The input eq settings are directly written in the fusion without the need of filterdesigner. These setttings are not stored in the Filterdesigner project file. When you open pre eq the settings are read from the plate. With Copy you can save the presets to clipboard and with write upload these to a other plate. One big remark: In Filterdesiger this input eq is layed on top of the summation curve when the plate is connected!
So, the project files just do those low pass, high pass thing.
And use the EQ to touch up.
Does anyone know how the processing is done bitwise on the Fusion? I assume the input bit width is 24 bit. And the DACs are 32 bit data, though naturally this precision is not achieved noise wise. I read somewhere that the internal processing is 64 bits, is this correct? The question is then how these bits are allocated, as headroom or as additional precision (or more likely: a combination of both). I currently listen around -52..-36 dB. So I would assume that roughly every 6 dB you loose a bit precision, so I lose 6..9 bits. Is there any way to know how hot the DAC is driven? I tried the new VU feature, but very often it just stays at -72 dB...
Is there any good practice how the set up the filters? E.g. can I just boost 30 Hz with 10 dB on channel 1 without worrying about clipping? Or should I bring back the gain of channel 1 with -10 dB at the same time to avoid clipping? Or is it better practice to attenuate instead of boosting, e.g. with a shelf above 30-40 Hz?
Fedde
Is there any good practice how the set up the filters? E.g. can I just boost 30 Hz with 10 dB on channel 1 without worrying about clipping? Or should I bring back the gain of channel 1 with -10 dB at the same time to avoid clipping? Or is it better practice to attenuate instead of boosting, e.g. with a shelf above 30-40 Hz?
Fedde
Fusionamps uses a 32-bit ADAU1450 DSP chip. So it works at 32-bit, I think.
Although the new EQ is a nice feature, but we now have more difficulty to track the digital sound level to not exceed the 0dB.
Or exceeding the digital 0dB is not a problem distortion wise? Some info would be nice about this.
Although the new EQ is a nice feature, but we now have more difficulty to track the digital sound level to not exceed the 0dB.
Or exceeding the digital 0dB is not a problem distortion wise? Some info would be nice about this.
The datasheet of the ADU1450 reads:
"The serial port input and output word lengths are 24 bits, but eight extra headroom bits are used in the processor to allow internal gains of up to 48 dB without clipping."
(https://www.analog.com/media/en/technical-documentation/data-sheets/ADAU1452_1451_1450.pdf page 89)
So it seems that internal clipping is nothing to worry about, unless you do something extreme, but you can run into clipping at the output. I guess the channel gain can be used to avoid this.
The effects of the EQ looks to be included in the blue curves shown in Filter design, so what you see in there is the total output of your channels.
"The serial port input and output word lengths are 24 bits, but eight extra headroom bits are used in the processor to allow internal gains of up to 48 dB without clipping."
(https://www.analog.com/media/en/technical-documentation/data-sheets/ADAU1452_1451_1450.pdf page 89)
So it seems that internal clipping is nothing to worry about, unless you do something extreme, but you can run into clipping at the output. I guess the channel gain can be used to avoid this.
The effects of the EQ looks to be included in the blue curves shown in Filter design, so what you see in there is the total output of your channels.
If your output meter barely moves, you have a big problem with gain staging. Your listening levels are very low or you speaker has a very high sensitivity.
Or actually both...
In the meantime I built MDF boxes around my Fusion 253 plate amps and started evaluating them. Noise level is okish, not objectionable from a distance, though I presume the noise will still reduce micro detail. But regrettably, the sound quality is less than the NC400 / Soekris DAC combo at low listening levels. The overall sound is more flat and there is less instrument realism. Root causes could be too much bit loss, but perhaps the jitter level of my spdif source is too high as well (RME AIO soundcard). Switching the card to Pro level helps a bit (probably overdriving the spdif somewhat, but this helps to reduce jitter).
Please note that I do not make effective use of all the Fusion capabilities yet, I just started to comparing the old and new setup. This while using a combination of PC based EQ and passive filters. So I expect I can win a lot yet by using the full Fusion approach and removing the other EQ and passive filters (except for a bit of tweeter protection that is).
I am seriously considering removing Rg fully on my plate amps to bring the gain back to 4.17x. This way, I can raise my master volume around 12 dB, bringing back some resolution...
(and on top of that I could still apply series resistors if I would want to)
Fedde
Or perhaps you just subjectively prefer the coloration of the Soekris DAC (a DAC using the ancient R2R architecture, and not known for accuracy or transparency)?
Happily. I designed my first R2R DAC back in the 70s - it is a very easy (almost trivial) architecture, but there is a reason we moved away from it 35 years ago.
85 dB SINAD is roughly equivalent to 14 bits.
Last edited:
The one who claimed that the Soekris DAC and NC400 combo represent higher sound quality.
Last edited:
YSDR,
The guy said: "the sound quality is less than the NC400 / Soekris DAC combo at low listening levels. The overall sound is more flat and there is less instrument realism."
Of course that's a subjective opinion.
But we are humans listening to music aren't we, not oscilloscopes....
Julf's remark on the accuracy/transparency thing is a gross generalization and oversimplification.
The main reason that chip manufacturers switched from R2R to DS was COST.
Today many of the best sounding (yes it's subjective...) dacs rely on R2R technology, and they measure extremely well.
The guy said: "the sound quality is less than the NC400 / Soekris DAC combo at low listening levels. The overall sound is more flat and there is less instrument realism."
Of course that's a subjective opinion.
But we are humans listening to music aren't we, not oscilloscopes....
Julf's remark on the accuracy/transparency thing is a gross generalization and oversimplification.
The main reason that chip manufacturers switched from R2R to DS was COST.
Today many of the best sounding (yes it's subjective...) dacs rely on R2R technology, and they measure extremely well.
Last edited:
Of course it is. I was simply suggesting that the cause of the subjective impression wasn't the Fusion.
If you say so. How many of them have linearity beyond 16 bits?
Sorry, but that part is simply not true, unless you interpret it very literally. The main reason was greater linearity and accuracy of delta-sigma technology.
At least that's easy to test by using the Soekris with the Hypex's analog input.
And here are the QA401 measurements posted at ASR.
C'mon....decades ago we had Analog Devices and Burr Brown chips featuring 18 bit or better. PCM1704 was 24 bit....
No.
DS is easier to reach the "perfect" specifications (at the cost of a lot of RF garbage...).
However, how come that so many prefer R2R dacs sonically?
All mad listeners?
Maybe something to do with the more benign character of R2R dacs wrt RF?
Funny subject, loaded with subjectivism, and not to be backed up with objective data.
I'd say thanks goodness!
Audio would become boring when "better measuring" would become identical with "audibly better"....
24 bit linearity? That I would like to see evidence of...
How come that so many prefer valves/tubes, vinyl, cassette, AM radio, boom boxes, Barry White or even flared trousers?