Interesting way of describing embedding 16 bit data in 24 bits. It's kind of a semantic point, but agree volumes between 0 and -6dBFS will hit all the bits. At -6, -12, -18, ... one zero moves from padding the least significant bits to padding the most significant bits.
Depends on SPL. At typical levels and power amp gains my experience is the DAC/preamp output and power amp input stage(s) are often an order of magnitude or more nonlinear than the drivers. That's largely mitigated by reducing the power amp gain from the traditional 25-30dB to something around unity. It's not hard to get 100dB out of a driver at unity gain, so there's little sacrifice here. Particularly for biamp/triamp/quadamp.
I've found reducing the gain is particuarly advantageous in digital crossover, DAC to power amp topologies operating at 16 bit. Padding 16 bit source material to 24 bit allows reach down to any DAC's IMD floor. However, basic audio DACs like the ones in the ADAU1701 MiniDSP uses are around 15 bit while a few really good ones approach 20 bit given quite a bit of attention to power supplies, board layout, and output buffers. Most of the top end audio parts are around 17.5 bit. Therefore, 24 bit doesn't usually offer more than another 6dB or so on the volume knob with respect to the IMD floor. So if you leave gain allocated to the power amp rather than the DAC the system's still likely to limit on the DAC up to oh, somewhere around 100dB SPL.
I suspect it's more Hypex's flat THD as a function of frequency (that implies greater full power bandwidth and/or Avol, which in my experience seems to correlate with a cleaner acoustic impulse even when amps measure pretty much identically into a dummy load). Did you ever get a chance to measure the impulse responses directly and compare?
Last edited:
Interesting point that last one... never occured to me that the impulse response might be different, but when you mention it I seem to remember that it looks better with Hypex. Will try to measure with both Hypex and ICEpower amps.
I have a DAD AX24 8-ch DAC. S/N is supposedly 119 dB, which is close to 20 bits. Its as good as it gets.... But at normal listening levels (100 db peak in listening pos) DAC output peaks at around -20 dB, so it safe to say that the real dynamic range is about 100 dB, which is far more than my listening room can provide anyway,
I have a DAD AX24 8-ch DAC. S/N is supposedly 119 dB, which is close to 20 bits. Its as good as it gets.... But at normal listening levels (100 db peak in listening pos) DAC output peaks at around -20 dB, so it safe to say that the real dynamic range is about 100 dB, which is far more than my listening room can provide anyway,
Yeah, I was hearing a difference between an LME49710 and an LME49990 and it took a couple hours to find a measurement which showed a difference between them. 
In my case the IR produced from the IFFT of swept sine acoustic data was identical---not surprising since both parts are quite linear and the IFFT is a linear transform. I had set up an RTA and measure the IR directly, which gets interesting as the dummy load response and acoustic response differ signficantly and the mic's IR isn't calibrated. I also found measurable differences on acoustic square waves too but find it's harder to reason about them than the differences in the impulse.
The noise floor for DnR's usually 10 or 15dB below the THD or two tone IMD floor. Multitone IMD levels tend to be somewhat higher than two tone levels so, for a constant PSD signal, it's reasonable to derate the DAC 2 to 3 bits from the resolution indicated by 0dBFS DnR. Music PSDs are more like 1/f^0.75 so the derating's actually proportional to frequency and a function of the bandwidth the DAC is handling. In biamp or multiamp with typical tweeter crosses around 2kHz it doesn't matter as much since the ear's sensitivity rolls off above 5kHz.
In my case the IR produced from the IFFT of swept sine acoustic data was identical---not surprising since both parts are quite linear and the IFFT is a linear transform. I had set up an RTA and measure the IR directly, which gets interesting as the dummy load response and acoustic response differ signficantly and the mic's IR isn't calibrated. I also found measurable differences on acoustic square waves too but find it's harder to reason about them than the differences in the impulse.The noise floor for DnR's usually 10 or 15dB below the THD or two tone IMD floor. Multitone IMD levels tend to be somewhat higher than two tone levels so, for a constant PSD signal, it's reasonable to derate the DAC 2 to 3 bits from the resolution indicated by 0dBFS DnR. Music PSDs are more like 1/f^0.75 so the derating's actually proportional to frequency and a function of the bandwidth the DAC is handling. In biamp or multiamp with typical tweeter crosses around 2kHz it doesn't matter as much since the ear's sensitivity rolls off above 5kHz.
