Hi,
At that time (early 80's) our instrumentation to measure THD and IMD consisted of specifically designed meters for this, with a nice needle pointer as readout. With this amplifier design, neither of these meters indicated the presence of any distortion they could reliably measure.
NOW, we have instrumentation that is by far more able to show subtle flaws, however it is hubris and intellectual arrogance to presume that these instruments are proof against the same thing that happened to our 1970's designed Distortion meters, that is to be shown to be useless to measure distortion that is clearly audible and objectionably so, while the empiricists and golden-ears you consider jokes are proven right (again, as so many times before).
Ciao T
At that time (early 80's) our instrumentation to measure THD and IMD consisted of specifically designed meters for this, with a nice needle pointer as readout. With this amplifier design, neither of these meters indicated the presence of any distortion they could reliably measure.
NOW, we have instrumentation that is by far more able to show subtle flaws, however it is hubris and intellectual arrogance to presume that these instruments are proof against the same thing that happened to our 1970's designed Distortion meters, that is to be shown to be useless to measure distortion that is clearly audible and objectionably so, while the empiricists and golden-ears you consider jokes are proven right (again, as so many times before).
Ciao T
Hi,
I know that the TDA1541(A) tolerates by far greater voltage on the output without degrading measured performance as far as an AP2 can tell. I also know that based on blind level matched listening tests exceeding this voltage threshold materially causes audible degradation. However I have no interrest to place this information in the public domain.
Nevertheless, it pays to not just blindly believe datasheet but to be empirical.
Nope, this is not the case. You can quite easily measure at what level of output voltage the DAC increases it's distortion. It varies for each DAC chip and also varies for different architectures.
This will depend strongly on your resistor value, the actual gain stage (e.g. are you actually forward biasing the J-Fet's junction and by how much). The PCM1704 also contains servo circuitry that we can in fact externally access to a degree and tweak to avoid it acting on the output signal.
I know a solid state output stage for PCM1704 that sounds and measures excellent and has a 100 Ohm input impedance (again, this is not available in the public domain and I will not place it there).
I am sure you can redesign the D1 as needed. Personally, I think the Sen/Cen approach is cuter and more interesting.
Sounds like a plan. Yes, the TDA1541 output offset can be pesky.
AMEN BROTHER.
Ciao T
I know that the TDA1541(A) tolerates by far greater voltage on the output without degrading measured performance as far as an AP2 can tell. I also know that based on blind level matched listening tests exceeding this voltage threshold materially causes audible degradation. However I have no interrest to place this information in the public domain.
Nevertheless, it pays to not just blindly believe datasheet but to be empirical.
Nope, this is not the case. You can quite easily measure at what level of output voltage the DAC increases it's distortion. It varies for each DAC chip and also varies for different architectures.
This will depend strongly on your resistor value, the actual gain stage (e.g. are you actually forward biasing the J-Fet's junction and by how much). The PCM1704 also contains servo circuitry that we can in fact externally access to a degree and tweak to avoid it acting on the output signal.
I know a solid state output stage for PCM1704 that sounds and measures excellent and has a 100 Ohm input impedance (again, this is not available in the public domain and I will not place it there).
I am sure you can redesign the D1 as needed. Personally, I think the Sen/Cen approach is cuter and more interesting.
Sounds like a plan. Yes, the TDA1541 output offset can be pesky.
AMEN BROTHER.
Ciao T
Hi,
First, the fallacy of generalising from a single data point should be obvious without needing to be pointed out, however it appears to be neccesary in this case.
Second, the claim that any digital filter algorithm can somehow magically deliver 20 Bit output with a 16 Bit input that replicates what a 20 bit input should have been is so completely ludicrous, it should not need pointing out that it is in direct contradiction with any known concepts and laws of information theory, however it seems again in need to be pointed out.
Third, given that the silicone chip layout and schematic is different, that the measured performance in many areas is different and that Philips felt the need to issue a new Datasheet that demonstrated these differences all argue that it in fact is a different device and that there is no magic involved.
The title of the thread states:
"Any good TDA1541A DAC kit?" and not "Is a modern OpAmp capable of delivering as I/V for TDA1541?". Did you read this wrong?
