And I for one agree with that.
With a footnote - 2 Ohm power shouls be regarded only as transient power, deliverable in peaks only.
It could be made steady state, but that would catapult the cost, imagine the size of the heatsinks one would need for steady state 2 Ohm power, and consequently, the size of the package. Not to even mention the dimensions of the power transformers and the filter caps.
So, 2 Ohm power - yes, definitely, BUT in short term peaks only, and with as little change in tonality as one can get out of it. I'm probably harping on again, but the best aspect of my old Marantz 170 DC power amp is that I can run it to the point where its peak LEDs start to show signs of life, which is over the 110W/8 Ohm line, and its tonality stays almost exactly the same as when I'm running it an average dissipation of 1 W/8 Ohms. That was quite a feat, my hat off to the designer, meaning as Nige says one Sid Smith. That man knew his beans.
I notice Cat 5 is becoming the speaker cable of choice . For years I have advocated similar cables . Resistance is about 93R / Km ( 8 x 24 AWG ) . There are 8 cores so many possibilities .
One version says copper over aluminum . Seeing as I was told copper is the worst sounding metal maybe I shouldn't reject this idea . I take that statement with a pinch of salt . However it makes me say why not aluminum .
Solid core does seem to mildly bandwidth limit the sound . As discussed by others it might be a nice thing . Tweeters do so , it is unavoidable .
An experiment I did years ago confirmed to me that the solid core is an effect and not a way to sound purity . In Belgium in the best French I could muster I ordered 100 metres of 1 mm solid-core installation cable . I said casually to buy a reel . Back he comes with 6 metres and a joiner . So we added it to the 2.5 mm 79 stand . The speakers Bose 901 . To my surprise adding the 1 mm removed the hash just like if no 79 stand used . 79 stand is not my cup of tea . Like instant coffee it makes no sense to use it . I am told I make great instant coffee ( some cold water to mix it first ) , I never drink it . Tea by the bucket . I drink Ethiopian coffee by choice . I can make real coffee quicker than instant . Old Nescafe is the better instant . It survives as the French demand it . It must be mixed if not to be ruined . It is pre-ruined really so it is the best of a bad thing . Alta-Rica is not bad . Cost more than I pay for real .
A friend has just replaced his cable with Cat 5 ( 2 metre = $1 ) . He says it is slightly better than previous $1500 cables . Madness in the right direction I say .
One version says copper over aluminum . Seeing as I was told copper is the worst sounding metal maybe I shouldn't reject this idea . I take that statement with a pinch of salt . However it makes me say why not aluminum .
Solid core does seem to mildly bandwidth limit the sound . As discussed by others it might be a nice thing . Tweeters do so , it is unavoidable .
An experiment I did years ago confirmed to me that the solid core is an effect and not a way to sound purity . In Belgium in the best French I could muster I ordered 100 metres of 1 mm solid-core installation cable . I said casually to buy a reel . Back he comes with 6 metres and a joiner . So we added it to the 2.5 mm 79 stand . The speakers Bose 901 . To my surprise adding the 1 mm removed the hash just like if no 79 stand used . 79 stand is not my cup of tea . Like instant coffee it makes no sense to use it . I am told I make great instant coffee ( some cold water to mix it first ) , I never drink it . Tea by the bucket . I drink Ethiopian coffee by choice . I can make real coffee quicker than instant . Old Nescafe is the better instant . It survives as the French demand it . It must be mixed if not to be ruined . It is pre-ruined really so it is the best of a bad thing . Alta-Rica is not bad . Cost more than I pay for real .
A friend has just replaced his cable with Cat 5 ( 2 metre = $1 ) . He says it is slightly better than previous $1500 cables . Madness in the right direction I say .
Unfortunately my attempt to make a headphone amp has resulted is a weird request . I have been asked to make one that sounds better than the original iPhone .
That might just be possible although my answer to him was there are 7 000 000 on the planet and not one of us can .
