Please allow me to do a little reality check here. The purpose is to sum L and R do a mono signal can drive a portable speaker (from Post #1).
Then you stated the sum needs to feed bandpass filters, but didn't say where or what they are. But they're either active at line level, or there's a power amp after the mono sum driving speaker crossovers. Good so far?
Then you're concerned that 10uF caps would put a filter at 30Hz. Why? What kind of load are they driving? A real 600 ohms? Because modern bandpass active filters would have an input Z of 5-10K, and that puts a 10uf first order filter at 1.5 to 3Hz.
But ignoring that, the purpose here is to drive a portable speaker. Powered, yes? So it's input is "high", again 5K to 10K. Or the bandpass filter's input is high. All you need, then, is a resistor to build out each channel, and join them at the first line level input for a perfect sum with none of your circuit. There is literally no point for it, unless I've missed something about your application that you're not explaining well. If there's something far more complex going on, share it so we can help. Othewise it looks like a solution in search of a problem.
Then you stated the sum needs to feed bandpass filters, but didn't say where or what they are. But they're either active at line level, or there's a power amp after the mono sum driving speaker crossovers. Good so far?
Then you're concerned that 10uF caps would put a filter at 30Hz. Why? What kind of load are they driving? A real 600 ohms? Because modern bandpass active filters would have an input Z of 5-10K, and that puts a 10uf first order filter at 1.5 to 3Hz.
But ignoring that, the purpose here is to drive a portable speaker. Powered, yes? So it's input is "high", again 5K to 10K. Or the bandpass filter's input is high. All you need, then, is a resistor to build out each channel, and join them at the first line level input for a perfect sum with none of your circuit. There is literally no point for it, unless I've missed something about your application that you're not explaining well. If there's something far more complex going on, share it so we can help. Othewise it looks like a solution in search of a problem.
sorry to break it to you but that's not possible....due to phase differences....early "stereo" recordings by the Beattle's uncovered that problem years ago!
Yes, playing a stereo mix in mono leads to phase cancellation and frequency masking as explained here:
https://www.sonible.com/blog/stereo-to-mono/
https://www.sonible.com/blog/stereo-to-mono/
The crossovers are active (butterworth) made with OpAmps that will share the same dual supply as the other OpAmps. Their input impedance is set to 6Kohm.
I used around 3Hz cutoff because i wanted frequencies from 30Hz and above to have the same voltage amplitude.
I read in a TI datasheet that for example, for 30Hz to pass at full amplitude the recommended -3db cutoff of that high pass filter is a tenth of that, so 3Hz. Thats why i made it 3Hz.
I wanted to use OpAmps to load the L and R outputs from the DAC as less as possible. The resistors to GND are to provide a minimum current to flow which reduces noise. The DAC output voltage amplitude will be 1V rms MAX so across 6.2Kohms the current will be well below 1mA, that my DAC can supply.
I found that connecting directly the outputs of the DAC with resistors directly to all three band pass filters will cause the DAC to be loaded too much and a little bit of distortion appears. With the buffer though, that problem goes away.
So those are the reason why my design is like this.
I used around 3Hz cutoff because i wanted frequencies from 30Hz and above to have the same voltage amplitude.
I read in a TI datasheet that for example, for 30Hz to pass at full amplitude the recommended -3db cutoff of that high pass filter is a tenth of that, so 3Hz. Thats why i made it 3Hz.
I wanted to use OpAmps to load the L and R outputs from the DAC as less as possible. The resistors to GND are to provide a minimum current to flow which reduces noise. The DAC output voltage amplitude will be 1V rms MAX so across 6.2Kohms the current will be well below 1mA, that my DAC can supply.
I found that connecting directly the outputs of the DAC with resistors directly to all three band pass filters will cause the DAC to be loaded too much and a little bit of distortion appears. With the buffer though, that problem goes away.
So those are the reason why my design is like this.
As a reference for those looking for a simple stereo to mono converter:
I believe the addition of the single 'to ground' resistor increases the available signal at the output.
I've used 2 x 1 K and 1 x 10 K with good results.
