At 1.2mA the transconductance of an ECC83/12AX7 will be around 1.8mA/V (the exact value depends on anode voltage), so rk is 555R. The load on the output is 89k, as I said above, therefore open-loop gain of the first stage will be 89k/555 = 160. This assumes, of course that the BJT collectors present an infinite impedance - which should be a reasonable approximation.
This does come out to ~17.6 dB of spare OLG, so your 15-20 dB guesstimate was bang-on.
0.2% of THD, even if mostly 2nd, isn't exactly blameless though. Harmonic distortion should just about make it below the human hearing threshold but the directly-related IMD won't. That's what gives the circuit its "character". It could obviously be worse, like 0.1% @ 1k but 0.4% @ 10k.
0.2% of THD, even if mostly 2nd, isn't exactly blameless though. Harmonic distortion should just about make it below the human hearing threshold but the directly-related IMD won't. That's what gives the circuit its "character". It could obviously be worse, like 0.1% @ 1k but 0.4% @ 10k.
Last edited:
Not feedforward. It just bypasses the NFB at higher frequencies, and turns the whole conglomerate into a darlington source follower. This gives great stabilty. Haven't you seen the squarewaves? Loaded with 1uF there is almost no ringing.
Member
Joined 2009
Paid Member
The input tube has a near vertical load-line but does that mean it will maximize distortion - if the plate curves are well behaved it should be fine. Not the lowest distortion maybe but there are much worse ways to operate a tube.
Member
Joined 2009
Paid Member
You have to be careful about this. This feed into the LTP tail will modulate the tail current through the LTP. When you modulate the current through the tail of an LTP you have signal a Mixer (a.k.a. Gilbert cell) - although the way you have it set up it's not quite such a simple thing because of the CCS load on half of the LTP and not the other half. However you choose to look at it I can imagine the possibility that this will produce intermodulation products. Since this 'feedback' circuit works more at higher frequencies the THD will be higher there, just like Wahab found in his simulations.
Last edited:
I'm glad you are debating this. Actually, the SS stage isn't so usual and it's perhaps not so intuitive.
Yes, one of the LTP mosfet's will draw slightly more current, but I have chosen to bias them so that this "correction" current will be very small compared with the bias current, and won't produce much dist. And remember that this LTP only will be engaged when the output devices makes dirstortion.
An example.
Since the amp is partially in class A, the un-corrected distortion amplitude will be less than 0.1V under normal circumstances. Then the correction voltage will be around 96% of 0.1V, we can say 0.1 V to make things simple. This means that there will be an offset current of around 0.1V / 470 ohm = 0.2mA. And that is only a tiny fraction of the bias current wich is set to 13mA.
But...
At 20khz the capacitive load from the output stage will cause the LTP to draw around 3mA.
But...
3mA on 13mA biased LTP will probably just create max 0.1% THD.
So...
This may be the reason to why wahabs simulation showed some increased dist at 10khz. But the rise was subtle. I think it was the H3 that rose from 0.007% to 0.03, which is about four times.
If we think we want better performance, we only have to bias the LTP a bit more.
Regarding IM. Actually, I have never really tried to learn the theory behind the various causes of IM. So I have nothing to say about that.
Can't wahab set up an IM experiment with his simulation? But on the other hand, what level of IM is acceptable?
And wahab, if you are reading this. Will you perhaps send me that sim-file of my circuit, that would be nice. My email is found at my site - Embuddy
And one more thing, my prototypes are biased 600 in total, that's why you got so surprisingly good figures. But on the other hand, a bias of .12A isn't too much.
Yes, one of the LTP mosfet's will draw slightly more current, but I have chosen to bias them so that this "correction" current will be very small compared with the bias current, and won't produce much dist. And remember that this LTP only will be engaged when the output devices makes dirstortion.
An example.
Since the amp is partially in class A, the un-corrected distortion amplitude will be less than 0.1V under normal circumstances. Then the correction voltage will be around 96% of 0.1V, we can say 0.1 V to make things simple. This means that there will be an offset current of around 0.1V / 470 ohm = 0.2mA. And that is only a tiny fraction of the bias current wich is set to 13mA.