I have allowed myself double the iPhone voltage . I will have a high input resistance . I might be able to sweet-spot the output a bit .
The buffer I saw yesterday made me think . Could I make in the style of the PE Texan ( Rega Brio ) an output stage with gain that has no obvious crossover distortion . Not liking the Brio means I would be cautious . As it might be collector output rail to rail is possible ( subtract Vce ) . Blomley amplifier where the input is via common base might be an idea ? Any thoughts ?
http://www.keith-snook.info/wireles.../New Approach to class B Amplifier Design.pdf
That might just be possible although my answer to him was there are 7 000 000 on the planet and not one of us can .
I have allowed myself double the iPhone voltage . I will have a high input resistance . I might be able to sweet-spot the output a bit .
The buffer I saw yesterday made me think . Could I make in the style of the PE Texan ( Rega Brio ) an output stage with gain that has no obvious crossover distortion . Not liking the Brio means I would be cautious . As it might be collector output rail to rail is possible ( subtract Vce ) . Blomley amplifier where the input is via common base might be an idea ? Any thoughts ?
http://www.keith-snook.info/wireles.../New Approach to class B Amplifier Design.pdf
Last edited:
I set about testing Nige’s hypothesis that adding bias to the buffer might improve things. Bat I was nagged by wondering about the potentials of the Philips preamp circuit.
So I thought – how would I do it? Sure, after the battle everybody’s a general, but then Philips has no copyrights for this circuit, it had been used before by others, as John rightly pointed out.
Anyway, here goes. Bear in mind that no real development was done on this, I literally slapped it together, making only the changes I thought would improve matters with an outlook towards a separate standalone model, not a modification of the Philips.
For a start, I took Demian’s suggestion from another matter to bias the input buffer FETs at somewhat higher currents than are usual in commercial gear. Nothing wild, mind you, just 4.21 mA per FET. Having made this buffer several times already, it was a given and it works really well.
The most obvious change from Philips is the addition of active CCSs for the input stage. I suppose a LED could have been used, but over the last 2-3 years, I’ve had LEDs which varied quite a bit in their voltages, while being nominally the same. So I stuck to the venerable 1N4148 diodes, they are far more consistent, and therefore better suited to DIY. The voltage difference above a LED is small enough to make no real difference in the requiredemitter resistors, 220 Ohms is still low enough not to introduce too much noise.
The second difference is in the addition of degeneration resistors into the differential pairs. I like the very idea of degeneration. They limit the gain to approximately 9:1, which is fine. Further development may only reduce this.
In line with my view that active devices should be decoupled from the supply lines by other active devices, both diff. Pairs got two diodes each. May be of no importance in real life, but I like to see them there, and the added cost is of no consequence.
I also added two BAV21 fast diodes (D6, D12), which serve to reduce settling time. Again, I cannot swear to their dramatic efficinecy, but it’s simple and cheap, and it certainly cannot hurt.
Philips did the bias circuitry with 2 diodes, I need 5 diodes to get what I wanted, and I wanted the output stage biased at around 20 mA. This is no random choice, BD 139/140 data sheet shows that by this current, it is at its best. On the practical side, they will be dissipating around 0.6W each, something they can do even under free air coditions, but small, cheap, easy to find local heat sinks will serve just fine and greatly simplify construction. Some smaller heat sinks will even allow them to be mounted back-to-back with the sink flat body in between, which is desirable, in my view.
In my version, there are no parallelled devices, but a classic driver to output stage connection. To the best of my knowledge, this is a good sounding combination which is well neigh impossible to beat overall, though some other connections may provide even better performance.
Next, I settled not for +/- 38V lines as Philips used, but for +/- 30V. This is, in my view, qute enough for just about anything, and I already have fuly developed, tried and used shunt power supply circuits at hand delivering this voltage with more than enough current at hand. So I’m lazy, so what?
