Switch analog audio signals

[obiwanjacobi]

Hi,

I'm looking for a way to switch analog audio signals. Now I know about the 4016 and 4066 chips but I was wondering if there was a better/other solution.

I'll give some context on what I'm trying to build. I want to change the order of several (4 or more) audio processing circuits (Guitar Effects). I could drop in a bunch of these analog switches but that didn't strike me as a very optimal solution. I also thought of using (miniature) relays, but that adds up real quickly since I probably have to construct a matrix. Everything is digitally controlled with an MCU.

Anyway, love to hear what you think would be a good solution.

Thank you,
Marc

[cybergibbons]

What's wrong with those?the 4066 is well suited. There are probably more expensive options with smoother frequency response.

[TerminalJack505]

Google "audio crosspoint switch" and find a part that meets your needs.

[obiwanjacobi]

FYI: Found some interesting reading material for switching audio.

Self on Audio (Chapter 12)

Hope it helps.

[free_electron]

Good analog switches are intersils ( actually siliconix , but they got bought by intersil ) DG40x series.

I have used those to make programmable gain amplifiers and tunable filters for ADSL signals. They work perfectly up to a few MHz. should be perfect for audio.
and they are not more expensive than your garden variety 4051/4052/4053/4066.

if you want to eliminate problems like crosstalk it may be required to not send a signal through teh same die. channel to channel isolation is not so good.
fortunately there is the '4066' in sot23 package : 74v1g66 is a single SPST-NO switch in 5 pin SOT23. ideal to plunk them exactly where you need them.

[kiyotewolf]

Might seem a bit archaic, but I remember hearing about someone switching audio off and on, using CDS cells opto-isolated to light sources.

Was a noiseless way of turning off and on audio sources.



~Paul

[T4P]

Might seem a bit archaic, but I remember hearing about someone switching audio off and on, using CDS cells opto-isolated to light sources.

Was a noiseless way of turning off and on audio sources.



~Paul

Yes i did come across it during my time in diystompboxes

[obiwanjacobi]

CDS == LDR?

I have found some LED/LDR opto's but they were very pricey.
And I was also concerned with the current you have to supply the LEDs for fully driving the LDRs.
Considering that I need 20-30 of these 'digi pots' it adds up really quickly (both current and price  )

What I really liked about an opto-isolated solution is the isolation between the digital circuit and the audio circuit.

[SeanB]

The most expensive item if you DIY is the CDS cells themselves. A high efficiency red LED will drive it down to sub 100R resistance, provided the spot from the LED covers the whole active area of the cell. The biggest problem is making a truly light tight enclosure, as it is hard to get 35mm film canisters any more, which were the perfect container to use. You probably will have to use black rigid pvc conduit, and cut short lengths and use end caps. LED in one cap and LDR in the other.

[T4P]

The most expensive item if you DIY is the CDS cells themselves. A high efficiency red LED will drive it down to sub 100R resistance, provided the spot from the LED covers the whole active area of the cell. The biggest problem is making a truly light tight enclosure, as it is hard to get 35mm film canisters any more, which were the perfect container to use. You probably will have to use black rigid pvc conduit, and cut short lengths and use end caps. LED in one cap and LDR in the other.

You can buck the LED and LDR together (as in place them exactly on each other) and heatshrink them OR you can pot them (how do i put it ... remember the acrylic headphone amp thing ? in the black version or as i mean like those epoxied IC's in cheap calculators and meters)

[obiwanjacobi]

Yeah, heat shrinking is something I considered too. But the LDR's I foudn so far do not go down to 100 ohms (or so). Most datasheets mention something like 27k at 10 lux.

Will a good LED produce more than 10 lux and will the LDR go down further?

BTW: I considered using led/ldr pairs for replacing manual pots and not switching audio signals (thats why I mentioned it in my previous post).
I am researching both at the same time, so I mix them up sometimes ;-)

[kiyotewolf]

What about using another resistor in parallel with the CDS?

Or something like that?

- is unfinished.

Felt like posting it anyways.

[obiwanjacobi]

Yes, putting a resistor in parallel is what I've seen others do too. I've just ordered a couple of LEDs and LDRs so I can experiment with them.

Thanks.

[Zero999]

Now I know about the 4016 and 4066 chips but I was wondering if there was a better/other solution.

Do you know the 4016 and 4066 also come in 74HC versions which have a lower on resistance?

[obiwanjacobi]

Now I know about the 4016 and 4066 chips but I was wondering if there was a better/other solution.

Do you know the 4016 and 4066 also come in 74HC versions which have a lower on resistance?

