If you fellas want to help me out maybe we can rearrange this amp properly. After I fix the oscillation of course. I haven't had a chance to look at it again. I plan on doing it tonight.
I am thinking maybe reduce that 470k (Rload) and possibly change the 1.5k resistors. I will have to read up on the math to figure out the optimum load. It looks like the amp has 12db attenuation at the input? Maybe this will help with some wiggle room in case I lose some gain from rearranging the SRPP.
DF96, I know you were not being racist, it's all good.
Maybe if we keep this topic going on the SRPP we can get Keit in here to argue with people and really shake things up 😉
I am thinking maybe reduce that 470k (Rload) and possibly change the 1.5k resistors. I will have to read up on the math to figure out the optimum load. It looks like the amp has 12db attenuation at the input? Maybe this will help with some wiggle room in case I lose some gain from rearranging the SRPP.
DF96, I know you were not being racist, it's all good.
Maybe if we keep this topic going on the SRPP we can get Keit in here to argue with people and really shake things up 😉
If this is what you want, then you'd do better with this design (attached). It still bootstraps the gain stage to a higher gain while making for a shallow, low distortion loadline. It also provides a Lo-Z output, and lower than the SRPP. It also allows for larger output swing for the DC rail voltage.
It can be worked in two ways: the bootstrap capacitor can be sized to make it effective at all frequencies of interest. It can also be made quite small ( ~20pF ) so that this operates as an output buffered gain stage. The PFB becomes effective just on fast transients, with the capacitor acting as a "hurry-up" capacitor.
Attachments
Yes, of course. The active load is also a modified CF, so you get lower output impedance - although not as low as an unmodified CF would give you.SY said:
Good idea, fix the problem first, then you will have time for "topology rolling"
Hint: Check/redo solders first, most common failure into a PCB at high temperature.
Optimum load for SRPP topology can be ridiculously low, depending on valve, and then you will need enormous coupling capacitors.
I would leave it as this, with some improvements, e.g. lowering 1.5K resistors a little bit.
SRPP topology was introduced into audio world in the 60's by Anzai, in addition that it sounds great, it is the most forgiving topology over there, as I am quite dumb, that's why is my favorite. 😀
Last edited:
Sorry, but I cannot see any SMPS inductor, on the upper left I can see an input inductor that seems to be part of an EMI filter.
Never seen so well built Chinese amp, nice.
From this side of the world it seems like an SRPP, but who am I to contradict all the experts. 😀
I usually ration what and how many threads I contribute to - usually no more than two on any one day. This thread is going nicely without me, and I've been contributing to another thread, where with the help of a couple of other contributors, the person in error has finally began to understand he's in error. If a thread no longer interests me, I drop out.
If you want my views on something, in a thread where I did contribute, you can always just ask, as in "Keit, what do you think..." or some such. The system will alert me.
Not quite. Distortions in the input stage, though maybe small, multiply and propagate thru the following stages, intermodulating with the distortions of each following stage. So it's always of benefit to cancel out distortions in one stage, rather than just pushing them down using NFB afterwards. I would even say that distortion cancelling stages are especially useful when using NFB.
SRPP is one distortion cancelling stage, LTP being the other major contender. Both have their pros and cons.
BTW, the Audio Note Ongaku uses an SRPP as input stage as well. Hardly considered some piece of chi-fi junk, I would say.
I guess if we use Merlin's definition, then it is not quite a SRPP, since it is both direct coupled to the next stage and it uses the anode of the lower tube for the output instead of the cathode of the upper tube. It gets even weirder, the cathode resistors are not the same value and only the lower tube's Rk is bypassed, wonder what Kondo-san was shooting for with this "SRPP" design?
That's wrong.
Essentailly, the distortions of each stage get summed, not multiplied.
This fact comes from the well known trig identity:
k . SIN(A)SIN(B) = k/2 COS(A + B) - k/2 SIN(A - B)
The relavence of this comes from the mathematical descrption of a sinewave:
V = Vpeak.SIN(2pi.f) often wrtten as V = Vpeak.SIN(2Pi.f + P) to indicate a phase shift.
In the trig identify,
k is a constant represnting the distortion fraction;
SIN(A) represents an input frequency;
SIN(B) represents a second input frequency;
COS(A+B) corresponds to the resulting sum frequency
COS(A-B) corresponds to the resulting difference frequency.