I believe the addition of the single 'to ground' resistor increases the available signal at the output.
I've used 2 x 1 K and 1 x 10 K with good results.
Last edited:
"The crossovers are active (butterworth) made with OpAmps that will share the same dual supply as the other OpAmps. Their input impedance is set to 6Kohm."
...there's the bit we didn't know. 3 individual devices, each at 6K. That changes the requirement a lot. That configuration didn't match my idea of what a "HiFi portable speaker" would be.
"I found that connecting directly the outputs of the DAC with resistors directly to all three band pass filters will cause the DAC to be loaded too much and a little bit of distortion appears. With the buffer though, that problem goes away."
What does "connecting directly the outputs of the DAC with resistors" mean? What resistors? What DAC? Any reasonable DAC should be able to drive a couple K. Did you measure the distortion?
This still seems like way too much "project" for what is actually very simple. But clearly, there is much more going on, and not all information is shared, so I'll jump out of this now.
...there's the bit we didn't know. 3 individual devices, each at 6K. That changes the requirement a lot. That configuration didn't match my idea of what a "HiFi portable speaker" would be.
"I found that connecting directly the outputs of the DAC with resistors directly to all three band pass filters will cause the DAC to be loaded too much and a little bit of distortion appears. With the buffer though, that problem goes away."
What does "connecting directly the outputs of the DAC with resistors" mean? What resistors? What DAC? Any reasonable DAC should be able to drive a couple K. Did you measure the distortion?
This still seems like way too much "project" for what is actually very simple. But clearly, there is much more going on, and not all information is shared, so I'll jump out of this now.
No the ground resistor does not increase the signal. And is unnecessary. The load will be provided by the destination device. The 470 ohms is a bit low, I'd suggest more like 1.5K. 1% resistors are nice, but not required. 5% is fine.
Welp. I'm not an Expert, sorry if i'm not very clear. I am still learning.
I'll try to explain this even better:
The left and right audio SOURCES are coming from a DAC that is driven by a Bluetooth module.
I wanted to unify those two to a mono signal. I also read the other posts, frequency masking is not a very big deal for me.
So i wanted to unify them, but FIRST, I wanted to add a layer of 'isolation' by using OpAmps as buffers (The inverting summing OpAmp + re-inverting OpAmp), so that i can load the audio SOURCES much less. This is also because from what i've been able to understand, this is a better way to unify the two signals (it does not cause interactions between the two signals and some other advantages). The re-inverting OpAmp brings the output signal in phase with the input one which makes things easier to test / check.
If i connect the audio SOURCES togheter without OpAmps (by using the simple way with just two resistors) and then connect this unified (with the two resistors) source directly to the three bandpass filters (whose three 6kohm inputs will add up and make a 2kohm input), one problem appears:
A resistive divider is formed because the input impedance of the active bandpass filters in in respect to GND. This will lower the voltage amplitude going to the bandpass filters, which i don't want to. (The bandpass filters are designed to work at best performance at 1V RMS.) I didnt want to ad a gain stage before because it would require more components.
As i said, i'm not an expert so i might do mistakes here. So i thought... To be on the safe side, i'll add decoupling capacitors, OpAmps end everything.
Like this the summed mono signal can drive whatever load without affecting the SOURCE signals that are coming from the DAC. They will always see the same 6.2K resistance, that this DAC can drive (Minimum 5K, it says on the datasheet).
This is the other reason i raised the values to 6.2K.
I'll try to explain this even better:
The left and right audio SOURCES are coming from a DAC that is driven by a Bluetooth module.
I wanted to unify those two to a mono signal. I also read the other posts, frequency masking is not a very big deal for me.
So i wanted to unify them, but FIRST, I wanted to add a layer of 'isolation' by using OpAmps as buffers (The inverting summing OpAmp + re-inverting OpAmp), so that i can load the audio SOURCES much less. This is also because from what i've been able to understand, this is a better way to unify the two signals (it does not cause interactions between the two signals and some other advantages). The re-inverting OpAmp brings the output signal in phase with the input one which makes things easier to test / check.
