I'd be doing a disservice to the 192K fanbois if they had to down convert from 192 to 176K. Having said that perhaps in a future design a switchable oscillator setting maybe available as an option.
cheers
I totally agree, I wouldn't suggest to configure all boards for 44.1kHz. But if it works it's an interesting option for a diy mod...
A switchable oscillator setting would be fantastic!
Impressive project. I'm building a 3-way open baffle active crossover system based on Hypex DLCP. Do some of you experienced designers / builders have any idea how I could reduce the risk of cone over-excursion /damage on my subwoofer driver due to a peak at the low end (20 Hz). If you want to have any low frequency response in open baffle you need to put a bit more gain in the DSP for the low frequency band but that increases the risk of over-excursion when a peak occurs. Is there a good way to include a limiter / compressor to reduce dynamic range for some frequency band like 20-40 Hz? Can that be done inside the hypex DLCP system (I didnt study the filter possibilities ind etail yet). Thanks in advance for any tips.
The best thing you could do might be to use Dipole Woofers from AE Speakers. They are designed for open baffle.
The natural solution would be to use a limiter, which is feasible in Audioweaver.
With Dlcp you can't do this and the only option is to use a subsonic filter.
Hi everyone
We are busy moving to a larger premises to work from as space is at a premium where we are at the moment so some delays are imminent. Hope to get current orders out the door ASAP so sorry for any inconvenience. Also there is a new version of Audioweaver on the horizon so I am kind of hoping to get hold of that and give it a thorough testing to make sure there are no issues.
Also many people have expressed interest in a completed fully working unit so I am going to see someone next week regarding getting a purpose built case that hopefully looks the part and where you don't have to spend hours drilling and filing holes - because there is a lot of holes in the back panel once it is fully optioned up
Regards
David
We are busy moving to a larger premises to work from as space is at a premium where we are at the moment so some delays are imminent. Hope to get current orders out the door ASAP so sorry for any inconvenience. Also there is a new version of Audioweaver on the horizon so I am kind of hoping to get hold of that and give it a thorough testing to make sure there are no issues.
Also many people have expressed interest in a completed fully working unit so I am going to see someone next week regarding getting a purpose built case that hopefully looks the part and where you don't have to spend hours drilling and filing holes - because there is a lot of holes in the back panel once it is fully optioned up
Regards
David
Audio weaver is great for this, you can set a limiter that increases it limiting based on input level, sonic you work out the maximum signal level which when amplified will cause maximum excursion of the cone you can then set a limiter to work as the level approaches that amplitude. Clever.
I have often wondered why we don't have a "shield" style enclosure for DIY amps - similar to the old ATX computer case design, with a swappable "shield" panel. For those cases unique motherboard connector patterns would just ship with a matching shield.
If you aren't familiar with this concept, basically you have a standard case or range of cases, and an insert (or back panel, front panel, etc) that can be pre-made to a variety of hole patterns. Then multiple amp designs could fit a variety of enclosures, with nothing more than a template shield available in whatever pattern is desired.
If you aren't familiar with this concept, basically you have a standard case or range of cases, and an insert (or back panel, front panel, etc) that can be pre-made to a variety of hole patterns. Then multiple amp designs could fit a variety of enclosures, with nothing more than a template shield available in whatever pattern is desired.
.jpg)
One room correction tool I have used is REW. It has a reasonably good room measurement function, although not an impulse. It then can generate biquad coefficients for parametric equalizing filters.
It works OK with a MiniDSP using the digital output from the MiniDSP to a better DAC. However, if you are using a SHARC the you would want to use FIR filters to also provide phase correction, and for that, I would. Think you would want to calibrate in the time domain using an impulse.
It works OK with a MiniDSP using the digital output from the MiniDSP to a better DAC. However, if you are using a SHARC the you would want to use FIR filters to also provide phase correction, and for that, I would. Think you would want to calibrate in the time domain using an impulse.
I looked at the ESS site..not much info there.
I have been pondering the volume control issue for a while.
If you figure you have a noise floor around -120 dB below the full range signal level, then attenuating in the digital domain will effectively reduce the bit resolution. The noise floor doesn't change, with reduced volume, so even if you have more that 120 dB of digital resolution, you will only realize 120dB of dynamic range. As you reduce the digital signal level, you are also reducing the dynamic range, since the full range signal will now max out at something less than the MSB of the digital words.
