LM386 audio amp circuit critique required

Discussion in 'The Projects Forum' started by Max12345, Aug 24, 2016.

  1. Max12345

    Thread Starter Member

    Aug 27, 2013
    63
    0
    lm386.jpg Greetings all
    I have built this circuit and am using it in a truck that I drive in the port of Cape Town, ferrying containers.
    We are allocated a different one on every shift.
    The environment is noisy and we are not allowed to use headphones as we have 2-way radios.
    The truck is equipped with a computer and therefore has a USB input.

    This circuit started off fairly simply, but as I read more and more about it, I added more and more sections to it.
    I do not use the stereo inputs, only mono and I do not use the switchable supply yet. I only use a 9V battery.
    It seems to work fine, but I'm wondering if there is a bit of overkill going on (lol).
    Pins 5 and 8 seem to be overloaded with the bass boost, the anti-hiss, the Zobel network, the output to the speaker and the gain control.
    I originally had a 220µ cap on the output, but changed this to 1000µ and inserted it directly on the speaker as it was too big for the veroboard. When I added the anti-hiss circuit, I didn't want to run another wire to the speaker, so I reverted to the 220µ on the board.

    Are all of these too much? Any critical remarks, comments or suggestions would be greatly appreciated.
    (I'm expecting lots of negative remarks, so please don't hold back).

    Thanks, Max
     
  2. ci139

    Member

    Jul 11, 2016
    341
    38
    the overkill would be going on when you pass the point where your satisfied with the operation of your amplifier . . .

    http://stephan.win31.de/music.htm#chipamps
    ... i have one of these (NJM386) laying around comparing to LM386 - they don't show output coupling capacitor greater than 250µF - the OpAmps usually do not like high capacitive loads on their output (depends on internal feedback mechanism) delays phase - i can't comment on this particular one

    (the inputs have 50kΩ to GND - SC-ing the pin2 to GND sets the OpAmp into maximum current mode - not necessarily good at all occasions -- input sensitivity v. frequencies , supply voltages)

    the datasheet's C7 (the C10 on this schem.) is shown to improve supply rejection at lower frequencies 10µF ~400Hz 50µF ~100Hz - i know better DC-OpAmps i don't want to go guessing about ~AC ones
     
  3. Kjeldgaard

    Member

    Apr 7, 2016
    73
    17
    My thoughts:

    R1 value appears to be far too high, the speaker current go through this resistance, and if the C7 should help to supply the current, so it should be moved to the other side of R1.
     
  4. ci139

    Member

    Jul 11, 2016
    341
    38
    you may be right about decreasing R1 but i won't recommend re-positioning C7 coz R1 & C3 disable LM386 to develop undesired oscillations (most likely)
     
  5. ci139

    Member

    Jul 11, 2016
    341
    38
    your src: http://www.eevblog.com/forum/beginners/inputoutput-caps-on-simple-headphone-amp/
    to set up the zobel network correctly you must follow next (i don't know how much sense would it make for noisy listening env. - the already implemented v. may be satisfactory enough - and the following - an overkill)
    http://diy-speakers-design.blogspot.com/2009/01/zobel-network-design.html
    http://sound.westhost.com/lr-passive.htm
    there is however more than the zobel to receive flat sound spectra involving distance from the speakers , speakers orientation, room acoustics . . . brain-physiologic response (your current health - vitamin contents) - we had a dude with ideal hearing (you tap a key on a piano - and he can tell the pitch and wheather it's in tune) - he had invested more than a year income to listenig room ??? - said it'll hurt it's ear otherwise - holding my breath :eek:
     
  6. ronv

    AAC Fanatic!

    Nov 12, 2008
    3,290
    1,255
    Just my 2 cents.
    I would get rid of R1 it is not needed (or wanted). The 386 is kind of sensitive to the battery voltage sagging which is what the 10 ohms will look like.
    The boost is nice.
    The anti hiss just lowers the gain at high frequencies. Kind of like a treble control. So if you are missing the high notes, take it out.
    Make C1 bigger. At least 1 Ufd. This will really improve the base.
    Set the volume control to maximum then adjust RV2 to the highest volume you want. Then replace it with a fixed resistor of that value. This will give you the best noise performance.
    The zorbel cap isn't the data sheet value, but probably makes no difference to the sound.
     