If i connect the audio SOURCES togheter without OpAmps (by using the simple way with just two resistors) and then connect this unified (with the two resistors) source directly to the three bandpass filters (whose three 6kohm inputs will add up and make a 2kohm input), one problem appears:
A resistive divider is formed because the input impedance of the active bandpass filters in in respect to GND. This will lower the voltage amplitude going to the bandpass filters, which i don't want to. (The bandpass filters are designed to work at best performance at 1V RMS.) I didnt want to ad a gain stage before because it would require more components.
As i said, i'm not an expert so i might do mistakes here. So i thought... To be on the safe side, i'll add decoupling capacitors, OpAmps end everything.
Like this the summed mono signal can drive whatever load without affecting the SOURCE signals that are coming from the DAC. They will always see the same 6.2K resistance, that this DAC can drive (Minimum 5K, it says on the datasheet).
This is the other reason i raised the values to 6.2K.
Last edited:
If you'd like a full and complete analysis of your system, I'd need specifics as to what the DAC is, what the filters are and what the speaker/amp is. Specifics mean make and model, so I can look up the actual specs. I'll admit to being curious what portable speaker needs 3 bandpass filters, while being fed from a Bluetooth module.
You're following theory correctly, but you may not be aware of the practical nature of this. Not your fault, and I'm willing to help, if you want.
You're following theory correctly, but you may not be aware of the practical nature of this. Not your fault, and I'm willing to help, if you want.
loading and buffering, isolation aside there's no way that mono summing does not change what is heard, it all comes down to the original stereo recordings content.
Don't worry, i dont need a full analysis of the entire system. It is already designed and it works flawlessly.
This speaker is actually... TRANSPORTABLE, more than portable. It has a total power output of 400W RMS.
The three bandpass filters are for the three speakers (Subwoofer, midrange, tweeter).
All of this works really really well, very good audio performance.
But i tested all of this with the L and R channels coming from the DAC directly linked togheter, and i discovered that was not the best thing to do. They were connected togheter directly to the inputs of the bandpass filters. No resistors, no anything.
And that is what bringed me here.
The only part missing from the system is the stereo to mono conversion and i was searching the best circuit that could do that without loading the L and R output of the DAC too much. OpAmps are the only answer, i guess.
I dont remember the part number of that DAC for now, but i remember i read that it can drive minimum 5kohms @ 1V RMS with best audio performance. Loading it more will cause distortion to rise.
I think i'm going to use the last circuit i posted. I think it should be ok now.
This speaker is actually... TRANSPORTABLE, more than portable. It has a total power output of 400W RMS.
The three bandpass filters are for the three speakers (Subwoofer, midrange, tweeter).
All of this works really really well, very good audio performance.
But i tested all of this with the L and R channels coming from the DAC directly linked togheter, and i discovered that was not the best thing to do. They were connected togheter directly to the inputs of the bandpass filters. No resistors, no anything.
And that is what bringed me here.
The only part missing from the system is the stereo to mono conversion and i was searching the best circuit that could do that without loading the L and R output of the DAC too much. OpAmps are the only answer, i guess.
I dont remember the part number of that DAC for now, but i remember i read that it can drive minimum 5kohms @ 1V RMS with best audio performance. Loading it more will cause distortion to rise.
I think i'm going to use the last circuit i posted. I think it should be ok now.
A zillion posts have appeared since...
I'd just say do you need R to be as low as 6.2k? I would go to at 15 to 20k if only to reduce loading and get the size of the coupling caps down. The single cap coupling the two stages isn't really needed. Far better to have a cap at the output of the final stage.
Fwiw this is the response with 1uF and 6.2k showing the -3db point.
I reduced the resistors to reduce noise as much as possible.
It might be exaggerated, but as far as i know its best to keep the impedance lower to reduce the noise.
I'm now using nichicon 10uF 25V FG series, the absolute best for audio. They are small.
I will remove the 10uF between the stages and put it on the last stage and that is finally it. Now it should all work properly.
It might be exaggerated, but as far as i know its best to keep the impedance lower to reduce the noise.