But...
At 20khz the capacitive load from the output stage will cause the LTP to draw around 3mA.
But...
3mA on 13mA biased LTP will probably just create max 0.1% THD.
So...
This may be the reason to why wahabs simulation showed some increased dist at 10khz. But the rise was subtle. I think it was the H3 that rose from 0.007% to 0.03, which is about four times.
If we think we want better performance, we only have to bias the LTP a bit more.
Regarding IM. Actually, I have never really tried to learn the theory behind the various causes of IM. So I have nothing to say about that.
Can't wahab set up an IM experiment with his simulation? But on the other hand, what level of IM is acceptable?
And wahab, if you are reading this. Will you perhaps send me that sim-file of my circuit, that would be nice. My email is found at my site - Embuddy
And one more thing, my prototypes are biased 600 in total, that's why you got so surprisingly good figures. But on the other hand, a bias of .12A isn't too much.
"The test of the machine is the satisfaction it gives you. There isn't any other test. If the machine produces tranquility it's right. If it disturbs you it's wrong until either the machine or your mind is changed." Robert M Pirsig.
Nice.
Empirically, my mind have never settled into a tranquil state if i'm not using low NFB and simple circuitry.
The only meaning with existence is happiness. If you are happy, everything is alright. If not you will fight fiercely to obtain it.
Nice.
Empirically, my mind have never settled into a tranquil state if i'm not using low NFB and simple circuitry.
The only meaning with existence is happiness. If you are happy, everything is alright. If not you will fight fiercely to obtain it.
There is one really bad way to operate a valve: wrong bias point. This design does not make that mistake - it has a good bias point.Bigun said:
The other bad way to operate a triode is to minimise the anode impedance - make it look into a short circuit. This completely eliminates the usual triode feedback, which is needed to linearise a triode. Hence, in the ideal case, you see the raw 3/2 power law response. This has all orders of distortion present, although for sufficiently small signals 2nd will dominate. That is why a cascode can only be used for small signals.
IM is caused by precisely the same mechanism as harmonic distortion: nonlinearity. If you can remember your algebra and trigonometry from school you will know that if you multiply a sine by a sine the result is two sines at the sum and difference frequencies. This is IM. If the two original sines happen to be at the same frequency then you get DC and second harmonic. That is all you need to know.Svitjod said:
Ideally, zero. In reality, somewhere near the noise level would be good. How much IM you can tolerate depends to some extent on what type of music you listen to. The more complex the music the less you want IM. By 'complex' I mean the number of simultaneously contributing instruments and voices, not the complexity of the tune!
Everyone has their own preferences for sound and circuits. Sound reproduction may require more than this.
Please enlighten an IM newbe. From the above definition by DF96, I assume that heavy IM normally would be worsened in an amplifier with many active elements after each other. The first element multiplies the signal with some factor and then the second stage takes over and multiplies that signal further. In this way, IM dist adds up.
I think this supports my usual ambitions to use simple circuits. With few active elements, there won't be much IM.
But let's not debate this further, now I think we must subject my SS stage for some IM tests. If the RC network really creates IM, then I have another solution in the back of my head that works equally good. But first some measurements! If my construction is flawed in some way, I'm very grateful for you help and inspiration.
Does anyone know the usual combinations of signals that normally constitutes an IM test?
I think this supports my usual ambitions to use simple circuits. With few active elements, there won't be much IM.
But let's not debate this further, now I think we must subject my SS stage for some IM tests. If the RC network really creates IM, then I have another solution in the back of my head that works equally good. But first some measurements! If my construction is flawed in some way, I'm very grateful for you help and inspiration.
Does anyone know the usual combinations of signals that normally constitutes an IM test?
No. As I said, IM is caused by nonlinearity. It doesn't matter where the nonlinearity comes from (lots of stages, a few poor stages).Svitjod said:
No.
I was not aware we were having a debate. I thought I was teaching you about IM.
