Op-amp class A bias circuits
Hi all,
Some replies on comments made:
a333bt [post #1182 #1194]
The bias resistor between the LM4562 and the minus of the power supply will modulate the bias current with the output signal and will pick up noise. I added some circuits that I experimented with, the cascode circuit (2 N-channel FET's and 2 resistors) performs best so far. The 1K series resistor is to reduce the effect of FET capacity
There is room for a fan, but even at low speeds and using the quitest types, the noise of the fan will become annoying. I can even hear the CD player servo noise, thats already annoying.
MGH [post #1197]
Copper conducts heat better, but it's very expensive at the moment and the heatsink needs to have a very even surface. I like the idea of the mesh pattern, but it would also reduce the top cover strength (it needs to support the tube protection grid).
SSassen [post #1198]
The accumulation of dust shouldn't be a problem at all, a DAC like this deserves to be dusted-off, cleaned and the stainless front panel to be polished each weekend.
tubee [post #1199]
The black color should emit heat better to the surrounding air, that's just what I like to avoid when conducting the D-I 8 core heat to the exterior of the housing. So at least the D-I core heatsink will be annodized in a lighter color.
adhoc [post #1200]
Since I am also slightly nitpicky, I will engrave "D-I 8 core" in the heatsink and annodize it with a gold color.
The TDA1541A shouldn't have the "heat" problem the TDA1541 seems to have, like Dragonmaster noted. Maybe the TDA1541A version has a thermal protection circuit that lowers current consumption at higher ambient temperatures. This would protect the chip until ambient temperature get's too high, but it would also mean performance would be affected by (high) ambient temperatures. By using heatsinks, the high die temperature would be avoided, preventing the TDA1541A's thermal protection circuitry from kicking-in. Perhaps this could be a explanation for the exceptional clear sound quality after placement of the large D-I core heatsink.
Hi all,
Some replies on comments made:
a333bt [post #1182 #1194]
The bias resistor between the LM4562 and the minus of the power supply will modulate the bias current with the output signal and will pick up noise. I added some circuits that I experimented with, the cascode circuit (2 N-channel FET's and 2 resistors) performs best so far. The 1K series resistor is to reduce the effect of FET capacity
There is room for a fan, but even at low speeds and using the quitest types, the noise of the fan will become annoying. I can even hear the CD player servo noise, thats already annoying.
MGH [post #1197]
Copper conducts heat better, but it's very expensive at the moment and the heatsink needs to have a very even surface. I like the idea of the mesh pattern, but it would also reduce the top cover strength (it needs to support the tube protection grid).
SSassen [post #1198]
The accumulation of dust shouldn't be a problem at all, a DAC like this deserves to be dusted-off, cleaned and the stainless front panel to be polished each weekend.
tubee [post #1199]
The black color should emit heat better to the surrounding air, that's just what I like to avoid when conducting the D-I 8 core heat to the exterior of the housing. So at least the D-I core heatsink will be annodized in a lighter color.
adhoc [post #1200]
Since I am also slightly nitpicky, I will engrave "D-I 8 core" in the heatsink and annodize it with a gold color.
The TDA1541A shouldn't have the "heat" problem the TDA1541 seems to have, like Dragonmaster noted. Maybe the TDA1541A version has a thermal protection circuit that lowers current consumption at higher ambient temperatures. This would protect the chip until ambient temperature get's too high, but it would also mean performance would be affected by (high) ambient temperatures. By using heatsinks, the high die temperature would be avoided, preventing the TDA1541A's thermal protection circuitry from kicking-in. Perhaps this could be a explanation for the exceptional clear sound quality after placement of the large D-I core heatsink.
Attachments
Hi John, nice class A bias circuits. Have you tried the solution with 2 BC560 transistors and 3 resistors too? How was current stability with it?
http://audio.nl/articlepic/fig4.JPG
http://audio.nl/articlepic/fig4.JPG
CD Engine
Hi John.
Are you still listening to the Sony CD changer? Or more to the silent Apple pc? Consider a real good Cd engine wich is not much depending on the CIRC error correction, a solid mechanism always enhance sound quality. It will play CD's quiet. You own now a top notch dac, it deserves a good CD engine too. Consider a Teac VRDS or Daisy Laser's CD pro kit. Or maybe a CD304(no mk2) and build a I2S digital out on. Can be found very cheap nowadays, has a solid CDM1 (and if you want to hear a very deep bass of that particular song, listen to it by its 14 bit dac )
Really really not offending ment here: i don't like Sony. My first Sony was horrible. Output stages going up in smoke few times, no starting cassette deck, blown fuses at startup, volume pots going full throttle while i lived above a barber shop.
