LM386 minimal circuit

[audioguru]

The circuit in post #11 produces One What! That is 1/2 Watt of audio and another 1/2 Watt of severe clipping distortion.
The 0.1uF input capacitor cuts frequencies below 380Hz (-3dB is 192Hz) so there will be no bass sounds.
The small output capacitor also cuts bass sounds.

 

[gophert]

I propose the datasheet

ak

Great idea! What page did you see the explaination of why the zobel network is needed?

 

[audioguru]

Great idea! What page did you see the explaination of why the zobel network is needed?

The LM380 came out before the similar LM386. An Applications document for the LM380 explains that the output without the Zobel network oscillates at an AM radio station frequency.

 

[AnalogKid]

The 0.1uF input capacitor cuts frequencies below 380Hz so there will be no bass sounds. .

I think that is a bit overstated. Note that the term "cutoff frequency" often is misunderstood by those with less experience or training.

First, by my math, the corner frequency is 159 Hz, not 192 Hz.

At 318 Hz the signal is attenuated by only 11%, and by 29% at 159 Hz. Input frequencies are attenuated by 50% or more only when below 80 Hz. So while the bottom end will be a bit thin, this hardly means there will be "no bass sounds".

ak

 

[gary350]

The circuit in post #11 produces One What! That is 1/2 Watt of audio and another 1/2 Watt of severe clipping distortion.
The 0.1uF input capacitor cuts frequencies below 380Hz (-3dB is 192Hz) so there will be no bass sounds.
The small output capacitor also cuts bass sounds.

I looked at several datasheets, 1 place claims 1.5 watts output. Most circuits that shows IC pin locations with parts shows 10K variable resistor pin 3 with no capacitor but some show .1 uf capacitor. Most datasheets show 10K resistor in series with .05uf capacitor on pin 5, only one circuit shows 10 ohm resistor. Output capacitor is not the same for pin 5 on all datasheets either. It is hard to know what is correct datasheets are not all the same. WHY such a variation in parts values on datasheets? I expect Google search for LM386 circuits to be different most of those circuits don't work. There is usually good information in, TTL books, CMOS books, Transistor Logic books, Semiconductor books.

Is there a chart somewhere that shows Hz for each capacitor value?

 

[Nigel Goodwin]

I looked at several datasheets, 1 place claims 1.5 watts output.

Do the simple maths, power output is directly proportional to power supply voltage and speaker impedance, although you'll destroy the chip is you try and push power too high.

Most circuits that shows IC pin locations with parts shows 10K variable resistor pin 3 with no capacitor but some show .1 uf capacitor. Most datasheets show 10K resistor in series with .05uf capacitor on pin 5, only one circuit shows 10 ohm resistor.

Looking at the circuits above none show a 10K zobel resistor (because it wouldn't work), although some show a 10K feedback resistor to alter gain.

A quick google shows this one:

https://www.ti.com/lit/ds/symlink/lm386.pdf

Which clearly shows 10 ohm and 0.05uF in the zobel network, personally I usually pick 10 ohm and 0.1uF (but I've never used an LM386), it's not at all critical. It also gives sensible power output figures, even though they are at 10% distortion.

In fact I was once asked to repair a bass guitar amplifier, which had gone up in smoke while my daughter was playing through it - rather bizarrely it still worked, even as it smoked. When I took it to pieces, I saw that the zobel capacitor (which was an electrolytic?) had gone up in smoke - as it was a locally made design I contacted the manufacturer to be told they had used numerous different values during production, including an electrolytic. IN the r=end I simply fitted my usual 10 ohms and 0.1uF

Output capacitor is not the same for pin 5 on all datasheets either. It is hard to know what is correct datasheets are not all the same. WHY such a variation in parts values on datasheets? I expect Google search for LM386 circuits to be different most of those circuits don't work. There is usually good information in, TTL books, CMOS books, Transistor Logic books, Semiconductor books.

Is there a chart somewhere that shows Hz for each capacitor value?

Simply work out the capacitive reactance for the frequency you're concerned with - you want the output capacitor to be less than the speaker impedance at the lowest frequency you're interested in. Bear in mind there's no advantage is having 20Hz capability for a 2 inch speaker, hence values such as 220uF.

 

[audioguru]

First, by my math, the corner frequency is 159 Hz, not 192 Hz.

Nope. The 10k volume control is parallel to the 50k input resistance of the LM386 and the 0.1uf capacitor feeding the resulting 8.33k produces -3db at 192 Hz.

At 318 Hz the signal is attenuated by only 11%, and by 29% at 159 Hz. Input frequencies are attenuated by 50% or more only when below 80 Hz. So while the bottom end will be a bit thin, this hardly means there will be "no bass sounds".

Nope again. The low frequencies begin dropping at 192Hz x 5= 960Hz but will be noticeable at 400Hz. 20Hz to 40Hz will be WAY DOWN producing no bass.

