More than one answer to that. Suppose your circuitry was DC coupled and supposing the source component you connected to it had a small DC offset present. That DC would be amplified and passed along to the speaker... not good. So we often AC couple to block DC. Having made the decision to do that we then ask "what is the lowest frequency we need to allow through" and calculate accordingly based on the other resistances around the cap.
Opamps love DC and work at their best at DC. As mentioned, its speakers that don't like it plus, some filters such as high pass by definition need to block DC.
Its also good practice to roll off the feedback network on an opamp to give a gain of unity (one) by introducing another cap in the feedback resistor that goes to ground (for non inverting configurations). Although opamps love DC they can be subject to DC drift if various conditions aren't met... rolling the gain off to unity at DC mitigates that issue.
A typical LF cut off point might be as low as 1Hz for audiophile small signal amplifier circuitry to avoid any possibility of phase shift at LF and distortion from the cap caused by significant AC voltage appearing across it as frequency falls.
Opamps love DC and work at their best at DC. As mentioned, its speakers that don't like it plus, some filters such as high pass by definition need to block DC.
Its also good practice to roll off the feedback network on an opamp to give a gain of unity (one) by introducing another cap in the feedback resistor that goes to ground (for non inverting configurations). Although opamps love DC they can be subject to DC drift if various conditions aren't met... rolling the gain off to unity at DC mitigates that issue.
A typical LF cut off point might be as low as 1Hz for audiophile small signal amplifier circuitry to avoid any possibility of phase shift at LF and distortion from the cap caused by significant AC voltage appearing across it as frequency falls.
that´s what i call an answer.
since the setup i build will never play below ~40hz, i guess that 'input cutoff' is not to critical for me.
the cheap tpa3116 boards i use seem to need some gain from the preamp especially the one, that drives the subwoofer.
in other words:
i was much more satisfied with maximum volume when i had a gain included.
i implemented that in the summer section.
maybe i should remove the volume control pot from the amp board (like recommended in the tpa3116-thread), use unity gain in all stages i have right now and add a decent volume control section to the preamp.(?)
since the setup i build will never play below ~40hz, i guess that 'input cutoff' is not to critical for me.
the cheap tpa3116 boards i use seem to need some gain from the preamp especially the one, that drives the subwoofer.
in other words:
i was much more satisfied with maximum volume when i had a gain included.
i implemented that in the summer section.
maybe i should remove the volume control pot from the amp board (like recommended in the tpa3116-thread), use unity gain in all stages i have right now and add a decent volume control section to the preamp.(?)
You can adjust the gain for a tpa3116!! It is done using a voltage divider (GVCC-GND) with the split/mid-point connected to the gain pin. Look at the data sheet if you need further information.
before i can progress the preamp i need to understand a thing or two...
in mooly's post #38 i see those diagrams.
i interpret the blue lines as not so good.
as far as i understand the phase is shifting in exactly the range of frequencies that should be hearable.
is it correct, that this is exactly, what i do not want?
what can i do?
without understanding that point, it is pretty pointless go any further...
in mooly's post #38 i see those diagrams.
i interpret the blue lines as not so good.
as far as i understand the phase is shifting in exactly the range of frequencies that should be hearable.
is it correct, that this is exactly, what i do not want?
what can i do?
without understanding that point, it is pretty pointless go any further...
Any RC filter networks generate phase shift. Its unavoidable but fortunately its not an audible problem. The steeper the filter characteristic and the more phase shift you see.
Just a note on your opamps popping,
Make sure that you are not using the TLC07x series as they are only good for 16v maximum and not +/-16v.
I learned that myself the hard way when all of a sudden a brand new opamp got very hot for some reason!! 😉
Also I would just use a 1/2 V Resistor divider to create your virtual ground system as this will keep the ground potential at exactly 1/2Vsupply as the voltage droops if you are using batteries to power the circuit.
This will keep the signal symmetrical as the supply voltage gets lower all of the way down to the opamps lowest operational voltage limit of just a few volts.
This is considering that even though you are using a 18V supply, your signal may only be no more then a a few volts or so.
Also I just read an article where a zener diode was used for a reference noise source as well.
Although this may or may not effect the performance of your circuit as it is on the ground side of the signal.
But it only took a gain of 10 to get a 1v level.
Here is that article,
White noise source flat from 1Hz to 100kHz | EDN
I use values of 1K to 10k to create my virtual grounds.
You can also use a unity gain opamp buffer biased at 1/2Vsupply for the Vref as well if you are using a lot of opamps in your circuit to increase the current capabilities and lower the impedance of your virtual ground system.
This method is also easier on battery drain as well than just using resistors.
As the lower the values of resistors creates the least amount of noise, but this also increases battery drain current.
FWIW
jer 🙂
Make sure that you are not using the TLC07x series as they are only good for 16v maximum and not +/-16v.
I learned that myself the hard way when all of a sudden a brand new opamp got very hot for some reason!! 😉
Also I would just use a 1/2 V Resistor divider to create your virtual ground system as this will keep the ground potential at exactly 1/2Vsupply as the voltage droops if you are using batteries to power the circuit.
This will keep the signal symmetrical as the supply voltage gets lower all of the way down to the opamps lowest operational voltage limit of just a few volts.
This is considering that even though you are using a 18V supply, your signal may only be no more then a a few volts or so.
Also I just read an article where a zener diode was used for a reference noise source as well.
Although this may or may not effect the performance of your circuit as it is on the ground side of the signal.
But it only took a gain of 10 to get a 1v level.
Here is that article,
White noise source flat from 1Hz to 100kHz | EDN
I use values of 1K to 10k to create my virtual grounds.
