Here is what I keep coming to for a "reasonable" amp. I don't claim anything greatly insightful, just that the SPICE testing of many combinations keeps coming to about this. Before I build something like this, I need to put the effort into the selection of driver and outputs. 2sc5200/2sa1934 outputs maybe? I don't claim I have sufficient understanding to select the perfect choice in standing currents yet. Clearly, no effort at all in protecting over-current or clipping recovery. This was just to understand the amplifier function and how it related to the discussion in this thread about why some amps ( Rotel 9 series) sounded better no my then less than desirable speakers, and why when I build much better speakers the much better amps (Parasounds) then sounded much better. Then when I modified my DH120 with what I first learned, it was most successful.
Any suggestions are most welcome. Please provide some insight on why as the goal is not to build a super amp, but to understand them better. (I have stockpiled enough HCA 1200's and 750s I am set for a while)
Any suggestions are most welcome. Please provide some insight on why as the goal is not to build a super amp, but to understand them better. (I have stockpiled enough HCA 1200's and 750s I am set for a while)
Attachments
Yes, fix the amp, not the speaker. I've found that quite "rubbishy" speakers can produce deafening levels of sound, and do it very cleanly - for me the fault has always been with the electronics earlier in the chain, though most people are happy to point at the "lousy drivers" causing the congested sound ...
That should be possible to do, but it's not the reality out there - I went to a recent audio show, and there was only one amplifier there that was clearly capable of that sort of performance. Having been aware of the typical deficiencies of amplifiers for many years now, I gave up testing the units I came across for subjective performance a long time ago - it was too depressing hearing the same weaknesses, over and over again ...
Hmmm, I found my wife and lesser me, to be far more sensitive to the tweeter distortion than the amp. Subjective. But when I stepped up the tweeters, she could "stand" my swing and jazz.
You should step-up and experience it for yourself and think about this , the D/A transfer is done at pressing , what is the transfer limit at this point , IMO here in lies the secret ....
Plus and if you only want to travel virtually , go into forums where their are other users of hi def digital and analog and ask the same questions, im sure the majority will give you the same answers i have and to be clear my (analog) direct to disc Lp's have the most realistic sonics and yes i do have some direct from digital master lp's that are very good , i do not have an issue with their sonics ( Telarc) and there are some that sounds like , well, bad digital, lean and unnatural ( DBX digital) they use 16 bit ,(soundstream) i have to look and see what the others are using ..
The analog just sounds more natural, using direct tape masters sounds even better , even with the limitations imposed by R/R , the difference is staggering you need to experience it for yourself ..
Last edited:
From only my beginner understanding, I am amazed at the choices, as I am not in a position to say "mistakes", but things as simple as omitting the current mirror in the input and not buffering the VAS as well as a host of layout and mechanical issues does seem inexcusable by this day and time. I have looked at well over 100 published schematics and there does seem to be a willingness to ignore prior art. I do have to put a time filter in it. As an example, comparing Erno's first published MOSFET amp, through the Haflers and into later updates, what I would suggest are errors in not compensating for the different capacitance between the N and P was quickly learned and corrected. Anything built after, oh say 1980, should make use of the insights learned.
Based only on SPICE, after a few more tweaks today, the above amp posts less than 100 dB THD + Noise @ 10Khz. "Reasonable" I hope. Then what would it sound like? I suspect I have not dealt with dynamic issues such as clipping recovery and I am concerned with bias stability just by how sensitive it is is picking a "fixed" value in the model. Will just clamping the bias and one driver transistor together suffice?
Provided physical construction is correct and the obvious protection/limit issues, have I addressed the majority of what you consider major discrepancies? I am going to post over in the solid state thread with some of my protection questions.
How often do you have your ears calibrated in order to make such observations that something sounds good or bad. My test equipment is calibrated by a standards institute every year in order to warrant traceability and fit for its purpose. In my honest opinion everyone claiming they hear this or that is only an opinion that is all. If you are stone deaf your claim of what equipment sounds like would be the most accurate.
I have thought about it and posted my line of reasoning to discover if I had missed out something vital. Seems I haven't coz you've not addressed that line of reasoning rather distracted attention onto some unknown concept of 'the transfer limit'. What does that mean?
So let's do a reality check one more (last!) time - the direct from digital LPs best the dCS right?
Transfer limit , unknown concept ? ......... Really !!!
I thought about it and posted my line of reasoning to discover if i had missed out something vital , seems like i have'nt because you are still asking the same questions...
Follow me now ,
Telarc's Carmina Burana Lp, betters their CD , so yes , for the 100th time , direct to disc analog betters the best Digital Lp, master tape copies played from R/R betters them all ....
The worst digital LP's i have is the DBX series , thin like mid-fi digital, this stands no matter if its 25k worth of dCs or the 2k of rega digital im currently using ..
Last edited:
exact, we are living in silicon world, tons of watts available everywhere
who needs 95+db boxes? design /waf/ is often priority, for average customers
Yes really - so you're not going to explain it?
CD played back on dCS - no surprise to me that LP betters it. I'm unimpressed with the technical (and most of the time, musical too) quality of Telarc, even the more so since they went over to DSD.....
You're more than welcome to say that 10,000 times - its not what I'm asking as I heard you the first time. This makes sense too, unlike 'transfer limit'
Disagreed!
It's just plain wrong to make the amp designer responsible for everything. A speaker manufacturer can thus throuw out any old junk and let the amp manufacturer cater for it. And then people bitch about the cost of amplifiers, when in reality they have to clean up the mess of a speaker as well.
There's no delegating of responsibility here, each should do his thing properly to make it all work well. To me, it's criminal to sell a loudspeaker as a nominally 8 Ohm speaker, when in reality it may actuall be more at 4 Ohms, with some dips even lower, to 3 Ohms or less. Some manufacturers (e.g. B&W, Focal/JM Lab, etc) at least quote minimum imprdance, as 3.3 Ohms or so. Most don't.
And that's just the bare impedance - we also need to consider VERY substantial phase shifts occuring, not infrequently reaching -60 degrees, and very commonly -45 degrees. Please Frank, don't tell me that ANY amp should be able to drive 3.3 Ohms with -60 degrees phase shift no problemo.
Your "solution" to the problem, buying/making more powerful electronics, is not a real solution, rather a makeshift one - that's just more practical than reworking the speakers XO network.
Is that so? I beg to disagree.
What about phase shifts? Let's say we have a phase shift of -60 degrees, which would look to the amp as halving of the impedance, since it would require twice the current.
So, what would you rather have halved, 6+ Ohms, or 3.3 Ohm? With the 3.3 Ohms behaving as if it were 1.65 Ohms?
As for the filters, which would you rather have, erratic behavior caused by dipping impedance, or relativly smooth behavor helped out by a relatively steady impedance?
The speaker makers are guilty too, they leave silly weaknesses inside, like junk connections between modules. But these are relatively easy to fix, just eyeball the obvious flaws and do a straightforward clean-up.
Well, amplifiers can be made robust, to drive difficult loads by improving the power supply side of things. To be blunt, I have not changed the topology of existing amps, I work to enhance the stability of the voltage rails, where it counts - I'm not interested in turning the 60W amp into a 300W unit - I just want those 60 watts to be genuine ones!! So, one of my "tests" is to run amps up to just below their clipping point, and listen for the telltale tonal changes which means the amp has run out of steam ...
Of course one solution is to alter the nature of the XO, so that it presents a benign, resistive load to the amp - Klipsch seem to understand this, I was impressed with the dynamics a nothing receiver was able to conjure up on the showroom floor.
That piece of congested music by REM, mentioned in the Blowtorch thread, is an excellent testing tool - on the PC plastic monitors it can be made to sound fine up to a certain volume, but any louder and everything starts to congeal. This is not the speaker's fault, the 'miserable' power supplies inside the actives have given their all, that much and no more is their cry!! Now, that can be "fixed" by diving inside and stiffening the rails by, say, a 10 fold factor - if there's enough room!
Yes, I agree, everything counts in the end, of course. And the factors combine, interact. What I want is for a system to be able to go to as loud as I feel in the mood for at any moment, with never a concern for any loss in quality at either end of the volume spectrum.
Last edited:
The reality is that digital can be made as "natural" as anyone could want, this is a playback refinement issue - the fact that many times digital replay has some sort of artificial, edgy or flat feel to it is merely distortion making itself audible. In the same way, vinyl is nearly always clearly vinyl, the distortion signature is quite clear - except that subjectively that distortion is a lot more pleasant. Digital distortion is nasty stuff, the only solution is to eliminate it completely from being an audible contribution to the sound ...
dvv,
I see your points. Let me limit this to a voltage source amplifier with sufficient current, not the rare current source or some big-box AVR with almost no power supply. A real amp as we expect. Basically, I am saying if the amp prefers a higher impedance ( inductive by nature) load, why reduce the impedance of the speaker to make it flatter? It could actually make it harder to drive and it will increase the distortion. Obviously, this assumes we can get the transfer function we want out of the crossover first.
No amount of great crossover will make any driver better than it is. We agree, a poor crossover will make any set of drivers worst than they need to be. Might I suggest your thoughts are that most cheap speaker systems have a cheap crossover too, so the drivers are being accused of being worse than they are due to bad system design? No amount of magic crossover will make a $28 SilverFlute sound like my $80 Seas or more significant, the Dayton $25 tweeter sound as smooth as the Seas $60 tweeter. A bad crossover could make the Seas sound just as bad.
My argument is that the cost of trying to tame a poor driver with the crossover very quickly exceeds the cost of buying a better driver. If you need more than one compensation network, maybe the driver choice was not the best for the application. Not a rule, just a hint.
I see your points. Let me limit this to a voltage source amplifier with sufficient current, not the rare current source or some big-box AVR with almost no power supply. A real amp as we expect. Basically, I am saying if the amp prefers a higher impedance ( inductive by nature) load, why reduce the impedance of the speaker to make it flatter? It could actually make it harder to drive and it will increase the distortion. Obviously, this assumes we can get the transfer function we want out of the crossover first.
No amount of great crossover will make any driver better than it is. We agree, a poor crossover will make any set of drivers worst than they need to be. Might I suggest your thoughts are that most cheap speaker systems have a cheap crossover too, so the drivers are being accused of being worse than they are due to bad system design? No amount of magic crossover will make a $28 SilverFlute sound like my $80 Seas or more significant, the Dayton $25 tweeter sound as smooth as the Seas $60 tweeter. A bad crossover could make the Seas sound just as bad.
My argument is that the cost of trying to tame a poor driver with the crossover very quickly exceeds the cost of buying a better driver. If you need more than one compensation network, maybe the driver choice was not the best for the application. Not a rule, just a hint.
- Status
- Not open for further replies.