Was that inside a DSP xover or an analogue one? I know from my time at a digital xover manufacturer that switching hash from the DSPs is extremely hard to prevent from getting into the analogue stages and I reckon some more attention to RF filtering would have improved its SQ for sure, not that I was involved in the design. I guess I'm just not that convinced that significant RF interference is present inside the earthed metal chassis, in my listening room, and in my house, at the edge of the city near the bush.
Purely an analog one. Have yet to find a DSP-based one to tweak....
These are class 2, unearthed speakers and were often connected to a desktop PC which is class 1. So there's a common-mode noise loop via the transformer inter-winding capacitance which is an RF aggressor.
That's both unfair and untrue. I described the other preamps I'm comparing it to much earlier on. I couldn't be bothered digging through the posts to tell you exactly where but I assure you the details are there. And as for no objectivity, I've gone to great verbal lengths throughout the thread to describe the measures I've taken to be as objective as possible about my assessments on SQ. If you regard listening as pointless and THD and noise measurements as completely characterising audio performance then you are of course entitled to that OPINION, but why keep butting in? In that case you must be convinced I'm making it all up so why waste your time reading it. But I know there are others out there who share my experience or at least are prepared to accept that I am hearing something real and that the extreme "rationalist" standpoint is somewhat simplistic.
Regarding your "suspicions" (which are actually on the record) the Tillbrook Ultra Fidelity discrete preamp is one of them. However I have changed the compensation and bias arrangements to fix an obvious stability problem with the original design, not to mention added in a track that was missing from the PCB in one channel, and this has greatly improved its SQ. I've written about that elsewhere on this forum. The reason I thought try doing this was that I could HEAR there was something not right about the sound. If I thought you were genuinely interested I would happily repeat the details of the others here too. The aspect of SQ of the Self preamp I'm unhappy about applies when comparing it to ANY of the other designs, so I'm not just nitpicking on minor variations of tonal colour or whatever. Besides, what are you proposing to do? Use your memory of the SQ of something you might have listened to 20 years ago to tell me that it's not an acceptable design to use for comparison? That's the kind of thing you seem to be complaining about isn't it?
I don't know who you are but I do know that this is supposed to be a forum for DIY hobbyists. I can understand professional designers getting frustrated by false assumptions and audiophile dogma and feeling the need to comment to help put us back on the "straight and narrow". But even if you are in this category, you're not contributing anything helpful along these lines. To sit there and demand that opinions on SQ are worthless and it is unacceptable to be unhappy about the SQ of a design that has good THD and noise measurements is unbelievably arrogant and extremely discouraging to those of us who enjoy the hobby for the sake of building something to enjoy listening to their music through. According to what you're saying, we're all wasting our time building stuff as we can just go out and buy "perfect" gear by checking the manufacturer's specifications, and if it turns out not to sound great we should adjust our ears because they are not reliable and the specs prove it.
Thanks Bonsai. I'm aware of the issues that apply at much higher frequencies, having designed 8-layer PCBs for digital video mux modems and other RF stuff in a previous job. Not that you will find chips for these applications available in TH packages
Believe it or not I'm not completely clueless about electronics, I just don't consider myself any kind of expert and prefer to try and maintain a little humility and perspective
I'm here to learn and was just curious to see if anyone knew of a sensible reason why there could be a problem at audio frequencies...Purely an analog one. Have yet to find a DSP-based one to tweak....
I have a spare DEQX PDC2.6 that needs tweaking if you're interested.
The 'why' is that such issues can inject low level distortion into the final output sound, which determines whether one setup sounds "better", subjectively, than another - the very thing that one uses to say that one opamp is better than another ...Believe it or not I'm not completely clueless about electronics, I just don't consider myself any kind of expert and prefer to try and maintain a little humility and perspective
Having myself spent years playing with audio gear, attempting to resolve the issues being discussed here, it's a no-brainer to have absolute integrity with every connection ...
Last edited:
Hi Frank
The problem here, as Owdeo has already pointed out, is that PCBs like this are not DIY friendly as regards modifications. Often as not, the copper lands are too small in most PCBs that are more suited to production line assembly. For DIY use, where modifications are a way of life, you need conservatively designed PCBs such as those from Silicon Chip magazine ,where you can play around if you are careful, without the risk of solder suckers etc. devouring fine copper pads, or the need to chop off components to avoid damage to a fragile PCB when removing them.
Kind Regards
Alex
The problem here, as Owdeo has already pointed out, is that PCBs like this are not DIY friendly as regards modifications. Often as not, the copper lands are too small in most PCBs that are more suited to production line assembly. For DIY use, where modifications are a way of life, you need conservatively designed PCBs such as those from Silicon Chip magazine ,where you can play around if you are careful, without the risk of solder suckers etc. devouring fine copper pads, or the need to chop off components to avoid damage to a fragile PCB when removing them.
Kind Regards
Alex
I have a spare DEQX PDC2.6 that needs tweaking if you're interested.