If you want a thread that debates "Is a modern OpAmp capable of delivering as I/V for TDA1541?" perhaps you should start one.
Your comments on Op-Amp's, output slewrate and many other things in this thread are not only (as I had to laboriously point out) in direct contradiction to established fact, but they are also at best extremely loosely related to the topic of this thread, though some, like the reference to the tale of the Fox and the Grapes are clearly entirely unrelated.
You behaviour is rather RUDE but more crucially, it breaks several of the rules posited here:
http://www.diyaudio.com/forums/site-announcements/167561-diyaudio-rules.html
Ciao T
First, the fallacy of generalising from a single data point should be obvious without needing to be pointed out, however it appears to be neccesary in this case.
Second, the claim that any digital filter algorithm can somehow magically deliver 20 Bit output with a 16 Bit input that replicates what a 20 bit input should have been is so completely ludicrous, it should not need pointing out that it is in direct contradiction with any known concepts and laws of information theory, however it seems again in need to be pointed out.
Third, given that the silicone chip layout and schematic is different, that the measured performance in many areas is different and that Philips felt the need to issue a new Datasheet that demonstrated these differences all argue that it in fact is a different device and that there is no magic involved.
The title of the thread states:
"Any good TDA1541A DAC kit?" and not "Is a modern OpAmp capable of delivering as I/V for TDA1541?". Did you read this wrong?
If you want a thread that debates "Is a modern OpAmp capable of delivering as I/V for TDA1541?" perhaps you should start one.
Your comments on Op-Amp's, output slewrate and many other things in this thread are not only (as I had to laboriously point out) in direct contradiction to established fact, but they are also at best extremely loosely related to the topic of this thread, though some, like the reference to the tale of the Fox and the Grapes are clearly entirely unrelated.
You behaviour is rather RUDE but more crucially, it breaks several of the rules posited here:
http://www.diyaudio.com/forums/site-announcements/167561-diyaudio-rules.html
Ciao T
Hi,
There are other ways.
AMR's designs do not use the LC configuration (we did try it initially, but found that we had to "tune in" each choke's direction very precisely to minimise hum pickup (and no two chokes and PCB's where ever exactly the same, so we could not cure it using just PCB layout)
Ciao T
There are other ways.
AMR's designs do not use the LC configuration (we did try it initially, but found that we had to "tune in" each choke's direction very precisely to minimise hum pickup (and no two chokes and PCB's where ever exactly the same, so we could not cure it using just PCB layout)
Ciao T
OK, I will say this for the last time:
This was a reply to your affirmation that only a resistive I/V can work (without saying what buffer will be able to pick that up without signal degradation).
IMO (shared by Philips and other companies), for a TDA 1541 (even for an "A"), at 4X OS, there is no need for faster I/V converters than good OpAmpms (not even exotic ones), because the signals are not changing that fast.
Tubes are just not delivering the 16 bit resolution needed in this application.
Last: Calling names shows that you cannot have a debate. "Ignore" bliss...
Last edited:
Hi,
For the last time, I discussed and pointed out many ways in this thread on how to use op-amp's for I/V and what to look out for.
Calling that "your affirmation that only a resistive I/V can work" is bending the truth a trifle, innit?
Well, that is your opinion, but it is not established fact. If it makes you happy to listen to your TDA1541 equipped player using such Op-Amp's is your prerogative. To then complain that the result sounds bad is also your prerogative.
It is however also the prerogative of others to point out that in their opinion, by doing so, you cripple the performance the TDA1541 is capable of and that your opinion regarding the sound quality of the TDA1541 is not based on a good implementation and hence not of use.
Can you provide proof of this? Well, proof of the contrary is in March 2012 Stereophile.
A DAC I designed with passive I/V and Tube gain stages is tested by JA in Stereophile March 2012 and shows appx. -135dB FFT Noisefloor with 24-Bit Signals (I am sure you can work backwards to the ENOB from this easily).
So tubes DO deliver 16 Bit resolution just fine (and yet another of your claims has been illustrated to be incorrect).
I did not call you names.