No. An RC network will be linear, so it cannot create IM. It may, however, allow adjacent active devices to generate IM.
Member
Joined 2009
Paid Member
We could collectively re-engineer this design to further reduce distortion, or accomplish some other goal but as you have also realized from reading my tag line - the goal I have is to enjoy the sound and there are many circumstances where this does not mean lowest distortion. That of course, is a popular debate and all I have to say on that topic in this thread is 'live and let live' - do not criticize the choice of others but help them achieve their goal.
Do you want to reduce the distortion or change it in some way ? - what happens to the sound if you remove your 'RC network' ?
Regarding IM. Actually, I have never really tried to learn the theory behind the various causes of IM. [/QUOTE]
I've never learned it either. But I understand that any non-linear element will a) distort a pure sine wave to produce additional frequency components that are harmonically related to, and higher in frequency to, the input signal, b) mix two or more pure sine waves to produce additional frequency components that are not harmonically related and will be higher and lower than the input signal frequency.
It seems to me that in general, the level of IM will rise when there are more cascaded non-linear stages (most generalizations have exceptions and if they exist, this forum will find them !)
By the way, I have also been thinking about no-feedback hybrids - just for the mental challenge rather than too much belief in it - some stuff here for example:
http://www.diyaudio.com/forums/tubes-valves/174889-cellini-2-hybrid.html
Last edited:
OK, I have studied a bit on Wikipedia about IM. IM is a result of non linear behavior, and so is THD. The interesting thing for me is what makes a particular amplifier showing unacceptable amounts of IM.
DF96 were correct - any non linear device creates IM. But what creates excessive or unacceptable amounts?
About the RC network. It both sets the level of NFB, and it also sets the frequencey where this NFB starts to diminish. Up to about 30khz, the NFB is linear and roughly 26dB, then it fades out. Without the network, the amp would be having a NFB that was limited only by Crss of the output mosfets, and this would probably cause instability.
OK, every active element creates IM. The following stage will create more IM both of the original signals, but also from the side bands that was created in the first stage. For me, that suggests that the signal will contain a more complex pattern of IM products when several active elements are stacked after each other.
Wouldn't this suggest that there after all is some point in making circuits with few active elements?
Many transistors normally doesn't worsen the THD or IM taken as a bunch, but the various overtones and IM products will be of a more complex nature.
Now we are in to the subject psycho acoustics. Do we perceive a more complex pattern more disturbing than simple ones? I don't think there have been very much research on this.
DF96 were correct - any non linear device creates IM. But what creates excessive or unacceptable amounts?
About the RC network. It both sets the level of NFB, and it also sets the frequencey where this NFB starts to diminish. Up to about 30khz, the NFB is linear and roughly 26dB, then it fades out. Without the network, the amp would be having a NFB that was limited only by Crss of the output mosfets, and this would probably cause instability.
OK, every active element creates IM. The following stage will create more IM both of the original signals, but also from the side bands that was created in the first stage. For me, that suggests that the signal will contain a more complex pattern of IM products when several active elements are stacked after each other.
Wouldn't this suggest that there after all is some point in making circuits with few active elements?
Many transistors normally doesn't worsen the THD or IM taken as a bunch, but the various overtones and IM products will be of a more complex nature.
Now we are in to the subject psycho acoustics. Do we perceive a more complex pattern more disturbing than simple ones? I don't think there have been very much research on this.
Member
Joined 2009
Paid Member
I find a SS amplifier with too much distortion sounds 'dirty' 'muddy' and although when this is reigned-in and everything is done just right it makes for an easy listening and relaxing sound at low-medium volumes it is limited in its use. I call it the "sound of the 70's". My very first amplifier project was based on one of Hugh's designs and it falls into this category and I like it very much.
