Looking for high-voltage analog matrix multiplexers

[SiliconWizard]

Hi all,

I'm looking for analog matrix-like multiplexer ICs, typically for signals up to +/-15V, up to +/-30mA, 8x8 or 16x16 channels.

Spec-wise, something like the ADG1438 would be a good fit. Unfortunately, to make an 8x8 matrix would require 8 of them (and a routing mess).
I've found a lot of ICs, but nothing that really gets close to this kind of specs.

If anyone can suggest any suitable part...

Thanks!

[Zero999]

Are you sure an analogue switch is what you want? 30mA is on the high side.

[SiliconWizard]

Well, yes. What would you suggest?

The reference specs I'm targetting, the ADG1438, are fine. Those are specified at 169 mA per channel at 25°C.

[jbb]

Well, if you need the full NxN then you will need N² switches.  Which starts to suck really fast.

For N = 16, that's 256 switches.  Which will cost you a pretty penny.

Do you need all the pins at the same time?  It might be possible to use an intermediate bus and subdivide the problem (which can help a lot where polynomial difficulty - i.e. N² - is involved).

[BrianHG]

Well, yes. What would you suggest?

The reference specs I'm targetting, the ADG1438, are fine. Those are specified at 169 mA per channel at 25°C.

Hmmm, you want 30ma across the ADG1438 21 ohm on resistance per switch = 0.7v loss = 21mw of heat.  Multiply that by 8 for sinking 30ma on 8 switches simultaneously = 168mw total dissipation.  Note that with a +/-15v supply, the on resistance drops to 10 ohm as you reach the rails & keep the IC at room temp.  Expect 20 ohm at 85 degrees.  I would say if you use the chip to 30ma/channel on all 8 channels simultaneously, add a small IC heatsink to play it safe.

Run the IC at full 169ma on all 8 channels simultaneously and it will cook without a big heatsink and some forced air cooling radiating a full watt or more.  And, as it warms, the internal resistance per channel rises as well, accelerating the effect.

[SiliconWizard]

As the structure of the ADG1438 (which is fine for my requirements) is a multiplexer, it means that the overall current will be divided amongst all the channels that are ON, since there is a common end. And I don't intend on having more than 30 mA max distributed to 0 to 8 outputs. I can guarantee (with the way the switches will be configured) that there will be +/-30 mA max per current path. So the limitation is fine with me.

As I said, a single IC that would basically integrate 8 ADG1438 would be perfect. I can do this with 8 chips, but it's a cost and huge real-estate waste.

I've found such matrices (often called 'crosspoint arrays') but usually with much higher Ron resistance (in the order of 100-200 ohm) and/or lower max voltage (such as 12/13V max, whereas I'd need up to +/-15V), such as the AD75019. With lower Ron, it would be almost perfect (I could live with +/-12V max if everything else fits).