Hi,
every single chip has got very different distortion.
They all sound different.
Even left / right channel.
I just put opa2604 in place of NE5532.
Not a big change compared to putting another 1541.
IMHO reclocking, video-opamps in output stages and stuff like that won't cure anything.
Also no wonder that one likes 1541, another 1543, it depends on the chip you have.
It may sound good or bad.
Even if you have a few they all may be bad.
Sorry
Bernhard
every single chip has got very different distortion.
They all sound different.
Even left / right channel.
I just put opa2604 in place of NE5532.
Not a big change compared to putting another 1541.
IMHO reclocking, video-opamps in output stages and stuff like that won't cure anything.
Also no wonder that one likes 1541, another 1543, it depends on the chip you have.
It may sound good or bad.
Even if you have a few they all may be bad.
Sorry
Bernhard
I really don't want to react to the cartload of threads you open with the truth but this is old news so I think I should react.
Pre 1989 TDA1541(A) had quality problems, later production was more consistent. 1985 non A versions are famous for bad quality. Fact not fiction. I avoid the older versions. It was at that time a real challenge to produce 16 bit chips. One of the reasons I avoid the non A versions is that they suffer a lot more from quality problems as the A versions.
This information is here and on the web, you should not believe what I say. It also depends if you use old used chips ( especially desoldered ones ). TDA1540 ( 14 bit ) is even known for wearout. After years of service they started distorting at power on. Earlier TDA1541's also failed more than normal, please read the service bulletins for that. I strongly suggest to treat them ESD safe although they're produced in bipolar technique and don't heat them up too long while soldering. Then leave them where they are. Wearing a wrist strap won't hurt you and the chips. Same goes for PCM63 that is even more sensitive. Why Bernhard do you think new unused S1's are becoming more and more expensive ?
I don't hear the extreme differences you describe with the more recent TDA1541A's I have. Only the S2 is *much* better than the others but it was produced at the end of the series ( 1997 ) at another production line. One can assume and experience that they knew at that time to produce them consistently. As a sidenote: I just yesterday read a story that I think was given here in a thread about technicians that had problems with linking SAA7220P/B with TDA1541A ( old story but very interesting ) and made some circuitry in between because in their opinion the two did not match although they were designed for eachother. The importance of the routing of the datalines was discussed. Too bad I don't remember which thread it was. These are sensitive chips, try shielding them and see the difference on your measurement rig. It's old technique of the beginning of the cd era so what does one expect ?
We know about higher distortion with non os. Higher distortion but lower influence of jitter. What's better ? And more important, the human ear is the final judge not the instruments. I heard low jitter high bit systems with technically much superior chips that sounded absolutely worse. What is it that you want to prove ? That we are all using inferior chips that are liked so much because they distort so much ? That the chips are not all the same ? Nothing is perfect, be it chips or ears. It's the experience of enjoying music, whether it has some minor distortion or not. And of course the chips differ, they're bipolar and old technique as explained. Compare it to the tube guys that do tuberolling.
Pre 1989 TDA1541(A) had quality problems, later production was more consistent. 1985 non A versions are famous for bad quality. Fact not fiction. I avoid the older versions. It was at that time a real challenge to produce 16 bit chips. One of the reasons I avoid the non A versions is that they suffer a lot more from quality problems as the A versions.
This information is here and on the web, you should not believe what I say. It also depends if you use old used chips ( especially desoldered ones ). TDA1540 ( 14 bit ) is even known for wearout. After years of service they started distorting at power on. Earlier TDA1541's also failed more than normal, please read the service bulletins for that. I strongly suggest to treat them ESD safe although they're produced in bipolar technique and don't heat them up too long while soldering. Then leave them where they are. Wearing a wrist strap won't hurt you and the chips. Same goes for PCM63 that is even more sensitive. Why Bernhard do you think new unused S1's are becoming more and more expensive ?
I don't hear the extreme differences you describe with the more recent TDA1541A's I have. Only the S2 is *much* better than the others but it was produced at the end of the series ( 1997 ) at another production line. One can assume and experience that they knew at that time to produce them consistently. As a sidenote: I just yesterday read a story that I think was given here in a thread about technicians that had problems with linking SAA7220P/B with TDA1541A ( old story but very interesting ) and made some circuitry in between because in their opinion the two did not match although they were designed for eachother. The importance of the routing of the datalines was discussed. Too bad I don't remember which thread it was. These are sensitive chips, try shielding them and see the difference on your measurement rig. It's old technique of the beginning of the cd era so what does one expect ?
