I am like a dog with a bone now and must find out . The output impedance and input loading will be checked , as few variables as possible . The output resistance can be whatever I like . 90 m ohm is where I start . I bet every headphone will have it's preference . In theory even 10R should suit a 32R load ( very old theory ) . Damping factors like that encourage a bit of bloom . It might also help the amp a bit .
One thing I might try is a null test . My analyzer will do that quite well as it isn't referenced to ground . Clip input ground to source and output to phones . Use tones and music . See what comes out . Try different phones also .
Ahhh ... this is the "great mystery" - but in fact it is not, as far as I'm concerned. This is one system in the 'zone', and the other is not; the fact that the systems only vary by the speakers being used is irrelevant - the second lot of speakers varies, improves, the overall sound just enough so that the ear/brain gets on board properly, and "understands" what the low level information in the sound it is hearing means.
I never, ever think in terms of speakers or amplifiers or sources; for me it is the system and always the system, breaking it down into component elements, and one of those being key, will just drag one away from better understanding, IMO.
Frank, 200 kHz open or closed loop?
I assume closed loop, getting it to run to 200 kHz open loop would bring up some heavy stability issues. 100 kHz is still managable in open loop even with off the shelf semiconductors and no fancy circuits. 200 kHz requires at least 4 to 5 times the work.
My favorite trick is to get it running up to 100 kHz open loop, and apply say 20 dB of NFB. Since NFB acts like a first order filter, my top end will not be 1 MHz, but it has a good chance of hitting 700-750 kHz closed loop at rated output -3 dB point.
If one is careful, and does not force issues, one can get away with distributed (instead of single point) compensation, and thus manage to keep one's voltage slew rate at or above the 100 V/uS mark. As John Curl rightly points out, it can be done without too much trouble, so do it. Also, it keeps the 100 kHz THD down at 0.2/04% into 8/4 Ohms.
After I have done all that, I include a safety precaution, a 6 dB/Oct. input signal filter at around 200 kHz, I honestly see no point in going above, based on my listening tests, nothing changes. This keeps the effective input-to-output amp slew rate at about 70 V/uS, or at about 1.75 times the minimum of 40V/uS (1V/uS per peak V of output for a nominally 100W/8 Ohms amp).
As far as I can fathom, this keeps my nose clean under any and all circumstances.
I assume closed loop, getting it to run to 200 kHz open loop would bring up some heavy stability issues. 100 kHz is still managable in open loop even with off the shelf semiconductors and no fancy circuits. 200 kHz requires at least 4 to 5 times the work.
My favorite trick is to get it running up to 100 kHz open loop, and apply say 20 dB of NFB. Since NFB acts like a first order filter, my top end will not be 1 MHz, but it has a good chance of hitting 700-750 kHz closed loop at rated output -3 dB point.
If one is careful, and does not force issues, one can get away with distributed (instead of single point) compensation, and thus manage to keep one's voltage slew rate at or above the 100 V/uS mark. As John Curl rightly points out, it can be done without too much trouble, so do it. Also, it keeps the 100 kHz THD down at 0.2/04% into 8/4 Ohms.
After I have done all that, I include a safety precaution, a 6 dB/Oct. input signal filter at around 200 kHz, I honestly see no point in going above, based on my listening tests, nothing changes. This keeps the effective input-to-output amp slew rate at about 70 V/uS, or at about 1.75 times the minimum of 40V/uS (1V/uS per peak V of output for a nominally 100W/8 Ohms amp).
As far as I can fathom, this keeps my nose clean under any and all circumstances.
Last edited:
While there is no useful signal up there for us, this does keep phase shifts at 20 kHz down below 5 degrees or so. Compare a square wave at 15 kHz for a low bandwidth with a wide bandwidth amp, and you can actually see the difference. The low bandwidth amp's will have a bigger tilt.
On the other hand, Frank, you are less likely to achieve synergy with a system which has serious flaws in it, such as a wildly varying, say +/- 5 dB, speaker response, probaly with some phase nasties added, etc.
It really boils down to finding the weakest link and doing something about it, then all that all over again, until you get it right. That usually takes years, even decvades, to achieve properly, and much hands on practical experience. Not to even mention patience.
Last edited:
Just to be clear this is a night and day difference . All I was doing was a test bed for some new drive units , a lazy test . I had a hunch it might work . I know the sad moment will be when they go in boxes .
If I owned a real cinema I would use 20 foot OB speakers and subs for the last octave . The nasty stuff as it is usually done ( 5.1 ) . I am sure it would remove the paper thin quality I hear in cinemas . If I am right the sound still goes through perfections in the screen ?
If I owned a real cinema I would use 20 foot OB speakers and subs for the last octave . The nasty stuff as it is usually done ( 5.1 ) . I am sure it would remove the paper thin quality I hear in cinemas . If I am right the sound still goes through perfections in the screen ?
"housekeeping" is a good way to put it. And it's true, I do pay much attention to that, and very probably overdo it sometimes. I tend to err on too much control, never too little. Probably my 25% of Austrian blood, and everybody knows Austrians are simply southern Germans.
In effect, it's really Otala & Lohstroh underneath it all, just done my way. Why amplify something with a factor of 99 dB and then slug it with 74 dB of global NFB, when you can have it amplify say 40 dB, make much less errors along the way, and use just 20 dB of global NFB, which in that case really serves only to make a good thing better, definition of keeping house, equivalent to mopping up the floor? Solve problems as much as you can when and where they appear, don't put them off for global NFB.
As for your comments, as ever, I welcome comments. I learn from them infinitely more than from back slapping.
In the ideal world you do use 74dB feedback . Never is it the ideal world and that's the problem . As far as I know amps with too much feedback are on the verge of oscillation . Square wave overshoot seems different and often is OK . Happy to be wrong about that .
Well, we could say one uses "centralized" and "distributed" NFB.
When one uses "distributed" NFB, one is referring to use of local NFB, which in turn reduces the need for global.
When one uses "centralized" NFB, then one uses little local NFB and leaves most of the work to be done by global NFB, which in turn then needs to be bigger.
Taking this into account, I would agree that in both cases we are in fact using about the same overall value compared to true zero NFB model, it's more about how we distribute it. It's not the neck, it's the throat.
Local negative feedback makes that bit linear rather than increase the dominant pole . As long as the end result is the best then it was the best thing to do . As I said yesterday you might win 3 dB at 20 kHz . In principle local feedback is a disaster . In REALITY it probably is the better idea . Where you have to be careful is when a new device comes along . Lets pretend it is a zero volt bias JFET with Ft at 100 MHz . By some quirk it has 100 pF input capacitance . No bias , not stability problems and loads of bandwidth . It MIGHT be a game changer . It might be as nonlinear as is possible . As long as it accepts loads of feedback it will work fine . The real world is different .
I am too lazy to rewrite something sent to a friend . Remember it is not for the forum so is different ( slightly ) .
" Usually there is no easy combination of valve and transistor . It would almost be like a steam cum petrol engine . Quite a nice idea now I think about it .
First thing to say is valves do not have a good sound . Mostly they have a very bad sound . Because of the way they work and tradition ( stinginess ) people try to get the least worst sound out of them . This happens to be better than 90% of transistor amps .
How so ? Easy . Transistor amps are like apple pie that has some fish in it . Fish are cheap and they can not resist putting fish in .
Jean Hiraga made an amp to try to resolve this .
I have been told the Missing Link sounds just like a valve design . Well it sounds more detailed than many . Richard Brice designed a valve phono stage that might challenge it . The Missing Link doesn't sound like a valve design . I just sounds OK and that's it .
The ideal valve transistor hybrid might destroy speakers . For that reason like a Quad 303 it should be AC coupled .
One thing I suggested to Art Audio is to make a transistor buffer for the EL84 . The 84 in triode gives 1.5 watts . With a buffer the sky is the limit . The amp would need an output transformer . So it would look puny and cost loads of money . So we didn't do it . The ideal transistor buffer has zero distortion if Early effect is avoided ( > -80dB ) . It should and would mimic the EL84 very well . EL 84 probably was the best sounding valve ever made . It is almost the best transistor also as it works almost like one . The old Heathkit EL84 amps amongst the best I ever heard regardless of device .Very open , fast, and white sounding ( white spectrum ) ."
I am too lazy to rewrite something sent to a friend . Remember it is not for the forum so is different ( slightly ) .
" Usually there is no easy combination of valve and transistor . It would almost be like a steam cum petrol engine . Quite a nice idea now I think about it .
First thing to say is valves do not have a good sound . Mostly they have a very bad sound . Because of the way they work and tradition ( stinginess ) people try to get the least worst sound out of them . This happens to be better than 90% of transistor amps .
How so ? Easy . Transistor amps are like apple pie that has some fish in it . Fish are cheap and they can not resist putting fish in .
Jean Hiraga made an amp to try to resolve this .
I have been told the Missing Link sounds just like a valve design . Well it sounds more detailed than many . Richard Brice designed a valve phono stage that might challenge it . The Missing Link doesn't sound like a valve design . I just sounds OK and that's it .
The ideal valve transistor hybrid might destroy speakers . For that reason like a Quad 303 it should be AC coupled .
One thing I suggested to Art Audio is to make a transistor buffer for the EL84 . The 84 in triode gives 1.5 watts . With a buffer the sky is the limit . The amp would need an output transformer . So it would look puny and cost loads of money . So we didn't do it . The ideal transistor buffer has zero distortion if Early effect is avoided ( > -80dB ) . It should and would mimic the EL84 very well . EL 84 probably was the best sounding valve ever made . It is almost the best transistor also as it works almost like one . The old Heathkit EL84 amps amongst the best I ever heard regardless of device .Very open , fast, and white sounding ( white spectrum ) ."
Last edited:
The forum member I visited recently has done that very thing too - just knocked up a super primitive OB to check out the units, and was so happy with the result that he's decided to dump the "in a box" solution; will soon tart up a decent looking OB arrangement. Of course, that it's fully active, and a DEQX unit can adjust the response with great accuracy helps; we ran a frequency sweep and subjectively it was very clean.
My friend John has offered to lend me his DSP used for his Pro Audio work . We found it could be configured the way I would want it to work . This would seldom be used for Pro Audio ( 1st order ) . As I said to John why not do it my way when classical music ? He does plenty of classical and seeks a proper sound . He uses the old Meridian M3 active speakers at home as he feels they are more to his way of thinking correct .
Many times I bottomed the main driver on the OB . My next step is to integrate the 15 inch woofer . If I can get 30 Hz - 6dB I will be happy . I hope to cut the main driver as low as 80 Hz . The 15 inch will take plenty of power so EQ should be possible .
Many times I bottomed the main driver on the OB . My next step is to integrate the 15 inch woofer . If I can get 30 Hz - 6dB I will be happy . I hope to cut the main driver as low as 80 Hz . The 15 inch will take plenty of power so EQ should be possible .
That depends entirely on the implementation. The challenges are greater, indeed, however to say they "fail" is patently false.
OB Theory
I've repeatedly read reports of incredibly good bass from proper implementations of OB, so when I hear reports of opposite, I suspect sub-optimal implementation.
I have yet to try it myself, but certainly will do so one of these years.
What is patently false is your assumptions .....
Perhaps the next time you try OB bass, you should add some of those Quantum BeeGees thingies. I am quite certain that'll make 'em work great!
- Status
- Not open for further replies.