LM386 amp circuits

[mhennessy]

And sometimes they are the opposite way around, which is confusing

As Paul said earlier, simply measure it. Turn to centre, and measure from the wiper to each end. Lin will be close to the same in both directions, and for log, it's normally 10%/90% or 20%/80%, give or take...

[Martin Bush]

Thanks. I will get the Avo out when I next get time and spend an hour or two puzzling over it

Unlike many on here it often takes a while for the penny to drop with things for me. All good fun and I figure with these cheap bits I can do a few bodges and make a few blunders in safety.

[mhennessy]

We all started somewhere

As a learning style, I'm a big fan of the "just do it" approach. Sometimes I see beginners agonising over Spice simulation results or theorising about some second or third order effect that someone has told them about (meaning well, or perhaps showing off), when really all they need to do is reach for the soldering iron and try something. Being afraid to make mistakes is probably the worst barrier to learning - and with the price and availability of components today, there's really no need to worry.

If something doesn't work, troubleshooting is where you'll learn the best lessons. If things just work, well, that's not much better than being able to assemble an Ikea bookcase - almost anyone can follow instructions

Do you have an oscilloscope?

[Ambientnoise]

Potentiometer code markings can differ old/new/europe/asia/usa so unless it says log or lin, it is best (and easy) to test as Paul said above.

Ken

[DangerMan]

There was a very good article in one of the enthusiast magazines, possibly Electronics and Wireless world, which did a very good overview/comparison of lots of small audio amplifier IC's.
It gave an overall summary of many popular types (including LM386 I think) and ISTR it included spectral analysis of their distortion products. Very interesting, but I can't find it now.

Anyone remember it?

Pete

[Martin Bush]

I'd be interested in that. I can't promise I will understand it all straight away though.

Mhennessy - I do a fair bit of 'just do it's type stuff. Not usually with audio, but that's a story for another thread. Let's just say iearn something each time, even if I dont get the rest I want.

As for an oscilloscope, I have one in order from a forum member, so I'm well on the way to having the gear if not the skills

[G6Tanuki]

Personally I'm not a lover of the LM386: while it 'works' I find them prone to strange oscillation-modes when driven, which can result in them running hot when at part of their drive-cycle they start delivering more power at 70KHz than at any audible frequencies.

My preferred DIL chip these days for low-power audio stuff is the Texas Instruments LM384

http://www.ti.com.cn/cn/lit/ds/symlink/lm384.pdf

having used its predecessor the 2.5W LM380 a lot back in the 1980s. Both the 380 and 384 were vastly better than the TI "SN7600x" series of IC amps from the late-1970s which blighted a lot of TVs and budget hi-fi 'music centres' of the era.

The SGS-ATES TBA800/810 series are also worth a look if you want 'classic' audio power amplification in a convenient-to-use DIL/quad-in-line package. They're easy-to-tame from an unwanted-supersonic-oscillation perspective.

[paulsherwin]

The LM386 is a great chip to play around with though - dirt cheap, easy to get going, tolerant of abuse, zillions of application circuits, reasonably high quality sound.

[G6Tanuki]

Yes - the 386 is great to play with but prone to ultimate disappointment. I see it like a child's tricycle whilst the other chips I mentioned are to me more like a noddy-bike with stabiliser-wheels which you can remove when you gain confidence and experience.

I also like the TDA2003: a bit more powerful but also *really* hard to destroy if you feed it from only 12V. There's a neat Velleman kit - http://www.velleman.co.uk/contents/e...224_k4001.html - which I've used in the past to replace the over-complex discrete-transistor output stage in a Roberts R707 [OK, this is heresy to some].

[MrElectronicman]

I had all kinds of problems with mine, I built it on strip board and ended cutting off all the excess strip to any connection to the IC and had to put a 10uF capacitor on the power supply line close to the IC as well.
Once that was done it sounds great to my ears, mind you I have a serious hearing problem, so that may not be much of a recommendation.

[Martin Bush]

Originally Posted by G6Tanuki

I also like the TDA2003: a bit more powerful but also *really* hard to destroy if you feed it from only 12V.

I may give that one a go.

Making up the little 386 kit gave me a bit of confidence and I like the idea of building my own amps, even if it is initially from kits.

