If only they didn't have to rush it to market. Remember that Philips brought us the compact cassette. It was improved by others.
We always expect too much. I'll bet they only thought of 20 Hz to 15 KHz. 😉
-Chris
How do they brickwall in studios??? 22nd order Bessels or what?
Or do they sample at higher frequecies with REASONABLE filters and then downsample mathematically?
😡
Or do they sample at higher frequecies with REASONABLE filters and then downsample mathematically?
😡
I was just reading an interview with Bob Stuart (Sp?) of Meridian, and he suggested that if they had choosen 55 kHz/20 bit we'd all be happy today...
Me i'd still like to see closer ro 400 kHz...
dave
Me i'd still like to see closer ro 400 kHz...
dave
Yeah I could go with that... specs that exceed the ears... and could be realized with simple (read low noise) filters.
Hell... it wouldn't have mattered if the ADC & DAC weren't ready yet. They could have left the placeholders in the format.
😡
Hell... it wouldn't have mattered if the ADC & DAC weren't ready yet. They could have left the placeholders in the format.
😡
Hi poobah,
True, but then it wouldn't have been cheap to produce. It came down to getting an album on a disc.
-Chris
True, but then it wouldn't have been cheap to produce. It came down to getting an album on a disc.
-Chris
OK...
But, what is the norm for filtering in digital studio recording gear?
I once had a Neumann 87 & 89. I know beyond 20 K is easy. Just curious, because I think the CD standard "painted us into a corner" on both ends.
😕
But, what is the norm for filtering in digital studio recording gear?
I once had a Neumann 87 & 89. I know beyond 20 K is easy. Just curious, because I think the CD standard "painted us into a corner" on both ends.
😕
Hi poobah,
I can't remember, but the sampling frequency is 48 KHz. I think you may find as much varience in recording / playback filters as there are brands of said gear.
Yeah. They burned us big time and consumers do not want to replace everything yet again! They are tired of being scr--ed and I can't blame them.
I believe the race for easy money did this to us.
-Chris
I can't remember, but the sampling frequency is 48 KHz. I think you may find as much varience in recording / playback filters as there are brands of said gear.
Yes, and albums cleared that easily too. An open reel machine can also get way up there. 22 KHz for a Teac is common. I've measured -3 dB points near 27 KHz at 15 ips. on Revox commonly (Ampex 456)
Yeah. They burned us big time and consumers do not want to replace everything yet again! They are tired of being scr--ed and I can't blame them.
I believe the race for easy money did this to us.
-Chris
Well...
It wasn't easy,and I respect all that. I remember the first discs I bought in '84 or so; you could hold them up to the light and see 1 mm or larger "holes"... the Hamming Algorithm took care of all that. You could sandblast a disc and my ReVox would still play it.
It wasn't easy,and I respect all that. I remember the first discs I bought in '84 or so; you could hold them up to the light and see 1 mm or larger "holes"... the Hamming Algorithm took care of all that. You could sandblast a disc and my ReVox would still play it.
Hi poohbah,
Thanks for your reply [post#862]
I have studied the Nyquist-Shannon sampling theorem. And I have been looking at alternatives since I ran the spectrum analysis on the D-I DAC, the thing is, the overall sound quality is still the best so far with this setup, but as I said I am still open for alternatives.
Thanks for your reply [post#862]
I have studied the Nyquist-Shannon sampling theorem. And I have been looking at alternatives since I ran the spectrum analysis on the D-I DAC, the thing is, the overall sound quality is still the best so far with this setup, but as I said I am still open for alternatives.
Hi MGH,
Thanks for your reply [post#898]
>1) I think in this case the DSP should be programmed from scratch.
>2) A transformer won't output DC, it only works with alternating current. The DC coupled system not only improves bass reproduction, due to the absence of non-linear components in the signal path, overall distortion is lower.
>3) The tube amplifier is both, differential amplifier (in order to process the differential I/V outputs) and output buffer. The combination of OPA627 I/V converter and the differential tube amplifier works very well. It also keeps noise levels low, even when using the tube amplifier only.
>4) I already designed the audio interface module to provide a SPDIF interface, it can be used with coax or a differential interlink. The newly developed shiftregister-reclocker can be used to minimize BCK jitter.
Thanks for your reply [post#898]
>1) I think in this case the DSP should be programmed from scratch.
>2) A transformer won't output DC, it only works with alternating current. The DC coupled system not only improves bass reproduction, due to the absence of non-linear components in the signal path, overall distortion is lower.
>3) The tube amplifier is both, differential amplifier (in order to process the differential I/V outputs) and output buffer. The combination of OPA627 I/V converter and the differential tube amplifier works very well. It also keeps noise levels low, even when using the tube amplifier only.
>4) I already designed the audio interface module to provide a SPDIF interface, it can be used with coax or a differential interlink. The newly developed shiftregister-reclocker can be used to minimize BCK jitter.
poobah said:
Brickwall filters for audio ADC's went out with the Berlin Wall. With perhaps one exception, they're all 64x and up delta-sigma types. They oversample then decimate. A first order R/C filter is all they need.
Hi philpoole,
Thanks for your reply [post#907]
Congratulations for building your first DAC.
> You could put a 11 or 16 MHz master clock (that matches the CDP clock frequency) close to the DAC and use it to synchrinously reclock the I2S signals. Then route it back to the CDP using coax or differential interface, I already tried it with 16 MHz using a RS422 differential interface (DS8922) and it works.
> The CDP clock is removed and the routed clock from the DAC is used instead. But if for some reason the CDP doesn't get the master clock, it could have a servo lock-out (spindle revs up). A switch circuit could be used so the CDP uses a local oscillator until it receives the synchronized external clock.
> Don't divide the 16 MHz.
Thanks for your reply [post#907]
Congratulations for building your first DAC.
> You could put a 11 or 16 MHz master clock (that matches the CDP clock frequency) close to the DAC and use it to synchrinously reclock the I2S signals. Then route it back to the CDP using coax or differential interface, I already tried it with 16 MHz using a RS422 differential interface (DS8922) and it works.
> The CDP clock is removed and the routed clock from the DAC is used instead. But if for some reason the CDP doesn't get the master clock, it could have a servo lock-out (spindle revs up). A switch circuit could be used so the CDP uses a local oscillator until it receives the synchronized external clock.
> Don't divide the 16 MHz.
Filterpro TI
Hi Bernhard,
Thanks for your reply [post#908]
I have a program for calculating analog filters, it's called filtrepro from TI, like I already noted at the start of this thread. To buid filters like this, always use 0.1% resistors and trimmers to accurately adjust capacitance values in the last filter sections. Use a accurate sweep generator and a oscilloscope to view filter response.
I calculated a filter for you:
Type chebychev
MFB topology (Multiple Feed Back)
Number of poles 10
Passband ripple 0.01%
Cutoff frequency 22KHz
Resistor seed 3K3, E96
Capacitors E6
I used a 10 pole filter since you need 5 op-amps anyway.
I added a printout of this example, with the program Filterpro from Texas Instruments you can experiment with different filter types and parameters. The Filterpro program cost $0
Hi Bernhard,
Thanks for your reply [post#908]
I have a program for calculating analog filters, it's called filtrepro from TI, like I already noted at the start of this thread. To buid filters like this, always use 0.1% resistors and trimmers to accurately adjust capacitance values in the last filter sections. Use a accurate sweep generator and a oscilloscope to view filter response.
I calculated a filter for you:
Type chebychev
MFB topology (Multiple Feed Back)
Number of poles 10
Passband ripple 0.01%
Cutoff frequency 22KHz
Resistor seed 3K3, E96
Capacitors E6
I used a 10 pole filter since you need 5 op-amps anyway.
I added a printout of this example, with the program Filterpro from Texas Instruments you can experiment with different filter types and parameters. The Filterpro program cost $0
Attachments
Hi John,
Thanks for the reply. I am very pleased with the DAC. It sound great. So much more detail than I am used to.
Last night, I built a reclocking circuit based on a 74xx174, but I was having trouble with my Kwak clock module. I'm hoping to rebuild it anyway (I'd prefer to have a simpler set of regulators for experimenting with).
Disturbingly though, I think I noticed with my Grados, some aliasing. If you listen to fairly simple music with maybe a single synth or something at a low signal level, and amplify it greatly, I reckon you can hear a high pitched 'noise', that changes tone in symapthy to the music.
I've seen aliased frequencies on a scope at university, but never heard it before.
I suppose it could be akin to the noise on vinyl. If you do notice it, you could get used to it.
So, an unfiltered DAC is typically fine for me (it sounds so good compared to my modded Marantz CD63!), but I wonder if the aliasing will be a problem (or I'll just know its there).
For now I'm more interested in improving what I have without a filter though. I2S reclocking, and maybe DEM reclocking.
Cheers,
Phil
Thanks for the reply. I am very pleased with the DAC. It sound great. So much more detail than I am used to.
Last night, I built a reclocking circuit based on a 74xx174, but I was having trouble with my Kwak clock module. I'm hoping to rebuild it anyway (I'd prefer to have a simpler set of regulators for experimenting with).
Disturbingly though, I think I noticed with my Grados, some aliasing. If you listen to fairly simple music with maybe a single synth or something at a low signal level, and amplify it greatly, I reckon you can hear a high pitched 'noise', that changes tone in symapthy to the music.
I've seen aliased frequencies on a scope at university, but never heard it before.
I suppose it could be akin to the noise on vinyl. If you do notice it, you could get used to it.
So, an unfiltered DAC is typically fine for me (it sounds so good compared to my modded Marantz CD63!), but I wonder if the aliasing will be a problem (or I'll just know its there).
For now I'm more interested in improving what I have without a filter though. I2S reclocking, and maybe DEM reclocking.
Cheers,
Phil
Hi Phil,
Just remember that your amplifier may not like the ultrasonic hash at all. Do yourself a favour and hang a 'scope on the amplifier outputs. The higher frequencies may not make it out of the amplifier but can still cause distortion, IM effects, slewing issues - you name it.
The energy may increase as you get closer to 44 KHz. A lot depends on your amplifier and it's input filter (if any).
-Chris
Just remember that your amplifier may not like the ultrasonic hash at all. Do yourself a favour and hang a 'scope on the amplifier outputs. The higher frequencies may not make it out of the amplifier but can still cause distortion, IM effects, slewing issues - you name it.
The energy may increase as you get closer to 44 KHz. A lot depends on your amplifier and it's input filter (if any).
-Chris
Hi Chris,
Thanks for the advice, I'll have a look into this. Its a very simple headphone amp I built. Its a complimentary feedback pair with current source biasing acting basically as a current amplifier for my Grados. Its possible it could work right up into the ultrasonics.
I haven't heard it with my main amplifier driving speakers.
Its very hard to detect audibly, for me anyway. I only really spotted it once last night with a specific track.
So, those with unfiltered NOS DACs, does this ultrasonic hash and resultant aliasing bother you?
Cheers,
Phil
Thanks for the advice, I'll have a look into this. Its a very simple headphone amp I built. Its a complimentary feedback pair with current source biasing acting basically as a current amplifier for my Grados. Its possible it could work right up into the ultrasonics.
I haven't heard it with my main amplifier driving speakers.
Its very hard to detect audibly, for me anyway. I only really spotted it once last night with a specific track.
So, those with unfiltered NOS DACs, does this ultrasonic hash and resultant aliasing bother you?
Cheers,
Phil
Hi Phil,
Remember, even if your amp doesn't pass the signal, it can corrupt the audio band signals. Think of it as AM interference (yeah, I know the frequency band is different).
To each his own ....
-Chris
It depends on what happens in the amplifier. I personally think this is a very bad idea. There is a reason all these manufacturers spent so much money on a circuit some people deam as usless or undersirable. Believe me, if Philips didn't need the part, it would not be there. The same holds true of all the other manufacturers.
Remember, even if your amp doesn't pass the signal, it can corrupt the audio band signals. Think of it as AM interference (yeah, I know the frequency band is different).
