I decided I could change the speaker setup so all four speakers went through one box. No more A/B and external switcher: one box. And that box could have a couple of indicator lights on it for signal. Perfect!
Except this is one of those projects that turned into kind of a science-related mimetic disorder episode. I had an idea, I came up with a circuit, I figured out I could make a way simpler version… then I fiddled with that and improved it again… then I realised I could add another bit of functionality over here… then I realised I could add another piece of functionality… until I ran out of parts. Then I got more parts and realised I could add something else… and I didn’t take any pictures for quite a while. But I have a couple.
In this photo, I’ve already done some work; those resistor pairs are replacements for lower-watt resistors of the same value. I didn’t really need to do that; they’re just buffers to prevent capacitance buildup in disconnected pins, and are out of the signal path.
It’s right about here that I really realised this wasn’t an audio switch, it was an audio-video switch. And I wasn’t using the video. And if I could isolate the video portion of the box… I could run a few volts of power indicator on those connections.
Which can also mean lines for speaker select indicators, because these switches are switching both audio and video circuits….
And that’s how these cascades get started. Follow along, eh?
Now, before you try to do this, it’s really, really important that you know you’ve isolated all parts of the video circuit from everything else. Including the ground line. Even if you’re only running three volts of DC on it, the speaker signal is AC, and speaker circuitry does not like DC. You can and will damage your speakers. I had to scrape away part of the circuit board to isolate the ground half of the video circuit; you probably will too!
Now by the time I’d started taking pictures, I’d already done a fair bit of work. Here’s the faceplate, with holes drilled for power, left signal, right signal, and speaker selection lights:
I thought I was done here, but that was a lie; I was just out of parts.
Here’s the faceplate, back on the board, with most of the LED lamps mounted. (See again: out of parts.) Here, I’ve also isolated the video circuit from all the others, while still leaving it connected to the switches. That resistor on the lower right is attached to what had been the video ground, and it’s the buffer resistor for the LED power circuit:
Now, with my multimetre, I’d figured out that the lower two posts on the left side (as seen in these photos) were the posts for completing the video circuit on each switch. Now they’re power switches, so let’s start connecting them to the LEDs!
Now here, I’ve built a return rail. Each switch is connected to an LED input, then all the LED outputs are connected together on that rail, and the rail connects back to the ballast resistor. If you’ve seen this circuit before done in isolated wires like this before, well, welcome to vacuum tube amplifier construction, circa 1956.
You might notice that we don’t have the power light attached, yet – because, again, out of parts.
Next up was to put in my tiny signal detectors. It’s nothing more than a high-value resistor and an LED. Remember: LEDs are diodes. You don’t need two separate parts (diode and lamp) if you have enough ballast to let the LED itself do the diode job:
The ballast resistor and LED are connected in series, picking off the signal line from the inputs, and going back to the grounds. I wired left and right grounds in parallel here… really just for symmetry’s sake. They’re the same ground. But as long as the signal side of the diode is isolated, you won’t get any signal intermix.
Note that I’m picking up the audio signal from the switches, like with the DC power. This is just for ease of access – there wasn’t a good place to pick up the signal anywhere else. Those were the input pins for each channel, so always on.
Now, overnight, that LED pickup bothered me. I couldn’t complete the box without the extra parts – this is when I was tweeting WHY ISN’T THERE A 24-HOUR RADIO SHACK IN THIS TOWN? SCIENCE DOES NOT CARE ABOUT YOUR NEED FOR SLEEP! – so the design was still foreground in my brain. And while it isn’t a big signal hit, every time that LED lights up, I’m picking off a little bit of output signal.
So I decided that I needed to add a “defeat” switch, to be able to have those LEDs out of the loop for ordinary listening.
I did that by taking the two returns from the signal-detect LEDs – the leads heading towards ground – and running them through an on/off switch. Easy-peasy. The most difficult part was a switch that would fit – I wanted to use a rocker switch I have several of, but it simply wouldn’t physically fit on the faceplate. Instead, I went with a push on/push off.
As soon as that ground is lifted, the circuit is isolated, and we’re all done.
To be honest, I don’t really hear any difference one way or the other, but I still feel better about having included it. ^_^
And, separately, I ran leads for the power indicator light. The way I wired it, it dims when a speaker is enabled. That was an accident, but I like it.
Finally – this piece of Radio Shack kit was built for line level signals, not speaker level signals, and I wanted more metal on the rails for the thicker signal. I’ve already been using a piece of kit a lot like this, but it bothered me more here. I ran that on the opposite side instead:
I was a little nervous about having the parallel rails – a tiny one on the top of the circuit board, a proper one underneath – but I did the time math and the difference is going to be inaudible to anyone. We’re talking less than microseconds.
So here’s what it all looks like before being put back in the case, with no power:
And here it is all connected up. One video connector is used for +3V input. The other video connects are filled with hot glue, to make sure nothing ever accidentally gets plugged in there. The left amber light is amp-has-power. The two red LEDs (which are brighter and less red in this photo than in real life) indicate strong speaker signal. The green LEDs indicate which speaker is selected. It is possible to have none selected, which is nice:
So. The old speaker-selector box is pulled; this replaces it. The old blue-rectangle amp-is-on power indicator is gone; this replaces that, too. The “A” and “B” switches on the amp can mostly be ignored. There are now lights to indicate which speakers are on, and there are loud-signal lamps which flicker as an additional layer of warning.
Hopefully, now, we’ll finally not have to worry quite so much about accidentally leaving the monitors on.