When the I/V stage is active then having filtering before the I/V is definitely a good thing for SQ, provided the noise levels are managed well. Adding the filter normally increases the noise. With discrete active I/V the input impedance needs managing too - which is how I came to use a transformer.
Seeing as I've gone back to passive I/V I guess I could try a filter between the I/V resistor and the buffer but then the buffer input resistance needs to be the filter termination... I would need to look for a filter that works into an effectively infinite termination resistance. Hmm, worth exploring a little I think.
Going back to the prototype DAC, I've listened for a while and found it an immensely enjoyable experience, which is what DAC development is all about for me
I was even able to listen to Bartok who for me is normally a turn-off. More to the point my wife has listened - her ears are far better than mine and she notices stuff I don't - and got a double thumbs up from her. She said so far she hasn't noticed any flaws in the SQ. Which is very encouraging.
Seeing as I've gone back to passive I/V I guess I could try a filter between the I/V resistor and the buffer but then the buffer input resistance needs to be the filter termination... I would need to look for a filter that works into an effectively infinite termination resistance. Hmm, worth exploring a little I think.
Going back to the prototype DAC, I've listened for a while and found it an immensely enjoyable experience, which is what DAC development is all about for me
I was even able to listen to Bartok who for me is normally a turn-off. More to the point my wife has listened - her ears are far better than mine and she notices stuff I don't - and got a double thumbs up from her. She said so far she hasn't noticed any flaws in the SQ. Which is very encouraging.
Last edited:
I've so far not played at all with 1541 but with 1387s plain passive I/V was too power supply sensitive for me. Even with 1F of supply capacitance made from hundreds of normal caps (not supercaps as their ESR is too high).
Speaking of the effects of transformers, this guy has it exactly right, mirrors my own experience very closely : Output transformers for DACs
Speaking of the effects of transformers, this guy has it exactly right, mirrors my own experience very closely : Output transformers for DACs
Last edited:
Here's a pic to whet your appetites...
Not sure why, but I really feel like I need this in my life
Love the work Abraxalito.
You ain't alone...
You ain't alone, toots!!Not sure why, but I really feel like I need this in my life
So do I. So do freakin' I!!!!!!!!
TDA 1387 is much more above the tda 1543 according almost everybody... according to me tda1545A was even better than the 1543.
Thorthen loesch who is not building the worst DAC on that planet has two models : one with a standalone tda1387 (I don't know if there was a difference with the Philips' and the ones made after by NXP elswhere... we have a lot of litterature as you know about tda 1541A years and countries where it was produced, but nothing about tda1387 years or origins).
His Flagship stays with a tda1541A (not crowned but from Taiwan while not the latest ones).
It's great that we've got guys like Abrax around whose mucking around with old school stuff.
Just looking at Red Book playback only, I feel that it has not been exploited to the max. Technology is moving ahead very quickly but that does not mean that the new breed of dac's will trounce old school stuff in SQ if properly implemented. My own findings is once you get the I/V & output stage correct, you must revisit the PS & fine tune from there. PS rules & it's not only about clean ripple free mega UF supplies. As for regs, at least for the TDA 1541, I find that they sound better with good old TL431 shunt regs over linear regs. Not sure about the new ultra low noise linear regs though.
Just looking at Red Book playback only, I feel that it has not been exploited to the max. Technology is moving ahead very quickly but that does not mean that the new breed of dac's will trounce old school stuff in SQ if properly implemented. My own findings is once you get the I/V & output stage correct, you must revisit the PS & fine tune from there. PS rules & it's not only about clean ripple free mega UF supplies. As for regs, at least for the TDA 1541, I find that they sound better with good old TL431 shunt regs over linear regs. Not sure about the new ultra low noise linear regs though.
The challenge (redux)
BTW ... I'm extending my gracious offer prev. granted to the DIYA 1387 fanboy community here:
lingDAC - cost effective RBCD multibit DAC design
Here's my test rig:
Contact me via PM.
You must ship the "Gross" unit(s) to me at YOUR OWN EXPENSE.
It will not be returned to you.
The full review will have its own, dedicated DIY thread.
BTW ... I'm extending my gracious offer prev. granted to the DIYA 1387 fanboy community here:
lingDAC - cost effective RBCD multibit DAC design
Here's my test rig:
Contact me via PM.
You must ship the "Gross" unit(s) to me at YOUR OWN EXPENSE.
It will not be returned to you.
The full review will have its own, dedicated DIY thread.
Last edited:
I've been spending the last few days exploring how much distortion the transformers generate - the result is that at low frequencies and high levels there is quite a lot. By 'a lot' I mean up to a few percent (below -30dB). This is fairly normal for transformers - the distortion they generate in the main depends on the flux level in the core, which gets to be the highest at the lowest frequency and the highest level (because flux is the integral of voltage). In my design it turns out this isn't the primary source of distortion, this took a fair bit of head-scratching to figure out though with 20/20 hindsight its obvious. There's distortion because the transformer (here I'm talking about the one between DAC and filter) isn't feeding into a very low impedance, rather it sees the I/V resistor at low frequencies and that's 5k6. So one solution is to reduce the I/V resistor - this does in practice work well but means changing the transformer ratio or alternatively adding more DACs to bring the output level back up again.
I've decided to change the transformers so that the design does still work with just a single 36 DAC board - this has put a delay in the project. The distortion isn't as bad in practice as the figures make it look for various reasons - the main one being that a 20Hz sine at full level is extremely unusual in real-world music - distortion drops as level goes down and also drops as the frequency rises. So bass is the most 'at risk' from the peturbations of transformers. Those who fear distortion most can reduce it by adding more DAC boards and swapping out the EP17s for something bigger - I've been experimenting with EE35s for this and I guess will report back on my findings after Xmas.
I've decided to change the transformers so that the design does still work with just a single 36 DAC board - this has put a delay in the project. The distortion isn't as bad in practice as the figures make it look for various reasons - the main one being that a 20Hz sine at full level is extremely unusual in real-world music - distortion drops as level goes down and also drops as the frequency rises. So bass is the most 'at risk' from the peturbations of transformers. Those who fear distortion most can reduce it by adding more DAC boards and swapping out the EP17s for something bigger - I've been experimenting with EE35s for this and I guess will report back on my findings after Xmas.
Hello,
Cool...
What if the dac board was connected to a pre instead a headphone, please ?
Could we accept not to glue to the 2 V or more at the output, instead accept a 0,7V to 1V and let the volume pot of the pre working in a more confortable zone towards twelve O'Clock ? I tried it after an friend"s advice on the tda 1541 dac. I had to reduce the I/V resistor value from 1/3. Worked great. According John from Ecdesigns (did you ask him finally?) who tested several dac chips to increase the current outputt with tda1543 and tda1541, the less dac chips you have the best the sound is. As I wrote he finally came back to a standalone dac chip to reduce all the problems such layouts give.
About c-core PS transformers, are there cheap options at Tobao or the chineese vendors ypou source in native language for good local prices ? Have you tested it, are they good enough factured ?
Last question please, did you measure the electricity qt outputt of the wall where you live ? Are there great variations ? I'm always estonished to see great differences with countries and electricity networks....
Great X-Mass and Sorgho liquors
Cool...
What if the dac board was connected to a pre instead a headphone, please ?
Could we accept not to glue to the 2 V or more at the output, instead accept a 0,7V to 1V and let the volume pot of the pre working in a more confortable zone towards twelve O'Clock ? I tried it after an friend"s advice on the tda 1541 dac. I had to reduce the I/V resistor value from 1/3. Worked great. According John from Ecdesigns (did you ask him finally?) who tested several dac chips to increase the current outputt with tda1543 and tda1541, the less dac chips you have the best the sound is. As I wrote he finally came back to a standalone dac chip to reduce all the problems such layouts give.
About c-core PS transformers, are there cheap options at Tobao or the chineese vendors ypou source in native language for good local prices ? Have you tested it, are they good enough factured ?
Last question please, did you measure the electricity qt outputt of the wall where you live ? Are there great variations ? I'm always estonished to see great differences with countries and electricity networks....
Great X-Mass and Sorgho liquors
I'm not following along with your questions @diyiggy. You can have a lower output level simply by specifying a different output transformer. I was indeed thinking of having only 1VRMS output instead of 2VRMS - I figure 2V is a bit excessive but 1V doesn't quite do it for my AKG240s.