I pointed out that your behaviour violates the rules that govern debate in this place and I called doing so rude, which it is, just as ignoring the rules for civil behaviour in a bar is rude or driving through a red light or cutting someone off.
You should perhaps review the rules and see just how many you have repeatedly violated in this thread.
Ciao T
For the last time, I discussed and pointed out many ways in this thread on how to use op-amp's for I/V and what to look out for.
Calling that "your affirmation that only a resistive I/V can work" is bending the truth a trifle, innit?
Well, that is your opinion, but it is not established fact. If it makes you happy to listen to your TDA1541 equipped player using such Op-Amp's is your prerogative. To then complain that the result sounds bad is also your prerogative.
It is however also the prerogative of others to point out that in their opinion, by doing so, you cripple the performance the TDA1541 is capable of and that your opinion regarding the sound quality of the TDA1541 is not based on a good implementation and hence not of use.
Can you provide proof of this? Well, proof of the contrary is in March 2012 Stereophile.
A DAC I designed with passive I/V and Tube gain stages is tested by JA in Stereophile March 2012 and shows appx. -135dB FFT Noisefloor with 24-Bit Signals (I am sure you can work backwards to the ENOB from this easily).
So tubes DO deliver 16 Bit resolution just fine (and yet another of your claims has been illustrated to be incorrect).
I did not call you names.
I pointed out that your behaviour violates the rules that govern debate in this place and I called doing so rude, which it is, just as ignoring the rules for civil behaviour in a bar is rude or driving through a red light or cutting someone off.
You should perhaps review the rules and see just how many you have repeatedly violated in this thread.
Ciao T
I am positive heckling is more offensive than name calling .
Be it 16 bit or 20 bit it does not matter what format it is still heckling .
I've obtained very good sound quality using the small (up to 1.6mH) 'Octa-Pac Plus' amorphous core toriodal SMD inductors from Cooper/Coiltronics. Be sure to get the 'Plus', as those are the armorphous core units. Copper also makes permalloy core versions of these inductors, but I compared them in-circuit and found the amorphous core units to be audibly more transparent. The permalloy units are simply 'Octa-Pac'. The armorphous core units also have an 'A' in their part number suffix, while the permalloy core units do not
While the inductance tolerance spec. is poor at 25%, and while I've not hand measured the units I've bought, I have measured the frequency response when used in a Zeroth-Order-Hold (aka, anti-sinc) correction EQ network with an NOS AD1865 DAC. These inductors are also rated to handle D.C. current, in fact, that's their intended application. The net response was flat to 20kHz +/- 0.25dB. I find the audible benefit of ZOH-EQ to be very worthwhile. These amorphous core inductors are available through digikey for less than $5, such as the 1.21mH unit I used and which is linked to below.
http://search.digikey.com/us/en/products/CTX300-4A-R/513-1732-1-ND/2458711
Last edited:
Thankyou, these look great, much better than the tdk smd inductors I was experiementing with years back. Added to my digikey BOM list, I don't remember anything like these 4 years ago when I was last tinkering with DAC building.
I don't know if I will implement ZOH-EQ immediately or experiement with trying to do it digitally with the computer, I think the old SA filters used to do some compensation digitally. The one thing I strive for with DAC's is absolute silent background, for they get attenuated and amplified (sometimes twice), any minute hum I can't deal with (headphone guy.)
Hi,
A DAC I designed with passive I/V and Tube gain stages is tested by JA in Stereophile March 2012 and shows appx. -135dB FFT Noisefloor with 24-Bit Signals (I am sure you can work backwards to the ENOB from this easily).
So tubes DO deliver 16 Bit resolution just fine (and yet another of your claims has been illustrated to be incorrect).
Absolutely incredible. Can you tell us the tube selected ? What was the 2H level ? Or I imagine we have to wait to the issue is in press, I guess its only January.