What I am not sure about is if this is really a drawback or not. If you run the DAC at full range (so the peak signal leve is all ones in the digital word corresponding to a sample at the peak), and then you are getting the full dynamic range of the system. If you then attenuate the analog signal, then the peak analog level goes down, but, as in the digital domain, the noise floor doesn't change. It seems that the only thing a post DAC attenuators might avoid is the effective lower sample resolution that occurs at higher attenuation levels.
At high digital attenuation levels, the overall resolution is lower. For example for a 32 bit system, full range would be xFFFF FFFF. Every 3 dB of attenuation drops a bit off the word, so, 12 dB attenuation would take that same peak signal to xFFFF FFF0, and we have lost 4 bits of resolution. If you figure you have a 250 watt system,and you plan to listen at maybe 1 watt, you will need about 48 dB of attenuation,so with a digital attenuators you will be effectively dropping 16 bits off the LSB end of the word ( which leaves you with only 16 bits of resolution (which is the point that Tranquility Bass was making about CD Redbook quality). What that means is that you either need 32+ bits of resolution, or you need to go to an analog (post DAC) attenuator.
Since in both the digital and analog cases the noise floor will be unchanged, it seems to me the issue is not DR, it is bit resolution and the resulting quantization that will occur for lower level signals...effectively the lowest level signal will no longer be represented by 16 bits, but will be represented by only one bit, the rest having been lopped off by the attenuator.
Thoughts on this?
I have been pondering the volume control issue for a while.
If you figure you have a noise floor around -120 dB below the full range signal level, then attenuating in the digital domain will effectively reduce the bit resolution. The noise floor doesn't change, with reduced volume, so even if you have more that 120 dB of digital resolution, you will only realize 120dB of dynamic range. As you reduce the digital signal level, you are also reducing the dynamic range, since the full range signal will now max out at something less than the MSB of the digital words.
What I am not sure about is if this is really a drawback or not. If you run the DAC at full range (so the peak signal leve is all ones in the digital word corresponding to a sample at the peak), and then you are getting the full dynamic range of the system. If you then attenuate the analog signal, then the peak analog level goes down, but, as in the digital domain, the noise floor doesn't change. It seems that the only thing a post DAC attenuators might avoid is the effective lower sample resolution that occurs at higher attenuation levels.
At high digital attenuation levels, the overall resolution is lower. For example for a 32 bit system, full range would be xFFFF FFFF. Every 3 dB of attenuation drops a bit off the word, so, 12 dB attenuation would take that same peak signal to xFFFF FFF0, and we have lost 4 bits of resolution. If you figure you have a 250 watt system,and you plan to listen at maybe 1 watt, you will need about 48 dB of attenuation,so with a digital attenuators you will be effectively dropping 16 bits off the LSB end of the word ( which leaves you with only 16 bits of resolution (which is the point that Tranquility Bass was making about CD Redbook quality). What that means is that you either need 32+ bits of resolution, or you need to go to an analog (post DAC) attenuator.
Since in both the digital and analog cases the noise floor will be unchanged, it seems to me the issue is not DR, it is bit resolution and the resulting quantization that will occur for lower level signals...effectively the lowest level signal will no longer be represented by 16 bits, but will be represented by only one bit, the rest having been lopped off by the attenuator.
Thoughts on this?
Last edited:
There is currently no I2S outputs on the board. You can route whatever you want to any channel you want.
Currently each board has to be configured independently but the volume control and source selection is always synced together on multiple boards.
May look at cascading operation in the future.
regards
david
Excellent explanation. And a very cool approach!!
The key here (that I had not really considered) is that the DAC noise floor depends on both the analog components and the quantization (the numerical errors described in the presentation). For a 16 bit recording the DAC noise will be fixed at about -96 dB (20log(2^16)). And that doesn't change if the volume control remains in 16 bits.
As long as the DAC is really doing D/A in 32 bits then this is a great solution!
I was using Cirrus 4398 24 bit DACS, but I think I'll switch my multi source system to use the ESS device.. (My SW guy is going to kill me!!)
Scott
Last edited: