Pushing the limits of TDA1543 NOS DAC - Page 12 - diyAudio
Go Back   Home > Forums > Commercial Sector > Vendor Forums > Audio Sector

Audio Sector Kits & PC boards from AudioSector

Reply
 
Thread Tools Search this Thread
Old 16th July 2011, 05:29 PM   #111
Bobken is offline Bobken  United Kingdom
diyAudio Member
 
Join Date: Dec 2002
Location: UK
Back in posts # 103 & 106 I mentioned the likely current requirements for the 3 shunt regs in this circuit, and in both cases as I clearly stated that this was *from memory*.

For anyone who wishes to depart from Peter's schematic and modify any values of components or adopt different active devices for themselves in these regs, my comments to re-calculate the current requirements and losses, should ideally be followed, of course.

However, I have just found the measurements I took on the prototype regs and using the values of resistors shown on the schematics, the 8+v reg actually supplied around 115mA on test, and the 5v reg supplied just over 50mA, for anyone who is interested.

Although not to the drastic extent of the 200% variation which I found with the alternative DN2540 devices, it must be remembered that due to component tolerances with BSP129s these currents could have been higher or lower with a similar reg built exactly the same and using the same value of resistor between gate & source.

In both cases, the measured currents are higher than the circuit demands which is not harmful except for creating excessive heat which generally is undesirable, but at least the resistor values shown should cover for any BSP129 devices which come out on the lower-end of the scale, so it should be fail-safe to adopt these values.

Regards,
__________________
Bob
  Reply With Quote
Old 24th July 2011, 05:45 PM   #112
slawney is offline slawney  Germany
diyAudio Member
 
Join Date: Aug 2003
Location: Frankfurt am Main
Thanks Peter and Bobken for publishing the fruits of your labors and the justifications behind them. At one point, Bobken, you said you had implemented some popular audiophile regulators in the design. Could you say which regulators you attempted (ALWSR, audiocom Sulzer, Burson, etc.?) in the early stages of your extensive research and what made them unsatisfactory from your perspective? After looking at your final circuit, I would expect, without having tried it yet, that the choice of all the components (active and passive) will probably play as much a role in the final sound as the topology itself. Just one other question, though: what is the measured output impedance of the final regulator you decided on, and how does that output impedance vary according to frequency?
  Reply With Quote
Old 25th July 2011, 11:34 AM   #113
Bobken is offline Bobken  United Kingdom
diyAudio Member
 
Join Date: Dec 2002
Location: UK
Quote:
Originally Posted by slawney View Post
Thanks Peter and Bobken for publishing the fruits of your labors and the justifications behind them. At one point, Bobken, you said you had implemented some popular audiophile regulators in the design. Could you say which regulators you attempted (ALWSR, audiocom Sulzer, Burson, etc.?) in the early stages of your extensive research and what made them unsatisfactory from your perspective? After looking at your final circuit, I would expect, without having tried it yet, that the choice of all the components (active and passive) will probably play as much a role in the final sound as the topology itself. Just one other question, though: what is the measured output impedance of the final regulator you decided on, and how does that output impedance vary according to frequency?
Hi slawney,

I fear that this response may not seem very helpful to you, but from the style of your questioning I think that you must have missed the essence of and the rationale behind my experiments with this NOS DAC design, although I believed that I had said enough before to make this quite clear.
I saw this as a fun exercise initially to occupy me in retirement and to discover for myself 'what all the fuss had been about' with these particular types of circuits which had been the subject of many years of development/refinement work by hundreds of other enthusiasts (like yourself), long before I came on the scene.

My initial expectations were low that the subjective results would be anywhere near to what I had enjoyed for a long while with analogue equipment, but knowing that Peter would not normally waste his time on a complete 'dog' of a result, and seeing the excellent review of his commercial version, I gave it a go. I also stated elsewhere that if I had seen the appalling measured results with sine waves at all frequencies (IIRC) above around 5kHz before I started playing-around with this circuit, I would doubtless have never even bothered! The puzzling matter for me was that I couldn't believe that the sonic results could still be so good subjectively, even though I could readily hear some shortcomings which I hoped I might perhaps improve upon.