But i know you a little by now and you choose allways you own solutions, its only an advice.
Hi John.
Are you still listening to the Sony CD changer? Or more to the silent Apple pc? Consider a real good Cd engine wich is not much depending on the CIRC error correction, a solid mechanism always enhance sound quality. It will play CD's quiet. You own now a top notch dac, it deserves a good CD engine too. Consider a Teac VRDS or Daisy Laser's CD pro kit. Or maybe a CD304(no mk2) and build a I2S digital out on. Can be found very cheap nowadays, has a solid CDM1
(and if you want to hear a very deep bass of that particular song, listen to it by its 14 bit dac ) Really really not offending ment here: i don't like Sony. My first Sony was horrible. Output stages going up in smoke few times, no starting cassette deck, blown fuses at startup, volume pots going full throttle while i lived above a barber shop.
But i know you a little by now and you choose allways you own solutions, its only an advice.
Digital sound source
Hi tubee,
Thanks for your reply [post #1203 / 1204]
> Yes I tried the current mirror circuit for OP-amp class A bias already, also posted it on this thread, I used a dual PNP transistor from Toshiba (2SA1349). The current stability depends on thermal drift and power supply stability (the current mirror output current matches the input current provided by the 30K resistor, so power supply ripple / noise is also added to the output current).
> No I don't use the Sony player much, after putting the LM4562 in both D-I DACs, it became clear, the iMac (USB) provided the clearest sound with the deepest soundstage. The Sony's direct I2S output seems to be no match for the shiftregister reclocker located close to the D-I core.
> I agree that a (modified) Sony CD changer isn't the best digital sound source, and with the LM4562 errors become audible that are not present when using the iMac as sound source using the same CD image. In other words, the Sony CD changer seems to modify the digital data being read from the CD, before it's output on the I2S.
But even when using the best transports available, read errors can and will occur, forcing the transport to start correcting these. It depends completely on the following circuitry how these errors are handled and if the digital bit-stream remains bit perfect or is altered in some way.
The transport being a opto-mechanical system, will have a limited life-time / wears out. When CD's are gradually being replaced by other media, the CD player will become obsolete sooner or later.
So in my personal opinion the best digital sound source would be a computer, generating exactly the same quality, every time. The data from the HD is buffered in RAM, so even multiple HD read errors or HD read delays will have no effect. It also provides the best platform for future developments.
Basically we only need a low jitter digital bit-stream of one's and zero's, without errors. If this criteria is met, it doesn't matter at all what kind of digital sound source is used.
When using a computer as digital sound source, the sound file can be created using CD paranoia (multiple reads to minimize read errors), or lossless CD tracks (not MP3) can be directly downloaded from the internet (in the future). The file is then stored on a hard-disk and is now fixed, it's integrity will be monitored by the operating system. So no matter if you play this file one time or 100 times, it will remain exactly the same. This is virtually impossible using even the best transport.
The many CD's that can be stored on a hard-disk using lossless compression, also means you have all this music available instantaneously, and can toggle between CD's / tracks within seconds, or create a play list. What's the use of a large CD collection if you can't quickly locate the CD / track you are looking for? The swapping of CD's in a single CD transport also increases the chance of damaging / scratching the CD surface.
When using lossless compression, the file size is approx. half of the original size. Suppose an average CD contains approx. 400Mb of data, this will result in approx. 200Mb / CD. Today HD capacity is rapidly increasing, let's take an average 250Gb hard-disk, this will hold approx. 1000 CD's in lossless format on a single 250Gb hard-disk. That's a lot of CD's! if this is still not enough, one could use a 500Gb drive storing over 2000 CD's!
There are some nice graphical interfaces available for browsing through a CD collection like coverflow. So it's very easy to find your favorite track quickly. This is a very big advantage that's impossible when using single CD's in a transport. The computer can be connected directly to both the D-I 8 and D-I 16 DAC without interface problems, just plug and play.
> no offense taken, I had some problems with Sony equipment too.
> I do listen to other peoples advice, many tips have already resulted in improvements made in the D-I DAC.
Hi tubee,
Thanks for your reply [post #1203 / 1204]
> Yes I tried the current mirror circuit for OP-amp class A bias already, also posted it on this thread, I used a dual PNP transistor from Toshiba (2SA1349). The current stability depends on thermal drift and power supply stability (the current mirror output current matches the input current provided by the 30K resistor, so power supply ripple / noise is also added to the output current).
> No I don't use the Sony player much, after putting the LM4562 in both D-I DACs, it became clear, the iMac (USB) provided the clearest sound with the deepest soundstage. The Sony's direct I2S output seems to be no match for the shiftregister reclocker located close to the D-I core.
> I agree that a (modified) Sony CD changer isn't the best digital sound source, and with the LM4562 errors become audible that are not present when using the iMac as sound source using the same CD image. In other words, the Sony CD changer seems to modify the digital data being read from the CD, before it's output on the I2S.
But even when using the best transports available, read errors can and will occur, forcing the transport to start correcting these. It depends completely on the following circuitry how these errors are handled and if the digital bit-stream remains bit perfect or is altered in some way.
The transport being a opto-mechanical system, will have a limited life-time / wears out. When CD's are gradually being replaced by other media, the CD player will become obsolete sooner or later.
So in my personal opinion the best digital sound source would be a computer, generating exactly the same quality, every time. The data from the HD is buffered in RAM, so even multiple HD read errors or HD read delays will have no effect. It also provides the best platform for future developments.
Basically we only need a low jitter digital bit-stream of one's and zero's, without errors. If this criteria is met, it doesn't matter at all what kind of digital sound source is used.
When using a computer as digital sound source, the sound file can be created using CD paranoia (multiple reads to minimize read errors), or lossless CD tracks (not MP3) can be directly downloaded from the internet (in the future). The file is then stored on a hard-disk and is now fixed, it's integrity will be monitored by the operating system. So no matter if you play this file one time or 100 times, it will remain exactly the same. This is virtually impossible using even the best transport.
The many CD's that can be stored on a hard-disk using lossless compression, also means you have all this music available instantaneously, and can toggle between CD's / tracks within seconds, or create a play list. What's the use of a large CD collection if you can't quickly locate the CD / track you are looking for? The swapping of CD's in a single CD transport also increases the chance of damaging / scratching the CD surface.
When using lossless compression, the file size is approx. half of the original size. Suppose an average CD contains approx. 400Mb of data, this will result in approx. 200Mb / CD. Today HD capacity is rapidly increasing, let's take an average 250Gb hard-disk, this will hold approx. 1000 CD's in lossless format on a single 250Gb hard-disk. That's a lot of CD's! if this is still not enough, one could use a 500Gb drive storing over 2000 CD's!
There are some nice graphical interfaces available for browsing through a CD collection like coverflow. So it's very easy to find your favorite track quickly. This is a very big advantage that's impossible when using single CD's in a transport. The computer can be connected directly to both the D-I 8 and D-I 16 DAC without interface problems, just plug and play.
> no offense taken, I had some problems with Sony equipment too.
> I do listen to other peoples advice, many tips have already resulted in improvements made in the D-I DAC.
How nice would it be if the CD player project gmarsh started wasn't dead : It was using an Analog Devices Blackfin DSP, which, running on linux or VisualDSP, would have extracted and buffered the CD's data in DAE mode just like cdparanoia and play it back on a jitter-free I2S output generated and clocked by the >400MHz CPU.
Thanks John for a small look in the future. Yesterday i considered for a slight moment to connect the soundcard to my pre-amp, but thats not the best way so i didn't (internal so "noisy" soundcard)
When everything is ready will try your USB/I2S connection, and see how music from a PC can sound. First i'll have to make the fans even more quieter as my first attempt.(finer blades on fan) Maybe passive cooling. The pc is now stable, thanks to a PROPER OS!)
Found, and will read this over too: http://www.diyaudio.com/forums/showthread.php?s=&threadid=88484&perpage=10&pagenumber=2
When everything is ready will try your USB/I2S connection, and see how music from a PC can sound. First i'll have to make the fans even more quieter as my first attempt.(finer blades on fan) Maybe passive cooling. The pc is now stable, thanks to a PROPER OS!)