And, finally, notwithstanding Nige’s comments, I did add a proper DC servo circuit. It’s simple and easy to make with off the shelf parts, and doesn’t cost too much either in labor, real estate or price. It keeps the DC offset at levels below +/- 1 mV, which is good enough for me.
Then I did some measuring, however only within the simulator, obviously, as I do not have a live working model. The results are quoted on the added sheet. Just one note – I limited myself to an output of 3 Vrms, and made the mistake of mixhing apples and oranges, since input signal is expressed in peak values and the output in rms values. That’s what happens when one is not careful enough, but on the other hand, that accidentally covers the entire territory and then some. I do not know of a power amp reuquiring more than 3.2V of input signal to drive it to full power, and here I am considering a 4.23V signal. Not to worry, it simply means that the results I got may be even better than specified in real life, and in many cases, WILL actually be better since most power amps do not need more than about 2V input to make the full power level.
It turns out that the circuit’s open loop gain factor is about 64:1 (or 36 dB), while its closed loop gain factor is about 17:1 (24.4 dB) into 600 Ohms and about 18:1 (25.1 dB) into a 20 kOhms load, correcting the 3 Vrms to 4.23 V.
BTW, there’s a mistake in my opening line on closed loop performance, so disregard the part „Vin 66 mV“ as a copy and paste mistake.
The spec sheet completes the story. While not the best I’ve ever seen, I’d say it is way better than just satisfactory. The frequency response measurements were made exactly as shown, including the R24 (1k) and C18 (100pF) input filter, so an open loop response out to 250 kHz at -3 dB is a very good result in my book. Closed loop it jumps to 650 kHz, which I think will merit a nod from Demian.
For me, the overall NFB factor is the key consideration, it’s nicely low just the way I like it. Few can argue that 12 dB of global NFB is a fearsome number.
P.S. Nige, use this and shock them to glory.
So I thought – how would I do it? Sure, after the battle everybody’s a general, but then Philips has no copyrights for this circuit, it had been used before by others, as John rightly pointed out.
Anyway, here goes. Bear in mind that no real development was done on this, I literally slapped it together, making only the changes I thought would improve matters with an outlook towards a separate standalone model, not a modification of the Philips.
For a start, I took Demian’s suggestion from another matter to bias the input buffer FETs at somewhat higher currents than are usual in commercial gear. Nothing wild, mind you, just 4.21 mA per FET. Having made this buffer several times already, it was a given and it works really well.
The most obvious change from Philips is the addition of active CCSs for the input stage. I suppose a LED could have been used, but over the last 2-3 years, I’ve had LEDs which varied quite a bit in their voltages, while being nominally the same. So I stuck to the venerable 1N4148 diodes, they are far more consistent, and therefore better suited to DIY. The voltage difference above a LED is small enough to make no real difference in the requiredemitter resistors, 220 Ohms is still low enough not to introduce too much noise.
The second difference is in the addition of degeneration resistors into the differential pairs. I like the very idea of degeneration. They limit the gain to approximately 9:1, which is fine. Further development may only reduce this.
In line with my view that active devices should be decoupled from the supply lines by other active devices, both diff. Pairs got two diodes each. May be of no importance in real life, but I like to see them there, and the added cost is of no consequence.
I also added two BAV21 fast diodes (D6, D12), which serve to reduce settling time. Again, I cannot swear to their dramatic efficinecy, but it’s simple and cheap, and it certainly cannot hurt.
Philips did the bias circuitry with 2 diodes, I need 5 diodes to get what I wanted, and I wanted the output stage biased at around 20 mA. This is no random choice, BD 139/140 data sheet shows that by this current, it is at its best. On the practical side, they will be dissipating around 0.6W each, something they can do even under free air coditions, but small, cheap, easy to find local heat sinks will serve just fine and greatly simplify construction. Some smaller heat sinks will even allow them to be mounted back-to-back with the sink flat body in between, which is desirable, in my view.
