parallelling DAC's / output voltage compliance
Hi, dddac
Thanks for your reply [post#199]
I have a question, I had a look at the DAC's you make (good job). As you already indicated you are using the TDA1543. The TDA1543 has a typical distortion of 0.018%, the TDA1541 has a typical distortion of 0.0018%, ten times less. Would this be a reason for the improvements you notice by using more DAC's?
The octal D-I DAC doesn't use the 8 DAC chips in "parallel" to reduce distortion or boost output current (these are only welcome side effects). The only reason is to create a very smooth unfiltered DC coupled output signal.
I am still a bit confused
about "high output voltages straight from de DAC" or passive I/V converters using resistors for I/V conversion, as these DAC's are specifically designed to "see" virtual ground on their output. Both the TDA1541A and TDA1543 datasheets indicate the following:
This is a quote from the TDA1541A datasheet (same for TDA1543A):
Notes to the characteristics
1. To ensure no performance losses, permitted output voltage compliance is ±25 mV maximum.
The TDA1543 datasheets also indicates: output voltage compliance ±25 mV
That is the reason why I use OP-amp based I/V converters, to maintain performance by keeping the output voltage at the DAC pin as low as possible. As I noted in an earlier post, a 33 Ohm resistor will cause an output voltage of approximately ±132mV! I have already build a hybrid setup (OPA627 I/V converter with triode differential amplifier), works very nice.
The SMD circuit board is a good idea for upgrading an existing DAC or prototyping. I would be very interested in the results of your listening tests.
Hi, dddac
Thanks for your reply [post#199]
I have a question, I had a look at the DAC's you make (good job). As you already indicated you are using the TDA1543. The TDA1543 has a typical distortion of 0.018%, the TDA1541 has a typical distortion of 0.0018%, ten times less. Would this be a reason for the improvements you notice by using more DAC's?
The octal D-I DAC doesn't use the 8 DAC chips in "parallel" to reduce distortion or boost output current (these are only welcome side effects). The only reason is to create a very smooth unfiltered DC coupled output signal.
I am still a bit confused
about "high output voltages straight from de DAC" or passive I/V converters using resistors for I/V conversion, as these DAC's are specifically designed to "see" virtual ground on their output. Both the TDA1541A and TDA1543 datasheets indicate the following:This is a quote from the TDA1541A datasheet (same for TDA1543A):
Notes to the characteristics
1. To ensure no performance losses, permitted output voltage compliance is ±25 mV maximum.
The TDA1543 datasheets also indicates: output voltage compliance ±25 mV
That is the reason why I use OP-amp based I/V converters, to maintain performance by keeping the output voltage at the DAC pin as low as possible. As I noted in an earlier post, a 33 Ohm resistor will cause an output voltage of approximately ±132mV! I have already build a hybrid setup (OPA627 I/V converter with triode differential amplifier), works very nice.
The SMD circuit board is a good idea for upgrading an existing DAC or prototyping. I would be very interested in the results of your listening tests.
DAC parts
Hi, MGH,
Thanks for your reply,
When the D-I DAC project is completed, I am planning to do the following:
1) Manufacture a professional printed circuit board set, as this project will be modular you can pick out the circuit boards you need from this set. It also enables future updates.
2) Provide schematic diagrams, partlist, mechanical drawings
3) Design a very special "custom" housing for it, that tops the D-I DAC design, I am planning to use annodized aluminum parts combined with high-gloss stainless steel. I can post some photo's of my high-end power amplifier that uses these materials as well, so you have an idea how it could look like.
4) Manufacture parts for this "custom" housing.
I am not planning to supply electrical components, only the special parts like printed circuit boards and housing parts.
Hi, MGH,
Thanks for your reply,
When the D-I DAC project is completed, I am planning to do the following:
1) Manufacture a professional printed circuit board set, as this project will be modular you can pick out the circuit boards you need from this set. It also enables future updates.
2) Provide schematic diagrams, partlist, mechanical drawings
3) Design a very special "custom" housing for it, that tops the D-I DAC design, I am planning to use annodized aluminum parts combined with high-gloss stainless steel. I can post some photo's of my high-end power amplifier that uses these materials as well, so you have an idea how it could look like.
4) Manufacture parts for this "custom" housing.