With a noise floor that low, it opens up a lot of possibilities. My measuring is rudimentary in comparison but when I measure my SSHV tube head amp with a decent 24bit soundcard I get the sense that a 16bit level noise floor analog dac section should be very attainable for a tube hobbiest. The toughest part is the 60hz noise, for a hobbiest like most of us dealing with the 60hz is a challenge cause really you can't get an accurate measurement of it with a sound card. Mains frequency noise can easily lead to a dominant imd character for system. But if I can't hear it or any harmonics of it playing full blast silence waves thru Grados I am happy
Also think the Passe pre-amp project/thread should be mentioned, I just looked at some of the measurements posted of it, and surely there is some learnings applicable to DAC's in that thread.
You might be interested in the simple (11 tap) FIR filter I devised for this function:
http://www.diyaudio.com/forums/digital-line-level/195791-open-source-dsp-xos.html#post2711741
Hi,
Not at all. Note that this is for FFT noisefloor, not for HD or SNR, SNR is around 110-112dBA.
I can do as much as 18dB better using rare and interesting tubes, but I have to use readily available (in quantety with good quality) tubes for commercial products.
Russian 6N1PEV. The noise BTW is not directly from the tube, but from the unbypassed emitter resistor used to linearise the circuit, though it is much less noise as I would have had using LED Biasing for example.
-60dB H2, -90dB H3, -105dB H4, ~-115dB H5 for 0dBFS
-76dB H2, -110dB H3 nothing else above -135dB for -10dBFS
The -10dBFS levels are around 20dB or more lower than the levels of H2 and H3 etc. in rather low distortion speaker drivers at around 1 Watt, so we have a point of comparison and this holds up with increasing level.
E810F or D3a can do nearly 20dB better...
My quietest Phono ever used a 1:20 Stepup transformer into an E810F, into a 600 Ohm LCR RIAA module and then into a D3a. Most of the measurable noise came from the high 600 Ohm impedance of the RIAA Module, not the tubes and even then I had major problems even measuring the noise.
Well, for this DAC JA measured, which has the power transformers on board, unshielded transformers (no metal can) and placed within a few inch of the tubes he found a few mains related spurs at 60Hz (-107dB in the noisier channel), 180Hz (-105dB in the noisier channel), 300Hz (-108dB in the noisier channel). The quieter channel had 60Hz at -112dB and the rest below -115dB.
In addition to the AP2 I use an EMU1616m based test setup. I have no problems reliably measuring below -120dFS with the EMU 1616m based setup.
I would say that numbers measured for this DAC (which are matched by the CD Player Joshua has and which is using mostly the same circuits for analog) are quite ordinary and can be attained without effort.
As there are some constrained introduced by commercial products that do not exist for a hobbyist (like it all has to be in one box, pass all electrical safety tests, conform with ROHS and WEEE and UL and FCC, use only components readily available from multiple sources and so on), I would expect anyone who puts in some effort and uses quiet tubes to get around 20dB better easily.
For example a PCM1704 DAC may use a CCS loaded triode wired D3a analogue stage and a 33 ohm I/V conversion resistor should allow around -130dB SNR re. 2V for the analog stage (the FFT noisefloor for such an analog stage would be around 24dB lower than the SNR), in fact way outperforming the PCM1704 on noise and/or exceeding the SNR of any pure multibit DAC ever made for audio (maybe also others).
Using this circuit would likely still do just fine for the latest DS DAC's with current output (scale the I/V resistors suitably of course and make it balanced) which approach (on paper anyway) 130dB SNR.
For anyone who does not want to add fuel to the cult of german Siemens and Valvo Tubes, the russian E180F analog (6Z9 IIRC) is only around 6dB noisier but has 4dB lower gain so we need to boost the signal which nearly evens out the score, but may cause issues with voltage compliance of some DAC's...
Of course, once we throw the kind of step-up and step-down transformers common in studio tube design any all tubes can be made to have excellent noise performance in the specific application...
Ciao T
Not at all. Note that this is for FFT noisefloor, not for HD or SNR, SNR is around 110-112dBA.
I can do as much as 18dB better using rare and interesting tubes, but I have to use readily available (in quantety with good quality) tubes for commercial products.
Russian 6N1PEV. The noise BTW is not directly from the tube, but from the unbypassed emitter resistor used to linearise the circuit, though it is much less noise as I would have had using LED Biasing for example.