Having discovered this shortly after the intitial construction, I paid very little attention to any later measurements although in the case of any new building-blocks like discrete regs (naturally before risking using them in the DAC circuit) I checked with a 40Meg digital storage 'scope that there was no obvious sign of instability and that the output was not unreasonable for ripple & noise etc. With my 'best' (so far!) empirical result which you refer to I have also already explained that this is not a sophisticated reg design and it is one which I would generally personally avoid in any analogue circuitry, especially where the current being drawn also fluctuated - unlike the situation with this DAC. Bearing in mind the massively distorted results anyway due mainly to the two (remaining) active components (also being used outside of the maker's original intentions) it would seem to me to be fatuous to even consider such matters as the various output impedances of the regs I tried, and whether they might rise with frequency or not. In a digital circuit like this with a steady current drawn in all three locations and with all of the enforced shortcomings, output impedances of regs was not uppermost in my thoughts, and I didn't make any attempt to measure or compare any of these in this manner.

It simply happened on trials, in exactly the same way back in 2004 you said on this Forum that you preferred Caddock TF020 resistors for the I/V resistors in those days, that I liked the subjective results with these simple shunt regs, and when Peter also tried them, he did too. I won't name the other regs/designs specifically because taken out of context this can provide the wrong impression to members about their individual performances. In the same way you have discovered in the past that certain components or building-blocks just appear to have some synergy or whatever in certain locations in certain circuits, I found this design to be subjectively the finest so far in this DAC. As I do with all of these experiments when concentrating on a specific region of the circuit, I continued to try alternative ideas until I was reasonably satisfied with the results, although I always re-check such fundamental decisons again after further changes elsewhere, and sometimes I change my previous choices in the light of this. As already mentioned, to me this is always an iterative process when after improvements in resolution etc achieved elsewhere in the circuit, it is frequently clear that an earlier choice may not still be the best option in another region which had been subject to 'optimisation' earlier.

I don't recall the individual sonic 'characteristics' I noticed at the time (probably around 3+yrs ago) with the two commercial reg designs I tried, and I don't have the inclination to look back through some 300 pages of notes to remind myself of those right now. I was simply not fully-satisfied at the time with what I heard then, so I moved on to another design.

Unfortunately, I have no practical experience of the ALWSR reg although I did see the schematic a long while ago, and IIRC this was rather complex and probably required (nearly?) as many components for a single reg as my entire stripped-down version of the DAC, before going over to trying discrete reg designs. Therefore adding the additional complexity for three of these particular regs was going entirely contrary to my wishes here, and it was not an attractive proposition for this circuit. I have little doubts that in a more-demanding application the ALWSR reg measures and also sounds extremely well from what little I have seen, but interestingly Peter mentioned earlier that when he tried this particular reg version he was not very enchanted with those results.

Finally, you are quite right in your assumption over the actual components dominating the overall sonic results with these simple reg designs, as I have already mentioned earlier-on in this thread. This is a position which I am more happy with in any 'subjective' development exercises like this has been, because with fewer components involved it is easier to 'optimise' these through extensive listening-trials which often take-up an inordinate amount of time and can be very taxing on occasions.

Regards,
__________________
Bob
  Reply With Quote
Old 25th July 2011, 11:35 AM   #114
diyAudio Member
 
Join Date: Jan 2006
Peter>

Do you think that 1uF output cap is enough with 22k input imedance Gainclone to get satisfactory bass extension?

Last edited by a_men; 25th July 2011 at 11:36 AM. Reason: mistake
  Reply With Quote
Old 25th July 2011, 12:10 PM   #115
Bobken is offline Bobken  United Kingdom
diyAudio Member
 
Join Date: Dec 2002
Location: UK
Quote:
Originally Posted by a_men View Post
Peter>

Do you think that 1uF output cap is enough with 22k input imedance Gainclone to get satisfactory bass extension?
Hi,

As I was already looking at this thread after posting another response, perhaps I can respond and save Peter some time.

"Enough" is a non-quantifiable expression here I think, but I doubt that you will experience severe bass curtailment problems with this arrangement.

At one time and because of the excessive costs of large-value Teflon coupling caps, I tried an experiment with a 1uF cap with a 'following' input impedance of 10k and the result wasn't as bad as I had assumed before I tried this out.

If my back-of-envelope calculation is correct (and no doubt someone will quickly correct this, if not!) the half-power (i.e. -3dB) roll-off will commence at around 7Hz. This is not too shabby, although in analogue circuitry I have always strived to get this down to 1Hz or lower because of possible phase-shift issues. The phase-shift commences at a decade higher than the -3dB point, but whether you will find this audible or not is uncertain in a circuit like this where other potential 'problems' may dominate the sonic results.

What I may suggest if you are also concerned about costs of a fancy and costly cap to be used here, what about trying a cheap plastic film cap of this value initially and (if you can, measure the results, but more importantly) listen to what the sound is like. Then if you are happy, you can decide on the costly cap you have in mind.

