[Dougigs]

I've been working on a small, simple circuit that will automatically turn on your headphone amp when it detects a line-in signal, and automatically turn it off after, say, 20 minutes with no audio. No switch needed, no more killing your batteries because you fell asleep with the switch on (I'm speaking from experience here).

Tell me what you think of the schematic below. I've adapted it from one designed by Rod Elliott of Australia to turn on and off AC-powered amps.

This sort of circuit is common for things like subwoofer amps, but I had dismissed it for battery use because it would require a battery-draining circuit to monitor the line-in. Then I realized this week that there are commonly available, inexpensive dual micropower op amps that draw as little as 1 microamp of current. It would take years to drain a battery. The Texas Instruments TLV2242 series are a good example.

The first opamp boosts the line-in signal; the high impedance input (and short lead lengths in a pocket amp) will prevent it from putting any load on the line. The second is a comparator that turns on the output if it detects anything above about 10mv on the line (you could adjust this). That charges C4 and turns on the MOSFET. With a signal gone, it takes about 20 mins to discharge C4 and turn off the MOSFET. Note that it uses a voltage divider and capacitor to generate a pseudo-ground for bias and comparator trigger-point setting (Mr. Elliott's idea); no need for rail-splitters etc here as it's all simple DC.

You could build this on a half-MINT-sized circuit board, using a dual opamp like the TLV2242 (among others -- search for 'dual micropower' at your parts supplier) and one of many DIP-sized MOSFETs (typically 1A, 60V). The largest items are the 3 big capacitors; you could use tantalum beads, or perhaps try to get away with smaller caps (easy to do with C4; just change the resistor value).

Tell me your thoughts -- have I neglected something? Am I reinventing a wheel that I hadn't noticed? -- and I'll toy with a PCB design (since I need practice at this).

 

[guzzler]

Looks ok to me, although my only concern is the extra size the the capacitors will take up. You could squeeze the resistors and opamps etc into a small space, and you could just use a ceramic for the input, no need to worry about quality there. Also would have to consider the Rds of the MOSFET as for portable use, all the volts matter

 

[Dougigs]

Yeah, good point about the MOSFET resistance. There are plenty out there with Rds of less than 0.01 ohms (some in nifty S08 packages), but best to specify one of these.

I think the big-capacitor problem can be worked around too. Of course, tantalum capacitors are just fine in a non-audio circuit like this, and there are tantalum-chip capacitors that are less than 4mm high in values of 100uf, 20V. That's about as high as an unsocketed DIP, so it could work.

On the other hand, there's no need for 100uf. All three capacitors could drop to 47uf and work fine (with a larger resistor on the third one). Not that 47uf is a lot smaller... though I suspect it would work just fine with 10uf, too.

 

[Dougigs]

OK, here's a schematic tweaked for pocket-sized use.

I've reduced the caps to nominal 47uf; it can probably work with even lower values. I'm doing PCB pads for SMD electrolytic/tantalum caps as well as conventional mini-electrolytics. SMD tantalums in these values are available, but they cost a fortune, so I think the best option here are SMD electrolytics, which are dirt cheap. A 47uf, 16v is 5.5mm high, while the same at 20v is just over 6 mm high, so this can be done in a MINT-sized format.

I'm designing it for the SOIC-8 sized power MOSFETS -- nominally the IRF7467 (30V, 11A, 12miliohm Rds, about $2.50), but there are lots of others in this pinout that would work fine.

I've put two 10M resistors in series for the off-timer (20m are not commonly available). This may sound unneccessary, as you'd think a cap would discharge into thin air or PCB resistance faster than through 20M. From playing with this circuit, I'd say not -- it would take hours and hours. It's probably worth playing with r and c values here to get the right balance of small size and the desired time.

Any feedback would be welcome.

 

[Dougigs]

And here's a PCB layout on a half-MINT-sized board (i.e. 0.75 inch by 1.9 inch). Critiques and comments here are especially welcome as I'm a real novice at automated board design (if you can call ExpressPCB automated).

