I would really love to know how you folks determine how much filter capacitance/energy storage you need in your projects.
As an example, take a preamplifier and a power amplifier rated at say 100W/8 Ohms nominally, but also to be very load tolerant.
And a special question for JC, since I know him to be involved in the indutry on the manfuacturing side (but also to anyone else who is in the same position) - how much of a factor in determining above is the price, if any?
As an example, take a preamplifier and a power amplifier rated at say 100W/8 Ohms nominally, but also to be very load tolerant.
And a special question for JC, since I know him to be involved in the indutry on the manfuacturing side (but also to anyone else who is in the same position) - how much of a factor in determining above is the price, if any?
Last edited:
I affraid this setup would lead to more noise (6dB), since the 'signal', to speak in SY's words, does not increase, but you will have two times 3dB noise addition.
vac
Correct me if I'm wrong, but you are a total tube guy. I do believe that a power supply for an extremely load tolerant transistor amp will be rather different, wouldn't it?
I mean, using say two tubes capable of 5A each and three pairs of transistors each of which is capable of say 15A per channel is not quite the same thing. Or am I wrong?
Hi,
Really?
Well, mine are in Electronics and Business Information Systems and I'm a part qualified accountant...
Math matters heavily in all three.
Now that sounds just like what is happening in a "discussion forum" I am familiar with, where "everything that doesnt support the current ideas held by some influential people as to how things ought to work is systematically diabolised and downplayed.".
Nothing new under the sun.
Ciao T
Really?
Well, mine are in Electronics and Business Information Systems and I'm a part qualified accountant...
Math matters heavily in all three.
Now that sounds just like what is happening in a "discussion forum" I am familiar with, where "everything that doesnt support the current ideas held by some influential people as to how things ought to work is systematically diabolised and downplayed.".
Nothing new under the sun.
Ciao T
Godfrey is correct.
The resistors are effecting an average of the four uncorrelated noise sources. If you think of any one of them having an effect on the average at the n.i. input of the opamp, its noise voltage is attenuated by a factor of 4 by the other three resistors. So the net noise at the opamp is the root-sum-of-squares of each contribution; if each noise source has the same rms noise voltage, but of course uncorrelated, the net noise assuming some noise bandwidth is ({[En/4]^2}*4)^0.5 (the r.m.s. sum of four sources each attenuated by 4). So each term is (En^2)/16, four of them are (En^2)/4, and thus the total is the square root of (En^2)/4, or En/2.
Signal averaging has helped, as we would hope.
Brad
Indeed.
My favourite math fast compilation was and still is the ingeenering
orientated memo by Bronstein and Semendiaev , wich you surely know
since you were in east Germany...
Obviously , economical theories cant be checked with experimentations
contrary to electromagnetical theories.
As such , it will be intenable to support a claim in electromagnetism
related matters without substanciating with measured effects in an
experience wich itself must be reproducible , that s why basically
the subjectivist position is branded more than doubtfull , as no valid
explanations are brought to explain effects that for the time seems
to be more the fruit of imagination than actual phenomenons.
Hi,
Usually I look at the space available, subtract that needed for the signal circuits (usually not a lot), for way, way oversized mains transformers (e.g. 1,200VA wound on a 1,500VA C-Core for a 180WPC stereo amp) and fill the rest with cap's, usually set up as CRLC filter chain, with R & L designed into the PCB.
My preamp's are generally passive, so a big fat nada for that.
But my old Arthur Loesch derived design had 6,800uF per channel AFTER the regulator, but that was for the Phono Stage, the line section was passive...
For the power amp in solid state, based on my current experiences, around 100,000uF, but I think more may still be better better. I would also still regulate the rails (yes, the output stage ones).
In my current "modify an Amp" project I'm quite limited by space, my demands on cap quality and all, this one will be around 150WPC/8R and each channel will get 4pcs 18,000uF Elna For Audio Cap's (they have measurable lower distortion than generic caps, lower microphonics too, incidentally) with 4pcs 1mH/10A Magnan/Iron core chokes plus two separate bridges (Hexfred Bridge Schottky enhanced), so 144,000uF in total...
I'd probably like more, but the transformer and heatsinks take a lot of space...
In Tube Amp's with around 430V rails and extrapolating from my 35 Watt'er I might put at least 2 * 2,200uF per channel and a massive 3H/0.4A Choke between the two.
In tube amp's I do however often use film cap's only and "finesse" circuitry instead.