I think working up to translating circuit diagrams successfully from scratch will take some time, and I have very little of that at the moment. But that's my ambition.

[Oldcodger]

Out of interest for anyone doing any telecomms cable tracing, a crude but effective tone kit can be made from a couple of 555 and a 386 with the gain set high via a pot and a small cap across the input to kill RF. I used a short metal spike on one leg of the input for pick up purposes, and he output into a pair of headphones. Not as good as commercial kit, but effective . It also worked as a guide to finding mains cables, but not something I'd trust too much.

[BigClick]

Originally Posted by Martin Bush

Those were great days, but I'm trying to do things the right way.

So, if I've got it right, the resistor method is the right and proper way regardless of the need to sort the amp out. Is that right?

My idea is to work with a standard input and speaker then work on the amp - trying some different designs until I get an acceptable combination.

Serious question what's wrong with just connecting L&R together?

[mhennessy]

If the source device has a low output impedance - such as a portable MP3 player expecting to drive headphones - then this is quite a cruel thing to do. Fine if the content is mono, but any stereo content results in differences between the two channels. In that case, the left and right output stages end up "fighting".

I ran into this many years ago when using a Sony RDS walkman (SRF-M48RDS) into a mono powered loudspeaker. I used this combination as a bedside radio in a room I was renting at the time. No, I wasn't collecting radios at the time!

Anyway, batteries were lasting much less time than I expected. So I took the radio home and measured the quiescent current, and sure enough, it shot up by many 10s of milliamps when receiving a stereo programme. So I added a couple of resistors (3k3 IIRC) inside the phono plug - problem solved. Batteries now lasted months again.

Of course, this depends on the device. A line-level output would have an appreciable source impedance - more than 100 ohms, perhaps - so that would be OK (though not ideal). But something designed to drive 32 ohm headphones might have no output impedance at all.

Mark

[BigClick]

Thank you for that explanation Mark. I haven't noticed any issue with my MP3 line in to a mono amp yet but I am aware of what's what now.

[ThePillenwerfer]

For years I got away with just joining the two signal wires together — I knew no better. I always got away with it but my 'Bush' DAB radio doesn't like it at all. I now use 10kΩ resistors and, as Mark says, they extend battery life if nothing else.

— Joe

[bikerhifinut]

Now theres an interesting thing.
I wonder if that's the reason my venerable Creative Zen media player doesnt like being connected to the audio input of the old Clarke & smith School radio. I used a standard stereo RCA socket to mono 6mm Jack adaptor and i was wondering if a lack of buffering was causing the sound to drop out and the player to glitch.
When we are moved after next week I shall try that idea of buffering resistors inline.
A.

[mhennessy]

Yes, anything is possible...

Another potential issue is mono 3.5mm jacks in headphone sockets, as these will short the right channel to ground.

Headphone amplifiers in these sorts portable devices don't have much voltage swing to play with, so can't have much output impedance, if any. And because they expect to drive a 32 ohm load, they have plenty of current reserve.

The resistors really aren't critical; even 100 ohms is better than nothing. 1k might be better. I used 3k3 because I had some in the small 1/8W package that fitted nicely in a phono plug, and 10k is also a good value.

Of course, if you go up in value, you lose signal level - which might or might not be desirable... How much level you lose depends on the input impedance of the following device. For example, it it's 10k, and you used a pair of 10k resistors, then you lose 3.5dB. In practice, input impedances tend to be much higher, but it's impossible to generalise...

[ThePillenwerfer]

Originally Posted by bikerhifinut

... causing the sound to drop out and the player to glitch.

That describes the behavior of my Bush very well: it'd play for a few seconds, then cut out for a while before starting again. The playing-to-not ratio was in inverse proportion to how high its volume was set. I've always assumed that it was triggering some sort of overload protection circuitry.

— Joe

[julie_m]

Or just pulling the battery voltage too low for the decoder IC to operate reliably

Fitting resistors somewhere between the output impedance of each channel and the imput impedance (ideally, the square root of their product; too small and you risk unwanted heavy currents, too large will swallow up the signal) in series with each stereo channel is the simplest way to prevent damage caused by connecting outputs together.