No, this was a guy who was trying to prove that a high distortion SE amp will result in lower system distortion because of interaction with speaker distortion. The data was pretty sketchy and it looked like noise artifacts completely overwhelmed anything he was trying to see.
Morgan Jones made a passing reference to the interstage distortion cancelling thing in his book. He pointed out that it won't generally work for cascaded gain stages. EC8010's interesting experiment would be an exception, but that's because of the cf's unity gain.
Morgan Jones made a passing reference to the interstage distortion cancelling thing in his book. He pointed out that it won't generally work for cascaded gain stages. EC8010's interesting experiment would be an exception, but that's because of the cf's unity gain.
EC8010 said:
I completely forgot having done the same with a zero-feedback two-stage EL84 SE. By connecting one input of an Audio Precision analyzer (thanks work!) interstage and the second input to the OPT secondary it was a simple matter to alter the CCS bias of the first stage and watch the OPT THD fall as the front end THD increased. My ear wasn't happy with the results and eventually the circuit went a different route.
More musings on tube/valve amps
There are some good points raised above.
Years ago Mr Partridge talked about "Transformer Distortion" and how distortion (predominantly 2nd harmonic) rises with lowering frequency as the impedance of the primary inductance becomes close to and then less than the rp of the output tube(s). That is a tube amp with an output transformer has higher levels of second harmonic distortion at low frequencies and this distortion reduces as the frequency increases. This gives the amp that valve "warmth" and impression of power and is, in general what has been traditionally referred to as that "valve sound". I am in full sympathy with those folk who say "if it doesn't have an output transformer then it ain't a proper valve amp".
Unfortunately limitations in the output transformer will always introduce limitations in the final amplifier. Output tansformer limitations in technical terms means:
a) limited primary inductance
b) excessive leakage inductance and shunt capacitance
Both of these things can be addressed with Local feedback at the output stage via Ultralinear connection, Cathode feedback, shunt feedback etc to effectively reduce the output tube(s) rp.
6dB of local feedback will halve the rp which will:
a) shift the artificial bass enhancment a full octave lower - in general it won't shift the low frequency roll off because you will run into core saturation BUT it will allow you to "tune" the low frequency harmonic content.
b) shift the high frequency roll off somwhat higher (not a full octave because its a 12dB/octave slope)
Of course - as an alternative to lowering output tube rp you can purchase top of the line toroidal output trannies with massive primary inductance (>500H) and small leakage inductance etc. and those of us who have done that will tell you that amps with these output trannies have an entirely different bass character - subjectively weaker but deeper and tighter. I lean toward cheaper output trannies which I have tuned via adjusting rp of the output tube(s) to get somewhere between the two extremes.
Example: I have a PPP KT88 Ultralinear 120 watt prototype I've run with both Hammond 1650T (Lp = 59H) and Toroidal VDV2100 (Lp=560H). In general I prefer the Hammond BUT only because I'm into low/no global feedback. As soon as you try to apply global feedback the massive bandwidth of the VDV2100 would make life MUCH easier.
The more local feedback you apply to the output stage to lower rp however the more demand you place on the driver stage and the more difficult it is to meet EC8010's ''harmonic content to remain constant with level" requirement.
The other issue raised above is speakers. We rave endlessly about total harmonic distortion figures of our amps BUT seem to gloss completely over the approximately 5% THD contributed by the typical speaker system. I'm not going to get into current drive vs voltage drive of speakers here except to make a point which is almost always overlooked.
If we talk about damping of speakers we talk about the speakers Qts. This is the parallel combination of the Qes (Electrical Damping) and Qms (Mechanical Damping). The way we drive the speaker and what we drive it with has absolutely no effect whatsoever on Qms. Amplifier/Speaker interaction has to do with the Qes ONLY and saying " My speaker is nominally 6 Ohms and my Amp Zout is 1.5 Ohms, therefore I have a damping factor of 4, MAY be entirely meaningless or at least may not correlate with actual damping of the speaker response). As for speakers cancelling the harmonic distortion introduced by the amplifier (sorry fellows), if you can show me a speaker with a transfer function identical to the amp or even the output stage of the amp (at all levels) which I can connect in antiphase I'll think about changing my mind BUT in the meantime I'll put it in the "pure bunk" audiophool basket.
Cheers,
Ian
There are some good points raised above.
