Case, if you want to see some magic, move V1 to the output of the opamp and short the inputs and measure the output.
AndrewT, it looks like this bridged configuration doesn't work with capacitors in the feedback loop.
AndrewT, it looks like this bridged configuration doesn't work with capacitors in the feedback loop.
Last edited:
AndrewT: those graphs are to show, how the caps influence the low pass. Should settle for a 220uF or 470 back to back. But for slave to work, they have to be like in the picture above, not between R11 and ground. I removed them all yesterday, now the slave works, but of course can get DC at the input. Works best with shielded cable, shield ground to the xover, and from there to chassis.
this way the caps cause trouble, as they stay charged and pull up pin 2.
in the simulation it does not matter, where the caps are, as per 1st or 2nd image
Last edited:
plus I could make some high pass so mid and high get the low frequencys filtered out, just to reduce the noise there. Its only a 6dB filter though and also means, there are 3 individual boards. Makes it hard to swap them.
Did you try connecting the signal grounds of the master and slave board? It should reduce some hum from the slave board.
Why back to back?
They are a DC blocking capacitor. They don't have any significant AC voltage across them. Measure an amplifier that is being tested to maximum power. Or predict by simulation.
The DC blocker can be exposed to DC if the amplifier fails and sends the output to either supply rail. In this fault condition you can protect the capacitor from excess DC voltage by placing a signal diode across it, in both directions, i.e. an inverse parallel pair of 1n4148
If you are concerned that these diodes might affect the sound quality of the amplifier you can double them up so that the capacitor has two diodes in series pointing in each direction.
The capacitor in the NFB lower leg can be low voltage typically 6.3V but there are reports that higher voltage here can result in better sound. I suspect these reports are from Builders who have installed a DC blocker that is too small and has significant impedance in the audio band. Fitting a big capacitor removes any effect from the audio band.
"Don't use 22uF !"
that is the value onST's datasheet and its just a single elco.Serious cost cutting there.
that is the value onST's datasheet and its just a single elco.Serious cost cutting there.
the simulation does not like my plan to put the capacitors the other way around.
edit: thats when the 100k resistor comes in. Now you see it
now you don't
reg connecting signal ground from master to slave, it helps a little, when I connect signal and ground to the xover. If I just connect signal, then it makes is worse. Overall, twisted or shielded cable, ground only connected to xover, is quiet.
I am aware that it cuts off at 1kHz but for now its just about, yes, can be done and I will not have that plop.
The back to back elcos are 25V. Setting C2 to 6.3nF gets me about 20kHz.
edit: thats when the 100k resistor comes in. Now you see it
now you don't
reg connecting signal ground from master to slave, it helps a little, when I connect signal and ground to the xover. If I just connect signal, then it makes is worse. Overall, twisted or shielded cable, ground only connected to xover, is quiet.
I am aware that it cuts off at 1kHz but for now its just about, yes, can be done and I will not have that plop.
The back to back elcos are 25V. Setting C2 to 6.3nF gets me about 20kHz.
Last edited:
R1, 100k, in the top diagram complicates the NFB. I don't know what that combination is doing.
As you can see the 100uF in series gives the equivalent of 50uF. That cuts off your low bass. Is that what you need?
For the very low DC and very low AC voltages across the DC blocker you don't need a bi-polar capacitor. Polar is good enough, if you have it oriented to suit the quiescent voltage. And then protect it from excessive reverse voltage with a signal diode.
At 35Vdc offset on the output from a failed output stage, the current passing through the diode is 35V/(22k+680r) = 1.54mApk
As you can see the 100uF in series gives the equivalent of 50uF. That cuts off your low bass. Is that what you need?
For the very low DC and very low AC voltages across the DC blocker you don't need a bi-polar capacitor. Polar is good enough, if you have it oriented to suit the quiescent voltage. And then protect it from excessive reverse voltage with a signal diode.
At 35Vdc offset on the output from a failed output stage, the current passing through the diode is 35V/(22k+680r) = 1.54mApk
Last edited:
maybe I leave the slave/bridge option alone. At 300mV peak to peak it drives an old 2 way speaker with a sinewave so my ears are bleeding from 2m away. Its really just sounding good not being loud
If I take the 100k ohm out, this happens:
The value of the um 100k ohm resistor seems does not matter
If I take the 100k ohm out, this happens:
The value of the um 100k ohm resistor seems does not matter
Last edited:
Have you calculated the minimum value for the NFB capacitor (C2)?"Don't use 22uF !"
that is the value onST's datasheet and its just a single elco.Serious cost cutting there.
NFB cap >= 0.47uF * 22k * sqrt(2) / 680r >=21.5uF
They meet that requirement for virtually no AC voltage across the NFB cap.
R3 (upper NFB) is tapped into the wrong position.maybe I leave the slave/bridge option alone. At 300mV peak to peak it drives an old 2 way speaker with a sinewave so my ears are bleeding from 2m away. Its really just sounding good not being loud
If I take the 100k ohm out, this happens:
in post 43 the first picture R5 and C6. It looks odd. R11 and R4 set the amplification. R 5 seems does the magic so its not ending up at -130dB.
"R3 (upper NFB) is tapped into the wrong position."
it is there because if it is on top of C5 it seems C3/C4 stay charged and put a voltage onto the inverting input. So when I switch mute off, pin 14 is at some decent negative voltage and I get DC onto the output, in bridge mode. Removing C3,4,5 solved the problem.
Building it like in this imageseems to work in the simulation and actually the 1st version was built that way. And worked without a plop when switching on, in bridge.
it is there because if it is on top of C5 it seems C3/C4 stay charged and put a voltage onto the inverting input. So when I switch mute off, pin 14 is at some decent negative voltage and I get DC onto the output, in bridge mode. Removing C3,4,5 solved the problem.
Building it like in this imageseems to work in the simulation and actually the 1st version was built that way. And worked without a plop when switching on, in bridge.
They look good because there is no DC at the input or output. You may see more if you remove C1_st from the input as you are now seeing several filters acting together to form the FR.
- Status
- Not open for further replies.