Ciao T
Usually I look at the space available, subtract that needed for the signal circuits (usually not a lot), for way, way oversized mains transformers (e.g. 1,200VA wound on a 1,500VA C-Core for a 180WPC stereo amp) and fill the rest with cap's, usually set up as CRLC filter chain, with R & L designed into the PCB.
My preamp's are generally passive, so a big fat nada for that.
But my old Arthur Loesch derived design had 6,800uF per channel AFTER the regulator, but that was for the Phono Stage, the line section was passive...
For the power amp in solid state, based on my current experiences, around 100,000uF, but I think more may still be better better. I would also still regulate the rails (yes, the output stage ones).
In my current "modify an Amp" project I'm quite limited by space, my demands on cap quality and all, this one will be around 150WPC/8R and each channel will get 4pcs 18,000uF Elna For Audio Cap's (they have measurable lower distortion than generic caps, lower microphonics too, incidentally) with 4pcs 1mH/10A Magnan/Iron core chokes plus two separate bridges (Hexfred Bridge Schottky enhanced), so 144,000uF in total...
I'd probably like more, but the transformer and heatsinks take a lot of space...
In Tube Amp's with around 430V rails and extrapolating from my 35 Watt'er I might put at least 2 * 2,200uF per channel and a massive 3H/0.4A Choke between the two.
In tube amp's I do however often use film cap's only and "finesse" circuitry instead.
Ciao T
100 wpc as "smaller amps" indeed.
Just shows how relative points of view can be; in Europe, 100 wpc is the borderline case between mid- and high power amps.
20,000 uF per channel for a 100 wpc is about where I'd put it as well, although I do tend to use 2 caps 10,000 uF each in parallel, and then add another same make, same type 4,700 uF for some speed.
I arrived at this value in two ways: one was by auditioning, my ears simply told me that sounded about right. The other was by working backwards in calculating.
Motorola/ON Semi literature tells me that practice has shown we need 1...2 Joules of energy for every 10 Watts of dissipated power, depending on how complex our load is. Since it's not the amp's job to choose speakers, I immediately took this as 2 Joules, period.
The rest I know. Since 1 Joule is a product of one supply line voltage (assuming a symmetrical power supply, + and -) squared times capacitance in Farads, since I know my supply line low will be 51V, 20.000 uF sould give me
[(51 x 51) 0.0247] = 64.24 Joules,
or 320 Watts of peak power into very difficult loads. This in turn satisfies the requirement od delivering 100/200 Watts into 8/4 Ohms by a decent margin even under the worst case conditions.
To boot, I tend to use separate toroidal transformers for each channel, each rated at 400 VA and full wave bridge rectifiers per each + and - line, with electronically regulated voltage amp supply lines, regulated at 55V +/- 0.1%. That seems to do the job well.
No, I am not as total tube guy.
I have diploma of Designer and Technologist of Radio and Electronics Equipment from TIASUR.
I am neither total Tube guy, nor total SS guy. I am familiar with pros and cons of both, and use both in my designs. And of course it does not matter which active elements are powered, despite of some beliefs. Like, people believe that push-pull tube amp tolerates higher PS ripples. This belief based on fact that class AB amp when well balanced cancels ripples in pauses. But ripples still affect sound, causing intermodulation between ripples and amplified signal.
Did I answer your question?
Thorsten, without wishing to interfere with your design project, I would strongly suggest you try it out as planned, using Elna caps, but then try again using Fisher & Tausche caps instead.
My money is on the side that says you'll never take those F&T caps out, and certanly not for Elnas.
Forget calculations and ripple factors, the simple fact is that they will give you whatever Elna gives you, PLUS a bass depth and control you will not believe. Especially since they are way cheaper than Elnas.
My money is on the side that says you'll never take those F&T caps out, and certanly not for Elnas.
Forget calculations and ripple factors, the simple fact is that they will give you whatever Elna gives you, PLUS a bass depth and control you will not believe. Especially since they are way cheaper than Elnas.
Hi,
I am not sure what you are referring to.
There are many areas where what is often brushed off as "audiophile nonsense" has very real, relevant and observable effects, IF (this is crucial) the system is evaluated as a whole.
For example parts microphonics are not usually observed on the test bench (except for those old metal can power transistors chirping like birds with signal), but they are very real.
As long as mains powered gear leads to circulating chassis currents, mains cables and interconnects will make clearly measurable differences in noise floor that a bench test will not reveal...