In my version, there are no parallelled devices, but a classic driver to output stage connection. To the best of my knowledge, this is a good sounding combination which is well neigh impossible to beat overall, though some other connections may provide even better performance.
Next, I settled not for +/- 38V lines as Philips used, but for +/- 30V. This is, in my view, qute enough for just about anything, and I already have fuly developed, tried and used shunt power supply circuits at hand delivering this voltage with more than enough current at hand. So I’m lazy, so what?
And, finally, notwithstanding Nige’s comments, I did add a proper DC servo circuit. It’s simple and easy to make with off the shelf parts, and doesn’t cost too much either in labor, real estate or price. It keeps the DC offset at levels below +/- 1 mV, which is good enough for me.
Then I did some measuring, however only within the simulator, obviously, as I do not have a live working model. The results are quoted on the added sheet. Just one note – I limited myself to an output of 3 Vrms, and made the mistake of mixhing apples and oranges, since input signal is expressed in peak values and the output in rms values. That’s what happens when one is not careful enough, but on the other hand, that accidentally covers the entire territory and then some. I do not know of a power amp reuquiring more than 3.2V of input signal to drive it to full power, and here I am considering a 4.23V signal. Not to worry, it simply means that the results I got may be even better than specified in real life, and in many cases, WILL actually be better since most power amps do not need more than about 2V input to make the full power level.
It turns out that the circuit’s open loop gain factor is about 64:1 (or 36 dB), while its closed loop gain factor is about 17:1 (24.4 dB) into 600 Ohms and about 18:1 (25.1 dB) into a 20 kOhms load, correcting the 3 Vrms to 4.23 V.
BTW, there’s a mistake in my opening line on closed loop performance, so disregard the part „Vin 66 mV“ as a copy and paste mistake.
The spec sheet completes the story. While not the best I’ve ever seen, I’d say it is way better than just satisfactory. The frequency response measurements were made exactly as shown, including the R24 (1k) and C18 (100pF) input filter, so an open loop response out to 250 kHz at -3 dB is a very good result in my book. Closed loop it jumps to 650 kHz, which I think will merit a nod from Demian.
For me, the overall NFB factor is the key consideration, it’s nicely low just the way I like it. Few can argue that 12 dB of global NFB is a fearsome number.
P.S. Nige, use this and shock them to glory.
Thanks Mate . I like this . Do try bypassing the buffer . Input JFET's might not need any degeneration . They should massively improve the sound if my hunch is right . Also they will tolerate no feedback cap as a possible upgrade . The supposed square law sound is as far as I can see nothing to do with it .
" The second difference is in the addition of degeneration resistors into the differential pairs. I like the very idea of degeneration. They limit the gain to approximately 9:1, which is fine. Further development may only reduce this. !"
That might do good in ways not thought of . It might help stability which seems marginal here . If you read Bob Stewart's Sept 1973 WW stuff about TID he clearly states why . Bob is basically saying negative feedback mostly is something to reduce output impedance and raise the input . It has some stimulating effects elsewhere . Sorry Bob that is the devastating reality of it as you rightly say . I don't go with that 100 % except to say it is dangerously like what Mr Dvv says with diagrams and PHD maths . Full paper sent to you Dvv . Read it for the thrill of it's clarity . I met Bob shortly after that and like him .
Bob States TID and crossover distortion sound similar . Thus a class A amp might sound worse than a class B . TID seems to be abandoned as it was too close to being why valve amps sound good . John seems to build amps that shouldn't have it . D Self wrote to me that all bad things show in standard distortion tests if looking carefully . I think that must be true . What might fool us is the effect might look very innocent in number terms . My NE5532 driving headphones might prove that .
I think I am right in saying only this simple Darlington you use clears out minority carriers when the output stage switches . This simple expedient solves the big bi-polar problem . That's why Audio MOSFET's sound better than people would guess is my guess . They do not have minority carriers . That is holes in N type and electrons in P type material as a charge . Majority carriers the the other way around and might as well describe wires .
