The New Hypex Fusion Plate amps

How can you listen to music at all if you are so concerned about quantization noise of the CD quality? :confused:
I am happy with CD quality, but I try to take clear is possible to configure FA253 in way that it will not degradate at least CD quality signal with needed OB correction.

One problem is still present. With bass dipole compensation about 20 dB there will be in bass range at 200 Hz end where compensation is 0, DAC output level 0.15V, this is 23 dB lower than DAC full level output, so THD+N in this rang will be 102.5-23= 79.5 what is lower than 16 bit signals maximum.

Question form this calculation: Did anybody know what opamp is used as buffer on MC252MF? If I searched internet, I not find any image where was possible to read out opamp type.
Plan is still to use partly analog compensation, with it I can win 13.5 dB what brings DAC output signal 13.5 dB higher and DAC output signal SINAD(THD+N) to about 93 dB.
If buffer schematic is same as Hypex suggested for his old buffer-less amplifiers, then for compensation is needed to add only 2 capacitors.
 
Internal processing in FA dac is 32bit floating point, AFAIK.

For OB/dipole loss compensation I try to use negative value (like xo filters do) and then , if necessary set overall gain of the "way" a bit higher.

Positive EQ in dsp will easily lead to digital clipping. I don't have an analyzer/oscilloscope, so I look at loudspeaker response and distortion. In my AINOs, there is total -20dB correction for the open 12" low-mid driver, but 1/3 of it is in the range where crossover's highpass is already counteracting.

If one has low bass dipole, need for 20dB compensation sounds like torture to the driver. Think about how much power you will need and what will the excursion of the driver coil be! Dipole bass is a real challenge!

Anyway, I'm not at all worried that you will loose bits or degrade S/N too much. Have you done a real world test already, or are you just tinkering about this?
 
Question is about initial 16 bit signal quantization noise.

I am happy with CD quality, but I try to take clear is possible to configure FA253 in way that it will not degradate at least CD quality signal with needed OB correction.

One problem is still present. With bass dipole compensation about 20 dB there will be in bass range at 200 Hz end where compensation is 0, DAC output level 0.15V, this is 23 dB lower than DAC full level output, so THD+N in this rang will be 102.5-23= 79.5 what is lower than 16 bit signals maximum.

Question form this calculation: Did anybody know what opamp is used as buffer on MC252MF? If I searched internet, I not find any image where was possible to read out opamp type.
Plan is still to use partly analog compensation, with it I can win 13.5 dB what brings DAC output signal 13.5 dB higher and DAC output signal SINAD(THD+N) to about 93 dB.
If buffer schematic is same as Hypex suggested for his old buffer-less amplifiers, then for compensation is needed to add only 2 capacitors.

This doesn't really make any sense to me. This is theorycrafting for theorycraftings sake. Maybe you should do a reality check and measure the THD+N of your current system at your target SPL at listening position and compare that to what you are theoring about and how it sounds to you.

FusionAmp has a really good DSP engine and filtering digital signals is what it's designed to do. While 20 dB is quite a lot, it's still within the range of what is plausible for any DSP.
 
FusionAmp has a really good DSP engine and filtering digital signals is what it's designed to do. While 20 dB is quite a lot, it's still within the range of what is plausible for any DSP.
You again did not understand my point. Problem is not in DSP, but in DAC output level, if signal output level is 23 dB lower than full range DAC output level, SINAD of this signal will also be 23 dB lower.
 
If one has low bass dipole, need for 20dB compensation sounds like torture to the driver. Think about how much power you will need and what will the excursion of the driver coil be! Dipole bass is a real challenge!
Yes, it is challenge. In my current system with 20 dB analog compensation without Hypex I had measured 20 Hz THD on comfortable listening level about 5%. Speakers in one channel had together 2032 cm2 piston area and 22 mm(p-p) Xmax.
 
You again did not understand my point. Problem is not in DSP, but in DAC output level, if signal output level is 23 dB lower than full range DAC output level, SINAD of this signal will also be 23 dB lower.

As far as I know, that is not exactly true. The noise floor can go lower if you are not outputting full scale signal, which can give you better SNR and SINAD. Also you surely cannot be listening to full volume at all times?! Normal listening conditions make amps produce power in milliwatts only. Given your high Sd, you shouldn't need to run even close to full power even with the compensation filter that you are designing. And another point, human hearing cannot detect distortion in -100 dB range. That kind of performance is required to ensure audibly transparent operation at all scenarios, like crazy OB filters.

Yes, it is challenge. In my current system with 20 dB analog compensation without Hypex I had measured 20 Hz THD on comfortable listening level about 5%. Speakers in one channel had together 2032 cm2 piston area and 22 mm(p-p) Xmax.

You got THD -26 dB at 20 Hz, that comes from the speaker, not from your amp and certainly not from your DAC. If you can loop back from DAC to computer, you can measure it.
 
Internal processing in FA dac is 32bit floating point, AFAIK.

For OB/dipole loss compensation I try to use negative value (like xo filters do) and then , if necessary set overall gain of the "way" a bit higher.

Positive EQ in dsp will easily lead to digital clipping. I don't have an analyzer/oscilloscope, so I look at loudspeaker response and distortion. In my AINOs, there is total -20dB correction for the open 12" low-mid driver, but 1/3 of it is in the range where crossover's highpass is already counteracting.