Found, and will read this over too: http://www.diyaudio.com/forums/showthread.php?s=&threadid=88484&perpage=10&pagenumber=2
Availability D-I DAC PCB's
Hi ak_47_boy
Thanks for your reply [post #1202]
1) Both D-I 8 and D-I 16 will have some minor changes, the PCB's will be modified to accept the LM4562 dual OP-amp instead of the OPA627 single OP-amp, and the D-I 16 power supply filtering will be improved.
2) I am planning to provide the PCB's first quarter of 2007. As noted earlier they will have (thick) flash gold plated traces, a transparent anti-solder mask and black silkscreen print.
3) The D-I DAC should have a control amplifier / buffer placed after it (volume / balance control), with unity gain and a input resistance of approx. 20...100K Ohms. RCA interlink between DAC and control amplifier should be kept as short as possible.
Hi ak_47_boy
Thanks for your reply [post #1202]
1) Both D-I 8 and D-I 16 will have some minor changes, the PCB's will be modified to accept the LM4562 dual OP-amp instead of the OPA627 single OP-amp, and the D-I 16 power supply filtering will be improved.
2) I am planning to provide the PCB's first quarter of 2007. As noted earlier they will have (thick) flash gold plated traces, a transparent anti-solder mask and black silkscreen print.
3) The D-I DAC should have a control amplifier / buffer placed after it (volume / balance control), with unity gain and a input resistance of approx. 20...100K Ohms. RCA interlink between DAC and control amplifier should be kept as short as possible.
LM4562 vs OPA627
Hi John,
Just my two cents about LM4562 vs OPA627.
I tested the smd LM4562 as a replacement for OPA627 in my DAC. The opamp is used as a unity gain buffer after passive I/V. I used one of my DACs with opamps in sockets. I have bypass caps soldered directly to the chips. Bypass caps were previously broken-in. I let the LM4562 burn-in for two days. I used one LM4562 for each channel, using only one of the two opamps per chip, the other opamp had its inputs shorted.
The LM4562 was very clear, maybe more so then OPA627 but was very thin sounding compaired to the golden sound of OPA627. After two days of LM4562 I am so much more in love with OPA627BP. The LM4562 is very stable with unity gain, just like OPA627.
The supply was +/- 18 volts.
I don't doubt your implentation sounds very good and possibly better then when you used OPA627. I am just reporting my findings in the event it will help you or anyone who reads this.
Cheers,
Brent
Hi John,
Just my two cents about LM4562 vs OPA627.
I tested the smd LM4562 as a replacement for OPA627 in my DAC. The opamp is used as a unity gain buffer after passive I/V. I used one of my DACs with opamps in sockets. I have bypass caps soldered directly to the chips. Bypass caps were previously broken-in. I let the LM4562 burn-in for two days. I used one LM4562 for each channel, using only one of the two opamps per chip, the other opamp had its inputs shorted.
The LM4562 was very clear, maybe more so then OPA627 but was very thin sounding compaired to the golden sound of OPA627. After two days of LM4562 I am so much more in love with OPA627BP. The LM4562 is very stable with unity gain, just like OPA627.
The supply was +/- 18 volts.
I don't doubt your implentation sounds very good and possibly better then when you used OPA627. I am just reporting my findings in the event it will help you or anyone who reads this.
Cheers,
Brent
LM4562 sound quality
Hi Brent Welke,
Thanks for your reply [post#1220],
First of all, the LM4562, just like the OPA627 needs to warm up / settle. When cold, sound quality is not optimal. It needs approx. 15 minutes to settle and produce optimal sound quality.
My first impression with the LM4562 was similar, thin but very clear high-resolution sound. So I switched to OPA627 again, but now in comparison the sound was rather laid back and muddy, covering up details, clearly heard with the LM4562. The crystal clear sound, typical for the LM4562 was gone. The thin sound is probably caused by the absence of distortion, producing less unwanted harmonics cluttering up the midrange (harmonics produced by the OP-amp that don't belong to the original signal).
I was used to listen to the warm laid-back sound of the OPA627, and had similar initial doubts about the LM4562 as you have. But now after listening to it for some time, switching between OPA627 and LM4562, the LM4562 sound quality is really superior to the OPA627. Even the D-I 16 sounds clearer than ever before with the LM4562.
I use the LM4562 metal can for I/V conversion and the SO8 version on a adapter socket for the diff amp (for the prototypes). Both run on a stabilized +/- 15V power supply. I also placed the LM4562 DIP samples in the control amplifier.
