The reason I asked is that balanced connections often add noise and balanced connections can end up in mismatch signal levels if not implemented properly.
Balanced connections are very good for noise immunity between devices, which is important in electrically noisy environments like entertainment venues with megawatts of dimmed lighting, or where the source and destination devices are miles apart. One handy aspect of balanced connections is that the signal does not need to be referenced to ground, which can make up for all sorts of indiscretions committed by audio equipment manufacturers who don't understand how to properly implement system grounds (which is most IMHO).
We live in an acoustically 'unbalanced' world, we experience 'single ended' sound, the acoustic sounds recorded are unbalanced waves and amplifiers and processors usually operate internally unbalanced except in extremely rare cases. So balanced connections require additional electronic stages or transformer conversions for each output and input. And no matter what is upstream the speaker always unbalances the audio so that we can hear it.
Don't overthink your setup. Get it to where it is working and not annoying you and then enjoy it. The objective is to listen to the music, not to the faults in your system, nor obsess about mythical or hypothetical improvements. Work on the fundamentals instead, because there are often huge gains to be had there through simply getting them right.
Balanced connections are very good for noise immunity between devices, which is important in electrically noisy environments like entertainment venues with megawatts of dimmed lighting, or where the source and destination devices are miles apart. One handy aspect of balanced connections is that the signal does not need to be referenced to ground, which can make up for all sorts of indiscretions committed by audio equipment manufacturers who don't understand how to properly implement system grounds (which is most IMHO).
We live in an acoustically 'unbalanced' world, we experience 'single ended' sound, the acoustic sounds recorded are unbalanced waves and amplifiers and processors usually operate internally unbalanced except in extremely rare cases. So balanced connections require additional electronic stages or transformer conversions for each output and input. And no matter what is upstream the speaker always unbalances the audio so that we can hear it.
Don't overthink your setup. Get it to where it is working and not annoying you and then enjoy it. The objective is to listen to the music, not to the faults in your system, nor obsess about mythical or hypothetical improvements. Work on the fundamentals instead, because there are often huge gains to be had there through simply getting them right.
You stated that you are running the crossover and amps at fixed level, and you are attenuating the signal in the digital domain, before the DAC.
What is your source, and how much do you need to attenuate the digital signal?
If you are attenuating more than a few dB at the absolutely maximum volume you are going to use, the gain after the crossover will raise the level of any hiss from the dac, and also raise the noise from the crossover. Now, some components are inherently noisy - I own a MiniDSP OpenDRC that makes a fair amount of hiss, but you need to move any constant attenuation as close to the speakers as you can. In effect, that means running the input level to the crossover as close to max as you can, and lowering the gain on the amps as much as you can.
Run your source as close to full tilt, and put line level attenuators on your amps (or use input pots, if they have them). This will maximise your S/N ratio. If this is ok and the noise is still too high - then you can start to look at the crossover and amps.
Johan-Kr
What is your source, and how much do you need to attenuate the digital signal?
If you are attenuating more than a few dB at the absolutely maximum volume you are going to use, the gain after the crossover will raise the level of any hiss from the dac, and also raise the noise from the crossover. Now, some components are inherently noisy - I own a MiniDSP OpenDRC that makes a fair amount of hiss, but you need to move any constant attenuation as close to the speakers as you can. In effect, that means running the input level to the crossover as close to max as you can, and lowering the gain on the amps as much as you can.
Run your source as close to full tilt, and put line level attenuators on your amps (or use input pots, if they have them). This will maximise your S/N ratio. If this is ok and the noise is still too high - then you can start to look at the crossover and amps.
Johan-Kr
Well yes, and no. Does a racing car drive need to know how the valve time advance in his car's engine works? Does a movie director need to know how many bits of data the CGI team need in their processor?
What they need to know is how to find and trust experts to sort these things out so they can focus their creative energy on what they are good at and/or paid to do.
Thanks for your reply. My source is a Fiio Q1 MKII.
Specs: Q1MKII_parameters
The balanced output has an SNR of over 109 dB (I use USB input).
I believe all the noise comes out of the crossover. The noise exists at the same level even if the inputs are shorted (balanced XLR IO, shorting hot to cold).
How much do you have to turn down the volume on the PC? If you have to turn down mmuch more than 10-15dB, you should attenuate your amps. Any attenuation of the amps will reduce the noise you experience from the crossover.
Since you are running single ended, you loose 6dB signal level that could have been used to improve your S/N by attenuating the amps.