You can also use a unity gain opamp buffer biased at 1/2Vsupply for the Vref as well if you are using a lot of opamps in your circuit to increase the current capabilities and lower the impedance of your virtual ground system.
This method is also easier on battery drain as well than just using resistors.
As the lower the values of resistors creates the least amount of noise, but this also increases battery drain current.
FWIW
jer 🙂
Last edited:
thanks gerald.
i´m not using those opamps.
i must have done something completely wrong wiring the whole thing on the breadboard.
the power supply with the zener which mooly suggested works perfect so far.
so i´m glad i´m beyond the popping opamps stage.
right know i´m sweating in the university and should be doing something totally different.
but one thing came to my mind.
i only can draw it with an online app where i am right now:
R4 should be a pot.
would this work?
how can optimize the ouput to feed a power amplifier with 26db gain and 30kohm impedance on its inputs?
i´m not using those opamps.
i must have done something completely wrong wiring the whole thing on the breadboard.
the power supply with the zener which mooly suggested works perfect so far.
so i´m glad i´m beyond the popping opamps stage.
right know i´m sweating in the university and should be doing something totally different.
but one thing came to my mind.
i only can draw it with an online app where i am right now:
R4 should be a pot.
would this work?
how can optimize the ouput to feed a power amplifier with 26db gain and 30kohm impedance on its inputs?
26db is a typical voltage gain for a power amp (X20) and so is 30k. Depending on the opamp just add a small (say 56 ohm) series output resistor.
R4 is OK as a pot to determine gain experimentally but is then better replaced with a fixed value.
R4 is OK as a pot to determine gain experimentally but is then better replaced with a fixed value.
ok, but what i really want is a sufficient volume control (i think that will need a pot anywhere).
the opamp will be a tl072.
the opamp will be a tl072.
i brought that to ltspice.
what i did is still based on mooly's suggestions and the file he uploaded.
i cascaded the filter stages to get 24db filters.
the high-out looks like what i expected.
i don't get what's wrong with low-output.
i would so much like bring a volume control onto this.
can someone please point me towards a good and simple solution.
what i did is still based on mooly's suggestions and the file he uploaded.
i cascaded the filter stages to get 24db filters.
the high-out looks like what i expected.
i don't get what's wrong with low-output.
i would so much like bring a volume control onto this.
can someone please point me towards a good and simple solution.
Your lowpass section starts out with two stages of highpass filters and then two stages of lowpass filters, this would cause an bandpass action and the the cause of the hump in the FR curve and then start to drop off.
jer 🙂
jer 🙂
For your volume control, if you want an "active" type then this is the proper way to do it,
http://www.ti.com/lit/ug/tidu034/tidu034.pdf
You could make R7 variable in your circuit above but its not as good an implementation as the full design.
Or you could add a normal pot after U6 followed by another opamp buffer to maintain a low driving impedance for the filters.
http://www.ti.com/lit/ug/tidu034/tidu034.pdf
You could make R7 variable in your circuit above but its not as good an implementation as the full design.
Or you could add a normal pot after U6 followed by another opamp buffer to maintain a low driving impedance for the filters.
i want a bandpass.
my english is not good enough to understand every bit of your post.
do you say that a bandpass like this doesn't work in general?
i don't understand what is happening around 115hz.
the resistor values are chosen for a 40hz hpf followed by a 90hz lpf.
i want seperated volume control for each output.
it would be nice to be able to mute the signal with the pot.
the very first circuit in the linked doc looks nice, but a bit too much.
the baxandall-circuit (fig. 4, page 5) looks great.
does it invert the signal?
would it make sense o implement it right before the outputs?
Okay , You didn't mention that it was supposed to be a Bandpass filter you just stated " I don't get what's wrong with the low output".
Sorry for the confusion. 😉
jer 🙂
Sorry for the confusion. 😉
jer 🙂
Yes, it does invert. If you want separate control for each output then I would consider just a conventional pot at the final opamp output. Use something like a 2k or 5k and there can be no real question of signal impairment or interaction with the power amp. Also, as you turn it down, you attenuate the noise from all that is present before. Its a win win situation.
i need your appreciated help once more mooly.
do you have an idea of what is wrong with the frequency response at 115hz?
and
when i feed a two-channel power amp with the high out, should i...
a) give the output for each channel its own buffer?
or
b) connect two rca-jacks in parallel right behind the last resistor?
do you have an idea of what is wrong with the frequency response at 115hz?
and
when i feed a two-channel power amp with the high out, should i...
a) give the output for each channel its own buffer?
or
b) connect two rca-jacks in parallel right behind the last resistor?
would this pot be fine?
mono:
2 x 5K Ohm Logarithmic Taper Potentiometer Pot A5K 5KA Free Shipping | eBay
stereo:
2 x 5K Ohm Logarithmic Dual Rotary Taper Potentiometer A5K 5KA Pot Alpha | eBay
mono:
2 x 5K Ohm Logarithmic Taper Potentiometer Pot A5K 5KA Free Shipping | eBay
stereo:
2 x 5K Ohm Logarithmic Dual Rotary Taper Potentiometer A5K 5KA Pot Alpha | eBay
Last edited by a moderator:
They look OK.
I've no quick answer to your filter problem, you would need to start from the begininning and analyse the response of each section and make sure it conforms to your own spec for what you want.
Feeding two identical power amps from one feed should be OK... although some will say its not ideal. I can't really see a problem though.
I've no quick answer to your filter problem, you would need to start from the begininning and analyse the response of each section and make sure it conforms to your own spec for what you want.
Feeding two identical power amps from one feed should be OK... although some will say its not ideal. I can't really see a problem though.
- Status
- Not open for further replies.