Question about Seven
hi Stig,
Im new here and i build a system with a computer setting very similar at yours.
For now i was running on XP but after a failure of the computer's motherboard i think i will not have any solution than put Seven OS on the computer ...
So did you alredy tried to run your config with PLParEQ and Console on a computer with Seven OS ?
And did problems appears ?
Jean Michel
hi Stig,
Im new here and i build a system with a computer setting very similar at yours.
For now i was running on XP but after a failure of the computer's motherboard i think i will not have any solution than put Seven OS on the computer ...
So did you alredy tried to run your config with PLParEQ and Console on a computer with Seven OS ?
And did problems appears ?
Jean Michel
StigErik what a system you have built! Thank you for sharing the process in such detail. To all the contributers, Gentlemen, thank you for the best audio thread I've ever read.
As I understand it, you are running a 5 way stereo system (10 analog out channels) yet the Fireface 800 and the DAD AX24 are 8 channel. How does the work, are you using 8 on the AX24 and 2 on the RME?
I've played with and listened to the PLParEQ software, very good sound very powerful. The fact that PLParEQ is doing both x-over and eq did not register at first read. That sunk in as I fell asleep and was w i d e awake after that. I need this. Great head up thanks yet again.
As I understand it, you are running a 5 way stereo system (10 analog out channels) yet the Fireface 800 and the DAD AX24 are 8 channel. How does the work, are you using 8 on the AX24 and 2 on the RME?
I've played with and listened to the PLParEQ software, very good sound very powerful. The fact that PLParEQ is doing both x-over and eq did not register at first read. That sunk in as I fell asleep and was w i d e awake after that. I need this.
Great head up thanks yet again.Hello StigErik,
It seems you already compared different speakers like the SA8535, the Beyma TPL-150 and other ribbon tweeters.
Do you know the AMT from Precide.ch ? And if it's the case do you have listen this last in comparison to the above mentioned league ?
The major issue is that most of these drivers need to be cut somewhere in between the 1K-3K bandwidth and this seems not to be a good idea... Some PMD would be preferred here, provided it'll go down to 200 Hz, don't you think so ?
But which PMD could do the job ?
Any comment would be appreciated...
It seems you already compared different speakers like the SA8535, the Beyma TPL-150 and other ribbon tweeters.
Do you know the AMT from Precide.ch ? And if it's the case do you have listen this last in comparison to the above mentioned league ?
The major issue is that most of these drivers need to be cut somewhere in between the 1K-3K bandwidth and this seems not to be a good idea... Some PMD would be preferred here, provided it'll go down to 200 Hz, don't you think so ?
But which PMD could do the job ?
Any comment would be appreciated...
I have and it's excellent. Point is I couldnt do any comparison and they are normally NOT sold at spare parts, but only as replacement for their speakers.
Precide
They own the intellectual property of Oskar Heil and use a proprietary membrane.
There are two models: XO point for the big one is 800hz, around 1,5khz (IIRC) for the small one.
They are mid-tweeters, extended to about 25khz.
Thank you for the input.
The Kithara is also an AMT you should find more info here :
Precide
http://www.precide.ch/eng/eheil/eheil.htm
http://www.precide.ch/eng/eheil/eheildetails.htm
It's a priced device, much expensive than the other cited.
Not in to horn tweeters Erik.
Horn loading gives very low distortion and linear response and there for clear sound. You have to pic out the good drivers and horns.
My current BR-speaker with horn high. Thd the 120dB is faulty it was about 95-100dB when measured.
On the thd figure you can see where the hornloading begins.
Attachments
Last edited:
They sell it as spare unit but it's very very expensive like 1045 CH Francs, more or less 1130 $ each...
Upto 30kHz. Now I use one 1" that is flat up to 18khz that is even better. I measured -55dB second and -80 third harmonic. At about 95-100dB can't say exactly doesn't matter its clear it is very good performing.
An externally hosted image should be here but it was not working when we last tested it.
Last edited:
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.