Speaker Volume of DIP vs SOIC LM386

Thread Starter

BackyardBrains

Joined Aug 26, 2010
33
I made a bio-amplifier circuit using the LM386N-1 as my audio amplifier. I attached the circuit I made below.

The first version of the board I made of this circuit used an 8-DIP package. I debugged everything, and decided to make an SMT version. This is where it got strange. The same signal going into both the DIP and SMT (SOIC) versions of the circuit produce dramatically different results. The SOIC chip is much, much quieter through the speaker then the DIP chip. ~2-5x quieter.

I checked the boards and everything appears to be the same up to the LM386 on both boards. The part numbers from DigiKey were LM386N-1-ND and LM386M-1-ND respectively. According to the datasheet, they should be identical.

Has anyone have an idea why this could be? I heard that some chips that have the same datasheets can behave differently from different manufacturers. Perhaps this could be causing the difference?
 

Attachments

mtripoli

Joined Feb 9, 2010
44
I'd offer two things: I know the part numbers are the same on Digikey, but the parts themselves actually are -1 parts? They didn't sub a N-3 or N-4?

If not, do you have a heatsink area for the SOIC part? The wattage of the package is smaller for the M part than the N part (the N part has a larger lead-frame so it has more inherent heat-sinking than the M part). It could be thermally limited. Though not ideal, try attaching a heatsink of some kind to the top of the SOIC (you really need some copper on the PCB to be effective). In a pinch, you *could try* squirting the part with some of the "coolant" from a can of air turned upside down. A short squirt; if it gets louder than you know it is in thermal limit. If you soak the board you run the chance of shorting the chip (might blow it out - might not).

EDIT: One other thing (not related to the LM386); if you have a scope, what does the output of each gain stage look like with the input grounded? If you don't have a scope you might be able to measure it with a good voltmeter. I suspect you have a large DC offset on the output (pin 6 of U1 and pin 7 of U4B) {besides the Vcc/2, of course}. Because the output is biased up you might not be aware of it. With the inputs grounded you should see Vcc/2. Remember, you are amplifying the DC offset as well and if not removed in each stage you loose headroom.

On the AD623N part you can use a pot on pin 5 to remove the offset (a 10 or 20 turn is a good choice for this). Why is the (+) tied to Vcc/2? Is this correct? Looking at the data sheet I don't think you need this.

On the TLC2272 the value should be R7||R8 (995.475) so a 1K will be close enough (again on the + leg).

EDIT #2:
Interesting project... Given that this is battery powered: You can use a "high brightness led" and a higher resistor value, this will cut down on current used. R1 and R2 can be much higher in value as well, 100K-200K is not uncommon. You have C1 and C2 on the output of the opamp; this severly limits slew rate. Usually one cap across R2 (or something like an electrolytic and a ceramic, say 10-47uF and 0.01-0.1uF) is all that is needed.

Mike T.
 
Last edited:
Top