From what I can fathom from a quick Googling, looks like its using sucky S-D DAC chips. So a complete re-work would be in order - probably better to start from scratch with a design conceived from the ground up for subjective SQ rather than measurements
Yes, sure! It has to do with the efficiency of the speakers... but, when you are listening to music 3 meters away from the speakers (I guess it is a fair asumption) and using a 86dB/W/m speaker excited by 4W, it is throwing 92dB SPL (and it is loud) at 1m, so.. perhaps you get about 88dB SPL at the listening position taking into account the "room effect".... and 88 dB SPL is quite loud. If you prefer a louder sound... well... you can send more power to the speaker.
Oh... yes.. that's true, but I don't need 100 dB SPL
... and I don't believe it is a very common SPL level in home listening.
Ciao,
...Yes, I know the dilemma with modifying in tight areas - I've done my fair share of ending up with bits of copper track lifting and fracturing, and parts of the circuit end up looking like the dog's breakfast. However, to truly assess whether a particular configuration is 'superior' to another I've found that the only way to make a final judgement is to restore full integrity every time and let it, and me, fully stabilise before making up my mind.
It's a right pain, and I give up in disgust at times, for a period -- but I've learnt this is the only way I can make decisions that are 'correct' in the long term ...
Cheers,
I was talking about distortion as a whole, not THD neither IMD alone, as they both occur at the same time when listening to a real sound (like music). But the resulting total distortion value is, at least, as high as the highest value.
I can't see how you can give a 1% figure for 'distortion as a whole' as no such measurement exists that I'm aware of.
This is of course the case yet if you read the Belcher article I mentioned you'll see that he mentions when the spectrum is rich in tonal content (as real music is) the IMD dominates over the THD by at least an order of magnitude.
I have read the Belcher article, linked to it several times myself - and I don't recoginze at all the interpertation and weight you seem to give it
it was significant for loudspeakers with % distoriton - the text claims correlation with listening test went away at better than ~ -30 dB "noise fill" numbers
I believe Scott Wurcer posted an example soundcard loopback with 30? multitones that showed little more than the window noise floor anywhere but signal bins - no exotic op amps, not even Class A bias required - very vanilla electronics doesn't show any new behavoir with the Belcher test for audio multitones
it was significant for loudspeakers with % distoriton - the text claims correlation with listening test went away at better than ~ -30 dB "noise fill" numbers
I believe Scott Wurcer posted an example soundcard loopback with 30? multitones that showed little more than the window noise floor anywhere but signal bins - no exotic op amps, not even Class A bias required - very vanilla electronics doesn't show any new behavoir with the Belcher test for audio multitones
Last edited:
I just did a "dirty addition"
of both output results, as it is the only point where I can do an addition (choose among linear, RMS, whatever). As you know, both THD and IMD generation are non-linear processes and I won't try to combine them to get a more correct result , even when they probably are a bit interdependent (hummm....nice task )
The operative word here is 'seem'. The interpretation and weight I am giving it diverges significantly from your perception of that interpretation and weight I conjecture, based on previous interactions we've had over the years.
I'm not proposing the Belcher test myself, just drawing on his understanding of the nature of distortion in real world music signals fed through electronics. What led to the choice of 30 tones? To me this doesn't look representative of a musical stimulus, at least not the kind of music I listen to (orchestral, choral).
The 'why' is that such issues can inject low level distortion into the final output sound, which determines whether one setup sounds "better", subjectively, than another - the very thing that one uses to say that one opamp is better than another ...
Having myself spent years playing with audio gear, attempting to resolve the issues being discussed here, it's a no-brainer to have absolute integrity with every connection ...
Ok, it's not really a technical argument then, just a vague speculation that "low level distortion" is being injected due to supposedly faulty contacts. I can see how with say a power amp's speaker terminal connections a soldered contact might be better than crimp lugs due to the impedances and currents there and a small non-linear contact resistance being a significant proportion of the total.
But with a preamp we are talking ~kR impedances where any slight non-linear joint resistance is a tiny fraction of that of the linear parts. And you haven't even explained why a turned-pin gold plated IC socket will not make good connections to the IC pins.
Are you saying that you're 100% sure that the character of sound of this design and of opamps themselves is purely a result of poor connections? Would the major hassle of unsoldering all those sockets and soldering in the opamps 100% make the SQ of this design equal to that of the others? Surely you can see there's a logical flaw in that argument - swapping opamps produces a different sound but they are in the same sockets! Furthermore, the other opamp preamps I'm comparing it to contain the same dreaded socketed ICs and yet they are ok
From what I can fathom from a quick Googling, looks like its using sucky S-D DAC chips. So a complete re-work would be in order - probably better to start from scratch with a design conceived from the ground up for subjective SQ rather than measurements
Yep, AD1955 from memory. And OPA2134 I-V opamps running from a switching +/-15V supply. And all SMD of course. Good decision
Not 'yucky' that would indeed be going too far, rather 'sucky'. Meaning the noise modulation bothers me subjectively.
Was your reference to 'AD DAC' (rather than the specific part number owdeo cited) to imply that the AD brand was some guarantee of subjective SQ incidentally? They not so long ago discontinued one of the last of their stellar sounding DACs, the AD1865.
Was your reference to 'AD DAC' (rather than the specific part number owdeo cited) to imply that the AD brand was some guarantee of subjective SQ incidentally? They not so long ago discontinued one of the last of their stellar sounding DACs, the AD1865.
Last edited: