(if you are using the classic 'thermionic valve stage'-stage) the I/V resistor refers to gnd, so not using the wisdom that current should flow to +5V...
Studiostevus you mean this?:
http://www.fortunecity.com/rivendell/xentar/1179/theory/vasfda/vasfda.html
that's what i meant... and IF that's what T uses, I am a bit surprised...
- I thought that 'the community' sort of agreed that a discrete stage is more transparent and neutral.
- Besides, my newly acquired knowledge tells me that currents should flow back into +5V
As AMR's designer and tda1541-guru, I would have thought that he would seek the optimum configuration for getting the best out of the dac
- I thought that 'the community' sort of agreed that a discrete stage is more transparent and neutral.
- Besides, my newly acquired knowledge tells me that currents should flow back into +5V
As AMR's designer and tda1541-guru, I would have thought that he would seek the optimum configuration for getting the best out of the dac
Last edited:
Hi,
Correct. Except I use this wisdom, but in a different way...
For the rest, I myself and many others like the way my solution sounds.
"The best" is often a subjective thing...
Ciao T
Correct. Except I use this wisdom, but in a different way...
For the rest, I myself and many others like the way my solution sounds.
"The best" is often a subjective thing...
Ciao T
Thanks QSS_T for the link to the "fortunecity" article of TLs.
One thing that puzzles me is the obvious good results obtained by simple applying the I/V resistor directly to the dac chip's o/p pins despite exceeding the 25mV compliance by a good margin (400mV).
Are the Phillips specs really that incorrect or am I missing something obvious?
Why is that
One thing that puzzles me is the obvious good results obtained by simple applying the I/V resistor directly to the dac chip's o/p pins despite exceeding the 25mV compliance by a good margin (400mV).
Are the Phillips specs really that incorrect or am I missing something obvious?
Why is that
The specs are not necessarily incorrect - they say in the datasheet 'performance degradation' - which presumably means the full level THD gets measurably worse. Since that's better than 0.002% there's some margin available. That reduction can happen without anyone noticing a change in sound quality. IME the sound quality of a DAC has more to do with spurious signals generated at normal average listening levels which would be at least 20dB lower than full scale. At such the compliance spec is only marginally being exceeded.
'The best' is most definitely a subjective thing as far as DACs are concerned.
'The best' is most definitely a subjective thing as far as DACs are concerned.
Hi,
Philips are conservative.
I would say that minimising both offset and I/V resistance is important. At AMR we have run extensive tests on the best compromise between low resistance (low distortion from exceeding the DAC's specified voltage compliance) and high resistance (minimises the needed amplification after th resistor and minimises Noise).
I can obviously not post any commercial information from AMR here, but AMR uses a 12AT7/ECC81 equivalent tube in a Gomez stage (this info is in the public domain) and outputs around 2V RMS... You guys can calculate back from that.
The Universal Tube Stage I designed for diyhifisupply uses 50 Ohm I/V with a TDA1541A and a 6922 Tube loaded with CCS, which gives around 30dB (30 Times) gain. The results are good.
Ciao T
Philips are conservative.
I would say that minimising both offset and I/V resistance is important. At AMR we have run extensive tests on the best compromise between low resistance (low distortion from exceeding the DAC's specified voltage compliance) and high resistance (minimises the needed amplification after th resistor and minimises Noise).
I can obviously not post any commercial information from AMR here, but AMR uses a 12AT7/ECC81 equivalent tube in a Gomez stage (this info is in the public domain) and outputs around 2V RMS... You guys can calculate back from that.
The Universal Tube Stage I designed for diyhifisupply uses 50 Ohm I/V with a TDA1541A and a 6922 Tube loaded with CCS, which gives around 30dB (30 Times) gain. The results are good.
Ciao T
Ah, forgot that the recorded levels aren't at 0dB, but well below it and so the single bit dacs of the max signal aren't in use all the time, or just for higher level transients in "normal" quality recordings.
So, for "normal" levels, we should be concentrating on the "10 bit dac" of the pins 13 & 18, yes?
So much for the benefits of 24 bit systems if most of the musical content comes from the least significant 10 bits of the the original 16bit/44k1 system and works so well! (What would a 24 bit version of the 1541A chip sound like, I wonder?)
Thanks for the info.
So, for "normal" levels, we should be concentrating on the "10 bit dac" of the pins 13 & 18, yes?
So much for the benefits of 24 bit systems if most of the musical content comes from the least significant 10 bits of the the original 16bit/44k1 system and works so well! (What would a 24 bit version of the 1541A chip sound like, I wonder?)
Thanks for the info.
No, that's a mistake - the MSB of the DAC is in use for (roughly, see below) half the time as its the sign bit.
Dunno where you got that from 😀 No, in a word. If you use the TDA1541A with the SAA7220 then that adds an offset of 0x0020 to its output. Thus the lowest 5 bits only are being toggled with really quiet signals. If you're using it in NOS then you might want to ask yourself the question 'why did Philips do this?'.
I'd expect a 24bit system to have a measurably lower noise floor than a 16bit one.
Thanks,
You can see that I don't understand how this thing actually works - if you have the patience ....
What is this "Sign bit" and what does it do? Does this set the o/p current to the 2mA level, and the signal swings another 2mA around that?
... and just to check, the 'lowest signal' bits are handled via the pins 13 & 18, yes?
I am using NOS and maybe will try an unused Aikido gain stage as a valve o/p stage with a simple resistor I/V stage (and the other R,C, L components) instead of a D1 fet o/p stage.
Thanks for the help abraxalito, Thorsten, - very much to the point - extremely useful.
You can see that I don't understand how this thing actually works - if you have the patience ....
What is this "Sign bit" and what does it do? Does this set the o/p current to the 2mA level, and the signal swings another 2mA around that?
... and just to check, the 'lowest signal' bits are handled via the pins 13 & 18, yes?
I am using NOS and maybe will try an unused Aikido gain stage as a valve o/p stage with a simple resistor I/V stage (and the other R,C, L components) instead of a D1 fet o/p stage.
Thanks for the help abraxalito, Thorsten, - very much to the point - extremely useful.
If you have the curiosity, then I'll bring along the patience to match it 😀
Audio signals are bipolar - they are both negative and positive (not at the same time of course). So when the DAC outputs a negative signal, the MSB is 0, and when it outputs positively, MSB = 1. Put another way, the bottom 32767 codes from the DAC correspond to something negative, and the top 32767 codes correspond to something positive.
Since for the 1541 the full scale current is 4mA, this is correct. The sign bit (or MSB) equates to 2mA of current.
Well 2mA around 2mA would mean 0 to 4mA so that sounds about right yep 🙂
Dunno, I'm not really an expert in the particular internal workings of the 1541, but I had always thought the caps were there for the coarse part (the top 6 bits) of the DAC, not the bottom part (which is hard wired in my understanding). Its the coarse part of the DAC that uses DEM and that needs caps (I think).
I'm a recent convert to NOS myself but have yet to graduate to the 1541. I'm still playing around in the small puddles of TDA1543 and TDA1545 and very much liking what I'm hearing 😀
Hi,
Gomes, si, I am doing Senor Gomes (from Portugal) a disservice otherwise...
The circuit bearing his name first appeared in l'Audiophile as input/driver stage for a 211 Amplifier and before they joined the Dark Side and started making Switched Mode Amp's Bel Canto had a rather good 845 Amp using 2 * 12AX7/ECC83 in such a circuit.
It is also known as "supertotempole".
I used a standard Gomes with 2C51 in my modifications for the Shanling CD T-100.
And I may have pioneered the "asymmetric Gomes" (which uses different tubes for Gain and Follower and needs some tricks) first with a lineamp that saw a short as a Diyhifisupply Kit... I also use it in the CD-77, with 12AT7 & 5687...
As usual John Broskie has what Charles Mordechai once described as "theory without practice is sterile". His findings are accurate, however if you can actually make a quiet PSU the Gomes stage in practice (using real tubes) measures better and sounds better...
Ciao T
Gomes, si, I am doing Senor Gomes (from Portugal) a disservice otherwise...
The circuit bearing his name first appeared in l'Audiophile as input/driver stage for a 211 Amplifier and before they joined the Dark Side and started making Switched Mode Amp's Bel Canto had a rather good 845 Amp using 2 * 12AX7/ECC83 in such a circuit.
It is also known as "supertotempole".
I used a standard Gomes with 2C51 in my modifications for the Shanling CD T-100.
And I may have pioneered the "asymmetric Gomes" (which uses different tubes for Gain and Follower and needs some tricks) first with a lineamp that saw a short as a Diyhifisupply Kit... I also use it in the CD-77, with 12AT7 & 5687...
As usual John Broskie has what Charles Mordechai once described as "theory without practice is sterile". His findings are accurate, however if you can actually make a quiet PSU the Gomes stage in practice (using real tubes) measures better and sounds better...
Ciao T
Hi,
I started with TDA1543 shortly after Ryohei Kusonoki published his design, probably in early 98. It was just for a lark, to see if this could work at all (I was extremely suspicious).
When it beat the stuffing out of my CS8412 & CS4329 DIY DAC (which I thought sounded worse than my modded Marantz CD-67) and out of my modded Marantz CD-67 I started experimenting more...
And ended up in loads of arguments on-line with the Technocrati who insisted "it cannot work", "it will blow up your tweeters", "It will cause the Sun to Nova" and similar stuff that rhymes with the name of european bottom feeding garbage disposal Fish...
Do try the TDA1541, use DEM reclocking, I2S attenuators and an open loop output stage (does not absolutely have to be toobs, but Mi-I likes Toobs...
Ciao T
I started with TDA1543 shortly after Ryohei Kusonoki published his design, probably in early 98. It was just for a lark, to see if this could work at all (I was extremely suspicious).
When it beat the stuffing out of my CS8412 & CS4329 DIY DAC (which I thought sounded worse than my modded Marantz CD-67) and out of my modded Marantz CD-67 I started experimenting more...
And ended up in loads of arguments on-line with the Technocrati who insisted "it cannot work", "it will blow up your tweeters", "It will cause the Sun to Nova" and similar stuff that rhymes with the name of european bottom feeding garbage disposal Fish...
Do try the TDA1541, use DEM reclocking, I2S attenuators and an open loop output stage (does not absolutely have to be toobs, but Mi-I likes Toobs...
Ciao T
- Status
- Not open for further replies.