I'm now using nichicon 10uF 25V FG series, the absolute best for audio. They are small.
I will remove the 10uF between the stages and put it on the last stage and that is finally it. Now it should all work properly.
That is true and lower resistance will generate less noise but I think you have to keep it all in context. A portable speaker application is all high level stuff really, its not like amplifying a signal from a tape head or phono cartridge. Its up to you 🙂 6.2k is no problem for an opamp to drive.
Wait a second.
It is not just a classic Bluetooth module.
It's QCC5125 from Qualcomm, supports LDAC audio up to 24bit and 990kb/s. Sampling rate is 96kHz.
And this will not be the only audio source.
There will be input jacks as well.
It is not just a classic Bluetooth module.
It's QCC5125 from Qualcomm, supports LDAC audio up to 24bit and 990kb/s. Sampling rate is 96kHz.
And this will not be the only audio source.
There will be input jacks as well.
Wait a whole minute. You're still doing it. You don't let out details until they're helpful in proving someone wrong...who is, by the way, trying to help you.
I looked up the QCC5125, max sampling rate is 48KHz, and while the "system" might be 24 bits, it depends on the input stream. I'm not going to bother to debate 24 bits here, but the thing is, like everything in audio, garbage-in, garbage-out. That thing doesn't define "Absolute Best Audio", it's part of a system, which you refuse to accurately define. (Portable speaker, etc.). Still don't know anything about the crossover (except they cannot all be band pass), and they certainly have model numbers at some point, or an entire circuit. We don't know what in the world the speaker is, amplfier is or are, what the actual signal source is, and what the application is.
Two resistors is all you need to sum to mono. Pretty much all anyone needs to sum to mono and maintain Absolute Best Audio. Unless there's more to it, which we still don't really know, but keeps changing.
"I need to buy a vehicle for a two mile trip every day". Ok. So...in all weather, or not? One person or 6? Carrying a backpack or a ton of equipment? Two wheels? Four wheels? Fuel? Comfort? Price? If you don't define the problem, there are an infinite number of perfectly valid solutions. I might pick a Vespa GTS300, you might want a Hummer. Both go two miles twice a day, and solve that problem. Right up until you redefine the problem.
Define your problem, by the details, every step in the chain, if you want suggestions that actually match your problem. If you don't want to do that, then please don't shoot down perfectly valid suggestions because you haven't defined the problem well.
I looked up the QCC5125, max sampling rate is 48KHz, and while the "system" might be 24 bits, it depends on the input stream. I'm not going to bother to debate 24 bits here, but the thing is, like everything in audio, garbage-in, garbage-out. That thing doesn't define "Absolute Best Audio", it's part of a system, which you refuse to accurately define. (Portable speaker, etc.). Still don't know anything about the crossover (except they cannot all be band pass), and they certainly have model numbers at some point, or an entire circuit. We don't know what in the world the speaker is, amplfier is or are, what the actual signal source is, and what the application is.
Two resistors is all you need to sum to mono. Pretty much all anyone needs to sum to mono and maintain Absolute Best Audio. Unless there's more to it, which we still don't really know, but keeps changing.
"I need to buy a vehicle for a two mile trip every day". Ok. So...in all weather, or not? One person or 6? Carrying a backpack or a ton of equipment? Two wheels? Four wheels? Fuel? Comfort? Price? If you don't define the problem, there are an infinite number of perfectly valid solutions. I might pick a Vespa GTS300, you might want a Hummer. Both go two miles twice a day, and solve that problem. Right up until you redefine the problem.
Define your problem, by the details, every step in the chain, if you want suggestions that actually match your problem. If you don't want to do that, then please don't shoot down perfectly valid suggestions because you haven't defined the problem well.
Thats why i said i'm not a expert...
The rest of the system is already designed (Active crossovers, speakers, speakers box, power amplifiers, power supply and battery protection) and it all works perfectly fine.
The audio sources i will be using wont be garbage. I know that if the source is garbage, then the speaker will also sound garbage, even if it's the best speaker in the world.
