Abraxalito, have you ever used an 'under-engineered' telescope? Heck, I have one in my closet! Tasco makes them by the thousands, YET the first Tasco that I owned, back in the mid '50's EXCEEDED its predicted performance. That, I presume, was initial 'over-engineering' or at least 'optimum-engineering' in order to get it noticed and bought by Americans. Later, the 'bean-counters' took over, I would presume that from my experience with these telescopes over the decades.
Last edited:
Perhaps you missed my earlier comment John in your rush to defend yourself. I said its not so easy to classify an individual product as 'over-engineered' without knowing the original constraints. If you go over them on here for the Blowtorch I might be able to help out.
Well - engineering is about optimisation, not compromise. From reading your posts over a number of years it seems jolly clear you do know what sounds good, and reviews of your products confirm this. But you do appear weak (and I've remarked on this in the past) on knowing why certain things sound good - without the 'whys' then optimisation is going to be a bit hit-and-miss. The solution adopted by many audiophile designers is what might be called the 'blunderbuss' approach - cover all bases and hang the cost.
Cost reduction is the accountants' stock in trade, not the engineers'.
I gave it here many times.
There are 2 criteria: voltage sag, and overheating. Calculate 2 transformers, according to both requirements, and select which transformer is bigger. Designing class A amp you can't calculate "undersized" transformer, it is one of the main reasons why people prefer to buy class A amps.
One secret of Wavebourn: for class AB it has to be calculated according to peak power consumption.
Secret of Japanese manufacturers: select "nominal" power (guess it in respect to peak power, according to the music your target auditory listens), and calculate the power supply for such requirements.
Hence the myth about "oversizing": size is always small. But it is not true for all amps. It true for Japanese mass production class AB amps only.
No reason to use transformers bigger than needed.
Sounds like a typical snobbish audiophile mantra. Care to clarify why you think most mid-fi is good enough? I find it sucks in terms of sound quality for money myelf.
Irrelevant question, John Curl. My Pyramid amps, for example, have power transformer designed for 360W power; output power is limited by optical compressors on 80W per channel, and it is the same on 4, 8, and 16 Ohm outputs. Difference in taps of secondaries of output transformers. Power transformer, as you understand, is one and the same. However, taps on secondaries is suboptimal decision. I would prefer to switch layers of secondary properly, but I did not design output transformers. I took for prototypes what was available.
Back to my comment about poor engineering: if it is truly the example suggested, of jitter and error rates being compromised audibly by poor power supply stability and noise (if I am recalling correctly) then I maintain that it is poor engineering --- or at the very least a failure of value-oriented engineering to put the expenditures on materials where they matter the most, or perhaps a failure/breakdown of negotiations between engineering and marketing.
One of my lessons-learned is to push back on supposed absolutes handed down by the suits, a subset of which is do not yield to temptation to make estimates before you're sure what a design target will require. Of course if you go overboard to make sure you're never going to underestimate, you may quash the whole program (or move the powers-that-be to get rid of you in favor of someone who is willing to give them the "answer" they want). This is a risk you have to take.
One of my lessons-learned is to push back on supposed absolutes handed down by the suits, a subset of which is do not yield to temptation to make estimates before you're sure what a design target will require. Of course if you go overboard to make sure you're never going to underestimate, you may quash the whole program (or move the powers-that-be to get rid of you in favor of someone who is willing to give them the "answer" they want). This is a risk you have to take.
Concur - the suits are by definition not knowledgeable about the whys of engineering decisions - they just go on the bottom line, which to them is one of the 'absolutes'. To do real optimisation and hence real engineering there can't be any absolutes, rather (somewhat elastic) constraints.
In my way of thinking, a 'suit' is by definition a person who knows the cost of almost everything but the value of practically nothing.
No, the jitter and error rates are not the culprit, they are just used by many as an easy thing to point the finger at, for why there are audible differences. I would say that the chain of events is that the DAC chip or circuitry, and the following analogue components are influenced by the supply stability and noise. Even if the analogue side is offloaded to a separate box, it is still tied into the bigger scheme of things via the power cord and digital link. And perhaps some RFI ...
Frank
If we're talking about the raw error rates (coming out of the data slicer from the photodiode) then for that to be zero is a sign of over-engineering because there are error correcting codes built in. So I'd hope the PSU wasn't engineered for such low noise or stability that the error rate went to zero.
bcarso,
What you've said, applies probably to 95% of the CD/DVD/SACD players on the market, maybe even more.
Take a look at US Patent #5329556; "Reproduction Equipment for Digital Audio", Meitner et al. - it contains some interesting reading on this subject.
This is the great big black hole of audio: certain "things" make the sound better, and no-one really has a good handle on why. Really. The theories and understanding are trailing the practicalities of getting good sound, badly, so the next best thing, in the meantime, is to just get on with getting decent results.
It's driven me mad - hah! - over the years, trying to get a comprehensive handle on such things, the 'why's; there are just too many factors "intermodulating" to make it a simple process of just having a checklist of cost effective techniques.
So in that sense, go the 'blunderbuss' technique: it works, and the refinement in understanding can slowly grow, after the fact, over time ...
Frank
Thanks, elektroj, for the Meitner ref ( a very old one by one of the true audio originals =] ).
I think my point is: decent power supplies are not that hard to do and not that expensive. Yes, you do have know what you are doing. But to imagine that this is so tough that we labor under onerous cost constraints, lacking which we would have glorious error-free ( or sufficiently error-free that we shouldn't need error correction strategies) data streams in low-jitter processions --- it just seems more a matter of competence than a matter of spending bucks.
And yes, tossing down a few LM317s may be inadequate. So maybe it isn't all that easy. But neither do I believe an oversized toroid (or whatever) is required.
I think my point is: decent power supplies are not that hard to do and not that expensive. Yes, you do have know what you are doing. But to imagine that this is so tough that we labor under onerous cost constraints, lacking which we would have glorious error-free ( or sufficiently error-free that we shouldn't need error correction strategies) data streams in low-jitter processions --- it just seems more a matter of competence than a matter of spending bucks.
And yes, tossing down a few LM317s may be inadequate. So maybe it isn't all that easy. But neither do I believe an oversized toroid (or whatever) is required.
This might sound somewhat controversial but here goes anyway - I think it suits the established players to keep the situation thus. What does strike me as slightly odd is the apparent lack of curiosity of some of the designers. By maintaining its arcane they can use the mystique factor in their marketing blurb. If there's science behind it I think they fear the 'romance' and 'aura' of the high-end might get lost.
There's more though - once it becomes science then it'll get commoditized (i.e. published) and anyone can have a go at making decent sounding kit. That'll most certainly eat into the established players' profit margins n'est ce pas?
- Status
- Not open for further replies.