Preamp noisefloor

[Michaelaudio]

Hello guys, im Michael and new to the Forum here, starting right into it with an annoying Problem.  So I designed a 2.1 Amp Board with 3 TPA3116. I soldered the parts on the pcb and everything works as expected BUT: there is a VERRY loud noise coming out of the amps. So if I turn the pots down (they are right are the input channels) the noise is gone so it hast to be Come from the preamp but I absolutley have no idea how to solve that, I hope u guys can help me! 

[magic]

So many parts

Could you give a specific example of one noisy signal path? Which input the signal comes in, which ICs it passes through, which output the noise comes out?

I see at least two pots, which of them is involved?
What if you turn the pot all the way up, does the noise disappear?

Is it white noise or hum or some other crap?

[Michaelaudio]

Basically the noise (white noise) appears on all inputs of the amp ic's. I can see the noise with my scope but dont know how to solve this. Tomorrow i can show a pic of my scope on the amp inputs when the pots are turned down (no noise on the outputs) vs when they r pulled up (noise increases as I turn the pots up, depending on how much i pull them up). Made a pic of the board, as u can see i had to make some improvements (schematic is Updates to the actually improvements) but I dont think thats the Problem

[Marco]

That D1/D23/D24 DC biasing network is silly.

[magic]

Basically the noise (white noise) appears on all inputs of the amp ic's. I can see the noise with my scope but dont know how to solve this.

That's easy mode - just track it down. Turn RV2 to zero to eliminate the possibility of amp IC current noise playing tricks on you. You will probably find that there is noise on the output of IC2D. See if there is noise on IC2A. See if it depends on RV1. Keep backtracking through that jungle until you see something interesting.

Could it be those zener virtual grounds? No idea...

[Michaelaudio]

@Marco yea but when u call something silly explain why it is silly. I know its not the best solutIon but I made a mistake on the pcb so thats my easiest solution for me.
@Magic  i checked the zeners and that doesnt seem so be it. But yea havent thought of that, ill remove step by step the 10u caps in the signal path and see when the noise is gone, will work on that tomorrow.

[Marco]

@Marco yea but when u call something silly explain why it is silly.

Well, it's an envelope detector ... eventually it will kinda sorta bias, but why do it like this? To speed up the charging of of C1/C15?

Normally there would be a electrolytic capacitor in parallel with the Zener and instead of D23/D24 there would simply be resistors. Say 100K, making the RC time to charge the input capacitors a second.

BTW, you are shorting the inputs when measuring noise right?

[Michaelaudio]

I got it!! It were the zeners! Replaced every zener with a 10k and.. silence!  Never thought of that, although i read in an AD article that zener biasing is better than just with resistors..
Anyway, now I move on to another Problem (of course) when I turn the bass boost on (sw2) there is a huge visible pop and AWFULL crackling on the sub. I put some caps right before and after Pin 2 and 3 of the switch to keep dc after way but that change nothing. When the pop dissppears and i switch in right on and off the pop is almost gone, so has to be a loading issue or something like that, anyone ideas ?

@Marco
I made a really stupid mistake on the schematic and therefo re on the pcb, so I had to make some New traces (as u might see pn the pcb pic) which would short L and R chanel together, so I used the diodes that L and R havent an effect on each other.  Shorting the inputs? U mean to ground? I did that when I turned the pots down.

[magic]

Output impedance of zeners is a few ohm. Your 1µF filtering capacitors are doing nothing, too low time constant.

As for bass boost, chances are that your filters are in some weird state (capacitors charged wrong, opamps saturated to the rail) when it's disengaged and then it takes a moment for things to get normal when you engage it.

[tooki]

Interesting project!

FYI (since I've been working on something similar), the TPA3126 is the newer, pin-compatible version of the 3116. It's lower power. Also, there's the 3128 (and its predecessor 3118), which is slightly lower maximum power, but has the thermal pad on the bottom, allowing you to use the PCB as the heatsink instead of needing hunks of metal. (Unless you're running at very high power levels, the chips aren't going to get hot at all.)

Also, I'm not nearly experienced enough with filters to identify everything that's going on there, could you maybe share a block diagram of what each op-amp section is doing? It looks like you're bi-amping the L and R speakers, so is that an active crossover?