The overall resolution with a substitute cap will almost certainly not be the same, of course, but it should give you a reasonable idea if the bass is rolled-off excessively for your requirements.

I hope that this helps.

Regards,
__________________
Bob
  Reply With Quote
Old 25th July 2011, 12:37 PM   #116
slawney is offline slawney  Germany
diyAudio Member
 
Join Date: Aug 2003
Location: Frankfurt am Main
Bobken, Thanks for your patient response. It is really quite a pleasure to see you posting once again after such a lengthy experimental and development phase. I know, from personal experience, this phase can go on for years, and ultimately the only person one is trying to please is oneself. Please understand that I am not a defender of the ALWSR or other commercial discrete designs, but I am trying to understand the circuit that you finally arrived at from a technical standpoint, which, given the technical imperfections of the TDA 1543 (you are entirely correct about the impossibility of reproducing a sine wave above 5 kHz with this chip in this NOS circuit), is perhaps not the right approach. It is true that the advantages of a low or stable output impedance will not matter as much for this circuit (I am not saying it is completely meaningless), given its relatively constant current draw. For the last few years, my work with the TDA 1543 DAC has been focussed on Peter's USB DAC for my computer and his standard gold DAC--both of which I eventually modified, sometimes in line with your own path with regard to the PLL, but mostly in my own way, starting from the circuit as Peter suggested, much of which remains the same (the Caddock I/V resistors have always stayed), some of which is different, mainly in the area of PSU and regulator topology. Using a separate transformer/smoothing circuit/regulator (discrete and compatible with the DACs needs and addressing whatever problems exist in the mains supply and surrounding components and environment) for each power pin is certainly one of the best ways to enhance the resolution and clarity of the design, and I am quite happy to see someone pursuing this path with such dedication and patience.
Given that you depend so heavily on listening impressions, could I just ask what type of music you listen to and what are the most important aspects of sound quality for you? You indicated that your listening habits are a little different from Peter's but that this difference was not an obstacle for both of you preferring ultimately the same final configuration.
  Reply With Quote
Old 25th July 2011, 12:39 PM   #117
diyAudio Member
 
Join Date: Jan 2006
>Bobken

Thank you for quick reply.
When I wrote "enough" I ment : I don't want thin sound with exposed details caused by too small output cap.
I don't wan't lost of detail and clarity caused by too big cap as well.

Local shop has two values of tinfoil caps 1uF and 2.7uF and I'm thinking about the best compromise.

My main amp is 2a3 Push-pull with 50K input impedance, but I'm going to build GC as a backup amp.

Now I think I was precise enough

Regards,
A
  Reply With Quote
Old 25th July 2011, 02:37 PM   #118
Bobken is offline Bobken  United Kingdom
diyAudio Member
 
Join Date: Dec 2002
Location: UK
Quote:
Originally Posted by slawney View Post
Bobken, Thanks for your patient response. It is really quite a pleasure to see you posting once again after such a lengthy experimental and development phase. I know, from personal experience, this phase can go on for years, and ultimately the only person one is trying to please is oneself. Please understand that I am not a defender of the ALWSR or other commercial discrete designs, but I am trying to understand the circuit that you finally arrived at from a technical standpoint, which, given the technical imperfections of the TDA 1543 (you are entirely correct about the impossibility of reproducing a sine wave above 5 kHz with this chip in this NOS circuit), is perhaps not the right approach. It is true that the advantages of a low or stable output impedance will not matter as much for this circuit (I am not saying it is completely meaningless), given its relatively constant current draw. For the last few years, my work with the TDA 1543 DAC has been focussed on Peter's USB DAC for my computer and his standard gold DAC--both of which I eventually modified, sometimes in line with your own path with regard to the PLL, but mostly in my own way, starting from the circuit as Peter suggested, much of which remains the same (the Caddock I/V resistors have always stayed), some of which is different, mainly in the area of PSU and regulator topology. Using a separate transformer/smoothing circuit/regulator (discrete and compatible with the DACs needs and addressing whatever problems exist in the mains supply and surrounding components and environment) for each power pin is certainly one of the best ways to enhance the resolution and clarity of the design, and I am quite happy to see someone pursuing this path with such dedication and patience.
Given that you depend so heavily on listening impressions, could I just ask what type of music you listen to and what are the most important aspects of sound quality for you? You indicated that your listening habits are a little different from Peter's but that this difference was not an obstacle for both of you preferring ultimately the same final configuration.
Hi slawney,

Thanks for your kind words and encouragement.

