Best Universal JFET Switch

[Jdansti]

me> There are opto-coupled back-to-back MOSFET packs

http://www.newark.com/vishay/lh1500at/ssr-mosfet-350v-150ma/dp/58K1788
http://www.farnell.com/datasheets/1809360.pdf
LH1500AT  MOSFET Relay, 350 V, 150 mA, 25 ohm, SPST-NO
$1.62

Thanks for posting this. This looks like it could be also be used where we want to chop the audio signal as we would in a tremolo. I've reviewed the data sheet here: http://www.farnell.com/datasheets/1809360.pdf. But I don't understand several things. Is it really as simple as using this as the diagram shows?

1) The internal LED's Vf is about 1.5V, so it looks like it has similar characteristics as the standalone LEDs we use and would operate at 9V with a 1K to 10,000K resistor in series. Correct?

I'm still learning about MOSFETs and don't understand how this device is connected to make the MOSFETs work. Maybe there are some internal connections that aren't shown on the digram.

2) Can the connections to S and S' be an audio signal on the level as what we normally see in a stompbox? 

3) Pin 5 is labeled "DC" (I assume a DC voltage), but I don't see any information in the data that describes what this voltage should be. Could it be 9V?

4) No other ground connections required other than the one for the LED?

[Groovenut]

1) The internal LED's Vf is about 1.5V, so it looks like it has similar characteristics as the standalone LEDs we use and would operate at 9V with a 1K to 10,000K resistor in series. Correct?

As far as I can tell , this is correct.

I'm still learning about MOSFETs and don't understand how this device is connected to make the MOSFETs work. Maybe there are some internal connections that aren't shown on the digram.

By connecting the drains of the two mosfets together, the body diode effect is canceled out. Otherwise you get signal rectification.

2) Can the connections to S and S' be an audio signal on the level as what we normally see in a stompbox?

Yes it appears so. The switch does need a load resistance to work against in order to function.

3) Pin 5 is labeled "DC" (I assume a DC voltage), but I don't see any information in the data that describes what this voltage should be. Could it be 9V?
I believe DC in this case is Drain Connection.

4) No other ground connections required other than the one for the LED?

Only the load which would need to be 10k-100k or so for good signal isolation.

This is only from my limited knowledge of using mosfet switches. I am sure there are others here more knowledgeable who will hopefully chime in to correct my errors

[Jdansti]

Thanks, Groovenut. 

[MetalGuy]

I've used this in noise gates and mute circuits and when in parallel with the sound chain it doesn't affect it in any way. Capacitance is low, ON resistance is ~5 Ohms, max voltage is 60V which is plenty for a 9V pedal. I haven't tried it in series with the sound chain.

http://uk.farnell.com/webapp/wcs/stores/servlet/Search?catalogId=15001&exaMfpn=true&mfpn=VO1400AEFTR&categoryId=&langId=44&searchRef=SearchLookAhead&storeId=10151&showResults=true

[DiscoVlad]

For the LH1500, according to this app note the S and S' connections are for switching AC loads (in this case DC terminal is left unconnected), while for DC loads S and S' are connected to the positive/high side, and the DC connection is connected to the negative/low side. This puts the two MOSFETs in parallel and gives higher current capacity.

[Jdansti]

For the LH1500, according to this app note the S and S' connections are for switching AC loads (in this case DC terminal is left unconnected), while for DC loads S and S' are connected to the positive/high side, and the DC connection is connected to the negative/low side. This puts the two MOSFETs in parallel and gives higher current capacity.

So could you connect an audio signal such as what we see in stompboxes to S and S' and have it switch the signal in and out?

[DiscoVlad]

I don't see why not... assuming the signal you're switching is within the device output current/voltage limits.

how they're connected depends on what you're switching. eg. For an effect bypass you'd need to use 3 of the devices. One each at the input and output, and another on the bypass path which needs to be switched opposite to the other two. Like RG describes here: http://www.geofex.com/Article_Folders/bosstech.pdf

[Jdansti]

^Cool-thanks!

[PRR]

Yes, the LED is just an LED, a high-quality red job (lower Vf than green purple white, and less parasitic resistance than 100 for a buck LEDs).