Yes, I do know the are also available as a 74HC - but I hadn't looked at the datasheet in detail, so I didn't know the resistance was lower.

But these switches will probably go between some input and output buffers (op-amps) that will run on +/-12V (or similar). So the ability of the 4016/4066 to run at higher supply voltage is more important I guess. Those buffers will also compensate for the higher on-resistance.

[Zero999]

But these switches will probably go between some input and output buffers (op-amps) that will run on +/-12V (or similar). So the ability of the 4016/4066 to run at higher supply voltage is more important I guess. Those buffers will also compensate for the higher on-resistance.

Are the signals AC or DC?

If it's AC, what's the maximum peak to peak voltage?

If it's within the 10V allowed by the HC versions, you could run the switch off a lower voltage (possibly even single rail) and AC couple the signal through the switch. The ICs have internal diodes, connected to either supply rail, so higher voltage input transients at power on won't be a problem, as long as the current is limited.

[free_electron]

Take a look at the dg4xx family from intersil / siliconix.

[kiyotewolf]

Is there any way we can see some scrap of a schematic, nothing, totally complete, just a small daveCad version of the, main function this device is doing?

If you must, you can use block diagrams, to further reduce the meaning down to something broad enough, to be sketched in 30 seconds or less.

I just feel I could make some nice schematics of my ideas, if I had a better view of how many inputs and outputs and controls there were, and what kind of target electronics we want to use.

::::

Do we want all passive component mixing and filtering, so it runs on zero power?

 [or]

Do we want active mixing and filtering, so it requires either single or dual rail power to operate?

 [or]

Will it run off a USB plug, with the two data lines 1 Meg. Ohm'ed off to ground, so we only use power off the bus, to run this thing?

 [or]

Option D * something I haven't thought of yet.



~Paul!

Thanks OP.  I await your reply.

[obiwanjacobi]

I'm sorry, I didn't realize there was need for a schematic.

This is a rough draft of what I'm thinking of:


Left the input buffer, in the center the audio switch and on the right the output buffer. These are the building blocks to make an audio matrix.

The idea is to build a matrix of at least 16x16. The switch enable will be driven by a couple of 4094 (at +/- 7.5V) - serial to parallel shift regs. Those will be driven by a MAXIM 1488E that will take logical 5V at the input and can output +/-7.5V. The logic serial signals will be generated by an MCU. Currently thinking in terms of some sort of AVR but that's just because I'm unfamiliar with the PIC. Have not done a detailed analysis of what I need, because I do not consider it a problem. There will also be a Midi interface that is handled by the MCU to allow for patch recall on a Midi Program Change.

The audio matrix  is usable in itself (and I will probably build one for my own home studio) but the initial idea was for use in a guitar (multi) effect rack, where the signal flow through each effect can be programmed.

The audio switch perhaps requires some extra explanation:
- I use 4x 4066 switches to make one audio switch (so thats one chip per audio switch).
- The reason for this is cross talk: there is no cross-talk if you route the audio signals through different die's.
- I use 2 parallel 4066 switches to lower distortion.
- I connect the signal to ground in off-state (better off-state, less leakage)
- Leftover switch is used as an inverter.

Hope this clears things up a bit.

[obiwanjacobi]

I await your reply.

And I yours... 

[codeboy2k]

looking at your schematic, you still have 4066's there.  forget those.. drop them on the floor.
 Free_electon got it right , he said consider the DG4xx series from Vishay/Siliconix.

The DG4XX series have better on resistance (20 ohm vs 50-80 ohm), better frequency response (flat to 200 mhz, vs cd4066 40 mhz), better channel to channel crosstalk (-70, -90 db vs cd4066 @ -50db) better switching speed (110ns vs 125 ns), better OFF leakage current (0.1nA@25C, 20nA max, vs 0.1nA and 500nA max) , etc etc...

They are almost universally used in audio.

oh.. and forget the audioheads trying to tell you to use a led with a LDS/CDR and heat shrink tubing.  Nobody does that, really. and you certainly don't want to do it 32 times for a 16x16 stereo cross-point switch.

[obiwanjacobi]

Thanks for your remarks.

I understand that they (DG4XX) are better, but they're also almost 10 times more expensive (17 cents vs 1,67 euros).
And this is only a first prototype (to  be more exact: it only exists on paper - my PC that is  ) so there are probably other issues I would want to fix once I get my first real experience with it.

I wasn't considering the LED/LDR coupler as an audio switching solution, but I was considering it as a "digital pot" replacement 
(BTW 16x16 is not 32!  )

[codeboy2k]

(BTW 16x16 is not 32!  )

I knew that