Cheers,
John
Hi Brent Welke,
Thanks for your reply [post#1220],
First of all, the LM4562, just like the OPA627 needs to warm up / settle. When cold, sound quality is not optimal. It needs approx. 15 minutes to settle and produce optimal sound quality.
My first impression with the LM4562 was similar, thin but very clear high-resolution sound. So I switched to OPA627 again, but now in comparison the sound was rather laid back and muddy, covering up details, clearly heard with the LM4562. The crystal clear sound, typical for the LM4562 was gone. The thin sound is probably caused by the absence of distortion, producing less unwanted harmonics cluttering up the midrange (harmonics produced by the OP-amp that don't belong to the original signal).
I was used to listen to the warm laid-back sound of the OPA627, and had similar initial doubts about the LM4562 as you have. But now after listening to it for some time, switching between OPA627 and LM4562, the LM4562 sound quality is really superior to the OPA627. Even the D-I 16 sounds clearer than ever before with the LM4562.
I use the LM4562 metal can for I/V conversion and the SO8 version on a adapter socket for the diff amp (for the prototypes). Both run on a stabilized +/- 15V power supply. I also placed the LM4562 DIP samples in the control amplifier.
Cheers,
John
D-I DAC digital interfaces
Hi MGH,
Thanks for your reply [post #1213]
First of all I want to wish everyone a happy new year,
I have been working on the D-I DAC digital interface and jitter performance.
The objective was getting equal sonic performance regardless of the type of digital interface used, I2S, USB or SPDIF, the aim was to get jitter values as low as possible.
After some major modifications, and completing the universal I2S interface prototype (SPDIF input) all different digital input sources now result in virtually the same sound quality, if there is still a minute difference, it's probably caused by different crystal jitter specifications. On the oscilloscope, no jitter is visible on the highest timebase setting (jitter was clearly visible on the CS8416, PCM2706 and direct I2S output from a LSI).
Now the D-I DACs can be connected to either USB, I2S (philips format) or SPDIF while maintaining virtually identical sound quality (provided the digital sound source outputs a error-free bit stream). Differences in (perceived) sound quality often relate to jitter / phase noise on the timing signal (BCK for the D-I DAC).
I also tested the D-I DAC using different computers as digital sound source. For mac and windows, iTunes can be used. For ubuntu, the Rhythmbox Music Player can be used. I also tested a active USB cable, this enables USB interlink cable length to be increased to 10 meters. This can be useful when a noisy computer needs to be placed in another room.
I am currently updating the PCB layouts.
Hi MGH,
Thanks for your reply [post #1213]
First of all I want to wish everyone a happy new year,
I have been working on the D-I DAC digital interface and jitter performance.
The objective was getting equal sonic performance regardless of the type of digital interface used, I2S, USB or SPDIF, the aim was to get jitter values as low as possible.
After some major modifications, and completing the universal I2S interface prototype (SPDIF input) all different digital input sources now result in virtually the same sound quality, if there is still a minute difference, it's probably caused by different crystal jitter specifications. On the oscilloscope, no jitter is visible on the highest timebase setting (jitter was clearly visible on the CS8416, PCM2706 and direct I2S output from a LSI).
Now the D-I DACs can be connected to either USB, I2S (philips format) or SPDIF while maintaining virtually identical sound quality (provided the digital sound source outputs a error-free bit stream). Differences in (perceived) sound quality often relate to jitter / phase noise on the timing signal (BCK for the D-I DAC).
I also tested the D-I DAC using different computers as digital sound source. For mac and windows, iTunes can be used. For ubuntu, the Rhythmbox Music Player can be used. I also tested a active USB cable, this enables USB interlink cable length to be increased to 10 meters. This can be useful when a noisy computer needs to be placed in another room.
I am currently updating the PCB layouts.
agent.5 said:
I must assume Johns DAC will sound less distorted, than Picasso's pictures look to me (taste, please no discussion)
On a serious note: John, I have some experience with offering designs to the DIY community and my advice, make sure it is almost idiot proof and don't try to make it perfect, as it will never see the world.... Just imagine Microsoft would do the same. We still would have DOS1.0
best regards
doede
PS: a VP of engineering of a well know company told me once that every time they made their products idiot proof, "they" invented a better idiot