Noise/Filter - Transistor selection for Infra Red

[eljay]

I've just built my Noise/Filter module and am pleased to say it's now all working and sounds good.

I did have a few problems finding a suitable transistor to generate the noise though, I tried 9 different BC547s (as recommended), each one seemed to generate an OK white and pink noise but none of them enabled me to set up the Infra Red. The White and Pink noise signals all seemed to be too high (+/- 12V) and the Level trimmer didn't bring the signal down to the recommended (+/- 6V), the Infra Red signal consequently seemed to be stuck permanently high and adjusting the ADJ trimmer had no perceivable effect. I tried a few other types of transistors and got the same results.

I finally tried an old C3371, this works. The White and Pink noise signal levels are now set at about +/- 6V and the Infra Red is now working as expected and has been adjusted with the ADJ trimmer to give a relatively even spread either side of 0V.

As Tony recommended the BC547s as the most likely to produce the best results I am wondering why they really weren't suitable on my board. I have used high tolerance components elsewhere, resistors are all 1% metal film except the 14k resistors which are 0.4W precision resistors because they are what I could get. Any thoughts?

[Synthbuilder]

The problem with using transistors for noise is that we're using them as they are not intended. Which means that no device will be specified for their noise performance when used in reverse bias. Most circuits do not heavily reverse bias their transistors because of the noise and possible long term side effects.

The noise we get from a transistor when used like this is essentially an unwanted side effect of the manufacturing process. This means that some devices will end up being better than others. I find I get the best sounding noise from transistors that make the least noise. Clearly those BC547s are noisy little beggars and the noise quality will probably be pants.

Unfortunately, this all means it's the luck of the draw. So just chose the ones you like the best. I had a whole bunch of BC549s that sounded perfect.

Tony

[Synthbuilder]

II have used high tolerance components elsewhere, resistors are all 1% metal film except the 14k resistors which are 0.4W precision resistors because they are what I could get.

Technically 1% resistors are low tolerance. High tolerance would be 5% and 10%. And if you are as old as me you may remember that these used to be the standard ones you could buy. 1% - luxury.

Nowadays we can get 0.1% for next to nothing and even 0.01% aren't that badly priced.

If you do need some odd value - and I don't tend to use odd values in my projects normally - remember you can put resistors in series to get different values. 14K can be made with the more common 13K and 1K resistors joined together one after the other. Just make a little triangle shaped arrangement with the joined ends sticking upwards.

Tony

[Pav]

Reopened to continue discussion

Hi, Have added the Noise/Filter to my collection.
Seems all working but i have hit the same calibration issues as elJay.
The only change from build guide is U7 where i have TL072ACP instead of LF412.

The problem is Level Calibration - I too am at maximum anticlockwise to get anywhere near +/- 6V.
Its more +8 to -4v.

I did find a BC547C on forth try that would oscillate the IR voltage by approx +/- 200mv
with ADj centered and CV at maximum +/- to give the pleasing mr Jarre wave effect to the noise.
The other transistors would only give me a positive shift or no shift in voltage.
Moving ADJ either way tended to freeze IR. Moving CV much lost the effect too.

Q1 could it be an incorrect resistor value to get some ability to set LEVEL trimmer correctly ?
Q2 is the +/- 200mv variance in IR Output voltage the expected range or should it be wider ?

btw I dont have a C3371 -wonder where eljay found that - i might have a rarish japanese
SC945 npn to try knowing pin order differs. Or a 2N4401 i could try.

Cheers
Pav.

[terjewinther]

I have built several noise modules over the years. My tip: use any old transistor you have in your parts bin for the noise signal. I don't care what it is called, as long as it makes noise. Yes, even non-standard transistors like Japanese or old ones or new ones that you randomly have lying around. I too have met the issue of transistors giving off too much noise, and as Tony said: it is better to choose one that doesn't give off too hot a signal.

[Synthbuilder]

As Terje says, use anything in there that will give you a nice sounding noise with no obvious crackles or occasional phuts. You can even try zener diodes too (cathode to pin 3, anode to pin 2). If so, use a zener diode with a breakdown voltage of 6.8V or over.

[Pav]

>>is the +/- 200mv variance in IR Output voltage the expected range or should it be wider ?

I would love for someone to check their IR out with scope to see if they get a different peak +/-
Then I know whether to stop trying transistors.

[Pav]

Hmm now im also getting some crosstalk / interference from a clock/lfo something creating a beating pattern with the noise. Ive unplugged everything save the mixer which fronts speakers and the noise module. Wonder what that might be ?

At least the last module build for 2021 - a FourMix works first time of asking for me.

[eljay]

I have measured the following results, having literally just tried out a selection of BC547s.

I started by measuring the output from the Pink noise socket on my scope. I tried five different Bc547s until I found one that gave a range in the vicinity of 12V peak to peak and then adjusted the trimmer to achieve 12VPP. Interestingly, when I then measured the White noise output, although the 'core' of the output appeared to be roughly 12VPP, the actual peak to peak readings were double that giving a range of -12V to +12V approx.

Finally I measured the Infra Red output and tweaked ADJ so that I got an even spread either side of 0V. The voltage range measured over several minutes is -2V to +2V give or take a few millivolts.

I don't know if these results are consistent with the experience of others but this seems to work ok.

[Pav]

ta...ill aim to get the same

[Pav]

One final question ... having tested with a pack of 10 transistors and made them pass current the wrong way (reverse bias ??) will they still usable as transistors ?

[Synthbuilder]

One final question ... having tested with a pack of 10 transistors and made them pass current the wrong way (reverse bias ??) will they still usable as transistors ?

I've wondered about this. Any degradation due to current limited reverse biasing will probably be gradual - over years I would think - and a quick test will probably not affect them.

That said, I separate them into a bag and use them only for switching applications rather than anything linear like amplifiers, ladder filters, etc.

[KSS]

I def separate them out. As opposed to throw them out.
As to how fast they might degrade? That's going to depend on the circuit -and transistor- used.

[euphonic]

I've just built my Noise/Filter module and am pleased to say it's now all working and sounds good.

I did have a few problems finding a suitable transistor to generate the noise though, I tried 9 different BC547s (as recommended), each one seemed to generate an OK white and pink noise but none of them enabled me to set up the Infra Red. The White and Pink noise signals all seemed to be too high (+/- 12V) and the Level trimmer didn't bring the signal down to the recommended (+/- 6V), the Infra Red signal consequently seemed to be stuck permanently high and adjusting the ADJ trimmer had no perceivable effect. I tried a few other types of transistors and got the same results.

I finally tried an old C3371, this works. The White and Pink noise signal levels are now set at about +/- 6V and the Infra Red is now working as expected and has been adjusted with the ADJ trimmer to give a relatively even spread either side of 0V.

As Tony recommended the BC547s as the most likely to produce the best results I am wondering why they really weren't suitable on my board. I have used high tolerance components elsewhere, resistors are all 1% metal film except the 14k resistors which are 0.4W precision resistors because they are what I could get. Any thoughts?

Hi Eljay, Thanks for sharing. My eye just caught your thread, as I am working on a Noise/Filter module right now. Good to know that other transistors (and even zener diodes) could be used as well. And I also learned to keep the tried transistors separate. Really helpful!!!!

[Noiseconformist]

As Terje says, use anything in there that will give you a nice sounding noise with no obvious crackles or occasional phuts. You can even try zener diodes too (cathode to pin 3, anode to pin 2). If so, use a zener diode with a breakdown voltage of 6.8V or over.

For the most recent replacement I picked an old BC337 from the drawer and I'm quite happy with it!