Look up basic MOSFET operation. Gate zero, MOSFET off. Gate at 5V-10V, MOSFET on. Instead of wiring a voltage to the Gate, this has a "solar cell" on the gate, the LED shines on it. This is off/on for DC, the MOSFET isn't bi-directional. To cover bi-directional current, two MOSFETs are wired back-2-back and I believe processed so this works smooth (discrete MOSFETs may not cross-over smooth this way). If your current is One Way, as Vlad says I *think* you can strap the S pins against DC and get parallel operation.

> assuming the signal you're switching is within the device output current/voltage limits.

And the part I pointed-at (there are many) is good for some things, if not the un-specified application in the first post.

This not good for tapping power off Hoover Dam. Or even for switching loudspeakers (even LM386).

Would work fine for headphones. Hi-fi line level. Guitar level. It is borderline for switching inside a vacuum tube guitar amp (but you would try to keep switching off the hi-DC nodes anyway).

These are specified as On/Off switches. Design LED current is 5mA, max is 16(?). As you get below 2mA, the switch switches but transition is undefined, VERY variable with temperature and part variation. While it must pass through every value from 16 Ohms to over 16 MegOhms, it does so in a narrow range of LED current and you can't pick what resistance it will be. While there are Duals, I don't think they are remotely "matched" because the pieces are not all from the same die. It "might" work as an audio limiter because that self-feedbacks itself (however stereo tracking is unlikely).

When you look at specs, note there are two "voltages". Both higher than most guitar-stuff, but if you are brave, know what you are facing. The "350V" is across the MOSFET when it is OFF. The "1,500V" is from any pin to any non-connected pin. You can use these parts between two circuits on VERY different power supplies. Controller on well-grounded 120V outlet, and furnace relay on jackleg 440V Delta. Stuff you probably do not want to play with.

[Jdansti]

^Thanks, PRR. This was very helpful.

[R.G.]

Back to back MOSFETs for switching have been around for a while. The new, all-integrated version is kind of new. Perhaps it's better.

I used MOSFETs and PV isolator-drivers for speaker switching once. Worked. There was an odd low-signal distortion that may or may not have been the MOSFETs, so I didn't let this version out of captivity.

In theory, MOSFETs are near-ideal analog switches. Their channels are resistors when properly driven. However, virtually all MOSFETs are made by a semiconductor process that isolates the channel inside a "tub" diffusion from the substrate, making for a reverse biased diode from the substrate. In separate MOSFETs this looks like a reverse biased diode from source to drain for normal operation. You get a decent variable resistor in the forward conduction mode, but the reverse diode is forward biased when a signal inverts the voltage direction on the MOSFET.

That means that a single MOSFET cannot block AC. The two-MOSFET version does block AC by setting the two substrate diodes up end to end so that they block in both directions, and the two MOSFETs in series can then act like two resistors in series. Works great for switching AC power.

I worry about what it sounds like when it switches audio signal. In theory, it should be great. My caution is that modern audio is remarkably intolerant of micro-distortions. The human ear is good at detecting those. I keep intending to run some distortion analysis on MOSFET switches as soon as I get another distortion analyzer. 

But they're probably good enough for rock-n-roll. Let your ears be your guide.

[PRR]

> MOSFETs ...for speaker switching once. Worked. There was an odd low-signal distortion...

There are thousands of these parts now. One I found claimed 16 Ohms (a bit high for speaker) but it *appeared* to be very linear up-to about half of rated current. However the draftsperson might have just rulered the graph. There is a distinct kink to a lower resistance around half of rated current; this would not be your "low-signal" distortion.

I too am curious how faithful it may be for very small signals. First-order approximation, non-linearities happen above part-Volt, which we don't want to approach when "on", so "it should be clean". LED power is large and LED-MOS coupling is very small, we don't expect MOSFET potential to change the light drive. But I am just not confident of very low voltages on MOSFET. And "discrete" MOSFET observations may be moot, because I see several parts and parameters which are "unusual". (Including a normally-on MOSFET, which is common in textbooks but has been rare in actual catalogs; also a current-limited device which implies extra circuitry.)

[KMG]

Only the load which would need to be 10k-100k or so for good signal isolation.

Very important warning!
In case of 1M or higher S-S capacitance is enough to pass HF part of signal trough closed switch.
With distorted signal you'll have "zzz" noise.