" The second difference is in the addition of degeneration resistors into the differential pairs. I like the very idea of degeneration. They limit the gain to approximately 9:1, which is fine. Further development may only reduce this. !"
That might do good in ways not thought of . It might help stability which seems marginal here . If you read Bob Stewart's Sept 1973 WW stuff about TID he clearly states why . Bob is basically saying negative feedback mostly is something to reduce output impedance and raise the input . It has some stimulating effects elsewhere . Sorry Bob that is the devastating reality of it as you rightly say . I don't go with that 100 % except to say it is dangerously like what Mr Dvv says with diagrams and PHD maths . Full paper sent to you Dvv . Read it for the thrill of it's clarity . I met Bob shortly after that and like him .
Bob States TID and crossover distortion sound similar . Thus a class A amp might sound worse than a class B . TID seems to be abandoned as it was too close to being why valve amps sound good . John seems to build amps that shouldn't have it . D Self wrote to me that all bad things show in standard distortion tests if looking carefully . I think that must be true . What might fool us is the effect might look very innocent in number terms . My NE5532 driving headphones might prove that .
I think I am right in saying only this simple Darlington you use clears out minority carriers when the output stage switches . This simple expedient solves the big bi-polar problem . That's why Audio MOSFET's sound better than people would guess is my guess . They do not have minority carriers . That is holes in N type and electrons in P type material as a charge . Majority carriers the the other way around and might as well describe wires .
Last edited:
Wayne, you are now seriously rambling.
Can you tell me what makes an output stage load tolerant and capable of very high impulse currents?
My point is, you can throw in 5 kVA transformers, 100,000 uF caps in any form you like, and still have only so-so current capability. You might also make an unassuming power supply and blow away the competition.
You oversimplify the whole matter. To you, big solves the whole problem. Well, it doesn't, it does help to solve it easier, but by itself, it doesn't do much, and in fact, quite a few mammoth power supplies kill all traces of emotion and life from the music. I daresay that power supplies are singly taken the most sensitive part of a power amp, it's so easy to fall short with them.
Can you tell me what makes an output stage load tolerant and capable of very high impulse currents?
My point is, you can throw in 5 kVA transformers, 100,000 uF caps in any form you like, and still have only so-so current capability. You might also make an unassuming power supply and blow away the competition.
You oversimplify the whole matter. To you, big solves the whole problem. Well, it doesn't, it does help to solve it easier, but by itself, it doesn't do much, and in fact, quite a few mammoth power supplies kill all traces of emotion and life from the music. I daresay that power supplies are singly taken the most sensitive part of a power amp, it's so easy to fall short with them.
Paper Tiger material D.....
And you know this - how? Have you measured it? Have you used it with several loudspekers, so you know from experience?
Pray explain to me how is 595W of impulse power into 2 Ohms "Paper Tiger Material", I'd really like to know.
Or do you base your views on steady state power delivery on your own, totally offbeat case?
It's like saying Rolls-Royce is the only real car on the planet, all the rest are just go carts in drag. Let's solve all our problems by throwing money at them.
D,
Why the centurion need Cojones ....
Heavy Load: How Loudspeakers Torture Amplifiers Page 2 | Stereophile.com
Pray tell how is this any different to what I have told you, in hard essence?
It tells you that 0.01% of the time you will need 4 times the average output into 8 Ohms, obviously referring to 8 Ohms being the nominal load impedance. I do all my testing with 4 Ohms as my nominal load impedance, so I'm ahead there.
In addition to which I make sure it can do impulse power bursts way over the nominal. 0.01% of 1 second is 10 mS, while my work assumes 50 mS as the time during which the protection circuitry is dormant, which is 5 TIMES longer.
With modern power devices, even the economy class ones such as Toshiba's 2SC5200/2SA1943, are capable of very large impulse currents, so assuming one hasn't severely underrated the required output potential and used too few pairs in context of desired results, ANYBODY who sets their mind on it can do it. It's not often done not for lack of know-how. but for purely economic reasons - that costs money.