So, I would suggest that often the failure to observe differences under electrical testing is more down to inappropriate and unrealistic test conditions coupled with an unwillingness to seriously investigate what is pre-judged, demonised and repressed as "nonsense" leading to experiments that are not designed to find the truth, but to provide data to support the "it's nonsense" prejudice.
Ciao T
I am not sure what you are referring to.
There are many areas where what is often brushed off as "audiophile nonsense" has very real, relevant and observable effects, IF (this is crucial) the system is evaluated as a whole.
For example parts microphonics are not usually observed on the test bench (except for those old metal can power transistors chirping like birds with signal), but they are very real.
As long as mains powered gear leads to circulating chassis currents, mains cables and interconnects will make clearly measurable differences in noise floor that a bench test will not reveal...
So, I would suggest that often the failure to observe differences under electrical testing is more down to inappropriate and unrealistic test conditions coupled with an unwillingness to seriously investigate what is pre-judged, demonised and repressed as "nonsense" leading to experiments that are not designed to find the truth, but to provide data to support the "it's nonsense" prejudice.
Ciao T
Last edited:
The problem with "audiophile nonsense" is that real effects often are assumed to present in contexts where they have no foundation. And since people want to understand what cause that effects, they find some explanations that satisfy them, no matter are they valid or not. Also, due to synesthesia their terms reflect images caused by sounds, instead of sound properties themselves, it is a quite different language, and sometimes it is hard to understand what they mean without meta-model questions.
AFAIK microphonic effects exist , for caps this is an evidence
and has surely been experienced by almost everyone by there.
Cables can be important if the amp is not stable , otherwise
if the amp is correctly designed their influence is close to zero
in respect of all other existant parameters in the audio chain.
That said , i m somewhat on the expectation about those
"clearly measurables differences in noise floor that a bench
test will not reveal"...
How can something be measurable yet not being revealable , that is
measurable i suppose , by a bench test ??...
No, I am not as total tube guy.
I have diploma of Designer and Technologist of Radio and Electronics Equipment from TIASUR.
I am neither total Tube guy, nor total SS guy. I am familiar with pros and cons of both, and use both in my designs. And of course it does not matter which active elements are powered, despite of some beliefs. Like, people believe that push-pull tube amp tolerates higher PS ripples. This belief based on fact that class AB amp when well balanced cancels ripples in pauses. But ripples still affect sound, causing intermodulation between ripples and amplified signal.
Did I answer your question?
Sorry, all I saw you posted about were tubes, so I assumed you were a dedicated tube guy.
As for your qualifications, that's all very nice, but as my late father used ot say, after 30+ years of teaching mechanical engineering, machine elements:
There are people who pass through school, and there are people through whom the school passes.
It's what's in your head, not in your pocket in form of papers certfying that you have completed some courses.
Yes, thank you, you have given me a fair idea of how you do it.
By the way, I never even saw the insides of the Faculty of Mechanical Engineering, yet I still built my own car from the ground up, using a hodge-podge of parts in what must truly be an international project (with parts from Serbia, Italy, France, Germany and Belgium).
Of course, no one man will ever know enough to do the whole job himself, and I was fortunate enough to have some VERY capable friends join in. The man who installed and mapped the fuel injection system goes to work at Bosch in Stuttgatr, and heads his own team in reseraching of new fuel injection systems. He is completeing his dcotorate in electronics, but his REAL value is in the fact thagt engine oil reaches his shoulders, he's not afraid to get his hands dirty and damn well knows how.
So, my outward appearing standard Yugo ripples from 0 to 62 mph (0-100 km/h) in 6.8 seconds, causing a massive headache to neighboring owners of Audi, BMW, MB and so forth owners. The look on their (дурак) faces in my rear view mirror is priceless!
BTW, I also have a degree in economics, but that didn't bother me much when I designed and made
my core product, a power line filter, or
my headphone amps, both as wholly bipolar, or with FETs in the input stages and MOSFETs for outputs. Both fully are fully complemetary. Shown is the MOSFET version, those are custom 50 VA transformers, and the power supply is a shunt type reguated power supply. The caps are Fisher & Tausche 4,700 uF/40V types, alternatives are Panasonic types of same specs.
Wayne would love it, headphone amps using 50W complementary power devices for the bipolar (MJE 15030/15031), or 75W for the MOSFET version (IRF 510/9510).
Last edited:
- Status
- Not open for further replies.