Years ago Mr Partridge talked about "Transformer Distortion" and how distortion (predominantly 2nd harmonic) rises with lowering frequency as the impedance of the primary inductance becomes close to and then less than the rp of the output tube(s). That is a tube amp with an output transformer has higher levels of second harmonic distortion at low frequencies and this distortion reduces as the frequency increases. This gives the amp that valve "warmth" and impression of power and is, in general what has been traditionally referred to as that "valve sound". I am in full sympathy with those folk who say "if it doesn't have an output transformer then it ain't a proper valve amp".
Unfortunately limitations in the output transformer will always introduce limitations in the final amplifier. Output tansformer limitations in technical terms means:
a) limited primary inductance
b) excessive leakage inductance and shunt capacitance
Both of these things can be addressed with Local feedback at the output stage via Ultralinear connection, Cathode feedback, shunt feedback etc to effectively reduce the output tube(s) rp.
6dB of local feedback will halve the rp which will:
a) shift the artificial bass enhancment a full octave lower - in general it won't shift the low frequency roll off because you will run into core saturation BUT it will allow you to "tune" the low frequency harmonic content.
b) shift the high frequency roll off somwhat higher (not a full octave because its a 12dB/octave slope)
Of course - as an alternative to lowering output tube rp you can purchase top of the line toroidal output trannies with massive primary inductance (>500H) and small leakage inductance etc. and those of us who have done that will tell you that amps with these output trannies have an entirely different bass character - subjectively weaker but deeper and tighter. I lean toward cheaper output trannies which I have tuned via adjusting rp of the output tube(s) to get somewhere between the two extremes.
Example: I have a PPP KT88 Ultralinear 120 watt prototype I've run with both Hammond 1650T (Lp = 59H) and Toroidal VDV2100 (Lp=560H). In general I prefer the Hammond BUT only because I'm into low/no global feedback. As soon as you try to apply global feedback the massive bandwidth of the VDV2100 would make life MUCH easier.
The more local feedback you apply to the output stage to lower rp however the more demand you place on the driver stage and the more difficult it is to meet EC8010's ''harmonic content to remain constant with level" requirement.
The other issue raised above is speakers. We rave endlessly about total harmonic distortion figures of our amps BUT seem to gloss completely over the approximately 5% THD contributed by the typical speaker system. I'm not going to get into current drive vs voltage drive of speakers here except to make a point which is almost always overlooked.
If we talk about damping of speakers we talk about the speakers Qts. This is the parallel combination of the Qes (Electrical Damping) and Qms (Mechanical Damping). The way we drive the speaker and what we drive it with has absolutely no effect whatsoever on Qms. Amplifier/Speaker interaction has to do with the Qes ONLY and saying " My speaker is nominally 6 Ohms and my Amp Zout is 1.5 Ohms, therefore I have a damping factor of 4, MAY be entirely meaningless or at least may not correlate with actual damping of the speaker response). As for speakers cancelling the harmonic distortion introduced by the amplifier (sorry fellows), if you can show me a speaker with a transfer function identical to the amp or even the output stage of the amp (at all levels) which I can connect in antiphase I'll think about changing my mind BUT in the meantime I'll put it in the "pure bunk" audiophool basket.
Cheers,
Ian
Having watched this poor suggestion get smacked around like the proverbial red-headed stepchild at a family picnic I feel obligated now to put up at least a minimal defense. Here comes the "ifs".
As I understand it, in any normal tube circuit the second harmonic generated by an SE stage is always anti-phase to the fundamental. If the same relationship holds true of a loudspeaker driver in theory cancellation would seem entirely possible. In fact I can't see what would stop it. Fundamental principles repeat themselves endlessly throughout the physical world, an electrical Butterworth is described by essentially the same set of equations as an acoustical for example, giving little obvious reason to dismiss the notion out of hand. Now, why is it necessary to assume that at an amp-to-speaker interface 2nd order cancellation must occur throughout the operating range? If the matching isn't perfect everywhere is there any reason it can't be close enough over a range of volume? At normal listening levels the 2nd order of amp and speaker might be close enough to permit significant (theoretical) cancellation. At higher volumes the transfer functions diverge and the sound becomes congested or strident or whatever the correct manifestation would be. At lower volumes the mismatch might fall below the audibility threshhold. There's also no reason to assume this 'matching' behaviour would be universal between all amps and speakers. Might explain system matching beyond the obvious efficiency/power criteria.
Seems to me the main questions are "do loudspeakers generate 2nd order distortion in anti-phase to the fundamental?" and "is there a voltage operating range over which the 2nd order distortion of the amp and loudspeaker are close enough in magnitude to permit significant operation?" .