I am not planning to supply electrical components, only the special parts like printed circuit boards and housing parts.
This is discussed in some other threads. The 1543 can be used without i/v, this is not conform the data sheets, paralleling them is not written about in datasheets either.
Its capable, even having higher distorsion, to deliver a very transparent sound. (Have here a Aristona SAA7220/TDA1543 player) Very dynamic for popmusic/jazz.
Another thing: when good pcb's are manufactured, there has to be some considering for different regulators. There has been a thread of the LM78** here, others diyers like the TL317/337 and/or shunt regulators.
For a good pcb layout, Guido Tent wrote an article with guidelines.
I will give some links to them tonight/tomorrow, no time left now.
Differential hybrid output stage
Hi, all
When I started designing the output stage for the D-I DAC, I choose an all OP-amp design, with one of the best OP-amps available at this moment. But since High-end applications are often tube based, I wanted to give the tube output stage a chance. Personally I was very sceptic about this "obsolete" technology as (discrete) OP-amp specs are far superior according to their datasheets.
So here is what I did, I build both, a differential OP-amp output stage and a differential tube output stage, and connected them both to the octal D-I DAC. Making sure both output signals had the same amplitude, and oscillograms of the signal shapes looked the same. FFT scans looked similar as well. Now toggling between both during listening tests should solve the question, tube or "tubeless".
So the race was on, first the OP-amp (OPA627), sounded perfect to me, detail, midrange, bass all there in a perfect balance, open sound, what more do you want. Then I switched to the tube output stage
how could this be? The sound was even more open, midrange was warm, the sound was reproduced totally effortless, in fact I listened to it for the rest of the day, one CD after the other, without ever experiencing the slightest listening fatigue, in fact it is still playing right now.
The D-I DAC and the differential tube output stage seemed to be the perfect match. So for High-end audio applications, it seems the tube has caught up with the superfast semiconductor to put it in tubee's words. Is the OP-amp really that bad? no, the OPA627 comes very close to the tube sound, but just lacks that little extra, bass reproduction is slightly better (DC coupled). The tube output stage really brings music to life, so my choice is made. But since the differential OP-amp stage only uses 1 OP-amp, I am planning to put both in the octal D-I DAC, so it has the unique feature to toggle between both semiconductor and tube output stage.
I have added the schematic diagram of the differential hybrid tube output stage, so you can have a look at it. The transistors and zenerdiode are used to stabilize and isolate the annode voltages so the "hum" is deleted. The differential stage uses 2 triodes, connected to "buffers" (cathode followers). The LM334 current regulator fixes the current for the differential stage. The filaments are put in series (12.6V). As I mentioned before, the I/V stage still uses the OPA627. The output is fully differential! but if you don't need that, just leave-out the second cathode follower. The very high amplitude output signal is attenuated to reduce noise, together with cable capacitance it creates a perfect signal. Power consumption is very low. I am sure the schematic can be optimized / improved, but it sounds very good as it is now. For 2 channels you need 3 tubes ECC82 for normal output and 4 tubes ECC82 for differential output. Both C3 and C4 should be high quality polypropylene cap's (I used Audyn cap's). I am not happy with cap's in the signal path, but in this case I am willing to make an exeption.
WARNING: if you intend to experiment with this tube output stage, be very carefull with the high voltages. Always switch off and wait for C1 to decharge before working on the circuit (check with voltmeter). Also make sure there are no short circuits as this might damage the I/V stages.
Hi, all
When I started designing the output stage for the D-I DAC, I choose an all OP-amp design, with one of the best OP-amps available at this moment. But since High-end applications are often tube based, I wanted to give the tube output stage a chance. Personally I was very sceptic about this "obsolete" technology as (discrete) OP-amp specs are far superior according to their datasheets.
So here is what I did, I build both, a differential OP-amp output stage and a differential tube output stage, and connected them both to the octal D-I DAC. Making sure both output signals had the same amplitude, and oscillograms of the signal shapes looked the same. FFT scans looked similar as well. Now toggling between both during listening tests should solve the question, tube or "tubeless".
