LM386 Instability Issue

[jjw]

I have made a detector based on NE612 with 4046 oscillator and it works fine.
Here is a simple detector with switching mixer:
**broken link removed**
It uses 74HCU04 inverter as amplifier and oscillator! I modified it to use electret microphone.
I removed the differential transistor pair, connected the microphone to one inverter, after that another inverter to make 180 degrees phase shift.
It runs about 50 hours with two Nimh batteries in series
I am using it to listen at fixed 30kHz frequency the Northern bat which is the most common species in Helsinki.

 

[rogs]

I have to say I think using that circuit , with a fixed frequency local oscillator and the 40 KHz transducer is very complex option for such a limited result?
For a fixed frequency device I would suggest a much simpler frequency division type detector... I built one of those many years ago, using this very simple circuit:
(which we discussed a while back in this thread: https://www.electro-tech-online.com/threads/lm386-bat-detector-stability.127619/ )

However, without using an electret mic and a variable frequency local oscillator I found it difficult to distinguish between different species, and missed a lot of the species altogether!
So a direct conversion heterodyne version seemed a better idea.... and using cheap, easy to find components, rather than specialised devices appealed to me!

But that is one of the good things about public forums like this. Different folk can put forward different ideas, and we can all benefit from gaining knowledge about alternative approaches to a specific task......

 

[jjw]

The frequency of the oscillator can be made variable by replacing one of the resistors with a potentiometer.
I am just using it for lazy listening on the couch, the detector on the window ledge and connected to stereo amp.
I made a youtube video with sound from this detector:

The sound is from detectors headphone output->stereo amp->loudspeaker->pocket camera microphone, so the quality is not very good.
I have the NE612 version with frequency range 30-60kHz which I use outside.

 

[rogs]

I have the NE612 version with frequency range 30-60kHz which I use outside.

If you have a sketch of your variable NE612 circuit that you could post, that might be useful to the original poster, as he is already experimenting with the NE612.
He might be able to use your circuit to help identify the problems he is experiencing with his present circuit?...

 

[jjw]

It is a copy of this:https://bertrik.sikken.nl/bat/ne612het.htm

 

[breakshift]

Right - I've come back to this now after a brief absence. I've built a Wien bridge variable oscillator that works quite well. I can vary the frequency from DC to way over my target of 100kHz. The waveform is quite distorted at the higher end of course, but is quite nice at the lower end. However when I connected it up to the NE612 mixer I was having some trouble. Rogs, you said that you used a switching mixer in your design so I looked it up on google. I discovered that the switching mixer is the typical operating mode for packaged mixers, and upon closer inspection of the NE612 datasheet it does appear to be of this operating mode. I also discovered that the local oscillator amplitude should be considerably higher than the signal amplitude. In my circuit this does improve the quality of the mixer output. However in the datasheet it says that the signal input should be 200mV and the LO should be between 200mV and 300mV, which is not 'considerably' higher. I'm driving the LO way over this amplitude as it improves the output signal significantly.

Now I'm not so sure why I bothered with the Wien bridge variable sinewave oscillator. If the NE612 is a switching mixer then the LO might as well be a square wave, right? unclejed613 stated in post #39 that it's not designed to operate with a square wave LO.. ? Your design, Rogs, is fundamentally the same as what I'm attempting to do, I believe, just with different components. Would you agree? I agree that using a Wien bridge LO may be slightly overkill, especially as I now believe it to be no better than a square wave LO in this application. You used a multiplexer for the mixing whereas I'm using the double balanced NE612 mixer - was your method balanced or did you not find this to be necessary?

Thanks for the info about bats in the UK. There are plenty of bats around where I live (yes, in the UK) but I don't know what species. Looking forward to investigating!

 

[rogs]

Your design, Rogs, is fundamentally the same as what I'm attempting to do, I believe, just with different components. Would you agree?........ You used a multiplexer for the mixing whereas I'm using the double balanced NE612 mixer - was your method balanced or did you not find this to be necessary?