Again, this is all said out of a sense of moral obligation to a severly whupped notion. I have no idea if it ever happens and in the morning might wonder what I was thinking. If it does occur though it suggests an experiment to examine, for example, Herb Reichert's notion that an SE amp is always voiced with a particular speaker in mind. Or the reputation of synergistic equiment matches. I wonder if it's possible to coax Rightmark into making the measurements.
As I understand it, in any normal tube circuit the second harmonic generated by an SE stage is always anti-phase to the fundamental. If the same relationship holds true of a loudspeaker driver in theory cancellation would seem entirely possible. In fact I can't see what would stop it. Fundamental principles repeat themselves endlessly throughout the physical world, an electrical Butterworth is described by essentially the same set of equations as an acoustical for example, giving little obvious reason to dismiss the notion out of hand. Now, why is it necessary to assume that at an amp-to-speaker interface 2nd order cancellation must occur throughout the operating range? If the matching isn't perfect everywhere is there any reason it can't be close enough over a range of volume? At normal listening levels the 2nd order of amp and speaker might be close enough to permit significant (theoretical) cancellation. At higher volumes the transfer functions diverge and the sound becomes congested or strident or whatever the correct manifestation would be. At lower volumes the mismatch might fall below the audibility threshhold. There's also no reason to assume this 'matching' behaviour would be universal between all amps and speakers. Might explain system matching beyond the obvious efficiency/power criteria.
Seems to me the main questions are "do loudspeakers generate 2nd order distortion in anti-phase to the fundamental?" and "is there a voltage operating range over which the 2nd order distortion of the amp and loudspeaker are close enough in magnitude to permit significant operation?" .
Again, this is all said out of a sense of moral obligation to a severly whupped notion. I have no idea if it ever happens and in the morning might wonder what I was thinking. If it does occur though it suggests an experiment to examine, for example, Herb Reichert's notion that an SE amp is always voiced with a particular speaker in mind. Or the reputation of synergistic equiment matches. I wonder if it's possible to coax Rightmark into making the measurements.
Thank goodness we've had a post talking about Qes and Qms instead of "damping factor"! Loudspeakers and amplifiers should ideally be designed to complement one another. Trouble is, that really will mean that one amplifier will suit one loudspeaker, and vice versa. Most modern drivers are electrically overdamped (Qes < 0.5), so driving them from an amplifier with a non-zero output impedance will help the bass. And if the driver's actually in the right size box, the bass might even be flatter and more extended, rather than just more of it. So that deals with < 200Hz. Above 200Hz we're into the realms of non-constant loudspeaker impedance affecting the voltage at the loudspeaker terminals. For a full-range driver, the solution is simple - add a Zobel network to correct the impedance. Loudspeaker crossovers usually assume zero source resistance...
Turning to the idea of cancelling 2nd harmonic distortion between loudspeaker and amplifier, regretably, I think I'm going to shoot it down. As rdf points out, just because the loudspeaker is an acoustical system, doesn't mean it doesn't follow the same shapes of equations as an electrical system, so that makes cancellation plausible. Unfortunately, the acoustical coupling efficiency from the cone doubles with doubling of frequency - if it wasn't for the fact that above resonance, the driver's motion is mass-controlled, this would mean a 6dB/oct rising response. As it is, the two effects cancel to give a flat response. However, it does mean that driver excursion falls by 6dB/oct above resonance. Since 2nd harmonic distortion is proportional to excursion, it must also fall with frequency. But the amplifier's 2nd harmonic distortion is constant with frequency. I suppose it might just be possible to juggle output transformer primary inductance so that distortion due to the output transformer matched loudspeaker distortion. Thoughts?
Turning to the idea of cancelling 2nd harmonic distortion between loudspeaker and amplifier, regretably, I think I'm going to shoot it down. As rdf points out, just because the loudspeaker is an acoustical system, doesn't mean it doesn't follow the same shapes of equations as an electrical system, so that makes cancellation plausible. Unfortunately, the acoustical coupling efficiency from the cone doubles with doubling of frequency - if it wasn't for the fact that above resonance, the driver's motion is mass-controlled, this would mean a 6dB/oct rising response. As it is, the two effects cancel to give a flat response. However, it does mean that driver excursion falls by 6dB/oct above resonance. Since 2nd harmonic distortion is proportional to excursion, it must also fall with frequency. But the amplifier's 2nd harmonic distortion is constant with frequency. I suppose it might just be possible to juggle output transformer primary inductance so that distortion due to the output transformer matched loudspeaker distortion. Thoughts?