Take an example: Say a stage has 1% distortion (rather high for an input stage)
Lets say the input from a previous stage is supposed to be pure 1V 1kHz but has 1% distortion.
The distortion products are, relative to the 1 kHz level:-
0HZ (DC): 0.01/2 COS(0);
2 kHz (2nd harm): 0.01/2 COS(2pi.2000);
So the output of this stage contains 0.5% distortion in addition to the 1% distortion from the previous stage.
This stage will also produce 1% 2nd harmonic, so, depending on phasing the distortion at the output is either 0.5% or 1.5%.
Since there is a phase inversion in each stage, in most cases it will be 0.5% . Now, send the signal on to a third stage, also distorting 1%.
The output of the third stage will contain:
0Hz (DC)
0.25% 2nd harmonic from intermoding the input;
1% 2nd harmonic from internal distortion;
We also have the 2nd harmonic beating with the fundamental. The resulting third is:-
0.005 x 0.01/2 ie 0.0025%.
Clearly, while 3rd harmonic is produced multistage amplifiers, the level is extremely low.
The rd harmonic level at the thrird stage output, depending phasing, is either:-
0.75%
1.25%.
In practice in tube amplifiers, the percent distortion rises with signal voltage level as the grid base is not scaled. Hence the distortion in an SE tube amp is essentially the output stage distortion. In stage order, 0.01% plus 0.1% plus 1% is near enough to 1%.
In a push pull amp, the voltage amplifying stage may contribute significantly if followed by a concertina phase splitter and the output stage is ultralinear or has local cathode feedback.
Tubes produce mainly 2nd harmonic and a tiny bit of 3rd. By using similar math, you can show that the 3rd is almost ano existent problem.
In solid state amps, it works differently, because device current are scaled to level and transistors are current drive. In a solid state amp, you need to get the distortion of each and every stage down.
Oh, yes, you're right, my fault. It's a half-mu variant, and no SRPP. Note the bypass cap is very small and only there to boost the top end of the audio band a little.
Not your fault at all, even Kondo himself called it an SRPP, and you are right about the small bypass cap, he must use it for shaping the high frequency response. But I think this one might be chucked into Broskie's "gratuious SRPP" bin.
Last edited:
What you forget is, that for each harmonic of every stage, intermodulations arise as well. And these do not simply add up or even subtract. Quite the contrary, for real music IMD quickly becomes the elephant on the dance floor, as Papa put it in his noteworthy write-up. Things become even more "interesting" when feeding these artifacts back to the start of the chain thru nfb..
Ok, of course, if all you do is listen to single sine waves...
Nope. It just doesn't work that way in tube amplifiers. My math doesn't imply it only works for sinewaves and not music. You can pick as complex an input signal as you like - you just end up needing a computer to do the calcs. The result is the same - the higher products of early stages don't get multiplied up in later stages, they get divided down.
If you take another look, you'll see that it was intermod I calculated.
Are you unfamiliar with high school trig? No prob if you are - I can explain it another way. Takes a lot more words though.
Re the effects of negative feedback: Its true that neg feedback creates higher order distortion products that would not be there without it. That doesn't change the fact that inherent distortion in later stages dominate in determining the system performance.
Since global neg feedback linearises the system, any neg feedback always results in less total intermod energy, regardless of the amount of feedback - except of coures more is always better unless you have stability problems or internal slew rate limitting.
Last edited:
No doubt, it does, but it would be swamp'd out by the output stage in this zero-feedback amplifier, so even if it is say 0.1% vs. say 0.5% for a plain old grounded cathode stage, would it be audible?
Sure(?) Maybe read this up if you don't believe what I type..
Percentage-wise maybe. But it means nothing for audibility. I'd take pure 5% H2 (on peaks) any day over a splash of complex IMD patterns..
This well known trigonometric identity is wrong.
The correct one comes from
2 sin(A) sin(B) = cos(A - B) - cos(A + B)
Which in turn becomes
k sin(A) sin(B) = (k/2) cos(A - B) – (k/2) cos(A + B)
Errk! You're right. That's what comes from relying on memory.
Doesn't change the conclusions though.
- Status
- Not open for further replies.