Analog bilateral switches

[darkfenriz]

Hi all

Not sure if that should be under 'solid state', but anyway:
I am working on a circuit that limits the feedback loop gain if any abnormal state is detected and need an analog switch.

The switch is going to be is OFF state during normal work, so the OFF characteristics (leakage, feedthrough, capacitance) are highly critical over the whole audio range and possibly above, the ON characteristics, like linearity are of secondary importance, because it switches ON only if amplifier is any sort of overloaded anyway.
The first obvious choice seems to be old good 4016/4066 CMOS chips, but I would like to know opinions from these , who tried them in audio, possibly in an application similar to mine. Some questions:

Do they seem good enough?
Any tricks with serial connection of switches or similar?
Care to suggest something better?

I need <200 ohms ON resistance at +5V /-5V and logic drive.

Any suggestions welcome!

Regards,
Adam

[drspiff]

Let establish my bona fides... I know next to nothing about electronics, but if you only need a switch with low on-state resistance and very high off-state impedance, why not use a small MOSFET? Do you really need the bidirectional structure of the 4066?

Regards,
Dr. Spiff

[darkfenriz]

Yes, I need bidirectional.
The switches are made of mosfets anyway, I think it's N+P in series.
Appreciete bona fides.

[ilimzn]

There are a couple of versions of these (eg. 7066), as well as ones from the HC(T) logic family which are quite vastly improved WRT switch resistance but operate on lower voltages. There are also 4051/52/53 which in their HC(T) version work with signals within +-5V.

Main problem with these is Ron modulation with current (i.e. voltage drop across the switch). You really need to load them with a very high impedance to make the changes in Ron negligible against the load, in which case they can actually work quite well.
If you look up the datasheet for these, they use parallel N and P MOSFETs.

There is a number of analog multiplexer families, several companies used to do them with the starting letters DG, like DG211, but most are long obsolete...

[soekris]

Check www.maxim-ic.com and www.vishay.com, those probably have the best audio quality analog switches.

Another possibility is the www.fairchild.com H11F1 jfet solid state relay, which might be the best if the slow ton/toff is ok in your application.

If you can use the slow solid state relays and need very low capacitance then there is the Vishay LH1541 or Panasonic AQV225.

You can most of them at www.digikey.com

Soren

[soekris]

Wtf, where did that "hot girls, sexy photos" comes from ???

Ok, apperently maxim is at Analog, Linear, and Mixed-Signal Devices from Maxim/Dallas Semiconductor


Soren

[mlloyd1]

soekris:

i think you mean a link to maxim-ic.com (not maxim.com)

mlloyd1

[mlloyd1]

back to the original question, take a look at max352 or max312 to see if those might be of interest.
higher supply voltage (in addition to lightest possible loading) helps mitigate effects of Ron modulation.
the maxim parts can use rails up to +/-15v and use logic level switch control signals.
for all cmos analog switches i know about, you have to watch out for inputs exceeding the supply rails, often mitigated with diodes to the supply.

mlloyd1

[event horizon]

Originally Posted by darkfenriz

Yes, I need bidirectional.
The switches are made of mosfets anyway, I think it's N+P in series.

If you want much lower than 200 ohm on resistance simply use 2 x N channel mosfets. Simply connect the sources together & the gates & use the drains as the input & output. 5V from both gates to both sources (which are connected together anyway) should switch the thing easily. With the kind of currents you'd be dealing with even the smallest mosfets should give you an on resistance of less than 10 ohms (assuming you don't go for 500V devices ).

In fact as far as i'm concerned it's bye bye output relay in a power amplifier

[brig001]

Could you use one open in series, and one closed to ground? The closed one would short out anything that passed through the first. If you used the 4016/4066, you have four switches, so you could use one open to block the signal, one closed to ground, second open to block any residual, and a second closed to ground the output of that. Doubt that anything could get through that lot. You could, of course apply the same logic to any of the higher quality switches mentioned above.
HTH,
Bri.