Ever since I commenced modifying audio circuits over 40 yrs ago, my goal has been to produce completed results which are as 'universal' as possible, and by that I mean that the individual equipment (DAC, in this case) should stand-alone sonically and not exhibit any obvious (untoward) characteristics. I don't like the idea which I have come across before where a particular front-end (for example) needs to be matched to say a particular-sounding power amp (or maybe speakers), or the overall sonic balance across the audio spectrum etc is not satisfactory.

Others whom I know have gone down this route because it was easier and less work involved, always hit the same problems sooner-or-later when they wish to substitute a new piece of equipment in their systems, and the sonic 'balance' is then unsatisfactory.

Havng been involved a while ago with a commercial project I did have the ideal opportunity to ensure that in the main I succeeded with this goal, and whatever other equipment I tried with my own gear seemed to work quite well together.

Unfortunately with this DAC circuit I have found this to be extraordinarily hard especially in the HF regions as already discussed, but I do have an acceptable sonic balance (most of the time!) nowadays. Of course, this is a lot to do with the choice of avoiding filters etc, but whatever I tried (and like quite a few other 'experimenters' have also found) nothing improved the overall subjective results without some other downsides.

This almost 'initiation' into seriously attempting to improve upon a lot of others' work before me and which had achieved quite surprisingly-good results already, meant that I had little choice in my mind other than to forget about measurements and simply concentrate on what I actually heard.

What amazed me in this case is that frequently when dealing with all-digital signals, one can modify the perceived results in many instances in a similar manner to analogue circuits. I can still see no possible reason for this since we are not talking about varying-frequency signals in the main, but nevertheless this is what I began to hear!

I have long since stopped losing sleep over this awkward phenomenon, but I just take advantage of these effects and try to make the best of any characteristics noticed. There are no 'absolutes' or right or wrong choices in my experience since I have come across hundreds of circuit 'variations' which all sound much better than they have any right to do, but just once-in-a while a particular combination of components sounds quite 'magical' in a particular parameter, so I note this and try not to lose this benefit.

Several years ago I advised Peter that I liked the 'sounds' with using naked Vishays for I/V duty, and explained the benefits I heard and went on to suggest that I would try the same type of resistor for the V/Ref region. He 'countered' by asking if I thought that the input shunt resistor might also benefit from being changed to a TX, to which I responded that I thought this to be very unlikely as we already used HQ Caddocks here and we were merely dealing with an all-digital soundstream after all. Well, I couldnt have been more wrong as we both discovered for ourselves on later trials!

I actually like the effect of adding caps across the V/Ref resistor and for a while I have been using a 0.1uF BG here, although I don't think that Peter is so enthusiastic about this addition. I spent nearly a year trying alternative constant-current sources for V-Ref and sometimes this was very impressive, but there are problems with setting currents as accurately as I prefer (for other sonic reasons) compared with resistors which can be paralleled so easily, and this went against the philosophy/theory that a high impedance to ground from pin 7 was desirable. I also tried many CRDs and fixed-resistor combinations in series to set the currents more accurately (I won't use any presets although the 26-turn Vishay 1280G bulk-foil series are the best I have used) and in the end the practical difficulties I encountered encouraged me to revert to a simple resistor.

Overall, I still prefer a TX2352 (or TX2575) here with this added BG cap, and that is how it is presently, although if I felt obliged to revert to a Caddock TF020 for V/Ref I might well prefer to avoid this bypass cap. Other resistors including alternative types of bulk-foil Vishays, are simply not as good and transparent here, unfortunately, as these naked types.

For me, the large Caddock TF020s are excellent but they have a signature which I prefer to avoid in that they fizz up the HF a little compared with a straight wire, and they also push-forward the presence-region a bit, which is why they sound more 'impressive', and they add a fraction of grain and mist to the results. However, as mentioned to Peter many times before, if my final results chosen for maximising other parameters like soundstage, focus, dynamics etc turned-out to be a tad on the bland side, I would have no hesitation in reverting to some Caddocks for I/V or maybe V-ref, because I know that this will spice-up the overall results a little and this might be a better overall 'compromise'.

As to my listening tastes, apart from head-banging type of music which I cannot handle, I don't really have many hard-and-fast rules over what I listen to during tests, but I generally try to avoid anything which I enjoy a lot!

Far too many times at the end of the track, and often an entire CD, I realise that I have become so engrossed in the sounds that I didn't do what I had set out to as a result of the enjoyment distracting me, so I needed to repeat the trial again! I do have several favourite tracks and some special test CDs for checking certain sonic effects specifically, but there are many of my CDs which I have listened-to so many thousand times that I know every squeak and pop etc on them, so I use these for individual checks after a modification when I believe that a parameter may have changed.