EC8010 said:
In my experience this isn't true, in an SE amp the 2nd rises rapidly at low frequencies due to the mechanisms outlined by gingertube above. For example, the OPTs used to prototype my heapdphone amp (vintage 7 kohm Hammond 354) are poor quality. It works reasonably well only because of low drive impedance (6c45) and lightly loaded secondaries. Still THD, overwhelmingly 2nd, is ~ 0.15% at 1 kHZ at tinnitus levels. At 50 Hz it's 5%. Seeing this was an ah-ha moment regarding SE's typically abysmal IMD figures, the standard mix frequencies are 60 Hz and 7 kHz.
I'm beginning to feel like an ACLU lawyer defending the 2nd.
rdf said:In my experience this isn't true, in an SE amp the 2nd rises rapidly at low frequencies due to the mechanisms outlined by gingertube above. For example, the OPTs used to prototype my heapdphone amp (vintage 7 kohm Hammond 354) are poor quality...
...I'm beginning to feel like an ACLU lawyer defending the 2nd.
But surely some SE amplifiers are made with decent iron? Or are you arguing that they all have nasty output transformers?
What's an ACLU lawyer?
Certainly, however it's not the absolute magnitude but the overall pattern of rising 2nd order distortion with lowering frequency which is potentially relevant. I think that behaviour's universal. The magnitude of distortion (obviously) would be circuit as well as transformer dependent.
ACLU = American Civil Liberites Union. Not famous for rushing to the defense of the 2nd Amendment, leaving that to the NRA.
ACLU = American Civil Liberites Union. Not famous for rushing to the defense of the 2nd Amendment, leaving that to the NRA.
There are high quality OPT C core transformers both PP and SE from Bartolucci and others with primary L of 300H or greater which is about equivalent to an xL of 38K ohms at 20Hz - arguably high enough that a SE output triode with an RP of 700 ohms would not see significant additional loading due to the thevenin equivalent of xL and the reflected load impedance on the secondary. Obviously there may be some trade offs with transformers with such high primary L such as increased leakage inductance and stray capacitance between windings.
Incidentally this is one area where a parafeed configuration can offer considerable advantages at both ends of the response range, particularly in low power applications.
It is quite possible to build SE amplifiers with relatively constant distortion across the whole frequency range, assuming good driver design what is required is a rather expensive and well engineered output transformer.
Incidentally this is one area where a parafeed configuration can offer considerable advantages at both ends of the response range, particularly in low power applications.
It is quite possible to build SE amplifiers with relatively constant distortion across the whole frequency range, assuming good driver design what is required is a rather expensive and well engineered output transformer.
I'm not able to test my SE amplifier right now (CV, could I have it back one day?), but I dug out my lab book. Maximum power (output stage onset of grid current); 6.8W at 3.2% THD. Distortion at 25Hz 1dB below full power 3.5% (H2 = -29dB, H3 = -54dB, H4 = -55dB, H5 = -79dB). HF response; -1dB at 14kHz, -3dB at 26.5kHz. The price paid for decent LF response is a less-than-perfect 10kHz square wave.
Attachments
This thread is interesting. You guys are obviously all intelligentand provide thoughtful well reasoned posts. But, what I find most interesting is the fact that the topic has turned to 2nd HD. Let me ask you guys, when the work day is done, when you go home and put on some music, I don't care if you sit in your special chair in the sweet spot and go swimming in the music, or if you listen 'critically' and try to identify the circular breathing of the flautist, or if you put on 80's long hair rock then go into the kitchen and dance while you make dinner, I want to ask you: how much difference do you think 2nd HD makes to the quality of the sound you hear?
I understand the temptation. As logically minded problem solvers we tend to look for problems we understand. I spent many years in a university learning to express problems in a way that made them easy to deal with. But, I gotta think that the road of 2nd HD is not going to lead anyone to a greater understanding of the question posed at the start of this thread. It never has before.
I do like the observation by EC8010 that in amps with NFB the feedback factor is frequency dependent. I like it because it's been my subjective observation that amps with NFB *tend* to be more two-dimensional while 3-D realism tends to be found more easily in amps without NFB. So what is it that the NFB is damaging? I see it as damage to the sound that makes it more difficult (less easy?) for our hearing to assemble the original experience.
My hunch, which is not very sophisticated, is that it has to do with subtle phase relationships between different parts of what are supposed to be the same sound. Not sure exactly where to go with that though.