If one has low bass dipole, need for 20dB compensation sounds like torture to the driver. Think about how much power you will need and what will the excursion of the driver coil be! Dipole bass is a real challenge!

Anyway, I'm not at all worried that you will loose bits or degrade S/N too much. Have you done a real world test already, or are you just tinkering about this?

The Fusion amp processing uses 8.24 bit fixed point. So 24 bit precision and 8 bits headroom. Not so easy to get it clipping thus, should be about 48 dB headroom...
 
Thank you for confirming fedde! 32bit floating point is almost same performance as 8.24 fixed bits

Single-precision floating-point format - Wikipedia
https://www.xmos.ai/download/Digita...he-XCore-XS1-for-Embedded-Developers(1.0).pdf

Sorry I said "in dac", but meant in dsp. But fedde, is FA dsp delivering 16/24 bits like original source, to the dac? And how about SNR with very low signal level then?

I am by the way sending my FA123 analog RCA signal most of the time, what bitrate is it's ADC feeding the DSP then? Anyway - no problems with noise or distortion.... Same scenario with my Minidsp 2x4HD and 4x10HD by the way.
 
As far as I know, that is not exactly true. The noise floor can go lower if you are not outputting full scale signal, which can give you better SNR and SINAD.

As I understand it depends how DAC is built, in some cases SINAD line is nearly direct line from 0 dB down.
But I did not know how it is in Hypex DAC.

You got THD -26 dB at 20 Hz, that comes from the speaker, not from your amp and certainly not from your DAC. If you can loop back from DAC to computer, you can measure it.
Sure it is coming from speakers not from amp or DAC. This is maximum distortion over all range and was bring as illustration for Juhazi that it is challenge to built bass with OB.
 
Last edited:
Subjective opinion coming.

I experimented a little with a series resistor vs fixed L-pad tweeter attenuation for a CD+horn to reduce a slight but audible hiss and it seems the series resistor is better to my ears.
Frequency response was compensated for the same target. With the L-pad in place, there was more WTF?? moments when listened to music, harsh transients popped out sometimes.
The series resistor just sounded right with no problematic tendency. I eveni inserted a series capacitor which acted as a part of the total filtering (like with the JBL M2), not just for protection function.
I liked the result. So now I’m going to use it this way for a while, I’ll see it in the long run.
The resistor is a Jantzen Superes, the cap is a Jantzen Superior-Z. The subjective sound character of this cap seems good for a (on its own) slightly dull sounding tweeter.
 
Subjective opinion coming.

I experimented a little with a series resistor vs fixed L-pad tweeter attenuation for a CD+horn to reduce a slight but audible hiss

I have this problem as well. I use the 111dB sensitive Radian 475PB-8 CD, and when I hooked up a new GAN amp, the background noise level was pretty high, too distracting for quiet passages.

System uses active crossovers, so the amp seems to have too high of a gain, since my volume slider is already loud at like 5 out of 100.

Which resistor and rating did you use?
 
I used a 4.7 Ohm in series with 2.7 Ohm in parallel for th L-pad (11dB attenuation). The series-only resistor is 6.8 Ohm for roughly 6dB attenuation. My amps are FA253 and these are very quiet, even without passive tweeter attenuation and a 109dB CD, the hiss is not audiible when music plays, so it's not distracting at all. I experimented out of curiosity.

Here is a good attenutor calculator, I used the minimun impedance value of the CD:
L pad calculator - attenuation dB damping impedance decibel loudspeaker speaker voltage divider - sengpielaudio Sengpiel Berlin

Attention, the series-only resistor changes the frequency response more than the L-pad, especially around the driver resonsnce frequency, so it probably requires more response shaping.
 
If I want to honest, I hate to say such a subjective attributes, but maybe the low volume listening are improved with the series resistor plus capacitor.

But as I said, I didn't had real problems with the noise when music was played, even with direct connection between the amp and the CD and I often listening around -50dBFS.
 
Since your tweeter is extremely efficient I'd suggest going further and try with only a large series resistor, say 30-40 ohm. It will attenuate the tweeter and also convert the amp into more of a current source amplifier than a voltage source. This in turn will probably reduce distortion by a fair bit as long as you can handle the bump at the impedance peak.

How much distortion is cancelled depends on the driver, in practice the current source mode cancels impedance variations since the output is now more dependant on the current rather than the voltage across the tweeter.

I do wish Hypex would sell commercial current source amplifiers though. The ideal case would be a toggle switch for the plate amps in the software. I'm periodically trying different mods to convert one of their amps into a current source but so far I haven't been successful. I do hope the next try is successful =)

So far I've only tried a passive loop around the amps, it works but there isn't enough gain to spare while also increasing output impedance. I can get it to ~ 10 ohm output impedance with reasonable gain but I want 100-200. Next try is to add an extra gain stage to the loop and see what happens.
 
Do not be too shy regarding subjective opinions, we are not on Audio Science Review here... :D

I would not go much above 8 ohm with series resistor, it would be better to go for gain mods then to save some power... Though you could try out what it brings. I think it makes sense to tune gains and attenuation so that the low, mid and high get comparable levels (for same or similar digital channel gain).