Is it possible for you to measure the responses at the output of the FusionAmp? That would eliminate any acoustic measurement influences.
Here is the woofer nearfield response. Darker yellow graph is without the filter, brighter yellow is filter added to the EQ tab, grey line is filter added to filter design tab. Filter parameters is F=50Hz, Q=9, gain=-10dB.
I did 5 measurements with each settings, the results was consistent.
I did 5 measurements with each settings, the results was consistent.
Got my 253s up and running finally.
I am getting some turn off thump regardless of how I turn them off (remote or power switch, muted or low volume with digital source running).
It seems to be coming from woofers only, but not sure cause its bass frequencies anyway.
I am running a 2way with the second nc252mp channel floating. Is that the cause?
Could that also be the cause of extra heat buildup? Shouldnt classd amps always run on a load?
I am getting some turn off thump regardless of how I turn them off (remote or power switch, muted or low volume with digital source running).
It seems to be coming from woofers only, but not sure cause its bass frequencies anyway.
I am running a 2way with the second nc252mp channel floating. Is that the cause?
Could that also be the cause of extra heat buildup? Shouldnt classd amps always run on a load?
I don't know.
Default settings is recommended. Experimenting with DAC filters if you are curious. Quality settings is best at default, no question.
Hi,
My current setup is : Raspberry Pi with Allo Digione > Minidsp Nanodigi > 3 IAMD V200 > 3-way speakers (SB NRX 10", SB Satori MW16P, SB Satori TW29R) about 88dB/1m efficiency
I'm considering replacing the Minidsp and the 3 V200 amps by a couple of FA123 or FA253.
I mainly listen at low to moderate volumes.
Now, I'm wondering about the bit depth used by the FusionAmps for the volume control and if I may have some signal degradation due to digital volume control when listening at low volume.
Per the block diagram in the FusionAmp manual, the volume control is managed by the DSP. Both DSP and DAC are 32 bits capable but the DSP datasheet seem to imply that the signal is truncated to 24 bits when it leaves and enters the DSP Sigma Core.
Page 89/195
Many of the signal processing functions are coded using full, 64-bit, double precision arithmetic. 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.
Does this mean the bit depth is actually only 24 bits and volume control only has 8 bits before it starts degading the 16 bits source signal ?
If yes, that would point towards going with the FA123 to reduce (a bit) the digital volume reduction at my usual listing volume. I don't really need extra power, but if I lose nothing at lower volumes vs FA123 I would go with the FA253.
What do you think ?
By the way, any feedback with the performance of the FusionAmps at lower volumes?
Thanks
Mathieu
My current setup is : Raspberry Pi with Allo Digione > Minidsp Nanodigi > 3 IAMD V200 > 3-way speakers (SB NRX 10", SB Satori MW16P, SB Satori TW29R) about 88dB/1m efficiency
I'm considering replacing the Minidsp and the 3 V200 amps by a couple of FA123 or FA253.
I mainly listen at low to moderate volumes.
Now, I'm wondering about the bit depth used by the FusionAmps for the volume control and if I may have some signal degradation due to digital volume control when listening at low volume.
Per the block diagram in the FusionAmp manual, the volume control is managed by the DSP. Both DSP and DAC are 32 bits capable but the DSP datasheet seem to imply that the signal is truncated to 24 bits when it leaves and enters the DSP Sigma Core.
Page 89/195
Many of the signal processing functions are coded using full, 64-bit, double precision arithmetic. 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.
Does this mean the bit depth is actually only 24 bits and volume control only has 8 bits before it starts degading the 16 bits source signal ?
If yes, that would point towards going with the FA123 to reduce (a bit) the digital volume reduction at my usual listing volume. I don't really need extra power, but if I lose nothing at lower volumes vs FA123 I would go with the FA253.
What do you think ?
By the way, any feedback with the performance of the FusionAmps at lower volumes?
Thanks
Mathieu
Hi,
I've received my Fusion plate FA252 to modify my RCF active speaker (to improve bad internal noisy digipro amplifier with bad ADC/DAC of RCF) :
Question about DSP headroom : (don't see in Hypex or HFD manual)
Woofer have -10/-8dB, sensitivity than the Tweeter.
I use AES INPUT (and i change the volume with MINIDSP DRC -DI + FIR)
if I feed 0dBFS signal into the Fusion DSP, and main volume of DSP is 0 dB, and PEQ EQ is at 0dB, what is the HEADROOM of CHANNEL GAIN in presets :
0 dB channel Gain is -24 dBfs of Headroom ? or -24 dB Channel Gain is -48 dB?
