| facundonu |
this circuit is intended to add both channels, and low pass at 12dB with fc~115Hz. its output will be conected to a chipamp input.
is it OK?
can this work? why? why not?
i need something to do this for a small subwoofer, and i quickly designed this circuit. tested fine in PSpice but i want your opinions.
 |
|
|
| AndrewT |
Hi,
for summing two channels you should use the opamp in it's inverting mode.
Then the virtual earth at the inverting input sums the signals coming through all the line resistors. Just perfect for a two channel into one sub.
You cannot hang an RF filter cap between the summing resistors and the inverting input pin, it may oscillate. But an extra resistor after the cap may be all you need to prevent the oscillation.
Look up some data sheets or application notes to find precise guidance on this.
Have you considered an MFB filter?
It allows inverting input.
and 2pole filter with the same opamp.
and you can adjust the gain.
and you can adjust the Q.
and you can adjust the Frequency as well as the summing function. |
|
|
| CarlosT |
Hey, Facundonu...would you be a gaucho and design a simple summing circuit for a fellow Argentino? Gracias, amigo. I got an iPod/MP3 player putting out stereo through the headphone jack but going into a single non-inverting LM3875 chipamp kit (i.e., summing both L-R channels into mono).
I like that...Facundonu...is that a play on words on our beloved facon? :D |
|
|
| neutron7 |
| wouldn't that cause a bit of crosstalk because the left side would leak through to the right through he summing resistors. (unless they are coming from a different opamp) |
|
|
| CarlosT |
| Why is crosstalk a problem when you're trying to sum up the channels anyway? |
|
|
| facundonu |
thanks a lot to everyone,
how about this one?
i guess the former circuit input section would require a 0 ohm source to keep its freq response.
the last 10k resistor would be the input impedance of the amp.
and MFB...
they are great, but far more complex and versatile than what i'm looking for. |
|
|
| facundonu |
| quote: | Originally posted by CarlosT
Hey, Facundonu...would you be a gaucho and design a simple summing circuit for a fellow Argentino? Gracias, amigo. I got an iPod/MP3 player putting out stereo through the headphone jack but going into a single non-inverting LM3875 chipamp kit (i.e., summing both L-R channels into mono).
I like that...Facundonu...is that a play on words on our beloved facon? :D |
facundonu is for Facundo Nuñez (my name).
in your case i'd go the easy way,
 |
|
|
| facundonu |
thanks a lot!!
i'll take a good look at it. |
|
|
| BobEllis |
| quote: | Originally posted by CarlosT
Why is crosstalk a problem when you're trying to sum up the channels anyway? |
Its not a problem for the subwoofer, but if the source has a significant impedance the main channels will have quite a bit of crosstalk. I prefer to buffer the L and R signals before mixing.
115 Hz. is awfully high for a 2nd order cross to a mono sub. Keep it fairly close to centered between the mains or you will hear the sub as a separate entity.
Edit: The catch with passive crossovers is the insertion loss. If the amp doen't have enough gain you'll end up needing an active stage anyway. |
|
|
| v-bro |
| quote: | Originally posted by facundonu
thanks a lot!!
i'll take a good look at it. |
For the buck it's worth experimenting to death with..:dead: :D
Not that that will be necessary, as it proved to be quite easy to tune .... |
|
|
| AndrewT |
Hi,
your second posting still makes the mistake of showing the summing resistors going to the non-inverting pin.
A summing opamp uses the inverting pin as the input.
Did you go and look at an MFB filter? It saves you having to use a dual. It only needs one. |
|
|
| facundonu |
| i used a non inverting summing circuit, it works as fine as the clasical inverting summing. |
|
|
| facundonu |
| quote: | Originally posted by BobEllis
115 Hz. is awfully high for a 2nd order cross to a mono sub. Keep it fairly close to centered between the mains or you will hear the sub as a separate entity. |
i know it is, i'm building a really low end system based on a pair of hewlett packard speakers that can't go below that frequency. |
|
|
| CarlosT |
| Thanks, Facundo. The folks here got me paranoid about going the simple passive way because I'm using a non-inverting LM3875 chipamp kit...something about virtual grounds BLAH BLAH BLAH...man...once I figure it out, I'm gonna write the "Chipamp For Dummies" book. |
|
|
| AndrewT |
| quote: | Originally posted by facundonu
i used a non inverting summing circuit, it works as fine as the clasical inverting summing. |
No it doesn't.
The inverting is a summer (or adder). |
|
|
| pinkmouse |
| quote: | Originally posted by AndrewT
No it doesn't. |
Exactly.
| quote: | Originally posted by CarlosT
...something about virtual grounds BLAH BLAH BLAH...man... |
I assume you can't use Google for some reason or another then? Yahoo will probably produce the same links if you try that. ;) |
|
|
| CarlosT |
| Al, I have been trying and the best thing I found was some dude New York Dave in the Sound on Sound forum's... |
|
|
| CarlosT |
Thanks, Al. I was looking for an apple and you wanted me to search for an orange. My keywords were "stereo channel sum summing mono" and so forth. Still a little over my head...but I'l keep leanring :D
I guess I could power this little "pre-amp inverting thingie" with the same power supply from the LM3875 power supply. |
|
|
| CarlosT |
Al:
Let me just make sure that I'm absolutely clear on something. If I had an amp that was configured as inverting, then I would not need this channel summing OP in front of it to combine stereo channels? It's because my LM3875 kit is a non-inverting kit that I have this complication? In your mind, would going inverting on the amp work out better? |
|
|
| pinkmouse |
| It would certainly be more elegant. You should be able to mod the pcb fairly easily, have a look at the datasheet to see the differences in schematic |
|
|
| Nordic |
| Speaking under correction, but the summing can be done by as simple unity gain opamp, with the left and right inputs connected to non inverted inputs trough a small 10k resistor... |
|
|
| victorio |
I recognize that...
