I’ve decided to try a valve amp in my system, I currently have a Peachtree Audio Inova driving some beautiful speakers designed by Troels Gravesen. Having built a few amps and speakers over the years I’ve decided to go with a Dynaco ST-70, and here is the question…
The ST-70 is a very simple design and I believe in the theory that the less we put into the signal path the better, how true is this really? Amplification is simply making the audio voltage larger…. The response of our hearing is a long way from flat so why introduce a whole lot of complexity into the signal path to make perfectly flat frequency response?
Would love your thoughts on this please, also any opinions on the Dynac ST-70. Sometimes (most of the time) the written audio reviews are just a whole lot of dribble written by people who obviously get paid by the amount of words written.
Many thanks. Dave
The ST-70 is a very simple design and I believe in the theory that the less we put into the signal path the better, how true is this really? Amplification is simply making the audio voltage larger…. The response of our hearing is a long way from flat so why introduce a whole lot of complexity into the signal path to make perfectly flat frequency response?
Would love your thoughts on this please, also any opinions on the Dynac ST-70. Sometimes (most of the time) the written audio reviews are just a whole lot of dribble written by people who obviously get paid by the amount of words written.
Many thanks. Dave
Encyclopeadia-sized books could be written on this, and have indeed been written.... And it isn't really about frequency response, it's about non-linear distortion. Getting a good frequency response in tube amplifiers is mostly about spending money on the output transformer and being sensible about coupling capacitors.
While the ear is certainly anything but flat, you need an amp reasonably flat for the same reason you need accurate colour reproduction in photography (the frequency response (colour sensitivity) or the eye is not much better than the ear) - so the reproduction is realistic. But it isn't critical, and getting it flat is easy anyway.
While having as few parts as possible in the signal path seems common sense, it is not anywhere as simple as that. Here's just a few points:-
1) A single vacuum tube (or transistor) generates mostly even order distortion. This distortion is not wholey unpleaseant to the ear but it does muddly the sound, make it difficult to identify individual instruments, and makes listening to music tiring.
This even order distortion can be largely cancelled out put using a second tube as an active load (ie instead on just using a load resistor).
However, an active load can, if you aren't carefull, introduce problems of its own - eg hum.
2. A pair of transistors used in cascade can be arranged to cancel out their individual distortions, by arranging their emitter currents in proportion to signal level - as the second one gets an inverted signal. For decades I used to think that this could not be done with vacuum tubes. But I learnt on diyAudio (one should learn one new thing every day if you keep your mind open) that indeed it can be done - by introducing a bit of resistive attenuation between the two stages, in conjunction with carefull tube selection, biasing, and anode load values.
3) An amplifier with three stages must alsys distort more than one with one or two stages - as each stage contributes. Actually, not necessarily. With three stages you have a lot more gain ad so you can make negative feedback a lot more effective, so distortion ends up less.
4) Push pull output and long tail pair operation offers cancellation of distortion.
5) Even with identical circuit configuration and tubes, one amplifier can clearly sound much better than another. That is, if some factors on how to choose component values were not understood by the designer of one amplifier (or the other got it right by chance/luck/experimentaion). Some of these factors are very subtle - eg the choice of resistors in series with grids. But theses subtle factors can add up, and sooner or later a badly designed amp will make it's defects known on particular music passages.
6) With more complexity, it becomes possible to work individual tubes less hard - good for tube life and good for sound quality.
Over time, a few amplifiers have become respected for impeccable performance. Generally, these are also the more complex - eg the General Electric "88-50" (a modified version of the Williamson.
Importantly, you should note that, for any given number of tubes, there are a limitted number of "sweet-spot" circuits that give the best possible performance for that number of tubes. For example, while the GEC 88-50 gives about the best that 6 tubes can do, the Quad II is one circuit in which 4 tubes can do really well. But a Quad II is not quite as good as the more expensive GEC 88-50. And since that best possible performance improves as the number of tubes (ie the sweet spot circuit complexity) increases, the real decision to make is this: "How much time or money do I want to spend?"