Some years ago, when I was researching headphones, I came across a study that partly explains why headphones will never sound like a pair of speakers (never mind the live experience.) What was revealed to me was that our ears detect different frequencies differently. If I remember right, low frequencies are detected directly by the ear drum while high frequencies actually travel around our head through the skull before they reach our ears. It seems obvious that our brain uses the various time delays to help assemble a 3D 'image' in our mind. Again, I don't know where this is going ...
In any case, I don't think that the most obvious engineering approach of reducing the kinds of distortion that are easiest to measure is going to get to the bottom of the problem. The challenge as I see it is to accept that the way the problem has been expresses in the past isn't working. We still need to be good competent engineers, really we need to be even more competent than before and find another way to express the problem that leads to a workable solution.
Thanks for letting me rant.
-- Dave
I understand the temptation. As logically minded problem solvers we tend to look for problems we understand. I spent many years in a university learning to express problems in a way that made them easy to deal with. But, I gotta think that the road of 2nd HD is not going to lead anyone to a greater understanding of the question posed at the start of this thread. It never has before.
I do like the observation by EC8010 that in amps with NFB the feedback factor is frequency dependent. I like it because it's been my subjective observation that amps with NFB *tend* to be more two-dimensional while 3-D realism tends to be found more easily in amps without NFB. So what is it that the NFB is damaging? I see it as damage to the sound that makes it more difficult (less easy?) for our hearing to assemble the original experience.
My hunch, which is not very sophisticated, is that it has to do with subtle phase relationships between different parts of what are supposed to be the same sound. Not sure exactly where to go with that though.
Some years ago, when I was researching headphones, I came across a study that partly explains why headphones will never sound like a pair of speakers (never mind the live experience.) What was revealed to me was that our ears detect different frequencies differently. If I remember right, low frequencies are detected directly by the ear drum while high frequencies actually travel around our head through the skull before they reach our ears. It seems obvious that our brain uses the various time delays to help assemble a 3D 'image' in our mind. Again, I don't know where this is going ...
In any case, I don't think that the most obvious engineering approach of reducing the kinds of distortion that are easiest to measure is going to get to the bottom of the problem. The challenge as I see it is to accept that the way the problem has been expresses in the past isn't working. We still need to be good competent engineers, really we need to be even more competent than before and find another way to express the problem that leads to a workable solution.
Thanks for letting me rant.
-- Dave
Dave's right, we didn't answer the questions.
EC8010 orignally asked:
There is a need most of us feel to justify it, technically. But what about the more obvious facts that tubes are easier to work with, generally run in simple circuits, and have a "look-and-feel" wow factor that no 2N3906 will ever have? I, for one, am not a believer in the religion that our tube offerings are superior sonically to anything the transistor guys could throw together.
Your other question was:
No. If you listen to a SET for a year, which by your own measurements is doing 3.4% distortion across the spectrum, and then switch to a feedback tube amp with under 0.5%, you're going to say it sounds weak, or sterile, or whatever. Point is, you're not going to like it - at first. But listen to that for a year, and then switch... now you'll hear the same thing in the SET that I do... grunge.
Joel
EC8010 orignally asked:
There is a need most of us feel to justify it, technically. But what about the more obvious facts that tubes are easier to work with, generally run in simple circuits, and have a "look-and-feel" wow factor that no 2N3906 will ever have? I, for one, am not a believer in the religion that our tube offerings are superior sonically to anything the transistor guys could throw together.
Your other question was:
No. If you listen to a SET for a year, which by your own measurements is doing 3.4% distortion across the spectrum, and then switch to a feedback tube amp with under 0.5%, you're going to say it sounds weak, or sterile, or whatever. Point is, you're not going to like it - at first. But listen to that for a year, and then switch... now you'll hear the same thing in the SET that I do... grunge.
Joel
I'm a believer...
Absolutely right, a 2N3906 has all the charisma of a wet weekend, but even without that head start how could the fiery glow of valve amplifier fail to captivate our hearts? Oh, and let's not forget that sublime midrange. Early Kate Bush through silicon? Oh, no!
As it happened, I listened to the SET for about a year. But its replacement with a PP (and yes, with a touch of feedback) was a breath of fresh air. But I hear what you're saying - you do become acclimatised to the sound of an amplifier.
To go back to the H2 argument. I'm afraid I take the purist engineer approach and say that I want as little distortion with as few harmonics as possible, and if there's an option on lower orders, then I'll have them. Not much of an answer, but I doubt if I'd allow H2 to rise unless it was already very low or H3 unpleasantly high - it's all a matter of degree. Regarding the stereo argument, I'm rarely convinced by stereo anyway...