DVV:
Your input follower is single ended, fine for a low cap load like this. Make the R in the source of the lower FET adjustable as an offset trim. Driving a pot you will need to get the offset down or you will get a number of issues (pops, breathing noises, contact degradation etc.).
12 dB of feedback may be a bad place. The anti-feedback crowd has measurements of distortion multiplication in that range. As I remember the problem is worst in the 10-20 dB range and goes away at higher feedback levels. I'm sure someone here can point you at the sources for this.
My concern around Darlington connections is getting both transistors switching on and off quickly. In this case its not such a big issue with a small but capacitive load. However the original circuit with no follower can't have crossover distortion (at least from the usual causes).
Below is a quick simulation of what is required to drive an IHF load (10K in parallel with 10 nF) for a preamp at 3V 1 MHz. Its pretty extreme but does highlight the issues. The peak current, 300 mA happens at the zero crossing so its maximum stress on the device. Its also where getting those minority carriers in and out is hardest.
a. wayne:
As DVV points out a big power supply and lots of transistors won't necessarily mean lots of drive. I would look at the voltage and current waveforms into a speaker at worst case before making any decisions about what is needed. The example above transformed for a speaker near a driver/crossover resonance could be really difficult to drive as it is all reactive. Reactive loads are where the transistors can't respond nicely. The VAS is near zero, a low current point and the load needs max current so the base drive goes to max and the VAS is not there. Even with a Darlington the issue exists. It suggests overbuilding at places that are not obvious.
Your input follower is single ended, fine for a low cap load like this. Make the R in the source of the lower FET adjustable as an offset trim. Driving a pot you will need to get the offset down or you will get a number of issues (pops, breathing noises, contact degradation etc.).
12 dB of feedback may be a bad place. The anti-feedback crowd has measurements of distortion multiplication in that range. As I remember the problem is worst in the 10-20 dB range and goes away at higher feedback levels. I'm sure someone here can point you at the sources for this.
My concern around Darlington connections is getting both transistors switching on and off quickly. In this case its not such a big issue with a small but capacitive load. However the original circuit with no follower can't have crossover distortion (at least from the usual causes).
Below is a quick simulation of what is required to drive an IHF load (10K in parallel with 10 nF) for a preamp at 3V 1 MHz. Its pretty extreme but does highlight the issues. The peak current, 300 mA happens at the zero crossing so its maximum stress on the device. Its also where getting those minority carriers in and out is hardest.
a. wayne:
As DVV points out a big power supply and lots of transistors won't necessarily mean lots of drive. I would look at the voltage and current waveforms into a speaker at worst case before making any decisions about what is needed. The example above transformed for a speaker near a driver/crossover resonance could be really difficult to drive as it is all reactive. Reactive loads are where the transistors can't respond nicely. The VAS is near zero, a low current point and the load needs max current so the base drive goes to max and the VAS is not there. Even with a Darlington the issue exists. It suggests overbuilding at places that are not obvious.
Attachments
D's ...
Firstly , i did not write the article gents, secondly, I did not specify what was or is necessary for good drive, but , surely a large PSU and output stage is a a good starting point.
Now,
DVV, stated 2 ohm peak , My point , thats a paper tiger and as clearly stated not only in that article, but in other real world situation , you need 2 ohm drive , not once in a while drive, if not, you are producing nothing more than a paper tiger.
Real world situation when building reference level audio will require such ...
Ohhh and most important SOA , Now Suited up ....
Firstly , i did not write the article gents, secondly, I did not specify what was or is necessary for good drive, but , surely a large PSU and output stage is a a good starting point.
Now,
DVV, stated 2 ohm peak , My point , thats a paper tiger and as clearly stated not only in that article, but in other real world situation , you need 2 ohm drive , not once in a while drive, if not, you are producing nothing more than a paper tiger.
Real world situation when building reference level audio will require such ...