You get what you pay for. What you pay for in parts, what you pay for in design time, what you pay for in expertise.
While the ear is certainly anything but flat, you need an amp reasonably flat for the same reason you need accurate colour reproduction in photography (the frequency response (colour sensitivity) or the eye is not much better than the ear) - so the reproduction is realistic. But it isn't critical, and getting it flat is easy anyway.
While having as few parts as possible in the signal path seems common sense, it is not anywhere as simple as that. Here's just a few points:-
1) A single vacuum tube (or transistor) generates mostly even order distortion. This distortion is not wholey unpleaseant to the ear but it does muddly the sound, make it difficult to identify individual instruments, and makes listening to music tiring.
This even order distortion can be largely cancelled out put using a second tube as an active load (ie instead on just using a load resistor).
However, an active load can, if you aren't carefull, introduce problems of its own - eg hum.
2. A pair of transistors used in cascade can be arranged to cancel out their individual distortions, by arranging their emitter currents in proportion to signal level - as the second one gets an inverted signal. For decades I used to think that this could not be done with vacuum tubes. But I learnt on diyAudio (one should learn one new thing every day if you keep your mind open) that indeed it can be done - by introducing a bit of resistive attenuation between the two stages, in conjunction with carefull tube selection, biasing, and anode load values.
3) An amplifier with three stages must alsys distort more than one with one or two stages - as each stage contributes. Actually, not necessarily. With three stages you have a lot more gain ad so you can make negative feedback a lot more effective, so distortion ends up less.
4) Push pull output and long tail pair operation offers cancellation of distortion.
5) Even with identical circuit configuration and tubes, one amplifier can clearly sound much better than another. That is, if some factors on how to choose component values were not understood by the designer of one amplifier (or the other got it right by chance/luck/experimentaion). Some of these factors are very subtle - eg the choice of resistors in series with grids. But theses subtle factors can add up, and sooner or later a badly designed amp will make it's defects known on particular music passages.
6) With more complexity, it becomes possible to work individual tubes less hard - good for tube life and good for sound quality.
Over time, a few amplifiers have become respected for impeccable performance. Generally, these are also the more complex - eg the General Electric "88-50" (a modified version of the Williamson.
Importantly, you should note that, for any given number of tubes, there are a limitted number of "sweet-spot" circuits that give the best possible performance for that number of tubes. For example, while the GEC 88-50 gives about the best that 6 tubes can do, the Quad II is one circuit in which 4 tubes can do really well. But a Quad II is not quite as good as the more expensive GEC 88-50. And since that best possible performance improves as the number of tubes (ie the sweet spot circuit complexity) increases, the real decision to make is this: "How much time or money do I want to spend?"
You get what you pay for. What you pay for in parts, what you pay for in design time, what you pay for in expertise.
Last edited:
Good choice, but make sure that your speakers will be happy with it. Can you borrow a tube amp to try?
What kind of driver stage do you plan to use?
Thanks guys
So the Dynaco is a good amp then? As for the source it's all .wav files and I'm looking at DAC options now. Have also posted in the DAC forum, looking at a DDDAC 1794.
I've built a bit of audio gear over the years, is it still an option in the modern digital world?
I love the satisfaction of building gear and can't justify spending $3-5K per component...
Cheers
So the Dynaco is a good amp then? As for the source it's all .wav files and I'm looking at DAC options now. Have also posted in the DAC forum, looking at a DDDAC 1794.
I've built a bit of audio gear over the years, is it still an option in the modern digital world?
I love the satisfaction of building gear and can't justify spending $3-5K per component...
Cheers
u
If you look at it strictly objectively, well, up to the late 1970's, building your own amplifier, if you had the skills, was easily justified - it was not that difficult to build something that was better in performance than than what you could buy in a hi-fi shop for the same money. The cost of parts back then to build an amp matching the famous British brands was a LOT less than the cost of the factory products.
That advantage has long gone. There's a lot of crap on the market, but but the good stuff can now be purchased for much less than the parts cost.