Based on that experience I thought this was the right approach, to engineer in some distortion. But the SS amps I've designed myself, to my ears, didn't respond to this approach very well. I found several ways that my amp designs could be easily set up to have a fair bit of dominant H2 harmonic distortion for example, but I preferred the sound without. I have come to the conclusion that tinkering with a SS amp to create the correct distortion profile requires some skill and understanding that is beyond me. The only option was to clone Hugh's designs and I chose instead to take my own path. I have a SS amp with gobs of feedback and it sounds glorious, but it is low distortion. I enjoy my clone of Hugh's amp and also my low distortion amp. Depends on music, speaker and other things as to how I use them.
Usually with a hybrid amp with tube at the front and SS at the output the designer makes the SS as clean as possible and relies on the tube pre-amp for the 'sound' signature. And that's why many hybrid amps have not been as successful as expected - because the 'sound' is to a large part, determined by the output stage and it's interaction with the speaker. Your design has the special benefit that you have tailored the SS output stage for 'sound' too.
p.s. just as a contrast, there are some hybrid designs by "Wavebourn" on this forum that deeply impress me - they are designed to use tubes and SS components to the best of their individual strengths and the resulting amplifiers are superbly engineered.
Based on that experience I thought this was the right approach, to engineer in some distortion. But the SS amps I've designed myself, to my ears, didn't respond to this approach very well. I found several ways that my amp designs could be easily set up to have a fair bit of dominant H2 harmonic distortion for example, but I preferred the sound without. I have come to the conclusion that tinkering with a SS amp to create the correct distortion profile requires some skill and understanding that is beyond me. The only option was to clone Hugh's designs and I chose instead to take my own path. I have a SS amp with gobs of feedback and it sounds glorious, but it is low distortion. I enjoy my clone of Hugh's amp and also my low distortion amp. Depends on music, speaker and other things as to how I use them.
Usually with a hybrid amp with tube at the front and SS at the output the designer makes the SS as clean as possible and relies on the tube pre-amp for the 'sound' signature. And that's why many hybrid amps have not been as successful as expected - because the 'sound' is to a large part, determined by the output stage and it's interaction with the speaker. Your design has the special benefit that you have tailored the SS output stage for 'sound' too.
p.s. just as a contrast, there are some hybrid designs by "Wavebourn" on this forum that deeply impress me - they are designed to use tubes and SS components to the best of their individual strengths and the resulting amplifiers are superbly engineered.
Last edited:
Bigun, the art of tailoring or voicing an amp is something that I too never had any luck with. These amps usually are some variant of a "blameless" amp and there are several ways to tweak around with the perceived sound. Personally, I think there is a major risk to get lost in the placebo smoke when applying this method, and it takes patience.
I'm more into Nelsons approach. Simplicity, low NFB, class A, mosfet and so forth. All the audiophile hobby horses activated at once. The problem is to finding out something interesting enough to build.
That philosophy usually pays off well. Perhaps the flavor doesn't suite you initially, but the inner quality is so good that you tend to stick to it. Amps designed for a specific "sound" are usually something that one gets tired off soon.
I have built some variations of "blameless" amps over the years. A recall a project from the Elektor magazine. It sounded rather warm and detailed. But after a while I found that there was something with the treble. It was very clear, but the amp sounded strained, as if it was designed for playing only military brass bands. I have noticed the same impression on many projects designed by "engineers", if you know what I mean. They are all made for military brass bands.
Constructors that are new in the game and ambitious tend to make amps that have that edgy feeling, but when they have gathered experience over the years, their products use to sound more digestible. The art is to make an amp that is pleasurable.
I'm more into Nelsons approach. Simplicity, low NFB, class A, mosfet and so forth. All the audiophile hobby horses activated at once. The problem is to finding out something interesting enough to build.
That philosophy usually pays off well. Perhaps the flavor doesn't suite you initially, but the inner quality is so good that you tend to stick to it. Amps designed for a specific "sound" are usually something that one gets tired off soon.
I have built some variations of "blameless" amps over the years. A recall a project from the Elektor magazine. It sounded rather warm and detailed. But after a while I found that there was something with the treble. It was very clear, but the amp sounded strained, as if it was designed for playing only military brass bands. I have noticed the same impression on many projects designed by "engineers", if you know what I mean. They are all made for military brass bands.