Ohhh and most important SOA , Now Suited up ....
Last edited:
Points taken, Demian, as I said, I literally slapped that together within an hour, not mcu thought invested. But of course, this is not to say there won't be. I am intrigued now.
On the graph, I note you used a 10V peak to peak signal. Is that much really necessary, or was that just for example's sake?
As for the point of your graph, now you see why I used Motorola's MJE 15030/15031 (140V, 8 A, 50W, Ft > 60 MHz) for my headphone amp/preamp, while the BD 139/140 were used as drivers.
On the crossover problem - I used a bias of 85 mA in those amps, which effectively made the whole range of voltages required for most cans a strictly pure class A affair. BUT, that also forced me to use an unusually large heat sink for such a device, and it wasn't amiss, trust me, no mere "if you've got it, flaunt it" affair.
On the global NFB. I had the same problem then as well. In general, my experience seems to suggest that having too little global NFB is not the best of solutions, and thus far, it SEEMS to me one should go for the 20-26 dB range as a very probable optimum (assuming other criteria are also met).
Thank you, Demian. I'll play with it a bit more coherently tomorrow, just to see how I fare, and get back on it.
On the graph, I note you used a 10V peak to peak signal. Is that much really necessary, or was that just for example's sake?
As for the point of your graph, now you see why I used Motorola's MJE 15030/15031 (140V, 8 A, 50W, Ft > 60 MHz) for my headphone amp/preamp, while the BD 139/140 were used as drivers.
On the crossover problem - I used a bias of 85 mA in those amps, which effectively made the whole range of voltages required for most cans a strictly pure class A affair. BUT, that also forced me to use an unusually large heat sink for such a device, and it wasn't amiss, trust me, no mere "if you've got it, flaunt it" affair.
On the global NFB. I had the same problem then as well. In general, my experience seems to suggest that having too little global NFB is not the best of solutions, and thus far, it SEEMS to me one should go for the 20-26 dB range as a very probable optimum (assuming other criteria are also met).
Thank you, Demian. I'll play with it a bit more coherently tomorrow, just to see how I fare, and get back on it.
………not a bad place and does not have the potential to become independent, therefore secure!!!
D's ...
Firstly , i did not write the article gents, secondly, I did not specify what was or is necessary for good drive, but , surely a large PSU and output stage is a a good starting point.
Which is ecactly as I said myself. A good starting point, but no guarantee. I have heard some amps which were made like the Rock of Gibraltar, with PSUs which would move a tram, yet they sounded disinterested and bleak, lifeless, despite oodles of power.
Still disagreed, albeit with a note - the usual criterium of a 10 mS (IHF) or 20 mS (IEC) burst is, in my view, too short. For example, the clicks and pops on LPs usually last 20-40 mS (read that somewhere), and are sometimes capable of destabilizing the preamp and the amp. Thus, to avoid that, I need bursts longer than 40 mS, which is why I use 50 mS as my own reference, and I need a healthy overhead voltage reserve, which is why I use higher than usual PSU rails.
You make no difference between what an amp can do in absolute and what it is effectively allowed to do, which is wrong. The Centurion iscabale of sustained 2 Ohm drive, but it is limted in comparison with 8 Ohm loads by the sustained current required (less so) and by the heatsink cooling capability (which is the main limiting factor). Because of the heatsink limitations, I need to use a current limiting ciruit to prevet overheating of the output transistor junctions, which will heat up a lot faster than the heat sink can draw from them.