If you look at it strictly objectively, well, solid state in the hands of competent engineers gives impeccable performance that is not easy to duplicate with tubes. And solid state keeps on working without any change in sound quality for years and years.... Yous can buy low cost integrated circuits which, with very few external parts, offer perfomance that top discrete-circuit engineers only dreamt about 30 years ago.
But I still build with tubes. I also do a lot of restoration work on gear made in the 1950's. I like using crafsman skills in my workshop. Like you I find it it satisfying. Very satisfying in fact. And a tube amplifier attracts interest and appreciation from visiting friends that a solid state job never does.
Building good tube amplifiers is not a cheap hobby. If your priority is saving money, go to a discount store and buy something factory made. But what really satisfying hobby is cheap? My wife paints in acrylics. She spends thousands of dollars on materials etc for that.
Satisfaction, and the challenge, why we do it. That's why we have diyAudio.
If you look at it strictly objectively, well, up to the late 1970's, building your own amplifier, if you had the skills, was easily justified - it was not that difficult to build something that was better in performance than than what you could buy in a hi-fi shop for the same money. The cost of parts back then to build an amp matching the famous British brands was a LOT less than the cost of the factory products.
That advantage has long gone. There's a lot of crap on the market, but but the good stuff can now be purchased for much less than the parts cost.
If you look at it strictly objectively, well, solid state in the hands of competent engineers gives impeccable performance that is not easy to duplicate with tubes. And solid state keeps on working without any change in sound quality for years and years.... Yous can buy low cost integrated circuits which, with very few external parts, offer perfomance that top discrete-circuit engineers only dreamt about 30 years ago.
But I still build with tubes. I also do a lot of restoration work on gear made in the 1950's. I like using crafsman skills in my workshop. Like you I find it it satisfying. Very satisfying in fact. And a tube amplifier attracts interest and appreciation from visiting friends that a solid state job never does.
Building good tube amplifiers is not a cheap hobby. If your priority is saving money, go to a discount store and buy something factory made. But what really satisfying hobby is cheap? My wife paints in acrylics. She spends thousands of dollars on materials etc for that.
Satisfaction, and the challenge, why we do it. That's why we have diyAudio.
Last edited:
First you have to decide what you mean by signal path. Most people who use this term don't understand it. For example, some of them claim to remove capacitors from the signal path by using a DC servo instead. They don't realise that the capacitors setting the DC servo time constant are very much in the signal path, but now the signal path is much more complex than it was.nzlowie said:
Good designs are often deceptively simple. They look simple at a superficial level, but a careful analysis (i.e. reverse engineering) will show that everything has been carefully designed to fit together. This is far from simple. People who don't realise this may believe that the design sounds good because it is simple; in reality it sounds good because it is well designed.
Two ways to achieve poor sound (while remaining blissfully convinced that the sound is good):
1. oversimplify - minimise the number of components "in the signal path", in the false belief that as everything distorts then less must be more
2. overcomplicate - solve all known audio problems by adding a new sub-circuit for each problem, use four valves where one would do, add lots of solid-state bias and load circuits, replace all voltage amps by SRPP, replace all cathode followers with White CF, use DC coupling and then have complex level-shifting circuits etc.
True, but you would be surprised how many people believe something different from this.
Getting a reasonably flat frequency response does not require complexity. The reason we want (well, some of us want!) a flat frequency response from our audio system is that the air (the medium which comes between us and the performers at a concert) has a reasonably flat frequency response. Those of us who want to reproduce sound require a flat frequency response; others have other aims, but sometimes don't realise or admit this.
Very true. Also true that some folk add complexity to solve problems that were never an issue or never even existed at all in the first place.
The tension is always between perfection and practicality. You with the ST–70 are choosing conventionally and wisely. Practical and reasonably perfected. Pentode section on the 7199 as voltage amplifier. Triode section as phase inverter. Push-pull final. Ultralinear tap on transformer for screen grid drive. 82 pF in series with 13 kΩ for oscillation / stability between stage 1 and phase inverter. All very simple.
Sounds very conventional?