Constructors that are new in the game and ambitious tend to make amps that have that edgy feeling, but when they have gathered experience over the years, their products use to sound more digestible. The art is to make an amp that is pleasurable.
I forgot. I have not tailored anything with this amp. I used the SS stage since it was an old idea, and I wanted to use it since it was justified as a current booster after the tube stage. I used it since the feedback loop was so very short, and my ambition was really to make it transparent.
I have no talents when it comes to tailoring sound.
I have no talents when it comes to tailoring sound.
Again that fanny thing about tailoring the sound. An amp is not musical instrument, it should reproduce the recorded sound with higher fidelity possible.
Member
Joined 2009
Paid Member
There are quite a few options for current boosters, the simplest I used so far in class A is the Sziklai feedback pair
http://www.diyaudio.com/forums/headphone-systems/275180-grasshopper.html
I find that with global feedback amps the character of the components etc. is lost because the feedback removes all personality. In other words, feedback reduces distortion (when used right) and the amp becomes neutral. My ultimate SS amplifier design (TGM8) uses a high feedback factor and sounds superb.
But without feedback my experience is that the ampler will have some flavour and you can optimize it for your liking. Often it is achieved by choosing certain brands of tubes, resistors and capacitors. My most recent experience of this was with a simple SPUD amp and it sounds very nice, allowing me to use a particular speaker that never sounded good with my SS amp.
Nelson is one person who is skilled in all these arts. His simple designs have different distortion profiles and hence a different sound. Some people prefer the dynamics of the F5, some say it was too harsh for them. A wise person on this forum once wrote something along the lines of 'let's not argue, there is an amplifier design for every person'
Hi Guys
Because IM introduces non-harmonically related frequencies, it is much more objectionable to our ear than simple THD. Anything that reduces THD also reduces IM and it is very important to decouple and separate circuits, their supply nodes and the currents that circulate through and between them.
I stated my view on distortion above, so lowest THD inhernently yields lowest Im and the most accurate amplifier. On the latter point, I agree with Dadod that for hifi the goal is accurate reproduction rather than tone altering or tone creation. Those things are up to the musicians and the producer.
Have fun
Because IM introduces non-harmonically related frequencies, it is much more objectionable to our ear than simple THD. Anything that reduces THD also reduces IM and it is very important to decouple and separate circuits, their supply nodes and the currents that circulate through and between them.
I stated my view on distortion above, so lowest THD inhernently yields lowest Im and the most accurate amplifier. On the latter point, I agree with Dadod that for hifi the goal is accurate reproduction rather than tone altering or tone creation. Those things are up to the musicians and the producer.
Have fun
Very Interesting, this discussion. Bigun, you think high NFB removes the character?
Then how come that opamps usually are easy to distinguish from each other? Despite the fact that they usually have a very high NFB and low THD?
These contradictions are what makes hifi so interesting. Personally I believe the human hearing is so complex and haven't been explored enough, and it might be impossible to do so since we are talking about subject impressions.
I think every amp has a "flavor", whether it is "blameless" or not. So, even if a neutral sound is desirable, it may be impossible to reach. Then, why not make the best of the situation and follow our ears and/or instinct and tweak our amps so that they will have a benign sound rather than a "strained" or boring sound?
The F5 sounds harsh? I would say that the treble is a bit "radiant" and the sound indeed is lean and dynamic.
My Amp667 sounds completely different.
Then how come that opamps usually are easy to distinguish from each other? Despite the fact that they usually have a very high NFB and low THD?
These contradictions are what makes hifi so interesting. Personally I believe the human hearing is so complex and haven't been explored enough, and it might be impossible to do so since we are talking about subject impressions.
I think every amp has a "flavor", whether it is "blameless" or not. So, even if a neutral sound is desirable, it may be impossible to reach. Then, why not make the best of the situation and follow our ears and/or instinct and tweak our amps so that they will have a benign sound rather than a "strained" or boring sound?
The F5 sounds harsh? I would say that the treble is a bit "radiant" and the sound indeed is lean and dynamic.
My Amp667 sounds completely different.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.