H11F1 observations

[tempus]

Hey all;

There's been a bit of talk around here concerning using the H11F1 as a possible substitute for a JFET mute circuit to help eliminate switching clicks. I did a little experimenting, and found the following:

1. It doesn't mute to silence - this was already known I think, and may not be a problem
2. the circuit wasn't transparent, i.e., even though the IC wasn't in the signal path (it just offers a shunt to ground for the mute), it attenuated the signal very slightly. This was a bit of a head scratcher, as I would have thought that there would've been enough isolation between the signal and IC

It would appear then, that this IC wouldn't be the ideal for a mute circuit, unless my design is flawed somewhere (if there's no way I should be experiencing #2, please feel free to give your input). I'm yet to do anything other than a cursory test with the TLP2222, but it may be the better choice.

[R.G.]

Numbers?

[tempus]

I'm sorry RG, I don't understand what you're asking.

[PRR]

> It doesn't mute to silence

No single part can. We have to use it in a voltage divider: two parts. One is the "switch" with high/low resistance, the other is typically a resistor.

For example, I could make that signal go AWAY. But the "on" position could be awful weak.

What's your resistor?

[Mark Hammer]

I'm sorry RG, I don't understand what you're asking.

I think he's asking for a more quantifiable sense of what the gap is between what you see as insufficient, and what one might achieve with other components or types of approaches.  Maybe there's a fix, and maybe the gap i s negligible under real-world circumstances...but we won't know until things get measured, right?

[tempus]

Right - I see what you're saying. When I'm testing things, (particularly in this case, where I'm modifying an existing project that would be a huge PITA to dismantle to do actual "proper" testing) I usually take a listen first. If the sound is audibly degraded, there's no real need to look at the numbers to tell me that it's not sounding right. As I mentioned, the signal is audibly attenuated, and my next course of action is to do a much closer ear test and possibly empirical test of a different part that I've found doesn't attenuate the signal (TLP222A). This component also mutes to silence on first listen.

I understand the need for empirical data, but your ears can tell you a lot too.

PRR - I used a 1K resistor, and I'm glad you asked, because it now occurs to me that with a lower value resistor, I may be able to get more muting. I kind of abandoned trying to find a solution when I heard the attenuation, but I'll give this a try when I get the chance.


Thanks

[PRR]

> I used a 1K resistor,.... with a lower value resistor, I may be able to get more muting.

Assuming the 1K is in series with the source, the FET is shunt to ground, and:

* H11F1M is 200 ohms

* TLP222A is 1 ohm

1K:200 is 6:1 or 15dB attenuation, not much.

1K:1 is 1000:1 or 60dB attenuation, a lot.

But why 1K? Most guitar-cord stuff is 50K or higher. We could use 10K series resistor. Now a 200 ohm FET gives 51:1 or 34dB attenuation. Not a whole lot, but a whole lot better than 6:1 or 15dB. Inside a box where buffering is possible, 200K series is fine and gives great attenuation.

[R.G.]

Horses for courses.

I've used the H11F1 for muting. It won't work all that well with a 1K source impedance, OK. The H11F1s I've tested have high "off-ness", being nearly an open circuit when the LED is really off.

An ordinary NPN transistor can do a decent job of shunting for low impedances, if you want to mess with the offset voltages and needing to be very careful about control feedthrough.

If the TLP222A works for you, fine. Use it.

[PRR]

TLP222A seems to be a much fatter part than H11F1M.

http://www.semicon.toshiba.co.jp/info/docget.jsp?type=datasheet&lang=en&pid=TLP222A

The ~~1 ohm is notable. It will block 60V with 1nA leakage (60 Gig Ohms?), it will pass 500mA with 0.4V drop and the drop appears linear. You get this with 5mA LED current, though 1.6mA may do for lesser tasks. LED drop is nearer 1.2V than the 1.6+ we see on visible LEDs.

"Fatter" also in that it is "slow". The off-state seems to have 130pFd capacitance, so impedance over 100K will begin to shave guitar-band; also the turn-on time can be 1mS-2mS and there's *relays* faster than that. (However ramping a linear switch is subject to drive load and interpretation; with better impedances and clever drive it can switch fast-enough for most any audio.)

I wish they'd characterized it for linear mode. Be an interesting feedback limiter device. The low impedance offers an order-magnitude lower hiss than an LDR, but how wide is the linear band when at medium resistance?

At ~~$3 for a dual device, >1K in-stock at Mouser, the cost is trivial.

[tempus]

Wow. I'm glad I posted, because I don't think I'm actually using this thing right. Here's a schem of my circuit:



The resistor I was referring to is the 1.5K (I erroneously posted 1K) current limit resistor from the uC output. As you can see, I was treating this more or less like a relay, with one end connected to the signal and the other to ground. Incidentally, I've been doing the same thing with the TLP222. This is also how I wired a previous BS170 (that RG helped me to discover was distorting my clean signal).

But you're saying I need a series resistor?

RG - I'm scratching my head at the lack of "offness" as well. Does the absence of the series resistor affect this?

Since the control voltage could be ramped up for the H11F1, I'd prefer to use it if possible. Plus I ordered about 10 of them.

Thanks for the help guys.

[R.G.]

The resistor I was referring to is the 1.5K (I erroneously posted 1K) current limit resistor from the uC output. As you can see, I was treating this more or less like a relay, with one end connected to the signal and the other to ground. Incidentally, I've been doing the same thing with the TLP222. This is also how I wired a previous BS170 (that RG helped me to discover was distorting my clean signal).

But you're saying I need a series resistor?

There is an implied resistor inside every signal source, as it will not supply its nominal voltage into an arbitrarily low resistance. What you want to do is to stick a resistor in series with the wire marked "from FX chain" that lets the H11F1 mute things well enough. Well enough is generally something over about 20db, so a resistor in the 2K - 4.7K region is enough for most purposes. 10K may well not be too big, depending a LOT on the unknown load that will be plugged into the jack at "gtr out". If this is a 1M input, 10K-47K may well be reasonable for the series resistor.

RG - I'm scratching my head at the lack of "offness" as well. Does the absence of the series resistor affect this?

Not really. From my reading of the H11F1 data sheets, it's "offness" is really, really good. They flatly state the off-state resistance as 300M minimum. That's pretty darned off. I'm not sure why you're not seeing it be completely non-attenuating when off. The on-resistance being limited to 200 ohms would make me believe that the on-ness is more of a problem.

[PRR]

You are trying to "short out" a signal source.

Is that wise?

> I need a series resistor?

There always IS a "series resistor". (As R.G. wrote while I wuz typing.)

See "Voltage divider" in the picture.

R(shunt) is the 200 ohm H11F1 (or 1 ohm TLP222).

But R(series) can vary all over the place.



A naked guitar can be about 5,000 ohms. Put a 200 ohm H11F1 across it, you get about 200/(5000+200) or about 0.04 of the original signal. dB notation may be more useful, -28dB. Which is a lot less, but not "mute".

It can be helpful to take things to absurd extremes. Say you wanted to "mute" a 12V car battery. (I been working on the truck's electrics.) Put 200 ohms across it. "Nothing happens." You can hardly measure the voltage drop.

More often we want to mute an effects device. We like transistors and chips because they can make a guitar signal stronger. OTOH, that makes them harder to "short-out". OTOOH, in guitar-cord work the strong chip is often "weakened" before the output jack. So R(series) can be 100 ohms or 47,000 ohms. 200 ohms R(shunt) will really clobber a 47K source, do very little for a 100 ohm source. And since you can't tell by looking at the box, you get inconsistent results.

Now you have the 1 ohm TLP222. It "shorts-out" 200 times better. In most audio cases the output will be super-weak. But many amplifiers distort when faced with a near-short. So the residual output is all buzzy garbage, annoying even when very small. Guitars and most effect boxes are low or limited power. Nothing smokes when you short them. However consider a TLP222 to "mute" a 12V battery. (Or if someone plugs-in the Speaker output of a 18-Watt amplifier.) 12V in 1 ohm is 12 Amps, 12V*12A is 144 Watts!! You get what you expect if you drop small conductors across a car battery: smoke and burn-up. Actually the TLP222 may die without drama; but it sure will die.

But R(series) has other effects. Perhaps a better way to "mute" is *two* switches. One shorts the output. One *breaks* the input. Technically we can't get a perfect break, or short. But assuming the "open" switch is a 200Meg leak, and the "closed" switch is 200 ohms, we get a million-to-one reduction, -60dB, which is "mute enough" for most stage purposes. (And as R.G. says, the "off" resistance is typically much higher than Meg-ohms.)

[tempus]

You guys are awesome. Really. Thanks for the detailed drawings and explanations.

RG - Yes, I knew that the gtr itself would already have a resistance, which is part of the reason I didn't think a series resistor was necessary. I understand better now though (it always seems like there's some little vital bit of information I'm lacking...). The output is going to a Boss NS2, which is specd for 1M input, so I can certainly use a higher resistor. Also, I'm going to experiment with this again; I'm still bugged by the attenuation that shouldn't actually be there. Maybe there was a bad connection on my breadboard or something.