Consider: the use of a small-signal pentode is by definition equivalent to the more-complicated-on-a-schematic cascode stage. (I'm going to get some flak here, but the analogy is true.) The mechanically shaded screen grid decouples the dynamics of amplification from influencing the control-grid's acceleration gradient. Just like cascode's 'upper stage' decouples an amplification triode's plate swing influence on amplification (and plate-grid capacitance). Etc.
So, in theory, you could substitute other topologies to effect similar ends. But why? The stage 1 pentode will work well, not being asked to deal with too large a voltage swing from the incoming signal.
The triode phase inverter is about as conventional as one might imagine. Because the triode is driven with such a high signal, we cannot expect brilliant linearity out of it. Makes for consideration but not concern to me. It'll work, and well. After all, any 2nd or even harmonics it adds might be viewed as “warmth”. As DF96 sez… 'others have different aims'.
Perfection? Well, in the end, it is not. But it is minimalist, and it has been engineered for reliable operation, practical affordability. As KEIT mentions, so much of a modest amplifier's performance depends on the quality of the output transformer, and investing prudently in the coupling capacitors. Nothing crazy, nothing crâp. After all, there really are only 2 ea. 0.1μF capacitors in the signal path. (The tiny ones are all high-frequency suppressors, and apart from being tiny (which implies HIGHLY linear), aren't even dealing with the critical sub–3500 Hz audio band.)
You could have a bit of fun experimentation swapping out one channels' worth of these for some other 0.1μF cap; then a little A-B switching might give your ears a bit of a learning experience.
Anyway, again: good choice. You will probably want to invest in an equally simple, modest and reliable preamplifier "and knobs" set. You know: knobs for switching sources. Knobs to adjust volume, tone contours. Linearity contours. Tiny blue LEDs. Glitz!
GoatGuy
Sounds very conventional?
Consider: the use of a small-signal pentode is by definition equivalent to the more-complicated-on-a-schematic cascode stage. (I'm going to get some flak here, but the analogy is true.) The mechanically shaded screen grid decouples the dynamics of amplification from influencing the control-grid's acceleration gradient. Just like cascode's 'upper stage' decouples an amplification triode's plate swing influence on amplification (and plate-grid capacitance). Etc.
So, in theory, you could substitute other topologies to effect similar ends. But why? The stage 1 pentode will work well, not being asked to deal with too large a voltage swing from the incoming signal.
The triode phase inverter is about as conventional as one might imagine. Because the triode is driven with such a high signal, we cannot expect brilliant linearity out of it. Makes for consideration but not concern to me. It'll work, and well. After all, any 2nd or even harmonics it adds might be viewed as “warmth”. As DF96 sez… 'others have different aims'.
Perfection? Well, in the end, it is not. But it is minimalist, and it has been engineered for reliable operation, practical affordability. As KEIT mentions, so much of a modest amplifier's performance depends on the quality of the output transformer, and investing prudently in the coupling capacitors. Nothing crazy, nothing crâp. After all, there really are only 2 ea. 0.1μF capacitors in the signal path. (The tiny ones are all high-frequency suppressors, and apart from being tiny (which implies HIGHLY linear), aren't even dealing with the critical sub–3500 Hz audio band.)
You could have a bit of fun experimentation swapping out one channels' worth of these for some other 0.1μF cap; then a little A-B switching might give your ears a bit of a learning experience.
Anyway, again: good choice. You will probably want to invest in an equally simple, modest and reliable preamplifier "and knobs" set. You know: knobs for switching sources. Knobs to adjust volume, tone contours. Linearity contours. Tiny blue LEDs. Glitz!
GoatGuy
Last edited:
I learned a lot from my first amplifier which was a 60s-era Dynaco 70. It is a good beginner amplifier with loads of available modifications - some good, some bad.
In stock form - and with aged components - it does suffer from a bloated bass and a rather soft top end. It also doesn't drive some (less efficient or complex load) speakers very well.
I don't know if it is the weedy power transformer and/or the single 5AR4 rectifier, but I've also noticed that my particular Dynaco 70 had a noticeable compression effect. Even with 93dB efficient speakers and a new SDS capacitor board, it seemed as if the instruments/vocals would "shrink" as the dynamics increased. Hard to describe but if sounded as if the vocalist, for example, was on a spring being pulled back if he sang too loud.