  7. Max12345

    Thread Starter Member

    Aug 27, 2013
    63
    0
    ci139, somehow I don't think that Mozart was entirely satisfied with his Requiem when he was done, or that Michelangelo was entirely satisfied with the Mona Lisa......
    The Japanese word Kaizen refers to any improvement, one-time or continuous, large or small, that can be made to almost anything...

    Kjeldgaard, thanks for the input

    ronv, this is what I was looking for, some constructive criticism. I'm going to implement all your suggestions:
    1. Remove R1
    2. Remove the anti-hiss circuit
    3. Replace C1 with a 1µ ceramic cap
    4. Replace C4 with a 0.047µ ceramic cap (can't find 0.05µ)
    5. Do the volume control/gain suggestion.

    Thank you
    Max
     
  8. ci139

    Member

    Jul 11, 2016
    341
    38
    i did some NJM386 research - /!\ what it seems (( i don fully grasp this thing yet )) is that : /!\ -- the bigger the output dc uncoupler capacitor - C9 - the better , at least in Spice i used 4700µF to avoid the Max. amplitude sine-wave to delay it's wave tops and it likely could be even bigger

    --- also . . . what i studied is that both inputs (( pins 2(-) & 3(+) )) are best to connect throuh 10u or greater capacitor and then set the zero balance by additional about 3kΩ resistor from input pin to Uee (the one that needs it) ((in parallel with the internal 50kΩ biasing R)) OR (( alternate )) ground the pin2(-) and →*→dc uncouple 10µF or higher + 3kΩ bias the pin3 if required (( on scematic cut between RV1-C6 and LM386 p3 otherwise the invading dc offsets the output - but as it's ~AC amplifier perhaps it's output should drift along with average input near Ucc !! ← if it does that - keeps near ground ?? i donno))
     
  9. ronv

    AAC Fanatic!

    Nov 12, 2008
    3,290
    1,255
    Let us know how it sounds after you finish.
     
  10. ronv

    AAC Fanatic!

    Nov 12, 2008
    3,290
    1,255
    If you run the AC analysis of the circuit in Spice you will be able to see the effects of the output capacitor on frequency response. Leave out the hiss and boost circuits until you get it working.
    The way the circuit is designed the output is set at 1/2 of Vcc.
     
  11. ci139

    Member

    Jul 11, 2016
    341
    38
    i don't run an anything at that point of time - though being a simple circuit - it's actually quite a complex one ← the LM386 -- i got it *working (read doing something resembles to*) at voltage range 9V to 16V the d.-sht spec.-s "• Wide supply voltage range: 4V–12V or 5V–18V" and my zero is drifting depending supply - however i found the next interesting . . . varying the vol with 3 synchronised potentiometers (see attc. or a sequential plot capture) ??? by the fq. resp. plot there's like nothing else required but the input and/or output bandpasses

    the task futher is quite simple o_O it seems that out of 8000000000 people we have luckily 1 who had mind to make an updated model - so it eighter verifies or not - good hint in that file is around NPN.h21e=150(to 400) PNP.h21e=40(to 200) ((ratio:: 3.75 to 2)) . . . it's all about fun! :p ((and shh..))

    "that file" actually suits and works for Spice frequency pass/analysis -- i don't see it in a scope of this thread to upload the results - all the gizmos at right of the Amp will /!\ quite sensitively /!\ (re-)shape the (loopback/re-)pass spectrum - if you can't hook your actual grid to a real frequency analyser - there's no point to suggest anything . . . also what is at left from the Amp adds to the "shaping" - you can have some passive or active filters there . . .
    . . . ups ... the original name was lm386.cir created by Helmut Sennewald - i downloaded multiple and "indexed"

    . . . i got my NJM working up to using it in this tread's grid - where it came out that above 1kHz it's responce goes chaotic - so i used the only candidate available :(:(:( - although the NJM is more complex it's setup is actually more intuitive also the LM actually has a CC or CV stabilizing it's differential input stage coz the resistance from pin6 to pin5 is not 15k+15k+150+1k35+15k but acually about 25MΩ ??? so they use a highly simplified "functional analog" on a datasheet . . . i just foud out that i have both chips laying around o_O
    PICT520G.gif
     
    Last edited: Aug 28, 2016
  12. Max12345

    Thread Starter Member

    Aug 27, 2013
    63
    0
    Hi ronv
    Just thought I'd let you know that my amp sounds great now, loud and crisp, with additional treble which I like.
    Thanks again.
    The only thing I didn't do is to change C1 to 1µ. I did but the amp sounded distorted so I reverted to the 10n.
     
    Last edited: Aug 28, 2016
Loading...