Microcontroller based switchbox

[bperboy]

I know that AMB has his Arduino-based system in the works, but I'm at a point in my EE education where I want to design something myself. I'm thinking of the following high-level scheme for a 4-input, 4-output audio switchbox.

TI MSP430 microcontroller

LCD display

pushbuttons

ADG1604 analog muxes

My question to the other EEs on Head-Case is this: most all of the electronic switchers I've seen so far in the DIY realm use relays. Is there any major downside to using analog muxes that I haven't realized yet? Any reason that I shouldn't pursue this technique? I was wanting to try something that I haven't seen already, and this seems like a very do-able project for me.

[Emooze]

From what I know, relays offer better sound quality as far as distortion goes compared to similarly priced analog muxes. Operating a FET in the triode region can't quite achieve the linearity nor the bandwidth of a decent mechanical switch.

Edited May 16, 2011 by Emooze

[bperboy]

Well, I've been checking out the characteristics of the ADG1604 on the datasheet, and THD is 0.007% over the audible range. Also, on resistance is less than 1ohm over the entire input voltage range for +-5V supply. The frequency response of the circuit is completely flat up to 500kHz, far beyond the audible range. At this point, I'm very dubious of any audible difference between a well-implemented solid-state solution vs a relay solution. Any other responses?

[kevin gilmore]

isolation between inputs may be an issue. feedthru does occur and for

some of the switches, protection diodes will short the input to near

ground with the power off.

[bperboy]

But if power is applied to the switches all the time, that shouldn't be an issue, correct? What would I look for on the datasheets for isolation properties? And how would that work if not all the connected devices were powered up at the same time?

[bperboy]

isolation between inputs may be an issue. feedthru does occur and for

some of the switches, protection diodes will short the input to near

ground with the power off.

Kevin, can you describe why feedthru occurs? I have the system breadboarded, and I do experience some feedthru. Say, if I've got an input playing on input1, but the firmware is using input2, I can still hear a very faint input1 coming through.

[bperboy]

See this link for my project.

Edited July 28, 2011 by bperboy

[The Monkey]

I will be keeping my eye on this. Would be nice to have an alternative to the Skipjack.

[swt61]

^

What ever happened to the group buy on TPA Darwin boards? I guess that didn't come to fruition?

[bperboy]

I will be keeping my eye on this. Would be nice to have an alternative to the Skipjack.

I am considering working up a PCB design for this, possibly in the fall. There are a few issues that I need to sort out first, notably the channel-to-channel crosstalk. Getting the audio performance of the system set is the most important priority. In terms of user interface, I'll definitely want to add an LCD display to any design I do.

EDIT: Whatever the final design ends up being, it certainly won't come close to the Skipjack's price

Edited July 29, 2011 by bperboy

[bperboy]

I have been searching the internet for suggestions on how to reduce the crosstalk effect on this system. It seems that if there is an active signal coming in that is not selected, it causes crosstalk. So I have seen an article that uses MOSFETS to ground any unused inputs. Does this seem like a simple and elegant solution?

[kevin gilmore]

The best solution for this is something texscan corp did about 25 years ago.

series mosfet switch to parallel mosfet switch to ground, to series mosfet switch.

Good for about 105 to 110 db when done this way.

[bperboy]

The best solution for this is something texscan corp did about 25 years ago. series mosfet switch to parallel mosfet switch to ground, to series mosfet switch. Good for about 105 to 110 db when done this way.

Can you give me a quick paint drawing? I'm having trouble visualizing..