We know about higher distortion with non os. Higher distortion but lower influence of jitter. What's better ? And more important, the human ear is the final judge not the instruments. I heard low jitter high bit systems with technically much superior chips that sounded absolutely worse. What is it that you want to prove ? That we are all using inferior chips that are liked so much because they distort so much ? That the chips are not all the same ? Nothing is perfect, be it chips or ears. It's the experience of enjoying music, whether it has some minor distortion or not. And of course the chips differ, they're bipolar and old technique as explained. Compare it to the tube guys that do tuberolling.
Hi ,
from the sinewave point of view , 1543 or 1541 in non os and single ended mode are both the same , they presents audible distorsions already at 15.5kHz .More than something is better in balanced mode wich I prefear.By the way is not a sordid sinewave that I am going to listen to - hopefully - music differs from it .
I/V conversion:
till ,
I have just implemented the Pass D1 at the 1541A and I am pleased with the results, even without the output buffers - it simplify the building life- the 1541A performs an round sounstage , defined very well from left to right ,high enough in a natural way , and ... above all ,the dinamics are not compressed in the bottom end with a clear and relaxed midrange .
I prefear the D1 above all I/V resistors and gainstages while it seems wasted with the 1543 .
Next step will be the balanced version for the 1541A.
Also 1543 sound more and more good in balanced mode: certainly if one pretends to use op amps and convert the balanced signal into single ended , the effort is lost .
To achieve the best from a balanced dac it is necessary to have after it a balanced circuitry .
********************
Regarding differences between channels 1543 is a champion expecially in volume , but the difference is coherent at the voltage level before the caps , and I find the cure in a trimmer .
from the sinewave point of view , 1543 or 1541 in non os and single ended mode are both the same , they presents audible distorsions already at 15.5kHz .More than something is better in balanced mode wich I prefear.By the way is not a sordid sinewave that I am going to listen to - hopefully - music differs from it .
I/V conversion:
till ,
I have just implemented the Pass D1 at the 1541A and I am pleased with the results, even without the output buffers - it simplify the building life- the 1541A performs an round sounstage , defined very well from left to right ,high enough in a natural way , and ... above all ,the dinamics are not compressed in the bottom end with a clear and relaxed midrange .
I prefear the D1 above all I/V resistors and gainstages while it seems wasted with the 1543 .
Next step will be the balanced version for the 1541A.
Also 1543 sound more and more good in balanced mode: certainly if one pretends to use op amps and convert the balanced signal into single ended , the effort is lost .
To achieve the best from a balanced dac it is necessary to have after it a balanced circuitry .
********************
Regarding differences between channels 1543 is a champion expecially in volume , but the difference is coherent at the voltage level before the caps , and I find the cure in a trimmer .
jean-paul said:
I apologize for that.
And I know, truth hurts.
jean-paul said:
Ok, I have another image.
My 1541 A S1 are all from 96 and all very different.
jean-paul said:
Some desoldered chips measure very good and also sound like that.
jean-paul said:
I should order one.
jean-paul said:
I don't know what chips you have.
Just I feel it is very important to know how the harmonics are distributed.
Philips may have stamped a chip with a crown, that has a low total THD, but that very THD may consist of very dominating 5th order harmonic solely
You will hear that.
I had just tested the new old S1 with opa2604.
Now after 1 hour I changed to NE5532 and restarted the analyzer.
Every single harmonic up to the 9th is within 0,2 dB compared to the opa2604.
All sound character belongs to the TDA chip.
I think your truth does not hurt anyone but yourself.
There are some factors to consider, did you measure/listen them all at the same time with the same setup ?
Were the S1's all new or from a used cdplayer ? If they are all new and from the same date, do they differ much in the same setup ? Of course they do differ, no need to discuss that as I tried to make clear in my post. The question is how much.
You can have another image but I changed enough TDA1541 non A versions because of defects like one channel distorting or dead. Quality is not about sound alone, it is also meant in the sense of how much they break down or have inconsistent variation within one batch.
We know that it is responsible for a large part of the results ( otherwise is would not be used so much I guess ) and the difference in chips but the extremity in difference in testresults of one batch are suspicious. BTW, do you have any numbers so that we know what we are talking about ?
jean-paul said:
Yes, all in the CD880 and I can repeat measurements as often as I want, the result is always the same.
Also from listening.
Best sound comes from chips with evenly distributed harmonics.
Even if the total is higher.
Sound that is not so good, comes from chips that have extremely low 2nd and 3rd order.
Copy from edited post:
I had just tested the new old S1 with opa2604.
Now after 1 hour I changed to NE5532 and restarted the analyzer.
Every single harmonic up to the 9th is within 0,2 dB compared to the opa2604.
All sound character belongs to the TDA chip.