So the race was on, first the OP-amp (OPA627), sounded perfect to me, detail, midrange, bass all there in a perfect balance, open sound, what more do you want. Then I switched to the tube output stage
how could this be? The sound was even more open, midrange was warm, the sound was reproduced totally effortless, in fact I listened to it for the rest of the day, one CD after the other, without ever experiencing the slightest listening fatigue, in fact it is still playing right now. The D-I DAC and the differential tube output stage seemed to be the perfect match. So for High-end audio applications, it seems the tube has caught up with the superfast semiconductor to put it in tubee's words. Is the OP-amp really that bad? no, the OPA627 comes very close to the tube sound, but just lacks that little extra, bass reproduction is slightly better (DC coupled). The tube output stage really brings music to life, so my choice is made. But since the differential OP-amp stage only uses 1 OP-amp, I am planning to put both in the octal D-I DAC, so it has the unique feature to toggle between both semiconductor and tube output stage.
I have added the schematic diagram of the differential hybrid tube output stage, so you can have a look at it. The transistors and zenerdiode are used to stabilize and isolate the annode voltages so the "hum" is deleted. The differential stage uses 2 triodes, connected to "buffers" (cathode followers). The LM334 current regulator fixes the current for the differential stage. The filaments are put in series (12.6V). As I mentioned before, the I/V stage still uses the OPA627. The output is fully differential! but if you don't need that, just leave-out the second cathode follower. The very high amplitude output signal is attenuated to reduce noise, together with cable capacitance it creates a perfect signal. Power consumption is very low. I am sure the schematic can be optimized / improved, but it sounds very good as it is now. For 2 channels you need 3 tubes ECC82 for normal output and 4 tubes ECC82 for differential output. Both C3 and C4 should be high quality polypropylene cap's (I used Audyn cap's). I am not happy with cap's in the signal path, but in this case I am willing to make an exeption
.WARNING: if you intend to experiment with this tube output stage, be very carefull with the high voltages. Always switch off and wait for C1 to decharge before working on the circuit (check with voltmeter). Also make sure there are no short circuits as this might damage the I/V stages.
Attachments
Thanks ecdesigns,
It shouldn't be too difficult to obtain 8 TDA1541As and chips for the timing board (hopefully you could point us to where we can get those chips). But I really hope you will consider a very high quality PCB as we discussed (posts 179, 182, 183).
I'm really happy to see you actually tried a tubed output stage as an alternitive to the OPA627. No matter how good solid state or OP amps are, I always prefer the sound of tubes in the end - like you said more alive sounding.
For us who don't have the experience as you in putting together DACs, will you include detailed, sequential instructions on how to put the parts together?
You said the housing for the DAC will look similar to your amp. Can you post a picture?
Keep up the great work.
It shouldn't be too difficult to obtain 8 TDA1541As and chips for the timing board (hopefully you could point us to where we can get those chips). But I really hope you will consider a very high quality PCB as we discussed (posts 179, 182, 183).
I'm really happy to see you actually tried a tubed output stage as an alternitive to the OPA627. No matter how good solid state or OP amps are, I always prefer the sound of tubes in the end - like you said more alive sounding.
For us who don't have the experience as you in putting together DACs, will you include detailed, sequential instructions on how to put the parts together?
You said the housing for the DAC will look similar to your amp. Can you post a picture?
Keep up the great work.
octal D-I DAC tube prototype
Hi, again
Well, this is called prototyping...As you can see it is time to develop both the main board and output stage REAL SOON as I am quickly running out of board space... . But this prototype has really fulfilled it's purpose. Now, just like me you must think there has to be a lot of crosstalk and interference with all those wires, well actually the output signal is kept very clean by using large amplitude signals that are attenuated at the output. (post#158). And where do these large amplitude signals come from? the dual 4 added DAC currents in combination with the OPA627 I/V converter stages, doubled by the differential amplifier!
Hi, again
Well, this is called prototyping...As you can see it is time to develop both the main board and output stage REAL SOON as I am quickly running out of board space...
. But this prototype has really fulfilled it's purpose. Now, just like me you must think there has to be a lot of crosstalk and interference with all those wires, well actually the output signal is kept very clean by using large amplitude signals that are attenuated at the output. (post#158). And where do these large amplitude signals come from? the dual 4 added DAC currents in combination with the OPA627 I/V converter stages, doubled by the differential amplifier!Attachments
Photo cascode MOSFET power amp
Hi, MGH
Thanks for your reply,
I got the 74HCT161 from RS components, Farnell should have them too. The HC version is more difficult to find. If the project is fully completed, I will certainly use high quality boards as this D-I DAC really deserves them. Of course I will add a detailed step by step intruction of how to build this DAC.