I have some high quality FLAC and DSD files which i will use to test my speaker. For music with even higher specifications than the Bluetooth module can offer, i will use the audio jacks.
And, another thing i didn't say... To me, this speaker does not have to sound 'absolute best' in every possible scenario.
I'm making it mainly for playing music outside. I wanted it to sound decently, but i also wanted it to be used in 'HI-FI mode' with even better audio performance...
Of course that is not an actual MODE you can set the speaker to. It's just that i will use the audio jacks instead of the bluetooth, which wont cause any audio degradation. The audio jacks will bypass the bluetooth module.
Now, everything is working perfectly as i wanted, but i was concerned on how to convert stereo to mono because from what i've been able to read on the internet, somebody says that its better to do it with OpAmps, others say to use just resistors and it would make no difference in audio performance. So i came here trying to get a solid answer.
I was concerned about this stereo to mono coversion ruining audio performance because all the rest of the system is working perfectly fine, and i didn't want the very good system to sound less good because of a badly made stereo to mono coversion. (Expecially when used in 'HI-FI' mode)
That is it.
I wanted help only in that small part of the entire system, because i already know with solid evidence that everything else works basically as well as it could.
How do i know?
I tested it but with only one channel and the music playing all only in one channel (balance set all to L or R).
It sounded amazing to me.
But of course that is not the way to do it. I needed to mix L and R signals in the circuit, not in the source.
The only thing i should have not written is probably 'absolute best way'.
I should have clarified that 'absolute best way to convert stereo to mono' was not meaning 'absolute best way to convert stereo to mono with absolute best audio performance'
All i wanted was a way to convert stereo to mono without causing weird interactions between the two channels that can cause artifacts and additional distortion (since this is what i've been able to understand according to the internet).
Now i finally discovered the best circuit that can sum both channels without loading them too much (Additional advantage) and that does not cause any weird interactions (Like for example backfeeding) between the L and R channels coming from the DAC.
So, in the end... I just wanted this stereo to mono coversion to preserve the quality of the audio content. Putting aside frequency masking for now since it's not a drastic thing, at least for me.
Again, sorry if i wasn't very clear about my main problem. I'm still new and learning. Electronics are though.
Thank you eveyone for all the help.
It is much apprecciated.
The rest of the system is already designed (Active crossovers, speakers, speakers box, power amplifiers, power supply and battery protection) and it all works perfectly fine.
The audio sources i will be using wont be garbage. I know that if the source is garbage, then the speaker will also sound garbage, even if it's the best speaker in the world.
I have some high quality FLAC and DSD files which i will use to test my speaker. For music with even higher specifications than the Bluetooth module can offer, i will use the audio jacks.
And, another thing i didn't say... To me, this speaker does not have to sound 'absolute best' in every possible scenario.
I'm making it mainly for playing music outside. I wanted it to sound decently, but i also wanted it to be used in 'HI-FI mode' with even better audio performance...
Of course that is not an actual MODE you can set the speaker to. It's just that i will use the audio jacks instead of the bluetooth, which wont cause any audio degradation. The audio jacks will bypass the bluetooth module.
Now, everything is working perfectly as i wanted, but i was concerned on how to convert stereo to mono because from what i've been able to read on the internet, somebody says that its better to do it with OpAmps, others say to use just resistors and it would make no difference in audio performance. So i came here trying to get a solid answer.
I was concerned about this stereo to mono coversion ruining audio performance because all the rest of the system is working perfectly fine, and i didn't want the very good system to sound less good because of a badly made stereo to mono coversion. (Expecially when used in 'HI-FI' mode)
That is it.
I wanted help only in that small part of the entire system, because i already know with solid evidence that everything else works basically as well as it could.
How do i know?
I tested it but with only one channel and the music playing all only in one channel (balance set all to L or R).
It sounded amazing to me.
But of course that is not the way to do it. I needed to mix L and R signals in the circuit, not in the source.
The only thing i should have not written is probably 'absolute best way'.