7 S1 are new and have same date code, one S1 is desoldered from the CD880 and one S1 is desoldered from another CDP.
Further tested 1 1541A and 3 plain 1541, all desoldered.
But, believe me, I don't rape the chips.
Every pin is heated for 2 seconds only.
Only the S1 from CD880 with double layer pcb, that was a fight
And this one got very good results.
from my own post :
So 7 new S1 chips with same datecode. What are the results of the measurements ? We are debating something very vague now.
I don't want to kick in an open door but just a question: are you absolutely sure you've bought real S1's ? Their popularity makes them interesting for forgers. There are fake S1's around that are rebranded normal ones and sometimes even non A versions. If all the S1's that are on the market now ( 10, 15 years after production ) are real Philips would have produced absurd amounts of selected chips
I reread your post concerning swapping OPA2604 with NE5532 without any difference in testresults. They sound absolutely different though, regardless of testresults. Your comment on being the TDA solely responsible for the sound character is not an absolute truth in that aspect. Some things in audio can not be proved by measurements alone.
So 7 new S1 chips with same datecode. What are the results of the measurements ? We are debating something very vague now.
I don't want to kick in an open door but just a question: are you absolutely sure you've bought real S1's ? Their popularity makes them interesting for forgers. There are fake S1's around that are rebranded normal ones and sometimes even non A versions. If all the S1's that are on the market now ( 10, 15 years after production ) are real Philips would have produced absurd amounts of selected chips
I reread your post concerning swapping OPA2604 with NE5532 without any difference in testresults. They sound absolutely different though, regardless of testresults. Your comment on being the TDA solely responsible for the sound character is not an absolute truth in that aspect. Some things in audio can not be proved by measurements alone.
jean-paul said:
I expected that...
One can never be sure if they are original.
But the S1 behave like A chips, so they are not plain, that is for sure.
Funny, but if you give me a 1541 with removed printing I can tell 100% if this is plain or A version.
Still, I have 1541 A S1 from CD880 which can not be a fake, same for the plain 1541 S1.
Opamps:
The spectrum of each TDA chip is very characteristic.
I measure up to 9th harmonic, so there are 8 harmonics which are in the range between -43,1dB ( by the way, this worst value was found in the 1541A ) and -66dB.
With a resolution of 0,1dB, there are 230 values per harmonic.
My math is not good but i guess this is 230^8
possibilities. The spectrum of a TDA chip is almost like a fingerprint, it is unlikely to find two identical ons, unless one tests a million.
If these subtle & characteristic ( about average 55dB below signal ) distortions are reproduced equal by two different opamps, how could they sound different ?
Original A-S1's are selected A ones Bernhard. There are even non A versions rebranded as A-S1 around, not necessarily meaning that yours are like that. If the S1 ( that simply can't be fake ) from the 880 is the best or one of the better ones you can draw a ( premature ) conclusion. Just as premature as some of the conclusions already drawn ( not excluding my conclusions )
Like comparing chips in different cdplayers in this thread :http://www.diyaudio.com/forums/showthread.php?s=&threadid=29058&perpage=15&pagenumber=7
End of discussion. If you aren't ready to accept that you have to learn. Equal *looking* signals can sound different if you like it or not.
Although I don't take off sleeves of caps I once heard a demonstration of that and heard the difference. Black magic ? You can measure as much as you want but there won't be a difference in results on the test rig.
OPA2604 and NE5532 simply have another sound character, just as interlinks can have or resistors. An accepted phenomenon I would think ( but hard to get I admit ).
stefanobilliani said:
Hello Stefano,
did you make any mod to the D1 stage to make it fit with the 1541A?
I'm very curious since I'm putting a great effort to make a good 1541 implementation, I have a PCB from Pedja (earlier version) with Elso's reclocker and want to try different I/V stages (I plan to compare Thorsten's OPA660 one, his tubed one with ECC88, Rudolf's super-pair and maybe the D1, even if I think that only half of it is needed).
After listening to them all I'll decide which one I like best.
From what you say the balanced version of the 1543 is really worth a try (too bad I don't have a completely balanced setup
)Cheers
Andrea
PCB???
Hi Stefano,
are you planning to make PCBs available? either SE or Balanced, (actually I would prefer SE). I think most people have problems when it comes to circuit boards for DACs. If no then do you have the track layout for the SE version so I could use the Press-N-Peel transfers? Please! Pretty Please!
Mozfet
Hi Stefano,
are you planning to make PCBs available? either SE or Balanced, (actually I would prefer SE). I think most people have problems when it comes to circuit boards for DACs. If no then do you have the track layout for the SE version so I could use the Press-N-Peel transfers? Please! Pretty Please!
Mozfet
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.