To each his own ....
-Chris
NOS DAC filter / reclocker
Hi philpoole,
Thanks for your reply [post#934]
The aliasing can be attenuated, by using a 8th order Butterworth filter (I attached the schematic diagram).
Now imagine the sound you got now, without the alising disturbance. The octal D-I DAC now compares to that setup as the NOS-DAC with a single TDA1541A compares to your modded Marantz CD63. You must have heared it to understand why I choose this setup.
I have a reclock circuitry now, (also on the attached schematic), that can be used separately to remove the effects of continuous jitter from BCK without complicated setups. It's the serial shiftregister reclocker, it uses a 48MHz clock module to reclock, when put in the DAC you can throw almost any continuous jitter affected signal at it, and it would still sound natural and open. The reclocker uses 4 components, power supply bypass components excluded. The unused SN74F00 inputs should be connected to GND. It's not expensive and I think you could give it a try, and let your ears decide.
Cheers,
John
Hi philpoole,
Thanks for your reply [post#934]
The aliasing can be attenuated, by using a 8th order Butterworth filter (I attached the schematic diagram).
Now imagine the sound you got now, without the alising disturbance. The octal D-I DAC now compares to that setup as the NOS-DAC with a single TDA1541A compares to your modded Marantz CD63. You must have heared it to understand why I choose this setup.
I have a reclock circuitry now, (also on the attached schematic), that can be used separately to remove the effects of continuous jitter from BCK without complicated setups. It's the serial shiftregister reclocker, it uses a 48MHz clock module to reclock, when put in the DAC you can throw almost any continuous jitter affected signal at it, and it would still sound natural and open. The reclocker uses 4 components, power supply bypass components excluded. The unused SN74F00 inputs should be connected to GND. It's not expensive and I think you could give it a try, and let your ears decide.
Cheers,
John
Attachments
Hi John,
Thanks for the reclocking idea. It was sort of my initial plan to have a self running clock, but only in order to reduce the clock frequency - as it might be less sensitive to jitter.
I think, at the moment, the best idea for me is to do things properly with the clock in the DAC driving the CDP.
First of all, I want to rebuild my Kwak Clock (it was in my CD63, but I was never really happy with its layout, this was a few years ago).
BTW, I was looking at the TDA1541A info thread, where HtP was suggesting a low pass filter and bias adjustment on all the I2S lines so the TTL voltage levels are converted to 1.3 - 1.6 v (or something like that).
Hifi apparently managed that, but has anybody else?
I think for a quad or octal dac, it would be a nightmare.
Cheers,
Phil
Thanks for the reclocking idea. It was sort of my initial plan to have a self running clock, but only in order to reduce the clock frequency - as it might be less sensitive to jitter.
I think, at the moment, the best idea for me is to do things properly with the clock in the DAC driving the CDP.
First of all, I want to rebuild my Kwak Clock (it was in my CD63, but I was never really happy with its layout, this was a few years ago).
BTW, I was looking at the TDA1541A info thread, where HtP was suggesting a low pass filter and bias adjustment on all the I2S lines so the TTL voltage levels are converted to 1.3 - 1.6 v (or something like that).
Hifi apparently managed that, but has anybody else?
I think for a quad or octal dac, it would be a nightmare.
Cheers,
Phil
Poobah, you are talking like an average japanese CDP-builder in the past mid 80's. Philips had (a short time though) the playing cards in hand with dig filter.
Still its a matter of taste: OS or Non-OS. On this very moment i play Toto in OS (i have the same album in vinyl also) there's still a sharp edge on treble. In Non-os i have to filter " in ear" and i get some tired after a longer hearing period. Nonos i tried with 1 dac, i can imagine the D-I dac with eight 1541's reduces distorsion a lot and makes music very convincing.
The experienced roll-of of treble in nonos is normal due the frequency output of dac alone. The TDA1541 has a straight amplitude in combination with the SAA7220. See also AN-207:
http://www.analog.com/UploadedFiles/Application_Notes/383160163AN207.pdf
I go back to my soldering iron again...