For example with no EQ/ no PEQ :
Woofer at + 8dBfs in CHANNEL 1 Gain
Tweeter at 0 dBfs in CHANNEL 2 Gain
and Channel 1 have 24-8 dB so 16 dB of headroom available for PEQ boost ?
and Channel 2 have 24-0 dB so 24 dB of headroom available for PEQ boost ?
or should i set
Woofer at 0dBfs in CHANNEL 1 Gain
Tweeter at -8 dBfs in CHANNEL 2 Gain
And 0 Channel Gain is 0dBFS ? or -24 dBfs ? or - 48 dBfs ? of headroom (with no EQ/PEQ actived)
or can i set :
Woofer at + 24dBfs in CHANNEL 1 Gain
Tweeter at +16 dBfs in CHANNEL 2 Gain
if i don't apply EQ/PEQ ?
Please Hypex enlight me, answer me, because it's no clear in Manual and spec.
How is Channel Gain et EQ/PEQ Headroom in DSP ?
(With examples, please)
Regards,
Olivier
I've received my Fusion plate FA252 to modify my RCF active speaker (to improve bad internal noisy digipro amplifier with bad ADC/DAC of RCF) :
Question about DSP headroom : (don't see in Hypex or HFD manual)
Woofer have -10/-8dB, sensitivity than the Tweeter.
I use AES INPUT (and i change the volume with MINIDSP DRC -DI + FIR)
if I feed 0dBFS signal into the Fusion DSP, and main volume of DSP is 0 dB, and PEQ EQ is at 0dB, what is the HEADROOM of CHANNEL GAIN in presets :
0 dB channel Gain is -24 dBfs of Headroom ? or -24 dB Channel Gain is -48 dB?
For example with no EQ/ no PEQ :
Woofer at + 8dBfs in CHANNEL 1 Gain
Tweeter at 0 dBfs in CHANNEL 2 Gain
and Channel 1 have 24-8 dB so 16 dB of headroom available for PEQ boost ?
and Channel 2 have 24-0 dB so 24 dB of headroom available for PEQ boost ?
or should i set
Woofer at 0dBfs in CHANNEL 1 Gain
Tweeter at -8 dBfs in CHANNEL 2 Gain
And 0 Channel Gain is 0dBFS ? or -24 dBfs ? or - 48 dBfs ? of headroom (with no EQ/PEQ actived)
or can i set :
Woofer at + 24dBfs in CHANNEL 1 Gain
Tweeter at +16 dBfs in CHANNEL 2 Gain
if i don't apply EQ/PEQ ?
Please Hypex enlight me, answer me, because it's no clear in Manual and spec.
How is Channel Gain et EQ/PEQ Headroom in DSP ?
(With examples, please)
Regards,
Olivier
I came here to ask the same think! I;d like your expert views on this too!
I can only send analog audio to the hypex and my listening volume is low. (97db woofer, 110ish db CDs - coax)
A digital volume control should be dithered before gain reduction to prevent loss of detail. It means the audio signal will recede below a fixed level noise floor, but that is probably better than just throwing away bits.
Regarding 16-bit source material, after DSP is applied bit depth may increase. If that happens then the signal should be dithered before bit-depth reduction is performed.
24-bits is not an unreasonable I/O limitation since the very best dacs available only have in the range of 21 - 22 effective bits (ENOB). The reason for using more bits than 24 in DSP is to minimize accumulation of calculational inaccuracies.
Don't know under what circumstances if any the DSP chips used employ dither.
Regarding 16-bit source material, after DSP is applied bit depth may increase. If that happens then the signal should be dithered before bit-depth reduction is performed.
24-bits is not an unreasonable I/O limitation since the very best dacs available only have in the range of 21 - 22 effective bits (ENOB). The reason for using more bits than 24 in DSP is to minimize accumulation of calculational inaccuracies.
Don't know under what circumstances if any the DSP chips used employ dither.
Last edited:
I've sent a mail to HYPEX : no answer. Where is the support ?
ADAU 1450 say :
see attachement and
https://www.analog.com/media/en/technical-documentation/data-sheets/ADAU1452_1451_1450.pdf
Headroom without clipping, but the Hypex DSP programming apply the same headroom than 24 bits -> 32 bits -> 24 bits of ADAU 1450?
Regards,
Olivier F.
Attachments
So it takes 24-bit input data, does the calculation in 64 bit, and outputs 24 bits. That means that any calculations/operations (including volume adjustment) are way below the noise floor. No need to worry.
If you search this thread you will find discussions on the gain structure of Fusion amps. With high efficiency speakers as above I would recommend to remove Rg to reduce idle noise and improve SNR and reduce truncation. Also serial resistors on the output could help, especially for the tweeter. Ensure that the analog input does have sufficient level to fill all the input bits, but leave a small margin to avoid clipping.
Also, I heard that processing of pre EQ is not double precision yet, but this is planned to be fixed. I did experiments with amplifying versus attenuation of channels, and I would recommend to amplify maximally as I believe it sounds better (but ensure you do not go over 48 dB gain). I have one channel at 15.9 dB (the max value) and one at 12.9 dB gain.
I would have preferred 32 bit input to the DAC, but with the measures above low level listening is possible with high efficiency drivers.
Fedde
Last edited: