mute / low signal sensing

[wmtunate]

I'm trying to come up with a circuit that will sense when a line level input drops below a certain threshold (say 100mV or so) and control a jfet switch (or something similar).

I'm using a TDA7240 20W amplifier chip, and it has a "standby" pin that when shunted to ground, puts the chip into a low-current mode.  I'd like to be able to have that happen automatically when the input signal goes low for a few minutes.

There apparently used to be ICs that did this for use with cassette tape players (BA336, BA338, etc.), but they're NLA.

I've been playing with the circuit found here:
http://sound.westhost.com/project38.htm

The problem is that the output is the opposite of what I want (i.e, the MOSFET source pin is switched to ground with the input signal is high).  I've tried implementing an inverting comparator at U1B, but it's not working.
I'm trying to keep this really simple for a small battery powered device running at +18VDC.  There's gotta be an easier way to do this with a comparator or a 555 timer or something.  Anybody have any suggestions?

 

[abbey road d enfer]

You just have to swap the + and - inputs of the second opamp

 

[JohnRoberts]

To get battery current low, a discrete design would probably be best.

Using  bare transistors you are stuck with 500mV thresholds, so perhaps use differentials made from matched fets.

If you can afford the current of a 1458 I would suggest something around the Lm339 comparator.  I guess it depends on how "off" the TDA is when idling. it if gets down to microamps, consider discrete to get in that same ballpark.

JR

 

[wmtunate]

To get battery current low, a discrete design would probably be best.

Using  bare transistors you are stuck with 500mV thresholds, so perhaps use differentials made from matched fets.

If you can afford the current of a 1458 I would suggest something around the Lm339 comparator.  I guess it depends on how "off" the TDA is when idling. it if gets down to microamps, consider discrete to get in that same ballpark.

JR

The TDA goes from 65mA at idle to 200uA at standby.  I've got a simple mic pre in parallel that draws about 2mA at idle and an LED drawing about the same amount that I won't be switching off, so I don't really need to get into the uA territory with this detection part of the circuit.  A constantly running LM339 drawing 2mA or so is fine.  The ~64mA that I can save from having this standby circuit is too much to ignore though.

Thanks for the LM339 suggestion.  I've got a few of those sitting around here somewhere, though I've never really experimented with comparators before.  Can I use one section of the 339 to gain-up the input signal like the first stage of that schematic?

 

[JohnRoberts]

No, but but you can use two 339 sections to detect pretty small + and - X mV of level.. I would have the open collector output of those first two comps, pull a cap down to ground (they can be tied together), with a few meg pull up r. A third section of 339 looks at this RC, and when no signal present for enough time that the cap charges up to X volts, the 3rd 339 turns off the amp.

It should turn on pretty quick and you can dial in an appropriate turn off delay. Current consumption is just the 339's internal draw and the voltage divider string to make your comparator thresholds.

JR

 

[PRR]

> below a certain threshold (say 100mV or so)

Lower. Much lower.

I had a mute which would not dial real low, and I had it cutting-out during a live show.

> goes low for a few minutes.

That helps. Mine was intended to "hide the hiss", so it muted real quick after the loud part stopped (or got quiet).

> easier way to do this with a comparator or a 555

You need to bring milliVolts up to Volts. You really need a HIGH-gain audio amplifier (or serious trickery).

Elliot's U1A is gain=1000, seems good to me. Since no fidelity is needed, you could use a thriftier/lamer amp than antique 4558 (TL072 is battery-thrifty and common, and actually better sound, not that you care).

U1A output idles at 6V DC. U1B's inputs are U1A (6VDC + audio) and 5.4V. If "audio" out of U1A  is over ~~0.6V, U1B flips state. Using 18V instead of 12V is no real change, threshold is 0.9V peaks. As abbey says, you want to reverse the inputs to get the opposite function. (What is an "inverting comparator"? 4558 is a comparator which can be used linear, but here it is just a comparator. Comparators usually have two inputs, you swap pins to "invert".)

D1 C4 R11 give a fast/slow time action.... ah, this is wrong for you. Replace RL1 with a 50K resistor, tie the gate of another MOSFET (or 2N2222) to there, and use the second MOSFET to clamp your TDA7240 pin.

You can maybe trim LM339 to directly sense milliVolt audio; IMHO it will be fussy.

You could maybe re-rig the time-constant to avoid a second MOSFET/BJT, but the brain-pain is more than the cost.

Seems like there IS a simpler way, but not coming to mind.

> The TDA goes from 65mA at idle

Yes, but at 4 Watts in 8 ohms it pulls ~~300mA. Working high-fi with ample headroom, perhaos the average draw is under 100mA; but you won't get 300mA peaks from a couple pocket-radio batteries. Gonna need a big cap. And battery life is not great.

> 2mA ... LED ...  won't be switching off

This should be in series with the feed to the muter. The muter does not need full 18V, and can probably be jimmied to pull a suitable current for the LED.

 

[Rochey]

could you simplify things with a transistor driven relay?
here's what I'm thinking, and feel free to shoot me down on this one...

rectify the input signal and buffer it. Drive it into a relay circuit, and scale the input resistor accordingly (to drop the current below the trigger point when the system is muted).

Alternatively, get your Arduino / MSP430 on. A $1 processor could sample a rectified input easily, and look for the average.

 

[JohnRoberts]

Yes, standard 339 is worst case 5 mV input offset so some gain before would improve sensitivity.  It depends on how low you need to go but note the input error will make it trigger sooner not later, so for one off just dial it in...  for production design, I'd optimize my gain structure so signal hot at that point, or do it discrete.  Any gain stage will consume current.

JR

 

[wmtunate]

Here's what I got so far.  Kind-of a combination of that original circuit with what John suggested.  Seems to work well in simulation.  It waits about 10 minutes to put the TDA in standby once the input voltage drops.  Once I get it breadboarded I'm sure I'll have to tweak the sensitivity.

Thanks for all the help guys.

 

[JohnRoberts]

You can use a second section of 339 to capture the other polarity of input swing.. just reverse the 339 input pins and tie the 339 outputs together, and use one more resistor in divider string to form a low reference threshold.

JR

 

[PRR]

> capture the other polarity of input swing

That bothers me too, but.... is there any likely case where it will fail to trip? Sax or male voice may trip 6dB sooner/later if phase is flipped, but this is probably NBD.

BTW, he shows LM393, not 339. Dunno what the difference is.

Omit R2. Make C1 0.01uFd, R3 1 meg if bass must be sensed, 100K if bass rumble should be ignored. C2 may be 10uFd or even 2uFd.

Total current consumption with TL071 and LM393 may so very nearly 2mA that you can put your always-on LED in series with the power feed.

The "big" problem I see is the Megs resistor charging an electrolytic. Old caps leaked so bad that this might never come up. Newer ones are much better; still you ought to get a case of caps from several vendors, cook and age, then test to see if the idea is manufacturable. Or for one-off with you nearby to re-fix, just go with any good healthy electro. If it is a problem (starts taking hours and days to shutdown), find a film cap (motor-cap).

Why do you want minutes of lingering? If audio stops, sleep. OK, you don't want to cut-off the tail of the reverberation(?), or chatter on very-soft sound, but minutes seems excessive (OTOH, 3 sec is too short IMHO).

 

[wmtunate]

LM393 is a dual, 339 is quad.  PCB space is at a premium here, but SMD is not an option right now, so if I can get away with DIP8 vs a DIP14, I'd prefer that.  The 072 will actually be a leftover from a 074 elsewhere in the circuit.  I'll tweak those values and see how it sounds when I get it mocked up.  If I need to add a third comparator, so be it.

You're right about the minutes thing.  I think 20-30 seconds would probably be fine.  I can tweak the upper threshold voltage of the second comparator to give me a few more seconds of wiggle room with a smaller cap value.