I should have clarified that 'absolute best way to convert stereo to mono' was not meaning 'absolute best way to convert stereo to mono with absolute best audio performance'
All i wanted was a way to convert stereo to mono without causing weird interactions between the two channels that can cause artifacts and additional distortion (since this is what i've been able to understand according to the internet).
Now i finally discovered the best circuit that can sum both channels without loading them too much (Additional advantage) and that does not cause any weird interactions (Like for example backfeeding) between the L and R channels coming from the DAC.
So, in the end... I just wanted this stereo to mono coversion to preserve the quality of the audio content. Putting aside frequency masking for now since it's not a drastic thing, at least for me.
Again, sorry if i wasn't very clear about my main problem. I'm still new and learning. Electronics are though.
Thank you eveyone for all the help.
It is much apprecciated.
Last edited:
Basic summing mixer/ analog adder is and always been around 10k.
Impedance is more than high enough for common capacitor values to have no high pass issues.
Mooly took it one step further and increased impedance to 22k which would be extremely compatible
with any signal chain seen in the wild. Also makes it high enough to exclude electrolytic values if desired.
Portable speaker application with small speakers wont have response over 80 to 100 Hz
unless you have some small " sub" might be 65 to 45 Hz
If you still require 1 Hz bandwidth, which caters to imaginary bandwidth
Calculate you min frequency high pass value
for 1 Hz 22k, capacitor value = 7.23 uF so nearest standard values 8.2uF even 10uF be more than fine.
for 20 Hz 22k ,capacitor value= 361.8 nF so standard value 470n more than fine as well.
8,2uf to 470n will basically work rather wide tolerance.
If our next imaginary issue is electrolytic in the path and 1uF non polarized have huge price jump.
820n would give you 8.8 Hz high pass
If opamps are non compensated for DC offset and it is good practice regardless to add error factor resistors to ground to non inverting pin.
Being a inverting amplifier for proper operation and stability. DC offset should be a low negative value
Suggest you read 500 page design application guide " Op Amps For Everyone"
Since it will answer almost any question possibly for any audio small signal op amp application you would ever need including the math
Impedance is more than high enough for common capacitor values to have no high pass issues.
Mooly took it one step further and increased impedance to 22k which would be extremely compatible
with any signal chain seen in the wild. Also makes it high enough to exclude electrolytic values if desired.
Portable speaker application with small speakers wont have response over 80 to 100 Hz
unless you have some small " sub" might be 65 to 45 Hz
If you still require 1 Hz bandwidth, which caters to imaginary bandwidth
Calculate you min frequency high pass value
for 1 Hz 22k, capacitor value = 7.23 uF so nearest standard values 8.2uF even 10uF be more than fine.
for 20 Hz 22k ,capacitor value= 361.8 nF so standard value 470n more than fine as well.
8,2uf to 470n will basically work rather wide tolerance.
If our next imaginary issue is electrolytic in the path and 1uF non polarized have huge price jump.
820n would give you 8.8 Hz high pass
If opamps are non compensated for DC offset and it is good practice regardless to add error factor resistors to ground to non inverting pin.
Being a inverting amplifier for proper operation and stability. DC offset should be a low negative value
Suggest you read 500 page design application guide " Op Amps For Everyone"
Since it will answer almost any question possibly for any audio small signal op amp application you would ever need including the math
Last edited:
Well... I dont think OPA1612's DC offset would cause problems at unity gain...
Its input bias current ranges from ±60nA to ±250nA MAX.
Maximum input offset voltage is ±500uV, minimum ±100uV.
And it should not cause instability, since the OPA1612 is unity gain stable.
I think the effect is negligible.
Honestly, i'm terrible at math and i don't really know ho to calculate those additional resistors.
That book is also long to read and it will take too much time for now.
But it could be useful in the future, so thank you.
Its input bias current ranges from ±60nA to ±250nA MAX.
Maximum input offset voltage is ±500uV, minimum ±100uV.
And it should not cause instability, since the OPA1612 is unity gain stable.
I think the effect is negligible.
Honestly, i'm terrible at math and i don't really know ho to calculate those additional resistors.
That book is also long to read and it will take too much time for now.
But it could be useful in the future, so thank you.
Last edited: