Noisy TB-303 clone, how to fix?

[Hashtag Octothorpe]

I just found Grumble's topic about a Moog-style transistor ladder filter with an "instrument amplifier" (I have never heard of such a thing) pulling the signal out of the top of the ladder.

Just last night I tried and failed to get a signal out of one of my 303 diode ladders with a TL074 op amp, because here's my problem:

The 303 filter is just so noisy! I'm getting good enough with my synth to actually care about noise, and turning the filter output amp gain up gives me just so much white/pink/infra noise, even with the resonance turned all the way down. It's made worse if there's Arduinos driving LEDs... those two circuits are noise monsters.

I didn't match any of the transistors. Will this cause noise? I don't care about mathematically precise response curves and I thought that's what matched transistors did.

I suspect that the output amplifier made of BJT transistors is where some of the noise is coming from, but I haven't fully integrated the knowledge of how transistors work, so that part of the circuit is tough for me to understand.

I'm following the debate about if a single jfet-input op amp like the TL07x will work to pick up the signal from the ladder, and, well... I didn't have any success. There's the multiple-op-amp replacement for the one "instrument amplifier" that Grumble drew without resistor values...? that I may have to try next.

[link added by Blue Hell] See : http://electro-music.com/forum/post-437952.html

---

tb303_comp_level.jpg
Description:	Modified from Tim Stinchcombe's schematic from his excellent project exploring the 303 filter
Filesize:	90.41 KB
Viewed:	370 Time(s)
This image has been reduced to fit the page. Click on it to enlarge.

---

[Grumble]

did i wrote instrument amplifier? it is actualy instrumentation amplifier, inamp for short.
here is the diagram if you want to make one with opamps:


you see that the inputs are very high impedance which you can change by having a resistor to gnd (1Mohm or something in that order) the way you use the opamp gives an input resistance of 100k on the - input and 1100k on the + input so that could be responsible for some of the noise.
also: are you sure it is noise and not oscillation?
i didn’t match my transistors for my first build, but i did for my second one, but i didn’t notice any difference.

[Hashtag Octothorpe]

nope, you wrote "instrumentation amplifier" my mistake. I know what they are now, but I had never heard of such a thing! And they're not TOO expensive -- less than $1 each from Aliexpress... but I am spoiled by $0.07 NE5532s or TL074s.

[Grumble]

i am more spoiled: i make use of the sample program from analog devices, so i get them for free 😀🤙🏻

[Hashtag Octothorpe]

LOL right? People with their VST and DAW "throw a compressor on these twenty tracks" while I brag about having TWO compressors.

[VA1]

Maybe you can post some audio ?

[VA1]

Still no audio from that noise ?
I made various filters with opamps like you try, they all still are noisey.

You are using a dual supply schematic for a single supply thing.
How i made it : both sides go into inverting amplifier :
4K7 input resistor and 10K feedback resistor, half volt in the non-inverting input.
Then both go to comparator with 10K input resistors and 330K feedback resistor, no half volt needed.
You can use the fourth opamp to generate half volt.

[ixtern]

It is hard to say what is making noise. There are some possibilities:

- resistors. Resistors, especially big resistances (100kOhm and more) and low quality (usually carbon types) may make substantial noise.

- transistors. Some transistors are more noisy than others. Lately I've been making noise generator. To test different noise sources I've mounted socket to change transistors and the winner was... BF423 (biggest noise). Try to use low-noise types, like BC109, BC239, BC549, BC550, etc.

- OpAmps. TL074 is not the low-noise one. Low-noise OpAmps are usually bipolar, not JFETs. E.g. typical noise at 1kHz for TL072 is 18nV/HZ, for NE5532 5nV/Hz. For lower frequencies is even worse for JFETS: at 10 Hz TL072 has 45 nV/Hz, NE5532 has 10 nV/Hz.
Notice that bipolar OpAmps like NE5532 have usually much larger input bias currents so some offset voltage may occur if resistors on + an - inputs are not equal.

- power supply quality. Use not only electrolytic caps, but ceramic ones in parallel also to filter high-frequency noise.

[VA1]

Transistors may also differ between eachother from the same type for noise generator, you always need to select one, or use a trimmer lateron.
You might make a noise generator that clips with a luck transistor.

About the noise, ( if the TS ever read this : you still need to post audio )
I like the noisey filter, i even use high eq to boost the noise, why :
because laptop/mobile phone producers dont have that.
Ofcourse some people who dont know analog complain.

[elektrouwe]

"... It's made worse if there's Arduinos driving LEDs.."
+5V rails is suspicious then. Is it filtered like +12Vis ?
try a 47uF || R24.

[VA1]

About the noise generator ( my favourite subject )
Some brands have stable noise over all transistors.
If you buy the same BC547B from another brand, the noise can differ per transistor.
There are no noise generator specefications in the datasheet so it dont matters.