I cannot tell how it will do with 0.45 volt reduction .... too much parameters to figure out .....
Reducing output by software is not the best solution ... will only reduce soung signal and not reduce noise floor ... as far as I know ... bu I have not much experience with digital audio ...
Reducing output by software is not the best solution ... will only reduce soung signal and not reduce noise floor ... as far as I know ... bu I have not much experience with digital audio ...
The noise is added after the gain set. I have checked the gain structure with help of minidsp support, but we couldn't fix it.
Lowering the gain through the software is just increasing the noise....
That's why I want to try the attenuation after the minidsp.
Lowering the gain through the software is just increasing the noise....
That's why I want to try the attenuation after the minidsp.
then possibly the amplifiers gains are set too high and you've reduced the signal too much in software? or you need amps with better S/N ratio's
Last edited:
the signal is from a mini DSP and changing the unit at this time is out of range so an attenuator is the easiest solution for now..
digital signal processing seems to give it's best performance when the absolute maximum signal within the processing chain is just below the 0dBfs.
That generally results in a fixed and high maximum output voltage.
This analogue signal needs to be attenuated to feed acceptable levels into the following analogue power amplifiers.
That's the way it is done for the Behringer dcx2496, the 6 channels of attenuation are between the power amp inputs and the digital processor.
It's the way we do it for the separates in a music replay system, where a digital source feeds into a volume control/pre-amp and then to the power amplifier/s.
Last edited:
you will not find 4k resistors easily.
E24 range are available from most retailers, some sell E96 and E192 ranges but even these don't include 4k
Use either 3k9, or 4k3, if you want/need near 4k.
That is my understanding too, Andrew.
However, there is the matter of music peaks. If the peak has to be just below 0dBfs ... then what do we take the 'normal' signal level to be? 😕
10dB down?
15dB down?
20dB down?
Andy
It varies with the type of music and the opinion of the recording producer.
For my purposes I apply 20dB from average to maximum for my listening and that seems to avoid voltage clipping on nearly everything I want to reproduce (analogue and digital).
But there is current clipping to be taken into account and many Builders completely forget this requirement. I champion current capability and regularly comment on it in my posts.
I have referred to chipamp as being current crippled and this conclusion has been arrived at by listening to typical chipamp implementations and they quite audibly cannot play as loud as same wattage discrete amplifiers. I believe this inability to be turned up is down to current clipping inside the very well protected chipamps.
For my purposes I apply 20dB from average to maximum for my listening and that seems to avoid voltage clipping on nearly everything I want to reproduce (analogue and digital).
But there is current clipping to be taken into account and many Builders completely forget this requirement. I champion current capability and regularly comment on it in my posts.
I have referred to chipamp as being current crippled and this conclusion has been arrived at by listening to typical chipamp implementations and they quite audibly cannot play as loud as same wattage discrete amplifiers. I believe this inability to be turned up is down to current clipping inside the very well protected chipamps.
Thanks, Andrew,
Yes, 20dB is a figure that has been recommended by others. But I fear this means the level set for the 'average' signal is too low for optimal listening (given that digital sounds best when driven just below its clip limit ... rather than 1/10th of this)?
Or am I worrying needlessly? 😱
This comment relates to power amps only, I presume?
Thanks,
Andy
If I am listening at 1W average then a 100W amplifier gives me 20dB for transients.
If I turn it down a bit and listen at 500mW average I now have 23db for transients (it clips less often) and if I turn it down a lot and listen at 200mW average, then I have even fewer clipping incidents.
I think you are needlessly worrying.
I think it applies to every source that has to drive a capacitive load (a cable and/or parasitics).
Thanks, Andrew.
Absolutely, Andrew! But you have taken an analogue example - lets take a digital example.
An ADC has a 2v limit. IOW, if the input signal is 1990mV ... all is well! 🙂 However, if the input signal is 2010mV ... I will get clipping. 🙁
If I want to allow 20dB headroom (10x) to avoid any clipping then the average input signal needs to be 200mV.
Given that for lowest noise, digital circuits should be run near the maximum signal level ... 200mV seems to me to be too low for good noise levels?
But if I make the average signal level, say, 500mV - to get good noise ... this only allows 4x (12dB) headroom for peaks. Which from all accounts is insufficient. 🙁
So the digital milieu provides a dilemma which the analogue environment doesn't! 😡
Regards,
Andy
Absolutely, Andrew! But you have taken an analogue example - lets take a digital example.
An ADC has a 2v limit. IOW, if the input signal is 1990mV ... all is well! 🙂 However, if the input signal is 2010mV ... I will get clipping. 🙁
If I want to allow 20dB headroom (10x) to avoid any clipping then the average input signal needs to be 200mV.
Given that for lowest noise, digital circuits should be run near the maximum signal level ... 200mV seems to me to be too low for good noise levels?
But if I make the average signal level, say, 500mV - to get good noise ... this only allows 4x (12dB) headroom for peaks. Which from all accounts is insufficient. 🙁
So the digital milieu provides a dilemma which the analogue environment doesn't! 😡
Regards,
Andy
You don't excessively attenuate the analogue signal before entering the ADC.
You attenuate the signal after leaving the DAC. It is this suitably attenuated analogue signal that you pass to the Power Amplifier.
Inside the digital processor you should be running at signals that never exceed 0dBfs and preferably run as close to that during the peaks of transients. Appropriate analogue level control before entering the digital processor should be set to achieve those dual goals, of approaching the 0dBfs and never exceeding 0dBfs.
Your digital processor does not run at mVac, it runs on bits of information. Only after passing through the DAC does it give analogue voltages.
You attenuate the signal after leaving the DAC. It is this suitably attenuated analogue signal that you pass to the Power Amplifier.
Inside the digital processor you should be running at signals that never exceed 0dBfs and preferably run as close to that during the peaks of transients. Appropriate analogue level control before entering the digital processor should be set to achieve those dual goals, of approaching the 0dBfs and never exceeding 0dBfs.
Your digital processor does not run at mVac, it runs on bits of information. Only after passing through the DAC does it give analogue voltages.
Last edited:
Understood. I will have an 8-channel attenuator just before the 8 channels of amplifier.
Exactly! 🙂
But:
* if the ADC has a maximum input signal level of 2000mV
* and I have to allow 20dB of peak (10x)
* the input signal peaks cannot exceed 2000mV
* which, with headroom of 10x
... means the average input signal level will only be 200mV. Which is breaking the desirable 'rule' of having the signal level close to the maximum for best digital sound.
Andy
I need the same kind of in-line attenuator but I need it to be 12dB, not 6dB
I have a pair of Rothwell 10dB on the way but they will serve a different purpose
here I need to integrate 12dB of attenuation at the end of a pair of diy interconnects
is it possible (=would it fit inside an RCA plug?) and wich values should I choose?
thanks in advance.
I have a pair of Rothwell 10dB on the way but they will serve a different purpose
here I need to integrate 12dB of attenuation at the end of a pair of diy interconnects
is it possible (=would it fit inside an RCA plug?) and wich values should I choose?
thanks in advance.
For 12 dB you need a voltage ratio of 4 ... ( 3.98 ) so upper resistor should be 3 times the value of the lower one .... 6K8 ohms & 22 K ohms may fit about 12.5 dB ..... but still have to know input & output impedance ... to make a final selection.
Andy,
As usual, you're over-thinking it. 🙂 The only thing that matters is that you not exceed 0dbFS with your musical peaks.
Unless you plan on compressing your music, the average level will be what it will be.
Dave.
As usual, you're over-thinking it. 🙂 The only thing that matters is that you not exceed 0dbFS with your musical peaks.
Unless you plan on compressing your music, the average level will be what it will be.
Dave.
- Status
- Not open for further replies.