Well, it's fundamentally the same - in that they're both mixers!..... My method is not 'double balanced', but it is balanced in the sense that the multiplexer needs a differential ('balanced') version of the linear component of the actual ultrasonic input signal, from the mic preamp.
Using a simple CMOS analogue multiplexer eliminates the need for worrying about relative signal amplitudes..... the LO source is simply used to 'switch' the analogue channels, as a 'rail to rail' voltage square wave.
I don't have any experience with the NE612 device.....I just used a simple off the shelf 'bog standard' CMOS multiplexer, to achieve what I wanted to do.
I wasn't really into making my project any more complex than it needed to be..... and using a 500MHz R.F. mixer - like the NE612 - does seem to me to be a bit of overkill, for a simple ultrasonic 'audio' type project ...
All IMHO, of course

 

[breakshift]

Yes I see your point and agree with you. I will probably get some multiplexers in and have a play with them, just out of interest and for comparison. The NE612 is quite cheap though and it does everything I need, in a small package. Plus I think the double balanced side of things is fairly important. If your mixer is not balanced, then your output signal will contain components at the LO freq and the signal freq which could prove to be troublesome. Was this the case for you? About eliminating the need for worrying about relative signal amplitudes - yes I see your point here also. The NE612 is a switching mixer type as I mentioned earlier but this is implemented though the use of a Gilbert cell, which provides the double balanced properties. So I believe this is why the output signal is significantly affected by the amplitude of the LO. So I suppose the primary reason for using the NE612 over a more simple multiplexer mixer architecture is the LO and signal rejection in the output.

I'm still unsure if there's any advantage in using the sinewave LO I built. I'll compare it to the square wave LO I was using previously.

 

[rogs]

I think you may be expecting a bit too much from a simple 'heterodyne' bat detector?
The only way to actually 'hear' the bats in any real detail is to record their original ultrasound transmissions as linear 'audio' sampled at - say - 192KHz onto a solid state recorder with sufficient bandwidth. And with a good quality microphone with an ultrasonic capability..
Then this recording can be analysed, by replaying the recorded file at a much lower speed.
There is special equipment available to do this, but it's very expensive. The only 'commercial' solid state recorder I know that could handle this is something like the Korg MR2 .. and I think even that's now been discontinued!

A heterodyne detector can only reproduce the difference (IF) signal between the ultrasound and the LO. These ultrasound 'bursts' do seem to have an FM component, in some species, and that can be heard as what is often described as a 'wet slap' type of sound.
A fixed frequency is just reproduced as a tuned 'click', similar to the simple 'frequency division' type of detector.
In neither case do you really care whether the 'audio' is above or below the LO frequency. A simple low pass filter following the mixer works just fine...

The link to the maths paper in my previous post (here it is again: https://www.analog.com/library/analogdialogue/archives/43-09/edch 4 rf if.pdf ) describes almost exactly what I did, in figure 4.4 and 4.5.

The actual device I built (based around the mixer circuit in post#42) was used to make the recording of soprano pippistrelles I linked to here: https://www.jp137.com/las/soppips.mp3

The LO was set to about 55KHz for that recording.

After a while, you get to recognise the characteristics of different species, both from the frequency range, and the type of 'sound'.
Noctules, for example make a sort of loud 'chonk' sound at around 25KHz. Daubentons (usually found near water) sound like a 'crackling' fire at around 50KHz... and so on......

But as these are only 'fake' sounds,there's no point (IMHO) in trying to make the device any more complex.
As you will discover, the microphone - and microphone preamp - are the most important part of the circuit ... they need to be quite sensitive to be able to find these little critters at distance of up to 30 or 40 metres.
Remember, they move fast, erratically, and it's in the dark!.....

 

[breakshift]

I'm probably over thinking things. I'm always doing it! But I'm not expecting much from this device, I'm just trying to understand exactly how things work and the pros and cons of doing things different ways. I built a variable relaxation oscillator as a square wave LO, and compared it to the sine(ish) LO. The sine LO produces a noticeably better output from the mixer, so I'll stick with that. So that's that sorted.