[Michaelaudio]

@magic  u mean if id add a bigger cap the noise would be refuced?
But how do I change that? Carl amps for example ofc dont have a krackling/pop when u turn bass boost on

@tooki yea I thought about taking the 3118 for the tweeters and mids 3116 seems so be a bit overpowered here. 
Yea its an active crossover, high pass for the high amp , high and low pass for the mid amp and low pass (with changeable crossover frequency) for the woofer amp. Can make a block diagramm later if u want.

[rkabz]

This is what I think is happening with S2. The input of IC6A (pin 3) is basically floating when S2 is open, so the input bias current will charge C52 to a voltage near to the positive supply voltage (to see why check the basic schematic of the NJM4558). When S2 is closed the voltage at the output of IC5A which is at around half supply, will discharge C52 until it stabilises at the new voltage, causing the transient you are experiencing. So basically it's a (lack of) biasing issue with IC6A pin 3.

[Michaelaudio]

I can imagine but can u tell me what "floating" exactly is?
So basically i can solve it, by biasing Pin 3 at Vcc/2 like on the other filters🤔

[rkabz]

Floating in this case means that the input pin is not connected to a proper DC biasing source, so yes, connecting it to vcc/2 will fix it.

[Michaelaudio]

So I tried biasing C51 to 12 V now, but the pop is still there.. I tried to Bias Pin 3 directly (not the right way to Do it) then there is no pop but the output signal is like 1/4 times as low as it should be..?

[magic]

With a scope you should be able to find out what's causing the pop...

[Michaelaudio]

Well I can see it on the output of the amp. There r some strange things going on. I Biased Pin 3 (not C51 ofc thats wrong )from the Resistor devider i go with a 10k to Pin 3 but then the output signal is verry low.. when i replace the 10k with a diode the output is as high as it should be but the Sinus is a lil bit deformed and ofc i have a pop, but I can nothing see on the scope there is 12V every where as it should be, im confused.

[Marco]

I made a really stupid mistake on the schematic and therefo re on the pcb, so I had to make some New traces (as u might see pn the pcb pic) which would short L and R chanel together, so I used the diodes that L and R havent an effect on each other.  Shorting the inputs? U mean to ground? I did that when I turned the pots down.

If you make the diodes 100K resistors even without a capacitor for the zener, crosstalk is limited. First the signal gets divided by the dynamic resistance of the zener over 100K and then that gets divided by the source impedance over 100K. Not a whole lot will remain.

The diodes make it an envelope detector ... it's silly.

[graybeard]

I got it!! It were the zeners! Replaced every zener with a 10k and.. silence!  Never thought of that, although i read in an AD article that zener biasing is better than just with resistors..
Anyway, now I move on to another Problem (of course) when I turn the bass boost on (sw2) there is a huge visible pop and AWFULL crackling on the sub. I put some caps right before and after Pin 2 and 3 of the switch to keep dc after way but that change nothing. When the pop dissppears and i switch in right on and off the pop is almost gone, so has to be a loading issue or something like that, anyone ideas ?

@Marco
I made a really stupid mistake on the schematic and therefo re on the pcb, so I had to make some New traces (as u might see pn the pcb pic) which would short L and R chanel together, so I used the diodes that L and R havent an effect on each other.  Shorting the inputs? U mean to ground? I did that when I turned the pots down.

Zener diodes are often used as noise sources.   Shunt voltage reference ICs are much quieter, but still noisy when compared with the input noise of many op-amps.

The resistors all contribute noise as well with their voltage noise be proportional to the square root of the resistance.  However how they contribute noise in a circuit is topology and sometimes bias dependent.

[Michaelaudio]

@Marco  ok ill do that, but why would it be an envelope detector, there is no signal just DC?

@graybeard  can u explain why zeners are so noisy?

[Marco]

@Marco  ok ill do that, but why would it be an envelope detector, there is no signal just DC?

Lets say the input before the blocking capacitor is at zero volt and there's some positive input leakage current and the opamp input slowly gets biased to one diode drop below the zener voltage.

Now the input voltage drops, does that get transmitted faithfully to the opamp input? No, because it will start pulling more current through the diode. Only once the voltage on the input goes up and the diode starts blocking is there again correspondence between the input voltage and opamp input. Every time the diode goes below the previous lowest envelope voltage this kind of discontinuity will keep happening.

[Michaelaudio]

Okay got it! So I replaced the 2 diodes with 100 now, and Biased the Pin 3 with a 10k too, and everything works now, thanks guys!