I also added a photo of the cascode 2 X 70W rms MOSFET power amplifier that is placed in a black annodized aluminum housing with a 5mm thick high-gloss stainless steel front piece to match the design. I manufactured all housing parts myself (exept the stainless steel piece) with the aid of 2 computer controlled milling machines. The logo text is actually a circuit board being backlit by LED block's. The push buttons are made from red transparent plexiglass, the icons for the backlit power and light functions are also small circuit boards. The amplifier is virtually indestructible, it is designed to work without output relays or DC protection (so it can play the track "Bela Fleck, a Celtic medly" without the DC protection kicking in. It can be short circuited or the input plug can be pulled with full power. The photo isn't really good, but I hope you will get an impression of how the octal D-I DAC could look like. I am planning to put the 4 ECC82's on top and fully integrate them into the design, protected by aluminum or stainless steel plate construction. It would also get similar backlit push buttons for power, lighting, input selection and tube / semiconductor mode. If you switch to semiconductor mode, the tube output section will be shut down and the output is re-routed to the semiconductor stage with the push of a button.
Hi, MGH
Thanks for your reply,
I got the 74HCT161 from RS components, Farnell should have them too. The HC version is more difficult to find. If the project is fully completed, I will certainly use high quality boards as this D-I DAC really deserves them. Of course I will add a detailed step by step intruction of how to build this DAC.
I also added a photo of the cascode 2 X 70W rms MOSFET power amplifier that is placed in a black annodized aluminum housing with a 5mm thick high-gloss stainless steel front piece to match the design. I manufactured all housing parts myself (exept the stainless steel piece) with the aid of 2 computer controlled milling machines. The logo text is actually a circuit board being backlit by LED block's. The push buttons are made from red transparent plexiglass, the icons for the backlit power and light functions are also small circuit boards. The amplifier is virtually indestructible, it is designed to work without output relays or DC protection (so it can play the track "Bela Fleck, a Celtic medly" without the DC protection kicking in
. It can be short circuited or the input plug can be pulled with full power. The photo isn't really good, but I hope you will get an impression of how the octal D-I DAC could look like. I am planning to put the 4 ECC82's on top and fully integrate them into the design, protected by aluminum or stainless steel plate construction. It would also get similar backlit push buttons for power, lighting, input selection and tube / semiconductor mode. If you switch to semiconductor mode, the tube output section will be shut down and the output is re-routed to the semiconductor stage with the push of a button.Attachments
Hi, Bernhard,
Well, lets put it this way. If I am working on a project, I often forget about time and realize I worked trough all night.
Electronics can be so fascinating, electronics is all I got, it consumes all my time. It's a continuous learning process that is rewarded when things work out the way you planned. I am very gratefull of being in a situation to be able to do all this. By posting information about this project I want to share the thrill and exitement of electronics design.
I have added another picture of the octal D-I DAC prototype in "action"
Well, lets put it this way. If I am working on a project, I often forget about time and realize I worked trough all night.
Electronics can be so fascinating, electronics is all I got, it consumes all my time. It's a continuous learning process that is rewarded when things work out the way you planned. I am very gratefull of being in a situation to be able to do all this. By posting information about this project I want to share the thrill and exitement of electronics design.
I have added another picture of the octal D-I DAC prototype in "action"
Attachments
Re: Photo cascode MOSFET power amp
Hi ecdesigns, it looks like Jameco has the 74HC161 version (as opposed to the HCT) version if in fact this is the one you mentioned and is preferable.
http://www.jameco.com/webapp/wcs/st...storeId=10001&catalogId=10001&productId=45452
Best - Stan
-ecdesigns- said:
Hi ecdesigns, it looks like Jameco has the 74HC161 version (as opposed to the HCT) version if in fact this is the one you mentioned and is preferable.
http://www.jameco.com/webapp/wcs/st...storeId=10001&catalogId=10001&productId=45452
Best - Stan
Housing / audio interface chip
Hi, MGH,
Thanks for your reply,
I already made some quick drawings of the octal D-I DAC housing, I will make a CAD drawing of it soon, so you get an impression. I need to do this now, so I can design the main board and addittional modules accordingly.