It's a SALLEN-KEY!!! :D
U can work good with it, but your cutoff frequency is not 115 Hertz in that config, but in 185 Hz.
Remembering that the -3 dB point in a Sallen-Key configuration is at the frequency
Fc = 1 / (2*pi * sqrt(R11*R12*C5*C6))
*sqrt is square root
*this is based on your second schematic (that with the buffer)
Yo could just change the resistors to 39K (cutoff frequency will be 105 Hz) or 33K (cuttoff frequency will be 123 Hz).
I would use the first , because with this values u choosed, this is clearly a BUTTERWORTH filter, and it doesnt decays so fast after the cutoff freq.
U could check the freq response on your simulator (the PSpice).
Or u could just use the passive config with just resistors and capacitors, but the freq response is worse.
Oh, and a BUTTERWORTH filter NEVER oscillates, so dont worry about that.
Hope i helped a little. ;-) |
|
|
| pinkmouse |
| quote: | Originally posted by Nordic
Speaking under correction, but the summing can be done by as simple unity gain opamp, with the left and right inputs connected to non inverted inputs trough a small 10k resistor... |
You can do all sorts of things in electronics, but the results may not be optimum... |
|
|
| facundonu |
| quote: | Originally posted by victorio
I recognize that...
It's a SALLEN-KEY!!! :D
U can work good with it, but your cutoff frequency is not 115 Hertz in that config, but in 185 Hz.
Remembering that the -3 dB point in a Sallen-Key configuration is at the frequency
Fc = 1 / (2*pi * sqrt(R11*R12*C5*C6))
*sqrt is square root
*this is based on your second schematic (that with the buffer)
Yo could just change the resistors to 39K (cutoff frequency will be 105 Hz) or 33K (cuttoff frequency will be 123 Hz).
I would use the first , because with this values u choosed, this is clearly a BUTTERWORTH filter, and it doesnt decays so fast after the cutoff freq.
U could check the freq response on your simulator (the PSpice).
Or u could just use the passive config with just resistors and capacitors, but the freq response is worse.
Oh, and a BUTTERWORTH filter NEVER oscillates, so dont worry about that.
Hope i helped a little. ;-) |
:bigeyes:
thanks a lot, i'll revise that part. |
|
|
| facundonu |
about virtual ground and stuff.
i know an inverting summing circuit would help isolate channels, but where am i wrong if i state this about my non inverting summing?
V+ = (V1-V2) x 100k/200k + V2 = (V1)/2 - (V2)/2 + V2= (V1)/2 + (V2)/2
V+ = V-
V- = Vout
thanks a lot! |
|
|
| facundonu |
the first one was generated with R=39k, the second with R=22k.
doesn't look like the results of that formula. :(
i agree to that formula, but simulation shows something different.
has it got anything to do with Q? I guess the "bode" lines cross in the value given by the formula, but the real line goes way below that. |
|
|
| AndrewT |
Hi,
expand your frequency scales to at least an octave above and an octave below the -3db turnover frequency. i.e. 40Hz to 300Hz.
An equal component value Sallen & Key uses a slightly different formula from the unity gain Sallen & Key.
The Butterworth must have Q=1/sqrt2 =0.7071
ECV S&K F-3db=1/2PiRC
UG S&K F-3db=1/2srqt2PiRC note the difference of the extra sqrt2, when the components are chosen to give the correct Q. |
|
|
| victorio |
Oops... :xeye:
Sorry, I calculated a wrong Q for the circuit. I thougth it was 0.707, but it is 0.5, so the frequency i gave you is the -6 dB point, and the decay is worse than the Butterworth.
I recommend you to make the feedback cap (C5) aproximately 2x the C6. This way u make a Butterworth.
Just to make things clear, i attached a figure.
The freq (Fc) i gave to you is called the NATURAL FREQUENCY.
When Q is above 0.707, the freq response has a peak of magnitude 20*log(Q) in the NATURAL FREQ.
When Q is equal or below 0.707, the magnitude in the NATURAL FREQ is 20*log(Q).
* log is log of base 10
** When Q=0.707, the magnitude at the Natural Freq is 20*log(0.707) = -3 dB, and the filter is called a Butterworth filter.
*** In your schematic, with Q=0.5, the magnitude at the natural freq is 20*log(0.5) = -6 dB.
Sorry by the tons of theoretical stuff... ithougth i talked a lot ;) |
|
|
| CarlosT |
| quote: | Originally posted by AndrewT
Hi,
for summing two channels you should use the opamp in it's inverting mode.
Then the virtual earth at the inverting input sums the signals coming through all the line resistors. Just perfect for a two channel into one sub... |
AndrewT:
Why wouldn't the non-inverting summing circuit in Fig. 3 of the attached not work instead of the inverting scheme?
http://www.analog.com/UploadedFiles...ets/SSM2141.pdf |
|
|
| AndrewT |
Hi Carlos,
I don't understand why you have referred us to a balanced line receiver.
But, go buy a 2141 and connect your two bass channels to the two inputs and send the bass output to your bass amplifier.
I think you will find that you can manage with a 10mW amplifier and it will never overload on ordinary stereo music signals. Potentially saving you a lot of money on big sub-bass speakers and amps. |
|
|
|
