Parallel analog switches.

why not use a fet?

NT

The body diode is a nuisance as is the Vgss max and the level translation from logic. All this is taken care of with an analog switch.

But yes, it may well turn out to be the best way.

A bit OT, but I was surprised to see that Microchip has acquired Supertex. They do some good high voltage stuff including switches albeit not appropriate for me.

Cheers

I chip-form, where I needed a zero-drop "switch", I've simply built a tri-state-able amplifier. ...Jim Thompson

On a sunny day (Mon, 2 Nov 2015 13:02:31 +0000) it happened Syd Rumpo wrote in :

I dunno of hand any analog switsh package that would qualify. you need to specify max freq throught it, max voltage drop allowed,

I think I could do that with bipolar high gain small signal transistors with a drop of a few hundred mV (same as you would have with many analog switches), where bias current would be 1/300 of the signal current, driving differentially. but have not worked it out.

let's see +Ib | PNP in -- c b e --- e b c -- out NPN | -Ib base (drive) currents would cancel to some point.

So drive with 2 current sources, their voltage needs to be a bit higher than the 30 V range. But just a quick idea, maybe for the spicy people to play with. Would stay clear of mosfets in this case, Cin too high. Do huge J fets exist?

But could be wrong, have not done any tronix for at least a few hours...

So we get HV PICs!

In fact that's what I'm trying at the moment (in simulation) - it's a class AB amplifier 10kHz to 100kHz with a PMOS and NMOS output pair. The difficulty seems to be avoiding a glitch when I switch, but I do have control over the amp input, so could probably turn off (in my case, set to half supply) the input a bit before tri-stating the output.

Cheers

I've heard of diode switches supposedly used in oscilloscopes: +Ib | ---- | | v v - - | | in-----* *-----out | | v v - - | | ---- | -Ib

(supposedly) good into the ghz range.

Here's an ECL version done as a sample-and-hold (designed ~1968):

...Jim Thompson

Not if the switches are in parallel, which is what Syd was talking about.

More a question for the group than a comment on your plans, but -- are there enough decent power FETs without the diodes that one could make this work with off-the-shelf parts and no great fear of part obsolescence?

I suppose one could always use four FETs, with two pairs connected drain- to-drain or source-to-source. It may not even be that bad from an electrical or board-space point of view compared to a multitude of analog switches.

Source-to-source makes for a bilateral switch, as in this example:

Just leave out the sensing stuff in the middle and tie the two gates together. ...Jim Thompson

2) If it's audio, why do you need microsecond switching? One of the standard "thump" avoidance techniques is turning on and off slowly.

Cheers, James Arthur

? James is there a diode drop in that switch? Or what's the middle lead connected to?

George H.

Without what diodes? If you could get p / n fets in pairs with external pins for the body diode, then you could wire 'em in parallel.. is that what you mean?

confused as always, George H.

The switch has no diode drop whatever. Drive the gates positive with respect to the center terminal and the MOSFETs turn hard-on. The switch then looks like 2 x Rds(on).

You need two FETs for a bi-directional switch to prevent reverse conduction through a single MOSFET's substrate diode.

Cheers, James Arthur

On a sunny day (Mon, 2 Nov 2015 17:33:35 -0800 (PST)) it happened snipped-for-privacy@yahoo.com wrote in :

OP specified 30 V range Your Vgs may be exceeded. There maybe switch feed through because of Cgs for mosfets capable of switching 500 mA. Coff (Cds + Cdg) may be a bit high, we still need to know the OP frequency specs. 'Positive relative to center terminal' when center terminal is defined by input voltage, you need several volts about the 30V range to open the mosfets.

But its early morning so .. if electrons decide otherwise those rule.

You can do that with an opto-fet, like LCA710S.

On a sunny day (Tue, 03 Nov 2015 07:40:39 +0000) it happened John Devereux wrote in :

Datasheet: Ton = 2.5 mS Toff = .25 mS No Go

That depends on the application doesn't it? I did not see a requirement on switching time. But you are right in that their low speed is their main drawback.

On a sunny day (Tue, 03 Nov 2015 11:26:23 +0000) it happened John Devereux wrote in :

OP mentioned us (micro seconds), and that a solid state relais bcould not be used because of speed.