Basically the DAC is housed similar as the Power amplifier, but, on top it get's really different. I plan to make 8 half-sinewave shaped plates that look as a "low resolution" step signal. They are then arranged basicaslly in the same manner as the signals in the D-I DAC itself. So when looking at the top, you see a high resolution sinewave, sideways you see the coarse shaped sinewaves. These plates also function both, as protection and screening of the 4 tubes. When you look at the DAC from the front, you see the 4 tubes trough the 8 horizontal plates. In the centre of the sinewave shaped plates that extend over the entire width of the housing is the logo. The sinewave plates would look awesome when made out of thin high-gloss stainless steel plates. The housing itself has all the backlit switches, and status indicators. I also plan to use a stainless steel plate on the frontpanel (similar to the power amplifier) to match the design. On the rear there would be connections for power supply, differential L and R outputs, 4 differential SPDIF inputs (SVHS socket), 4 coax inputs, a USB input and a (differential) I2S input.
The HC question is already answered,
I use the Crystal CS8412 audio interface chip. But since the audio interface is modular, different chips can be used like CS8414 (adapter socket) or DIR1703 with twin PLL system. I tried the DIR1703 in the past, it had a very stable output signal, but it is very sensitive to interference signals (it would immediately lock-out).
Hi, MGH,
Thanks for your reply,
I already made some quick drawings of the octal D-I DAC housing, I will make a CAD drawing of it soon, so you get an impression. I need to do this now, so I can design the main board and addittional modules accordingly.
Basically the DAC is housed similar as the Power amplifier, but, on top it get's really different. I plan to make 8 half-sinewave shaped plates that look as a "low resolution" step signal. They are then arranged basicaslly in the same manner as the signals in the D-I DAC itself. So when looking at the top, you see a high resolution sinewave, sideways you see the coarse shaped sinewaves. These plates also function both, as protection and screening of the 4 tubes. When you look at the DAC from the front, you see the 4 tubes trough the 8 horizontal plates. In the centre of the sinewave shaped plates that extend over the entire width of the housing is the logo. The sinewave plates would look awesome when made out of thin high-gloss stainless steel plates. The housing itself has all the backlit switches, and status indicators. I also plan to use a stainless steel plate on the frontpanel (similar to the power amplifier) to match the design. On the rear there would be connections for power supply, differential L and R outputs, 4 differential SPDIF inputs (SVHS socket), 4 coax inputs, a USB input and a (differential) I2S input.
The HC question is already answered,
I use the Crystal CS8412 audio interface chip. But since the audio interface is modular, different chips can be used like CS8414 (adapter socket) or DIR1703 with twin PLL system. I tried the DIR1703 in the past, it had a very stable output signal, but it is very sensitive to interference signals (it would immediately lock-out).
Octal D-I DAC sketches
Hi, MGH
Thanks for your reply [#219]
>The speed of logic circuitry becomes critical when the design is timing critical. In a critical design, these speed differences can cause circuitry to "trigger" at the wrong moment or causing erratic timing. BCK frequency is 2.8224 MHz / 354nS, a low or high level is 177ns. So if the logic used has equal then or more than 177ns you're in trouble. The HCT has a propagation delay of 20ns, the HC 15ns. So with both chips we are quite "safe".
>So why do I prefer higher speed? Suppose you have a slow logic circuitry in the SPDIF signal. Transients would look like ramp's rather than the ideal "vertical lines". Now suppose the SPDIF receiver schmitt trigger input circuitry threshold voltage varies a little bit, it then would trigger on different levels of the ramp shaped transient. So phase variation occurs, also called jitter. This is one of the things that is happening with optical SPDIF interfaces as they have lower speed / ramp shaped transients. So basically, slow ramp shaped transients are an invitation to jitter.
>Modern major brand power supply regulators, when used correctly, are not bad. They are cheap, produce a quite stable output voltage and by adding decoupling cap's of the right value close to the regulator you end up with a good power supply. Zener diodes are a bit of a problem (dissipation and power supply regulation) as TDA1541A supply currents can go up to 50mA. Bandgap reference diodes (LM336-5.0) are quite good but the maximum allowable current is only 10mA. So for the moment I will keep the existing regulators.
