Friday, May 31, 2013

Repairing a Synare Sensor

Picture grabbed from video ;)
I was asked to take a look at this strange thing and see if I could repair it.
Supposedly it made fun noises once upon a time, but not anymore. I had never heard of it before and I couldn't really find any schematics, but I decided to look at it anyway.
(I actually fixed this back in Decemeber 2012, but I've been too lazy to post about it until now ;) )

Investigating

This thing had some sort of clamp on the side (visible on the bottom on the photo above). It looks like you were supposed to attach it to something and it would then pick up vibrations and trigger sounds that way.

The only thing I knew was that it was silent. When I opened it, I found there was something mounted in the middle of the PCB. It looked like kind of an old toy microphone capsule. There was something rattling inside it...

This was actually the exact same type of microphone as the one found in the Space Drum I repaired just a couple of weeks earlier!
Since replacing the microphone in the Space Drum with a piezo element seemed to work, I thought it might work in this case as well.




Fixing

So, I attached a couple of wires to a piezo element and soldered them to where the microphone used to sit.

It worked! :)
I temporarily taped the trigger to the case of the unit and tested it and the response was OK after I had tweaked the sensitivity a bit, using the trimmer located in the center of the board.
Satisfied with the results I put some hot glue on the wires attached to the piezo element to make it a bit more robust and used one of those small spongy adhesive poster things to attach it to the inside of the case.

I was going to call it a day but noticed the sounds was a bit weird. As I turned the volume knob down I heard what sounded like distorsion as the pitch of the sound dropped. Hmm?

Improvement?

There was a CA3080 on the output so I assumed it was being used as output amplifier somehow. I don't really have any idea how they work so I read a bit and think I got a reasonable understanding of it.
It looked like the signal was clean until it reached the CA3080 on the output, so there must have been a problem with the output stage.
I was wondering why it would distort when you turned the volume down, but sound good when it was at max. Very strange. Then I realized that the CA3080 has a current output and tries to maintain a certain current, regardless of the load.
To maintain a constant current when the load increases, you have to increase the voltage. The problem is that it's difficult to increase the voltage above the supply voltage.

On the output there was a 10k resistor in parallel with the output jack. At full volume, the amplifier input impedance and the 10k resistor were put in parallel with the volume potentiometer's full resistance.
As the volume was turned down, this 10k and amplifier impedance was connected in parallel only with part of the volume potentiometer, thus increasing the resistance on the output of the CA3080.
The CA3080 had to increase the voltage to keep the same current flowing, but it can't go above the supply voltage obviously, so the voltage clips, the current clips and the audio output clips.

To prove to myself that this was the case, I built the output circuit in LT Spice, assumed it was plugged in to an amplifier with 20k input impedance, and simulated it.

Note the Volume potentiometer is drawn as 4 resistors in series.
The bottom waves are the voltages across the inputs on the CA3080 for 4 different frequencies.
The top waves show the same frequencies on the output, or what the amplifier it's connected to sees.

Now, imagine you turn the volume down to half.
In the schematic it is simulated by connecting R10 to the middle of the volume potentiometer resistor ladder.
Since the parallel load of the output only affects a smaller part of the potentiometer, the total resistance will be greater.
Now look at the simulation for that:


Volume turned down to 50% (R10 connected to R8+R11)
For the same input signal as before, you can see on the top waveforms that the high frequencies look OK but the cyan 550Hz waveform looks a little bit clipped at the top. The purple 50Hz wave is a complete disaster and is clipping like mad!

So, what can be done do avoid this?
Well, if the output resistance seen by the CA3080 is constant, the response will be the same.
My plan was to put a buffer after the volume potentiometer to make sure that no matter that you put the potentiometer at, the resistance would be both constant and low enough for the CA3080 to be able to drive the desired current through it.

Now, I'm not very good at designing circuits. I had a TL061 at home and was hoping that if I used it as a simple buffer it would be enough.
The simulation did actually support my theory quite well!
Original vs buffer-modified
In this picture, the original schematic is to the left and the original output is the BLUE waveform in the topmost window. The buffer-modified schematic is seen to the right with the output waveform visible in RED in the topmost window. I didn't choose those colours...
But, as is clearly shown here, the original blue output clips really badly at low volume, while the buffered red output looks perfectly smooth.

One problem though... I managed to get the buffer op amp to go into self-oscillation at times. I'm not sure if I induced this myself with my crappy USB oscilloscope or if it was just too unstable in unity configuration.

I told the owner of the Synare Sensor about my findings but he seemed happy the way it was working before and wasn't really interested in the modification.
It's a bit of a shame that I didn't get the chance to put my mod to use, but as I wasn't 100% sure how to prevent the oscillations I had seen, perhaps it was for the best.
Anyway, perhaps my little modification can help someone else who has noticed this and want to avoid this annoying effect!

All in all, a successful repair! :)
Here's a small video I made to show the owner that it was working:

No comments:

Post a Comment