Previously I was having real trouble with the NE612 but it seems to be much more reliable now I'm driving the LO harder (albeit against the datasheet recommendation..). So hopefully I'm getting somewhere now!

I always over think things - just the way my mind works, I have to understand every little detail. That's probably why this is taking me a while to finish..!

 

[chemelec]

I was looking at this article and noticed a few things that didn't seem correct or needed.
33K across the 4069, Normally recommended as 10M.
Also I don't see a need for the LP Filter.

His Design:
**broken link removed**

I wanted to contact the Designer in Japan about this, But I don't see any Contact Info for him.

So I did a Redesign of it, also using a 9 Volt Battery.
I also tested it, using a 24 Khz pickup. (.0022 Cap)
See My Design, ALSO Works REALLY GOOD.

 

[audioguru]

I was looking at this article and noticed a few things that didn't seem correct or needed.
33K across the 4069, Normally recommended as 10M.

It is not a low current higher voltage 4069, instead it is a higher current low voltage 74HC04 that works fine as a linear amplifier with a 33k negative feedback resistor.

Also I don't see a need for the LP Filter.

It removes high frequency "hash" from the mixer.

His Design:
**broken link removed**

I wanted to contact the Designer in Japan about this, But I don't see any Contact Info for him.

So I did a Redesign of it, also using a 9 Volt Battery.
I also tested it, using a 24 Khz pickup. (.0022 Cap)
See My Design, ALSO Works REALLY GOOD.

 

[breakshift]

I wonder if anyone could help me understand the sounds I'm getting from my circuit. At the moment I've got the LM386 removed entirely, and I've just got some headphones connected directly to the mixer output to simplify things for the time being (low volume of course but sounds are quite clear via the headphones). Basically the circuit is performing more or less as expected.

I have a 40kHz transmitter I'm using to test the receiver. With the transmitter off, the output is quiet across the LO range, as expected. With the transmitter on and the receiver LO between 30kHz and 50kHz I can clearly hear the difference frequency as an audible tone sweeping from high pitch to low pitch and back to high. Above and below this range however I shouldn't hear anything, but I can hear further tones sweeping up and down. My best guess is that these unwanted tones are the result of LO harmonics mixing with the signal frequency and its harmonics to produce further sounds in the audible range. I recorded the output sounds to upload but it appears that I can't upload WMA or MP3 - is it possible to upload audio files?

So the circuit is working in the sense that it'll pick up ultrasound signals and downconvert to the audible range. However the unwanted sounds render the circuit ineffective at determining the received frequency, because there are audible sounds generated when the LO is way off from the received signal. In fact there are audible sounds pretty well right across the LO range, with varying magnitude/volume.

Chemelec when you were doing your bat detector did you test it with a transmitter? Could you hear the sounds as described above?

I've taken a snap of my circuit as it currently for further info.

 

[JimB]

You spurious signals sound as though they are a result of overdriving the mixer.
Try reducing the amplitude of the signal from your test transmitter.

JimB

 

[chemelec]

I Haven't seen a Ultrasonic Pickup, like the one your showing in that Picture?
"That Black Thing, With the Silver Metal End?
Where did you get it?
What Specifications?

I am not sure What sound your hearing, But on mine I believe I am pretty Much Just getting the correct sounds.
Yes I also used a Signal Generator (With Sine, Square or Triangle Wave) and a 40 kHz Transmitter for Testing purposes.

The Most Interesting Test I did: Was just Rubbing my "Thumb and Fore-Finger" Together.
I can Easily hear it CLEARLY at over a 2 Meter Distance from the Reciever.

 

[breakshift]

Reducing the Rx signal amplitude results in the same audible characteristics at the output just quieter. At some point of course the volume of the output is reduced to the point where the unwanted tones become inaudible leaving only the correct tones. What I'm not sure of is if this is reducing the spurious frequencies amplitude relative to the intended signal. I will add amplification to the end of the mixer to see if this is the case.

I'm just using a bog standard 40kHz piezo receiver, seen in the top right underneath a headphones jack (with headphones plugged in) which I believe you may be mistaking for something else. I intend to return to the MEMS mic once I'm satisfied with circuit performance using the 40kHz piezo mic. For information, top-right: mic and op-amp buffer, top-left: op-amp relaxation oscillator with trimmer pots to adjust freq and amplitude, middle-bottom: mixer, very-bottom: 5V regulator and supply decoupling caps.

Currently I've not got any input amplification, just the op-amp buffer before input to the mixer. This is because I can just move the test transmitter around to achieve the signal amplitude I want.

I don't have a sig-gen but I'm considering purchasing one. My scope has a built in sig-gen that I could pay to have enabled. I think it would be really useful. It would allow me to test the circuit with a sine wave LO and see if the spurious signals disappear. And also to test the circuits freq response across the entire range as opposed to just 40kHz.

I tried rubbing my fingers together but didn't hear anything coming from my receiver! Before the test transmitter I was just jingling my keys and I could hear that, but the receiver output was drowned out by the direct noise of it.

 

[chemelec]

EDIT:
I Assume your using the NE612 Circuit.
It is designed for an Electric Condensor Mic. "MCE2500"
It has a Much Higher Output and Wider Frequency Response than a 40Khz Reciever.
"Available from a company in the Chech Republic"
(The Author clams bandwidth from 20 Hz to about 60 Khz.)
I have a few doubts on the that 60, but it does look pretty good.

And 40Khz Recievers Like what your using, only have a Bandwidth of about 2 Khz, either side of Center Frequency.

And What Exactially is your Transmitter?

You Don't say: "Where you Live?" (COUNTRY?)

But you Might consider getting or making the PCB and building up my circuit.
(or a Complete Kit)
http://chemelec.com/Projects/Ultrasonic-2/Ultrasonic-2.htm

On My Circuit, You should DEFINATELY Hear the Fingers Rubbing Together Both at a Close Distance and also at a Distance of OVER 1 Meter away. (Possibly even 3 Meters.)

 

[breakshift]

Yes I'm pretty much using the direct conversion NE612 circuit. I will eventually be using a MEMS mic with a decent bandwidth into the ultrasound region, but for now I'm just using the 40kHz transceiver (can be used as transmitter or receiver) because the MEMS mic is a tiny SMD device (intended for hearing aids) and was really fiddly to prototype with. So for convenience I swapped for the time being. The transmitter is the same device, and like you say its very narrowband at just 40kHz really.

You guessed right earlier in the thread - I'm in the UK.

Why do you use a 1k resistor to ground across the mic? Also, probably a simple answer to this, but why do you convert the input signal to differential mode?

 

[chemelec]

In one of your Posts, You Mentioned the BBC, So My guess was you were in the UK.

It is a 10K Resistor, Lowers input impedance, so reduces Hum/Noise pickup.
The Origional Designer of this used the differential input.
I didn't change that, But I suspect it also helps in reducing Noise.

Today,I Just Ordered a couple of These Recievers from Mouser.
They Should be Pretty Good Also and may also do OK at 60khz.
https://www.mouser.com/catalog/specsheets/KT-400244.pdf

Not Sure if the Frequency Response of the reciever really Changes much of anything.
I also tried a 24 Khz Reciever, and Detection of SAME Sounds DIDN'T Change much at all.

My point was if you really wanted to build my Circuit, I would consider sending you the PCB at No Cost.
The PCB just costs me a Few Cent, It is the Mailing cost that is somewhat Expensive.

 

[jjw]

I was looking at this article and noticed a few things that didn't seem correct or needed.
33K across the 4069, Normally recommended as 10M.
Also I don't see a need for the LP Filter.

Hi
**broken link removed**

I wanted to contact the Designer in Japan about this, But I don't see any Contact Info for him.

---------------------------------------------

His Email address is on the lower left corner on his page.