No I haven't asked ECD John anything - your response to my question for clarification left me lacking curiosity. As for the number of DAC chips, it all depends on how the design is using them. Try telling Stavros (designer of the well regarded Aries Cerat range) that he'd do better to ditch his multiple chips in favour of a single one. See what he says.
C-cores I haven't tried, I did get some R-cores made many years ago. I haven't monitored the mains voltage so can't answer about variations.
No I haven't asked ECD John anything - your response to my question for clarification left me lacking curiosity. As for the number of DAC chips, it all depends on how the design is using them. Try telling Stavros (designer of the well regarded Aries Cerat range) that he'd do better to ditch his multiple chips in favour of a single one. See what he says.
C-cores I haven't tried, I did get some R-cores made many years ago. I haven't monitored the mains voltage so can't answer about variations.
Don't wanted to bother you. So... sorry.
Hello,
Uh...wow !Hey, I'm not trolling !
Sorry not to have answered your question about the precise page where John talked about that issue. Indeed if you answer a question with a question and say after it belongs to me...well I will not go for a dialectic competition, have no time for that and it's not the purpose. You perfectly understood that I said lowering the voltage was to avoid more chips and too complex layout and not lowering voltage for just lowering the voltage. As for the traffo, yes I imagine you can change the ratio to increase the output voltage but at the cost of some other problems or huge parts.
OK it's mostly a headphone dac... I understand and respect that, it's indeed a limitation in the freedom to design the output stage.
Finally sorry my curiosity makes yours smaller... I was about asking you if you prefered c core design vs other styles for a simple design (I mean no isolation and or symetryseur voltage PS traffos)...but I will not. From your answer you picked what you had on hands and I have nothing to say against that.
Btw I'm not a fan of Aries Dacs... So my understanding is here it's about multype grossly dac chips and headphone dacs and that we can't discuss that blocked choice and the goal is to find tips around that concept to cunterbalance all of that (from the digital inputt to the dac chips and layoutt) in the outtput stage...
which is sending me after the answer in the territory of no curiosity anymore as well. Each time I asked you and made some inputs it seems you was not interrested, my only humble goal was always about to acheive the best sound around your good ideas and you foccusse to the outputt stage which seemed to me the real purpose around your filter primart idea, but who am I to tell it's greener there to a green duck .
I will just lurk as I stay open, at what it seems to me to be now a walk longer with chains thread (no offense nore attack, just a little chocked about your post which dictated mg
Y certainly too fast answer).
Anyway cheer, it's Xmaths yet, it's me in the kitchen this year (no inputt on that please, lol)
Hello,
Uh...wow !Hey, I'm not trolling !
Sorry not to have answered your question about the precise page where John talked about that issue. Indeed if you answer a question with a question and say after it belongs to me...well I will not go for a dialectic competition, have no time for that and it's not the purpose. You perfectly understood that I said lowering the voltage was to avoid more chips and too complex layout and not lowering voltage for just lowering the voltage. As for the traffo, yes I imagine you can change the ratio to increase the output voltage but at the cost of some other problems or huge parts.
OK it's mostly a headphone dac... I understand and respect that, it's indeed a limitation in the freedom to design the output stage.
Finally sorry my curiosity makes yours smaller... I was about asking you if you prefered c core design vs other styles for a simple design (I mean no isolation and or symetryseur voltage PS traffos)...but I will not. From your answer you picked what you had on hands and I have nothing to say against that.
Btw I'm not a fan of Aries Dacs... So my understanding is here it's about multype grossly dac chips and headphone dacs and that we can't discuss that blocked choice and the goal is to find tips around that concept to cunterbalance all of that (from the digital inputt to the dac chips and layoutt) in the outtput stage...
which is sending me after the answer in the territory of no curiosity anymore as well. Each time I asked you and made some inputs it seems you was not interrested, my only humble goal was always about to acheive the best sound around your good ideas and you foccusse to the outputt stage which seemed to me the real purpose around your filter primart idea, but who am I to tell it's greener there to a green duck
.I will just lurk as I stay open, at what it seems to me to be now a walk longer with chains thread (no offense nore attack, just a little chocked about your post which dictated mg
Y certainly too fast answer).
Anyway cheer, it's Xmaths yet, it's me in the kitchen this year (no inputt on that please, lol)