Why do most ccts use the non-inverting input for this device? Surely any possibility of instability is greatly reduced if the o/p is 180 degrees with the i/p.

Even the spec. sheet shows the non-inverting input usually used, except for the AM radio amp.

Also, is there any reason why I can't use the pins 1 - 8 10uF gain boost cap as well as the bass boost (HF cut really) 1k8/33nF pins 5 - 1 components?

I expect these questions have been answered before, but I couldn't find exact answers after a forum search.

Thanks

 

Sure you can boost the gain then reduce the highs. But boosting the gain reduces the negative feedback then distortion is increased.
1.8k with 33nF will cause only very high audio frequencies to be reduced then it will not sound like bass boost, instead it will simply sound muffled with the highs reduced.

 

Thanks Audioguru. I'm not designing it not for great audio anyway, it's to reduce the audio bandwidth of a direct-conversion CW receiver.
Still confuse about the use of inverting/non-inverting inputs though.

 

Speech consonants are important tor understanding what is said. Consonants go up to about 14kHz. When you reduce the bandwidth then everybody says, "Huh, what did you say?" FAILING or SAILING or millions of other completely different words sound the same. That is why narrow bandwidth communications use spelling, numbers and FOXTROT CHARLIE ETC.

Some people think that reproduction of sounds should be non-inverting and other people like me do not, as long as everything has the same phasing.

 

Thanks Audioguru. But it's for CW, (ie. Morse), so bandwidth need only be 500Hz. The signals can be very weak, but initial tests suggest 2 to 5uV at the antenna via a crude direct conversion receiver can give an audible signal in the headphones. This means a lot of audio gain is required.

For anyone interested it's for the Pixie transceiver. This design has its own limitations, so all I'm trying to do is increase the again and reduce the audio bandwidth.

It's pointless trying too hard - there are better ways of doing it, so it's a fine balance between acceptable performance and complexity.

With just one transistor between the detector and the LM386, sensitivity increases significantly. But stability is a problem with poor layout, hence the differences between inverting and non-inverting IC inputs.

 

Still confuse about the use of inverting/non-inverting inputs

The choice of which to use depends upon your application.

• They both have feedback to the negative input, so stability is not a factor in the choice.
• The non-inverting configuration has a higher input impedance and bandwidth [bandwidth is GBW/gain for non-inverting and GWB/(1+gain) for inverting].
• The inverting configuration has a virtual-ground summing junction that allows the summing of more than one input.

In many cases, either configuration will work equally well for the circuit requirements.

 

Hello,

In the sprat-pixie-file there are several options given.
One of the options uses an opamp before the LM386 for filtering.

Here is also an other page about the Pixie 2:
http://www.circuitswamp.org/projects/pixie2.html
There is a PCB layout give on that page.

Bertus

 

Thanks bertus. I found those files. Like many of the ccts, it doesn't use pin 7 to mute, but instead relies on a 1k resistor on the supply line. I can't see why LA3ZA's idea of using pin 7 and a diode doesn't get used. It allows the LM386 to have a proper supply.

Using an op-amp for filtering is starting to get into the law of diminishing returns, particularly as some filtering can be achieved with the LM386 itself.

Using a BPF rather than a LPF may also help reduce the broadcast interference.