 

[audioguru]

I looked at several datasheets, 1 place claims 1.5 watts output.

There is only one datasheet, not several. The LM386 was made by National Semi who was purchased a few years ago by Texas Instruments.
The datasheet shows 0.7W into 8 ohms when the supply is 9V and the clipping distortion is horrible at 10%. A graph shows that clipping begins at 6Vp-p which is 0.56W and some calculations show that there is low distortion at 0.45W and at lower power.
The datasheet shows 1W into a 16 ohm speaker that is not used today and a 12V supply. A graph shows a small increase in output power with an 8 ohm speaker with a 12V supply but much more heating.

Most circuits that shows IC pin locations with parts shows 10K variable resistor pin 3 with no capacitor but some show .1 uF capacitor.

If the signal source has DC then the input capacitor blocks the DC from upsetting the volume control and the IC. If you do not want bass then use 0.1uF as I show above but the calculation is simple for the input and output capacitor values for the lowest frequency you want.

Most datasheets show 10K resistor in series with .05uf capacitor on pin 5, only one circuit shows 10 ohm resistor.

The single datasheet shows that the Zobel network at the output is 0.05uF in series with 10 ohms.

Output capacitor is not the same for pin 5 on all datasheets either. It is hard to know what is correct datasheets are not all the same.

The value of the output capacitor is simply calculated with the speaker impedance for the lowest frequency you want because the capacitor and its load are a highpass filter.

WHY such a variation in parts values on datasheets?

The input and output capacitor values are different because some people want good bass and other people have tiny speakers that cannot produce any bass or have a telephone input signal that has no bass.

I expect Google search for LM386 circuits to be different most of those circuits don't work. There is usually good information in, TTL books, CMOS books, Transistor Logic books, Semiconductor books.

Some websites post errors but the datasheet and education are always correct.

Is there a chart somewhere that shows Hz for each capacitor value?

When the level is reduced -3dB then everybody notices the small reduction in level. -3dB is half the power or is 0.707 times the voltage or current.
The -3dB cutoff frequency calculation involves the capacitor value, the load impedance and pi. The calculation for the capacitor value is 1 divided by (2 x pi x f x R). For example, a 91Hz cutoff happens when the speaker is 8 ohms and the capacitor is 220uF.

 

[AnalogKid]

The 10k volume control is parallel to the 50k input resistance of the LM386 and the 0.1uf capacitor feeding the resulting 8.33k produces -3db at 192 Hz.

Only at with the volume control at the max setting. At lower settings the corner frequency shifts downward.

The low frequencies begin dropping at 192Hz x 5= 960Hz but will be noticeable at 400Hz. 20Hz to 40Hz will be WAY DOWN producing no bass.

I know how single-pole rolloffs work, but didn't want to clog up the response with the responses at 5, 4, 3, 2, and 1 time constant. Also, very few musical instruments and even fewer human voices produce significant energy at of below 40 Hz. I think ther general concept of "bass" sounds is a bit higher than 40 Hz.

ak

 

[gary350]

What is zobel ?

I did Google search LM386 datasheat I can not find one that says 10K. ??? I guess your right they say 10 ohms.

I did find this.

 

[audioguru]

The students obviously copied the datasheet circuit wrong. YOU should copy the datasheet, then you will not copy student errors.

 

[ChrisP58]

What is zobel ?

I did Google search LM386 datasheat I can not find one that says 10K. ??? I guess your right they say 10 ohms.

I did find this.

You need to understand the difference between the recommended application circuits from the manufacturers datasheet and the many (often wrong) third party application circuits that litter the internet.

 

[gary350]

What is zobal?

 

[Nigel Goodwin]

What is zobal?

Have you never heard of google?

 

[JimB]

What is zobal?

An incorrect spelling of Zobel.

JimB

 

[MichaelaJoy]

https://lmgtfy.com/?q=zobel+network

Direct Link: https://en.wikipedia.org/wiki/Zobel_network

gary350: In your schematic of Post #11, the Zobel Network is R4,C5, and C6.

 

[Nigel Goodwin]

gary350: In your schematic of Post #11, the Zobel Network is R4,C5, and C6.

I wouldn't agree with C6 - which isn't even present for a split supply amplifier. The zobel network is just R4 and C5.

 

[AnalogKid]

From Wikipedia:
"This network is more akin to the power factor correction circuits used in electrical power distribution..."

DING - light just came on. Deep down inside I knew this, but had never seen it in words.

ak

 

[MichaelaJoy]

I wouldn't agree with C6 - which isn't even present for a split supply amplifier. The zobel network is just R4 and C5.

I stand corrected.

 

[unclejed613]

with the values shown in the data sheet, the zobel network's corner frequency is 318 khz