I actually try to listen to every parameter involved in the results but there have been many times when I have got through many dozens of tracks believing that the results are quite superb, only to discover disappointingly that say some castanets are not as crisp as before on a track which has these recorded on it, usually suggesting a HF issue which affects the 'rise-time' of the sounds. This means more work to attempt to recover this, or sometimes the need to abandon this particular latest combination because I am not happy to lose this benefit which I have enjoyed before.

Regards,
__________________
Bob
  Reply With Quote
Old 25th July 2011, 03:08 PM   #119
Bobken is offline Bobken  United Kingdom
diyAudio Member
 
Join Date: Dec 2002
Location: UK
Quote:
Originally Posted by a_men View Post
>Bobken

Thank you for quick reply.
When I wrote "enough" I ment : I don't want thin sound with exposed details caused by too small output cap.
I don't wan't lost of detail and clarity caused by too big cap as well.

Local shop has two values of tinfoil caps 1uF and 2.7uF and I'm thinking about the best compromise.

My main amp is 2a3 Push-pull with 50K input impedance, but I'm going to build GC as a backup amp.

Now I think I was precise enough

Regards,
A
Hi,

Please don't think that I was intentionally criticising the "precision" in your question before because I fully-understood what you had said then.

However, this is a question which I cannot properly answer for you except to do what I already did in telling you how to calculate the results theoretically, and only you can decide whether or not a phase-shift matters in your set-up, for example.

As some plastic film caps 'are as cheap as chips', I also suggested a way in which you could fairly-cheaply resolve the issue to your satisfaction empirically, so I am at a loss to know what else I can add here.

I said before that I did not wish to solve other's (off-topic) design issues for them and I have done my best to advise you how this can be easily and cheaply achieved. Also, this is DIY where sometimes one has do some of the work for yourself.
Perhaps Peter will add something to this query which you put to him anyway, but unless you cannot afford the higher-value cap out of the two you mention, it seems fail-safe to me to go for the larger value cap
in this instance, unless you are prepared to experiment for yourself.

There is little question in my mind that using the larger value should ensure that the LF regions should be to your liking, and if you feel that for whatever reason there is any loss at extreme HF with a larger cap, this can be ameliorated with a smaller bypass of course. Alternatively, if you do the trial that I suggested with a lower-value cheap cap to assess the LF response and this is to your satisfaction, then you should be confident that the HF should not cause you (so) many problems. In any case I think that I said before that with 1uF caps of the finest type I am aware-of (i.e. the very costly Teflon V-Caps) I still prefer to add a smaller bypass for the ultimate in sonics at extreme HF, although the maker doesn't actually recommend this approach as far as I am aware.

Regards,
__________________
Bob

Last edited by Bobken; 25th July 2011 at 03:10 PM. Reason: spelling of aware!
  Reply With Quote
Old 28th July 2011, 11:51 AM   #120
slawney is offline slawney  Germany
diyAudio Member
 
Join Date: Aug 2003
Location: Frankfurt am Main
Thank you Bobken. I also found that the Vishay TX2352 is indeed more natural sounding than the Caddock TF020, avoiding the presence bump around 5KHz and the treble haze and fizz created by TF020. When supplying TDA 1543 with 5V, however, I still prefer TF020 over TX2352 as the dynamics get bland with the TX2352--my only criticism of the TX2352 actually, besides the price and fragility.
  Reply With Quote

Reply


Hide this!Advertise here!
Thread Tools Search this Thread
Search this Thread:

Advanced Search

Posting Rules
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is On
HTML code is Off
Trackbacks are Off
Pingbacks are Off
Refbacks are Off


Similar Threads
Thread Thread Starter Forum Replies Last Post
FS: NOS TDA1543 DAC 2A3SET Swap Meet 0 24th March 2011 07:10 PM
Pushing Aleph P to limits - Passive Tri-amp Alves Pass Labs 4 12th September 2006 01:56 AM
TDA1543 and CS8412 NOS DAC Igla Digital Source 3 7th March 2006 07:27 PM
Nos Tda1543 Dac. IpsilonSound Digital Source 0 4th June 2004 06:25 PM
TDA1543 nos dac question... setmenu Digital Source 13 2nd June 2004 12:03 PM


New To Site? Need Help?

All times are GMT. The time now is 02:01 PM.


vBulletin Optimisation provided by vB Optimise (Pro) - vBulletin Mods & Addons Copyright © 2014 DragonByte Technologies Ltd.
Copyright 1999-2014 diyAudio

Content Relevant URLs by vBSEO 3.3.2