>The current DAC circuit board is designed with decoupling and suppressing interference in mind. Each DAC chip has it's own power supply, Each DAC chip has a separate analog and digital ground connection, tracks are kept very short. Analog main board design has to be done very carefully to get the lowest interference possible. I am also planning to put all digital modules on a separate digital mainboard. I use a dual stabilizing method (cascading regulators) for stable output voltages. For decoupling, tantalium cap's seem to do a good job in filtering out interference, ceramic SMD capacitors take care of the very high frequencies (tantalium cap's are horrible for analog audio applications, but quite good for power supply decoupling, and they remain stable over time). Even with the worst-case test setup I used, the output signal is very clean. The sound is so good, it is very difficult to "switch off the set" once you start listening, so that is one of the things I have been doing the past days, the other "thing" is attached to this post, hope you like.
The upper sketch is the top-view of the octal D-I DAC, showing how the 8 coarse protection and screening plates form a "smooth" half sinewave shape, reflecting the operating principle of the octal D-I dac in the housing design. Text is engraved in the aluminum plates.
The lower sketch is the front view, status indication is done trough orange LED's behind the 4 tubes. When warming-up the LED's flash, if the SPDIF signal is received correctly, the LED's spread a orange glow around the tubes. Using high-gloss stainless steel plates would give wonderful effect.
Hi, MGH
Thanks for your reply [#219]
>The speed of logic circuitry becomes critical when the design is timing critical. In a critical design, these speed differences can cause circuitry to "trigger" at the wrong moment or causing erratic timing. BCK frequency is 2.8224 MHz / 354nS, a low or high level is 177ns. So if the logic used has equal then or more than 177ns you're in trouble. The HCT has a propagation delay of 20ns, the HC 15ns. So with both chips we are quite "safe".
>So why do I prefer higher speed? Suppose you have a slow logic circuitry in the SPDIF signal. Transients would look like ramp's rather than the ideal "vertical lines". Now suppose the SPDIF receiver schmitt trigger input circuitry threshold voltage varies a little bit, it then would trigger on different levels of the ramp shaped transient. So phase variation occurs, also called jitter. This is one of the things that is happening with optical SPDIF interfaces as they have lower speed / ramp shaped transients. So basically, slow ramp shaped transients are an invitation to jitter.
>Modern major brand power supply regulators, when used correctly, are not bad. They are cheap, produce a quite stable output voltage and by adding decoupling cap's of the right value close to the regulator you end up with a good power supply. Zener diodes are a bit of a problem (dissipation and power supply regulation) as TDA1541A supply currents can go up to 50mA. Bandgap reference diodes (LM336-5.0) are quite good but the maximum allowable current is only 10mA. So for the moment I will keep the existing regulators.
>The current DAC circuit board is designed with decoupling and suppressing interference in mind. Each DAC chip has it's own power supply, Each DAC chip has a separate analog and digital ground connection, tracks are kept very short. Analog main board design has to be done very carefully to get the lowest interference possible. I am also planning to put all digital modules on a separate digital mainboard. I use a dual stabilizing method (cascading regulators) for stable output voltages. For decoupling, tantalium cap's seem to do a good job in filtering out interference, ceramic SMD capacitors take care of the very high frequencies (tantalium cap's are horrible for analog audio applications, but quite good for power supply decoupling, and they remain stable over time). Even with the worst-case test setup I used, the output signal is very clean. The sound is so good, it is very difficult to "switch off the set" once you start listening, so that is one of the things I have been doing the past days, the other "thing" is attached to this post, hope you like.
The upper sketch is the top-view of the octal D-I DAC, showing how the 8 coarse protection and screening plates form a "smooth" half sinewave shape, reflecting the operating principle of the octal D-I dac in the housing design. Text is engraved in the aluminum plates.
The lower sketch is the front view, status indication is done trough orange LED's behind the 4 tubes. When warming-up the LED's flash, if the SPDIF signal is received correctly, the LED's spread a orange glow around the tubes. Using high-gloss stainless steel plates would give wonderful effect.