Lastly, we need to take into account what is our base value according to which we want to compare capabilities. It's one thing to drive a relatively clean speaker like mine (worst case phase shift -25 degrees, lowest actual impedance 6.5 Ohms, 1W/8 Ohms wproduces 92 dB SPL at 1m), and a really difficult speaker, such as B&W 802, the one mentioned in the article. It will straight off the bat need 3-4 times the power I need for the same SPL at 1 m. For a "typical" (whatever that is) loudspeaker today, in a room, the AVERAGE dissipated power is not likely to go above say 10W into 8 Ohms. This is to say that under bad conditions, it will peak at 4-5 times that, which will put that at around 100-125W/8 Ohms, even if for that 0.01% of the time. That becomes 500W/2 Ohms with the very worst of cases, and if you are not into PA, the HK 680 integrated will cover your butt, in peaks only, of course, Since the Citation is even nominally more powerful, and has a bigger PSU section all around, I sincerely doubt it will even hiccup, although I haven't tried that, I have nothing to actually try it with.
But, you double my budget and I'll have it delivering max power into 2 Ohms for hours on end - but at a price, and at a size too.
Real world situation when building reference level audio will require such ...
Wayne, I am NOT trying to build anything of reference value. I like myself, but not that much to delude myself that I am capable of building reference standard gear. I know my limitations (well, mostly).
Ohhh and most important SOA , Now Suited up ....
No flames from me, I just get irritated on rare occasions when you or anyone else start throwing numbers around like they were peanuts. Hey kiddo, what's a coupla kilowatts between friends?
Do I tell you how you should make a loudspeaker? No, and I never will, because I know you know a lot more about that than I do, but I know enough about them not to make flippant remarks about them.
You know, I always say about myeslf that I am one smart dude, not because I know everything, but because I know who to ask what I don't know and they do.
Bredren ,
Never stated Kilo-watts, stated Low-Z drive. Now if you plan on building something and seeing you are no Spring chicken then why not SOTA, why build OK, i just dont get it D, it doesn't have to be a mega-watt amp, but why not something to say SOTA, you have no market restrictions, design away buddy.
I would not even attempt to do a speaker system unless it's SOTA, why waste time building some little BS avg speaker, my time is more precious than that , might as well be listening to PC speakers like Frank ...
It's like that smarter than everyone dude, building himself an 8watt toooby, then claiming nirvana ...
Never stated Kilo-watts, stated Low-Z drive. Now if you plan on building something and seeing you are no Spring chicken
then why not SOTA, why build OK, i just dont get it D, it doesn't have to be a mega-watt amp, but why not something to say SOTA, you have no market restrictions, design away buddy. I would not even attempt to do a speaker system unless it's SOTA, why waste time building some little BS avg speaker, my time is more precious than that , might as well be listening to PC speakers like Frank ...
It's like that smarter than everyone dude, building himself an 8watt toooby, then claiming nirvana ...
Said partially in jest, unfortunately many people don't appreciate what the goals are - which is to extract the most satisfying sound from any particular system, no matter how 'cheap' and 'incapable' it is,I would not even attempt to do a speaker system unless it's SOTA, why waste time building some little BS avg speaker, my time is more precious than that , might as well be listening to PC speakers like Frank ...
.Just recently fired up the Aldi TV for audio duties - bizarrely, it's capable of 'bigger' sound than the PC speakers, courtesy of using higher power, class D amps. But the speakers there are even more tiny bits of nothing, and they're just plunked into a rattly plastic frame, have to watch the volume. Yet the PC setup produces cleaner, more engaging sound from a cold start, because I've done a first round of inner "tweaking". And they're both less capable than the Philips HT - which is still out of action - which could go louder, with better "real" bass - but required a very extended warmup period to sound decent. Each is an experiment in itself.
I heard some of my demo CDs on 'SOTA' speakers, at the audio show - and talk about dreadful value for money, at a snap judgement level: major chunks of detail in the recording were completely missing, very lopsided tonal balance, and as soon as volume was asked for they went very messy ... if that is what you're chasing, good luck to you,
!As regards amplifier power, it's not the numbers, but how intelligent the designer was - I've heard the dead sound of powerhouse amps, and this says to me that in their efforts to physically construct something that could do the rated power, the manufacturers introduced far too many weakness that pulled the SQ down. Take one of those beasts, do major butchery on them, and then respectable sound should be possible ...
- Status
- Not open for further replies.