Of course these days the Dynaco 70s that are being sold as kits often feature much more beefy power transformers and even solid-state rectification.
In stock form - and with aged components - it does suffer from a bloated bass and a rather soft top end. It also doesn't drive some (less efficient or complex load) speakers very well.
I don't know if it is the weedy power transformer and/or the single 5AR4 rectifier, but I've also noticed that my particular Dynaco 70 had a noticeable compression effect. Even with 93dB efficient speakers and a new SDS capacitor board, it seemed as if the instruments/vocals would "shrink" as the dynamics increased. Hard to describe but if sounded as if the vocalist, for example, was on a spring being pulled back if he sang too loud.
Of course these days the Dynaco 70s that are being sold as kits often feature much more beefy power transformers and even solid-state rectification.
Everyone has a different process by which they determine (whether by analytical thinking or subconsciously "what feels right") what their objectives are when preparing to start a build project.
I personally can't relate to this simplicity v. complexity as a starting point. In my process this particular dilemma manifests itself rather indirectly; in the form of "can I justify this expense (of money or space inside chassis)".
As an consideration in itself, complexity means nothing to me. If the added thing isn't expensive, and has any measured and/or perceived benefit, and there is space for it - I will always add it. Why not, if there is added fidelity? If there isn't, then, why add it?
That is only true if the circuits are exactly the same. Open loop, using tubes with good curves, a two stage resistor loaded system will always distort more than a three stage system using MOSFET gyrator plate loads or similar.
And, a CCS tailed LTP will always distort less than a SE stage using the same good curved tube and producing same output.
The argument from "human hearing is inherently nonlinear" or similar arguments (like the "think of all the op amps in studios!") is in my opinion a reason to have better performing (more transparent) amplifiers, not less. Because there are inevitable bottlenecks, one should maximize fidelity in all other parts, not "ease up".
Distortion on distortion sounds the worst. If you can lessen only one of the distortion sources, you should. In my philosophy.
How? Only thing I can think of is if you use a tube as the top device, and don't properly reference the heater supply to ground (which is standard procedure).
I personally can't relate to this simplicity v. complexity as a starting point. In my process this particular dilemma manifests itself rather indirectly; in the form of "can I justify this expense (of money or space inside chassis)".
As an consideration in itself, complexity means nothing to me. If the added thing isn't expensive, and has any measured and/or perceived benefit, and there is space for it - I will always add it. Why not, if there is added fidelity? If there isn't, then, why add it?
That is only true if the circuits are exactly the same. Open loop, using tubes with good curves, a two stage resistor loaded system will always distort more than a three stage system using MOSFET gyrator plate loads or similar.
And, a CCS tailed LTP will always distort less than a SE stage using the same good curved tube and producing same output.
The argument from "human hearing is inherently nonlinear" or similar arguments (like the "think of all the op amps in studios!") is in my opinion a reason to have better performing (more transparent) amplifiers, not less. Because there are inevitable bottlenecks, one should maximize fidelity in all other parts, not "ease up".
Distortion on distortion sounds the worst. If you can lessen only one of the distortion sources, you should. In my philosophy.
How? Only thing I can think of is if you use a tube as the top device, and don't properly reference the heater supply to ground (which is standard procedure).
Why?
It might seem like a daft question, the reason I ask is because a good SS amp should be just making the signal louder.
That's what you said you thought an amplifier is..are you looking for something else? other than flat frequency response.
I have to add this can you tell a tube amp from a SS and if you can is it distortion.
Then ask yourself are you looking for this or this?
Or phase splitter or SE..just thought provoking for fun.
Regards
M. Gregg
Attachments
Last edited:
This and some other comments by your good self show that you have not understaood a point I made: Distortion plus distortion does NOT NECESSARILY add up to a worse situation. If the topology is such that the phase of the distortion components in a 2nd stage is opposite to the the phase of the harmonics of a 1st stage, then there is some distortion cancellation, the same as occurs in LTP's, active loads, and push-pull.
The harmonics of a two stages of a two stage cascade amplifier will of course always be opposite in phase, due to signal inversion - the problem is to equalise the amplitudes.
So, while a two stage amplifier will usually distort more, it can be arranged that it distorts less than either a single stage or a three stage circuit.
Why are tubes used as active loads prone to hum (and noise)?
Simply because tube are designed on the assumption that their cathodes are at a low impedance point in the circuit. There is always some degree of leakage from heater to cathode (except in specially made tubes) and there is always some capacitance between the two.
When a tube is used in a conventional circuit, the cathode is either bypassed to earth with the cathode bypass cap, or it goes to earth via a low value resistor. And/or its a push pull output stage where signal level is high and hum is cancelled in teh output transformer. However an anode circuit is a high impedance point, allowing coupling of hum from the heater to make its presence felt.
The problem is completely solvable of course - but it is something the designer must consider.
Last edited:
I like the distinctive mechanical design shown in your first picture.
However, in the second picture, it's hard to see, but it appears that the output pentodes (EL34's) are mounted without their cathode planes vertical. That is definitely contrary to tube manufactuer's advice and may lead to a high failure rate.
Output tubes should be vertical, socket underneath. But where they must be horizontal, they should be installed with the cathode plane vertical, so that grid sag does not alter characteristics or cause shorts to cathode.
Perhaps its interesting to look at design philosophy..
I tend to find there are engineers amps and HiFi ideas.
The more I look the more I'm not sure what a tube amp should be..
I remember someone saying to me what's the point of a tube amp that sounds like a SS design..
I found it hard to respond at the time, because music is music right?
How can it be anything else?
But of course if you want the music to be portrayed like ghosts in your living room and 3D holographic image perhaps that's a different thing.
Because its not the music its the presentation of it. We want to "See the music". (sounds mad doesn't it..)
Regards
M. Gregg
Last edited:
I understand that distortion cancelling is a thing. However, in the scenario I mentioned, using resistor loads of 2 to 5 times rp, the effect does not surpass the performance of a 3 stage gyrator loaded system. Furthermore in practise I am highly skeptical of distortion cancelling schemes; it relies (in my opinion) too much on too many variables.
If the 2 stage resistor loaded system uses a distortion cancelling scheme, then it is not the same circuit topology. It should be compared to a 3 stage gyrator loaded system that is also using a distortion cancelling scheme (whatever that would in practise be). Otherwise we could compare a three stage CCS tailed LTP system to a two stage resistor plate loaded cathode RC biased system. The 3 stage system would then win.
I have no experience with cascodes, so I leave them outside my previous comment.
Hm. You're of course completely right, it's a high impedance circuit node.
I mainly use headphones, which are brutally unforgiving to noise. When I used tubes as the top device in mu followers (12AU7, 6SN7, 6C45P and some others) I never got any hum from those at all.
But I do believe you it could happen. Best to use MOSFETs as top devices for a bunch of reasons.
It's a little hard to dispute that.
However, I will have a go...
Just as there is satisfaction in styling or mechanical design, there is satisfaction in devising a tube amplifier that audibly matches what a good solid state design can do.
We don't do it for cost reasons, or other entirely objective reasons. If we were completely objective, we would just go down to the nearest discount store and buy an SS amp.... That's is NOT to say we should waste money on things that don't do anything usefull though. Like DF96 alluded to.
I always find HIFI a strange subject, the pursuit of excellence or the mad chasing of the unachievable. I have spent many happy and frustrating times tweaking amps building things then stripping them to bits looking for a holy grail that moves every time you hear something better.
Even SS amps that people buy to only scrap them and go up in cost and power until they have a Maserati. Then they fall silent and it never gets turned on because there is something missing/not quite right..
I remember being at a HIFI convention and after hours of going from room to room listening to different kit stumbled into a room with vintage HIFI equipment and listening to a small blue Pye amp with albatross speakers the relief was instant and I wanted one..why who knows..
Regards
M. Gregg
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.