-60dB H2, -90dB H3, -105dB H4, ~-115dB H5 for 0dBFS
-76dB H2, -110dB H3 nothing else above -135dB for -10dBFS
The -10dBFS levels are around 20dB or more lower than the levels of H2 and H3 etc. in rather low distortion speaker drivers at around 1 Watt, so we have a point of comparison and this holds up with increasing level.
E810F or D3a can do nearly 20dB better...
My quietest Phono ever used a 1:20 Stepup transformer into an E810F, into a 600 Ohm LCR RIAA module and then into a D3a. Most of the measurable noise came from the high 600 Ohm impedance of the RIAA Module, not the tubes and even then I had major problems even measuring the noise.
Well, for this DAC JA measured, which has the power transformers on board, unshielded transformers (no metal can) and placed within a few inch of the tubes he found a few mains related spurs at 60Hz (-107dB in the noisier channel), 180Hz (-105dB in the noisier channel), 300Hz (-108dB in the noisier channel). The quieter channel had 60Hz at -112dB and the rest below -115dB.
In addition to the AP2 I use an EMU1616m based test setup. I have no problems reliably measuring below -120dFS with the EMU 1616m based setup.
I would say that numbers measured for this DAC (which are matched by the CD Player Joshua has and which is using mostly the same circuits for analog) are quite ordinary and can be attained without effort.
As there are some constrained introduced by commercial products that do not exist for a hobbyist (like it all has to be in one box, pass all electrical safety tests, conform with ROHS and WEEE and UL and FCC, use only components readily available from multiple sources and so on), I would expect anyone who puts in some effort and uses quiet tubes to get around 20dB better easily.
For example a PCM1704 DAC may use a CCS loaded triode wired D3a analogue stage and a 33 ohm I/V conversion resistor should allow around -130dB SNR re. 2V for the analog stage (the FFT noisefloor for such an analog stage would be around 24dB lower than the SNR), in fact way outperforming the PCM1704 on noise and/or exceeding the SNR of any pure multibit DAC ever made for audio (maybe also others).
Using this circuit would likely still do just fine for the latest DS DAC's with current output (scale the I/V resistors suitably of course and make it balanced) which approach (on paper anyway) 130dB SNR.
For anyone who does not want to add fuel to the cult of german Siemens and Valvo Tubes, the russian E180F analog (6Z9 IIRC) is only around 6dB noisier but has 4dB lower gain so we need to boost the signal which nearly evens out the score, but may cause issues with voltage compliance of some DAC's...
Of course, once we throw the kind of step-up and step-down transformers common in studio tube design any all tubes can be made to have excellent noise performance in the specific application...
Ciao T
Well for a LSB of 24 bit bit audio we are talking about amplifying tenths of microvolts with passive IV. What does NP say "its the first watt that matters" I think that applies here and am not too concerned about 24 bit. Its great for recording and mastering but I'm not building a collosium to listen to 144 db dynamic range. Hey this is a TDA1541 threas, sorry for getting off topic.
I've experiemented with the little studio mic stepup transormers. The results were poor I think because they are designed for a mic output imedance which is much lower than a DACi-out. I was using a cinemag CMMI-10B (1:10). Sound was hollow using a pcm63k, 4.7ohm i/v, step-up, then tube gain stage, dropping the transformer and using a 47 ohm i/v resistor and sound was great.
I think I am going to stick with the D3A idea, although if I had a stepup transformer of even 1:4 it would help with the holding the voltage compliance spec as my plan is for this to feed a very low gain spud DHT headamp. But I will probably play it safe an put a line level stepup in the amp.
No current production DAC chip has 24 effective bits.
It seems to me that there is no point in having the analog output section having a noise floor far below the actual effective bits of the DAC chip itself. (Unless someone will correct me).
AFAIK, the best transformer in the audio signal path is no transformer.
There are tube-based (or hybrid) MC phono-stages without any step-up transformer.
Well I find transformer coupling better than RC coupling and dc servo for headphones, I used to be biased against transformers learned not to be close minded with audio as in the right application a quality transformer can be a good choice, for i-out DAC's I don't know.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.