Your soundcard has a 2,5mm balanced output that could help you there, but you’ll need an adapter for it.
https://www.amazon.com/2-5mm-Cable-Balanced-Headphone-Adapter/dp/B07WTMFQ4G
Johan-Kr
Actually the specification for the Q1MKII is >110dB 'A' weighted signal to noise. S-N is the difference between maximum output level and noise floor, with the low frequencies filtered out somewhat. The same specification for the 10B is >128dB unweighted, and would be some dBs more if the same 'A' weighting was applied.
There is not enough output from the Q1MKII to drive the 10B properly. You need a gain stage between the two, otherwise the 10B will always be working in its high THD+N region.
Personally I would not use the balanced headphone output, even though that could give 6dB more level. If you want to maximise quality, minimise circuitry, especially consumer audio outputs that are not designed as line drivers.
I would use the line output and then add a consumer audio level interface. As you can't bypass the balanced inputs to the 10B without modifying it, you might as well use an unbalanced to balanced interface, which will also remove ground reference and prevent the introduction of another potential noise source (i.e. the USB supply ground noise) so you can probably use the Q1MKII plugged in to a power source.
There is not enough output from the Q1MKII to drive the 10B properly. You need a gain stage between the two, otherwise the 10B will always be working in its high THD+N region.
Personally I would not use the balanced headphone output, even though that could give 6dB more level. If you want to maximise quality, minimise circuitry, especially consumer audio outputs that are not designed as line drivers.
I would use the line output and then add a consumer audio level interface. As you can't bypass the balanced inputs to the 10B without modifying it, you might as well use an unbalanced to balanced interface, which will also remove ground reference and prevent the introduction of another potential noise source (i.e. the USB supply ground noise) so you can probably use the Q1MKII plugged in to a power source.
There's nothing wrong with that, but the driver needs to ask the right questions to get the desired result. There's no point adjusting valve timing if the problem is the injector timing is out.
Thanks for the info. I didn't know about the caveat with S-N measurements.
How would I calculate how much gain I need from the Fiio to the 10B?
If I'll be using an amplifier anyways, why not use the Fiio's balanced output? The amplifier can be used as a line driver. It's not that long of a connection though, the cable is only 15 cm, and the 10B has high input impedance as well. What other issues could result here from the fact that the balanced output is not meant as a line driver, as you say?
It's not strictly a balanced output; it would be more correct to call it a push-pull output. It uses bridged amplifiers to get more voltage swing from the tiny supply voltage the the Fiio runs on.
Personally I prefer not to use headphone outputs where a line output is available. All circuitry is a compromise and for headphone drivers the compromises have been chosen by the designer to be the least impactful for the intended purpose, in this case headphones. The line output usually comes from upstream of the headphone driver (so less added N+D) and is designed to perform best into a 10kΩ+ resistive load, not an 8-150Ω reactive load.
We have established that you ideally would have 20dB pads between the crossover and amplifier for optimum THD+N, so to keep the same gain as you have it now, put 20dB of gain between the Fiio and the crossover.
If you find in your current setup you have too much gain, then put less gain in from of the crossover, but keep the 20dB pads on the outputs.
A balancing adapter for connecting consumer audio equipment to professional audio equipment would normally have ~20dB of gain, whereas a balancing adapter for connecting consumer to consumer audio equipment with balanced inputs would normally have ~14dB. Either would be fine, as long as competently made.
Personally I prefer not to use headphone outputs where a line output is available. All circuitry is a compromise and for headphone drivers the compromises have been chosen by the designer to be the least impactful for the intended purpose, in this case headphones. The line output usually comes from upstream of the headphone driver (so less added N+D) and is designed to perform best into a 10kΩ+ resistive load, not an 8-150Ω reactive load.
We have established that you ideally would have 20dB pads between the crossover and amplifier for optimum THD+N, so to keep the same gain as you have it now, put 20dB of gain between the Fiio and the crossover.
If you find in your current setup you have too much gain, then put less gain in from of the crossover, but keep the 20dB pads on the outputs.
A balancing adapter for connecting consumer audio equipment to professional audio equipment would normally have ~20dB of gain, whereas a balancing adapter for connecting consumer to consumer audio equipment with balanced inputs would normally have ~14dB. Either would be fine, as long as competently made.
Last edited:
I should just correct a comment I made in a previous post. The signal to noise is the difference between the nominal or rated signal level of a device and it's noise floor. The difference between maximum undistorted output level and noise is actually the dynamic range, also measured in dB. The 10B has 128dB of dynamic range, which is roughly inaudible to painful if gain structure is correctly set up.