While I've designed it for the smallest possible headphone-amp package (and I've determined that it will fit in my mint tin if I use SMD electrolytics, since it will eliminate the on-off switch and I'm not using input coupling capacitors since it's an IPod hookup), this is a handy circuit for amps of all sizes -- by attaching the MOSFET to a small relay, it's a great circuit for turning on and off AC-powered audio components when there's a signal.

I've got a p-to-p-wired version of this circuit turning on and off my big KT88 Class A amp (since I was tired of forgetting to turn it off and heating my office to sauna-like temperatures). In an ideal scenario, you could simply press 'Play' on your CD player or Itunes window and have your whole audio chain spring into life from the audio signal... and then turn itself off 20 minutes after you press 'pause.'

 

[guzzler]

Print Screen it into your favorite image editor, and save as GIF. Paintbrush even works. I you can, move straight to Eagle, especially seeing as you're in the UK. It will export an image very easily and is vastly more powerful than ExpressPCB

 

[rickcr42]

T Quote:

he first opamp boosts the line-in signal; the high impedance input (and short lead lengths in a pocket amp) will prevent it from putting any load on the line

10K is not even close to a high impedance and the summing junction no different than a typical passive 2:1 audio mixer network.I would seriously consider another way all things considered

(especially if your amp is low also in the input impedance area.A 10K input Z in parallel with that puppy gives you a total of 5K and that would be extremely low)

 

[Dougigs]

rickr42 -- the summing network could be a problem with regard to crosstalk. This could be pretty easily solved by using only one channel (there's no reason to have both hooked up in this circuit, so best to leave it out).

But is it really a 10K impedance? Those 10K resistors aren't referenced to ground -- they're series input resistors. Sorry if I'm way off base here, but wouldn't the input impedance of the op amp swamp the series resistor value?

 

[Dougigs]

Guzzler - thanks for the help with that. The PCB images are now up on that post -- I will take your advice and try Eagle. The limitations of ExpressPCB were driving me crazy.

I'll probably take rickr42's suggestion to heart and rejig it so it uses only one channel on the input... Or whatever seems best to avoid loading the audio circuit with a low impedance.

 

[rickcr42]

Quote:

rickr42 -- the summing network could be a problem with regard to crosstalk. This could be pretty easily solved by using only one channel (there's no reason to have both hooked up in this circuit, so best to leave it out).

That was my initial thought when seeing this for the first time.


Quote:

But is it really a 10K impedance? Those 10K resistors aren't referenced to ground -- they're series input resistors. Sorry if I'm way off base here, but wouldn't the input impedance of the op amp swamp the series resistor value?

The opamp is 'flipped" so on first glance it looked like the 10 was at the suming junction (-) of the opamp.My bad.
I still do not care for what is a non standard stable impedance on the actual input of the amp.
I personally would do a google on "VU meters" and choose a nice high isolation input circuit/precision rectifier from those available (do the image search for saving time

) which can then be used to drive pretty much any mechanical or optical relay.
Dealing with basic electronic devices you can pretty much go a straight formula route for design but when dealing with audio devices you need to look specifically at any possible consequences of any addition from an "is it audible and will it add distortion" point.i am not slamming your choice but I personally would not be comfortable with that right at the input of an audio device.

BTW- maybe search "Jukka Tolenen Peak Readng Meter" which from memory has an input and DC rectifier circuit is ideal for this use with maybe a minor modification

 

[amb]

Quote:

Originally Posted by Dougigs the summing network could be a problem with regard to crosstalk. This could be pretty easily solved by using only one channel (there's no reason to have both hooked up in this circuit, so best to leave it out). But is it really a 10K impedance? Those 10K resistors aren't referenced to ground -- they're series input resistors. Sorry if I'm way off base here, but wouldn't the input impedance of the op amp swamp the series resistor value?

Headphone amps typically handle load impedances far lower than 10KΩ (or even 5KΩ), so the loading effects of this circuit on the output is not really a concern. However, crosstalk is a problem with the current circuit, and I don't think connecting only one channel is a good solution. You might have a recording that starts playing primarly on the unconnected channel for a while and it won't turn on the amp.

I suggest changing the first stage opamp in your circuit to an inverting topology, so that it becomes the classic opamp summing amplifier. The summing junction then becomes a "virtual ground" and that would eliminate the crosstalk.

 

[rickcr42]

Quote:

Headphone amps typically handle load impedances far lower than 10KΩ (or even 5KΩ), so the loading effects of this circuit on the output is not really a concern.

Missing my point I think amb.I think you will agree that you want a driving stage (source) to be as low in impedance and as high in current as is reasobaly possible which then goes to an input impedance that ideally does not load down that source.
As you parallel devices on the driven stage input impedance you lower the overall Z by a adding up the multiples of figures then dividing the impedance by the amount of devices in parallel.That is the reason we use an overall input buffer/driver for any audio device that goes to multiple devices as in a distribution amp where any mismatches and possible high end attenuation could take place if not compensated for.
My main concern with this particular circuit is it does not provide a predictably stable impedance thus making any compensation a shot in the dark at best fro my read of the circuit.

It may work fine in practice but to say I would be comfortable with that at the input of an audio device for high fidelity use would be off the mark.Far too much that is not optimal going on there .just personal opinion as always and as always again YMMV

ricksta

 

[RickG]

This might be of interest.

"The PushButton

Press ONCE to turn on the power for at least ONE hour.
Press THREE times -- " -- THREE hours (Blue led flashes 3 times in response)
Press SIX times -- " -- SIX hours (Blue led flashes 6 times in response).
Press SEVEN times -- " -- FOREVER (Blue led 7 flashes; unit must be turned off manually)


Press ONCE to display battery voltage (The Amber Led flashes the battery voltage. For example, if the voltage is 8.5volts, the Led will flash 8, then 5 times.
Press TWO times to toggle Flashlight mode
Press FIVE times to toggle the Blue Led brightness


PRESS and HOLD for 3 seconds to Power OFF.
When pressing for the above functions, you must do all the presses within a three second window.

The most-commonly used functions are, of course, ON and OFF, so we've made these the simplest button-pushes: Push ONCE for ON, and Push and HOLD for OFF. "

 

[amb]

Quote:

Originally Posted by rickcr42 Missing my point I think amb.I think you will agree that you want a driving stage (source) to be as low in impedance and as high in current as is reasobaly possible which then goes to an input impedance that ideally does not load down that source. ...

Agree in principle, but what is acceptably low or high depends on the application. For a power amp meant to drive 8Ω speakers, its output impedance should optimally be far less than 1Ω and an additional 100Ω load on its output is all but negligible. For a preamp meant to drive the input of a power amp, several hundred ohms output impedance is common and the load shouldn't be less than several KΩs. This is a headphone amp we're dealing with here, which resembles the latter case and since it's intended for a portable amp application I think it's fairly safe to assume that the output impedance of the "source" is no more than that of a typical preamp.

Quote:

My main concern with this particular circuit is it does not provide a predictably stable impedance thus making any compensation a shot in the dark at best fro my read of the circuit.

How is the impedance not "predictably stable"?

 

[Dougigs]

amb -- I like your inverting approach and don't see any negatives to it. I've always used these as unity-gain circuits; I'd just have to set it with a gain of 100 but I'm assuming this is calculated the same as for any inverting circuit (and this certainly doesn't need to be a precision circuit). Are 10K input resistors on a summing circuit sufficient to prevent any crosstalk? Since its input is a virtual ground, I would imagine this to be the case.

But oddly enough, are we bringing the impedance-load problem, which hadn't really existed in the non-inverting circuit, back to life this way? Inverting amps have a much lower input impedance...