PRR - nice diagrams man - I really appreciate all the trouble you went to to whip those up. As I mentioned earlier, I knew the guitar already had a series resistance, but I hadn't thought of the different conditions that the signal might be hitting the mute from. When my pedals are bypassed, it would essentially be the guitar's output resistance, but with pedals engaged, I suppose it could be as low as 1K (or lower), so the resistor would help out there.
Manly, full flavoured circuit...lol

When I get some time I'll mess around with this again and post my findings.


Thanks again.

[R.G.]

I do like the idea of trying to mute a car battery. One of my truck batteries can't be muted by multi-hundred-ampere starter motor.  

But yes, you need a series resistor. Fortunately, Mother Nature has already insisted that there be a series resistor there. But adding another resistor gives you more control. A guitar is typically 4K to 18K in series with 2Henries to 6Henries of inductance. So you get much more attenuation at treble than at bass because of the inductive nature. Pau's circuit with a buffer gets rid of this problem.

It took me some time to catch on to the idea that every circuit has parts that are not shown in the schematic; the input source impedance and the output load impedance, at least. They're always there.

There are some other issues that touch on this.
1. You're trying to mute switching clicks, not mute the audio. Switching clicks from relay coupling are coupled through parasitic capacitances.

You don't have to have a dead short to make these inaudible. You just have to lower the loading on the signal line from the maybe-1M that lots of amplifers and effects have, down to a lot lower. Even 10K quiets relay clicks. A lot. You can even put a resistance between the H11F1 and the signal line so the H11F1 can only pull the line down to a loading of 5K-10K to ground and this will eliminate the capacitor coupled clicks from all but the most beastly relays. I have personally done exactly this, and with the H11F1. It can be made to work well.

2. If you must mute all the audio, you can use two H11F1s, one in series and one in shunt. They are activated oppositely. When signal is to flow through the line, the series F1 is on, the shunt is off. When signal is to be muted, the series F1 is off (and > 300M) and the shunt is on. The attenuation is then 200/300M, or over 120db. That's pretty darned mute.

3. The only reason to use an H11F1 is to get optical isolation from the control signal. Otherwise, a JFET works well, even if it takes some design effort.

4. It is possible that the TLP222 is an even better mute. I haven't tried it. But I'm leery of the linearity of two MOSFETs in series. It's probably fine, but still, could have funny effects. This is just me being conservative.

[tempus]

Hey all;

Well, I tried the H11F1 as a mute again, and I'd like to withdraw all of my initial comments:

1. In my original experiments, I was using a 1.5K resistor as the current limit to the control side of the H11F1. When I lowered this to 390 ohms, the mute was almost total, even with no series resistor. Clearly I wasn't hitting the LED in the H11F1 with enough current originally.
2. I couldn't reproduce the attenuation that I had in my initial experiments, which had led me to the conclusion that the H11F1 didn't turn fully off. In this round I noticed no difference in the sound with the device in vs. out, so there must have been some weirdness going on in my hookup wire or breadboard.

I haven't actually installed it yet (that's going to take a fair amount of work), but it should be fine for a mute.

Sorry for the mixup and thanks for the comments.

[R.G.]

Ah. That matches my experiences. We've learned something (and I didn't know this, either) - you have to hit the LED hard enough to fully turn on the FET. I've apparently done this OK by accident before. Good work.

[Beo]

I'm also experimenting with AVR uC controlled relay switching and H11F1 muting. I have all the major pops and clicks out, but I'm still getting a small click just from the mute circuit (i.e. after disconnecting the relay trigger and testing the mute itself). I've tried caps to slow down the optofet transition, hitting the led less with higher resistors, and various in-series resistors before the optofet. I have a DC blocking cap and signal pulled to the center of the 9v battery just before the optofet shunt. Digital and audio grounds are separate, and I've disconnected the bypass led. What I'm getting is that with no signal, no clicks when I mute, but if I strum a chord, I get clicks proportional to the volume of my chord. Any insights?

Travis

[bool]

It doesn't "detect" the zero-crossing? So, it will probably have to "fade to zero" - and your "fade" is too fast so it "clicks"? Just a guess from your description.

[Beo]

It doesn't "detect" the zero-crossing? So, it will probably have to "fade to zero" - and your "fade" is too fast so it "clicks"? Just a guess from your description.

I'm not sure what you mean by detecting zero-crossing. If you are suggesting to slow down the optofet turn on (i.e. a soft mute), I haven't had much luck getting that to happen using a higher led resistor or a cap across the led.

[midwayfair]

Question: Why didn't you set it up so that the muted position turned off the internal LED instead of trying to dump signal to ground? That would stick 300M in series with the signal. Your guitar's pickups won't make a peep through that. That's the method used for optical bypass switching, also.