Digital devices have no overload margin (maximum bits is maximum bits) so quoted S-N is either close to the dynamic range or is the dynamic range. Fiio don't specify how the measurement was made so it is not clear how much headroom there is above 110dB. Given that the device runs on a 5 volt USB supply and probably a 3.7 volt battery inside, it is unlikely to have much headroom, if any.
The maximum output level is given as 4.4Vpp = 1.5Vrms or ~+4dBv. Note that is the clipping level. I would take a punt that because the frequency response curves are at -5dB, there is 5dB of headroom, giving a total of 115dB A weighted S-N and a noise floor around -110dBv, A weighted.
Digital devices have no overload margin (maximum bits is maximum bits) so quoted S-N is either close to the dynamic range or is the dynamic range. Fiio don't specify how the measurement was made so it is not clear how much headroom there is above 110dB. Given that the device runs on a 5 volt USB supply and probably a 3.7 volt battery inside, it is unlikely to have much headroom, if any.
The maximum output level is given as 4.4Vpp = 1.5Vrms or ~+4dBv. Note that is the clipping level. I would take a punt that because the frequency response curves are at -5dB, there is 5dB of headroom, giving a total of 115dB A weighted S-N and a noise floor around -110dBv, A weighted.
Did you want a good circuit to build yourself, or a good product off the shelf?
Thanks for the correction.
Up to $500 for a fully finished box (mechanical assembly required is OK but not too much soldering), I'd buy it. Above that, I'd build it myself. Either way I think it would be a fun experience to build something like that.
(I'm very good with soldering, just defining what "off the shelf" means here)
Up to $500 for a fully finished box (mechanical assembly required is OK but not too much soldering), I'd buy it. Above that, I'd build it myself. Either way I think it would be a fun experience to build something like that.
(I'm very good with soldering, just defining what "off the shelf" means here)
OK. So I am retired now and not up to date with what's on offer out there. Maybe someone else can chip in. I'll have a bit of a gander on the net and see if something stands out. Did you want the same, more, or less gain in the system than you have now? Where are you running the output level on the Q1?
BTW, someone asked me before, so I just did a test since it's day.
I'll normally be listening to music at ~ -22 dBFS, and if I want to turn up, say, a dance track to listen to it at higher volume, then I'll do -6 dBFS, -2 dBFS, or even 0 dBFS.
That's with the Bryston power amplifiers set to 2V (which is the lower sensitivity setting) and with the differential output from the Fiio, and the Fiio set to Gain=H and Bass=Off.
Gain on Fiio Q1MKII can be set to L or H. On the 3.5mm SE headphone output L gives -2.5 dB gain. H gives +3.2 dB gain. On the 2.5mm BAL headphone output which I'm using L gives +3.2 dB gain and H gives +9.1 dB gain.
Bass is a bass emphasis circuit that can be switched in or out.
The Fiio specifies peak output voltage as > 4.4 Vpp for the 3.5mm SE headphone output and > 7.4 Vpp for the 2.5mm BAL headphone output which I'm using.
Bear in mind the Fiio is only a stop gap. It's the best DAC I currently have. Eventually I'll buy a studio DAC that's superior to it, but the Fiio is actually an insane bang for the buck proposition so it won't be cheap, especially if I want to do a digital crossover (which I do).
I'll normally be listening to music at ~ -22 dBFS, and if I want to turn up, say, a dance track to listen to it at higher volume, then I'll do -6 dBFS, -2 dBFS, or even 0 dBFS.
That's with the Bryston power amplifiers set to 2V (which is the lower sensitivity setting) and with the differential output from the Fiio, and the Fiio set to Gain=H and Bass=Off.
Gain on Fiio Q1MKII can be set to L or H. On the 3.5mm SE headphone output L gives -2.5 dB gain. H gives +3.2 dB gain. On the 2.5mm BAL headphone output which I'm using L gives +3.2 dB gain and H gives +9.1 dB gain.
Bass is a bass emphasis circuit that can be switched in or out.
The Fiio specifies peak output voltage as > 4.4 Vpp for the 3.5mm SE headphone output and > 7.4 Vpp for the 2.5mm BAL headphone output which I'm using.
Bear in mind the Fiio is only a stop gap. It's the best DAC I currently have. Eventually I'll buy a studio DAC that's superior to it, but the Fiio is actually an insane bang for the buck proposition so it won't be cheap, especially if I want to do a digital crossover (which I do).