My perception is that global feedback can take away some of the "air" of the sound. Maybe I just like H2?
Absolutely right, a 2N3906 has all the charisma of a wet weekend, but even without that head start how could the fiery glow of valve amplifier fail to captivate our hearts? Oh, and let's not forget that sublime midrange. Early Kate Bush through silicon? Oh, no!
As it happened, I listened to the SET for about a year. But its replacement with a PP (and yes, with a touch of feedback) was a breath of fresh air. But I hear what you're saying - you do become acclimatised to the sound of an amplifier.
To go back to the H2 argument. I'm afraid I take the purist engineer approach and say that I want as little distortion with as few harmonics as possible, and if there's an option on lower orders, then I'll have them. Not much of an answer, but I doubt if I'd allow H2 to rise unless it was already very low or H3 unpleasantly high - it's all a matter of degree. Regarding the stereo argument, I'm rarely convinced by stereo anyway...
My perception is that global feedback can take away some of the "air" of the sound. Maybe I just like H2?
Well, I’m pretty sure I do SE for the music. I listen to a lot more now, of all types. My previous main amp for near twenty years, an MC240, sits under the TV looking beautiful but dusty. Decent enough professional audio systems are playing around me in studios all day long so I'm constantly aware of that sound. Every now and then an SS amp comes home for a few days. It's something which has been on my mind a lot since building that first SE and in the hope of gleaning some insight I ordered a Peter Daniels Gainclone kit. It should make an excellent alternate reference when done and from a measurement vs. research into audibility thresholds perspective is arguably impeachable.
Looked at from a different angle, of my two EL84 SE amps one is the venerable Mullard 3-3, a direct-interstage coupled high feedback (for a SET) design. At normally listening levels, say ~2-3 volts P-P into a pair of 12" Tannoys, the mid-band distortion measures about 0.1% or less across a wide range. The second amp is a self-design and uses much the same parts (Russian EL84M, Mills resistors, ASC PS caps), has feedback only around the OPT stage via the secondary as a cathode load, and measures THD five times worse than the Mullard. The Mullard is far and away the more abrasive and hard. Spectral analysis of the harmonics afterwards almost certainly revealed why. The Mullard achieves low single numerical THD by clamping down the second harmonic at the expense of a dense series of higher harmonics into the teens. The home-brew's distribution is strong second, a third 10-20 dB further down (lunch break, graphs at home) and very little else at the same operating level. From an engineering perspective the Mullard would normally be considered five times better everything else being equal.
Looked at from a different angle, of my two EL84 SE amps one is the venerable Mullard 3-3, a direct-interstage coupled high feedback (for a SET) design. At normally listening levels, say ~2-3 volts P-P into a pair of 12" Tannoys, the mid-band distortion measures about 0.1% or less across a wide range. The second amp is a self-design and uses much the same parts (Russian EL84M, Mills resistors, ASC PS caps), has feedback only around the OPT stage via the secondary as a cathode load, and measures THD five times worse than the Mullard. The Mullard is far and away the more abrasive and hard. Spectral analysis of the harmonics afterwards almost certainly revealed why. The Mullard achieves low single numerical THD by clamping down the second harmonic at the expense of a dense series of higher harmonics into the teens. The home-brew's distribution is strong second, a third 10-20 dB further down (lunch break, graphs at home) and very little else at the same operating level. From an engineering perspective the Mullard would normally be considered five times better everything else being equal.
That doesn't mean it isn't unstable at the frequency extremes, or that it doesn't have large amounts of distortion there either. I've built "textbook" circuits from the 50's/60's (literally out of textbooks) that had 6dB or more peaks at 2Hz and 80kHz. Guess how those amps sounded?
Never assume.
As for spectrums, they are only one bit of information. You need all the others to put it into a valid context.
Joel
Yes, it's well worth checking the stability of a circuit with your output transformer and your layout - otherwise there's a real danger of simultaneous HF and LF instability. A 10kHz square wave quickly shows up HF problems, but if your oscillator can't do square waves, then a quick sweep to check that there isn't a peak in the response is almost as fast. At LF, a slow sweep is needed. (The typical Wien bridge oscillator takes "x" cycles to stabilise, and while that's fast at HF, it can be quite a while at LF.)
PS I liked your mirror analogy, Dave.
PS I liked your mirror analogy, Dave.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.