Class A hybrid power amp up and running

Joined
Sep 7, 2016
Messages
148
Reaction score
69
The unit I've mentioned here and there over the past few weeks as being a WIP is playing in the background as I type this:)

For now it's powered by two bench PSU's, a large commercially built unit (physically large anyway but only 2 x 30V @2A output. Used here in parallel to get 4A max @ 30V), and my DIY valve hybrid variable regulated PSU giving 250V and 6.3V for the valve stage. The 30V max of the bench PSU limits output to 12.5WPC for the moment but that's no real issue for me as my Leak Stereo 20 is adequate in terms of power. I have a large toroidal transformer ready for it which should give 18WPC (higher powers can be obtained with larger heatsinks and higher PSU volts obviously) and 176,000uF of smoothing capacitance waiting in the wings.

It passed all bench tests this afternoon including the important square waves into reactive load. A 10KHz squarewave looks really nice!

Only been on for half hour and new components running in etc so early days to give any idea of the sound but certainly not bad so far:)
 
No.I have learnt from bitter experience after posting photos of prototypes in the past that those who may be interested in such photos generally have one of two motives:

1/ To try and reverse engineer your new baby, if photos are detailed enough. I consider everything I design proprietary and whilst I'll probably never sell any or make a penny from it I'm damn sure no one else is going to!

2/ People who don't know their arse from their elbow criticise it because they haven't a clue what good engineering practice actually is and think all components should be laid out symmetrically, in perfectly parallel rows, and with tightly bundled wiring looms all bent at precisely 90 degrees..... even on a one off prototype! (as an aside here, I've noticed loads of equipment these days where the innards have obviously been designed to look nice to the non technical [with compromised performance because of this], as many different brightly coloured capacitors as possible, red, blue or black PCBs and everything laid out completely symmetrically even when by doing so they've had to run the tracks from the transformer, carrying huge current spikes, for 10" in order to get the smoothing caps in a nice neat row down the middle etc! Oh then there's the "go faster" legends like "super linear sound transformer" written on top of said article:rolleyes:)

Oh and to add a third one, people who don't know a resistor from a radish but have drank the hi fi kool aid will start criticising/offering advice based entirely on what brand of capacitors or wire you have used!


Must be just me but I can't see any point in photos of hi fi gear on forums in general unless they illustrate a particular point ie show someone which gear wheel in a cassette mech to replace. I find it bizarre that people seem to want to look at simply a pic of an amplifier sat on a table etc:confused:
 
In my case, just interested. I know what my rat's nest prototypes look like, and get a mild kick seeing others'.

As to components, indeed!

By the way, I loved the resistor and radish. I'll definitely use that one myself.

S.
 
Was going to post pics of the DIY variable regulated supply for valve kit as I'm quite chuffed with it but it seems one must use an image hosting service. It's "finished" and anything I want kept out of public view is inside... It's also hybrid and has a 6550 output valve driven by an ECC81. High voltages and solid state will usually end in tears with something like a bench PSU where shorts can and will occur eventually, hence the valved output and driver stages.
It is built "old school" on an aly chassis, neon power light, Elstone transformer etc but inside it is SS and precision. 100 - 350V in 3 ranges with 100mA max output, very low output Z and noise, output voltage is set by a 10 turn vernier dial and you can dial in lets say 312V and this is what it will output to within around +/- 20mV... yes that took some doing!
Heater supply is regulated by a very low drop-out discrete regulator of my own design and can give 3.5A. Getting 6.3V regulated DC from a 6.3V AC winding was erm... "fun". All SS power electronics are inside the unit so from outside it looks like a 50's - 60's all valve bench PSU.... well if you ignore the modern smoothing caps anyway!
 
After many experiments and subsequent listening it's no longer class A! Tried op amp input stage in place of valve and also tried with and without overall NFB. All made only slight changes to the sound as something else was dominating what I was hearing. As another possibility with this versatile topology is using it in class A/B rather than class A I thought "yeah what the heck let's try it"... Well i'll go to the foot of our stairs! Big improvement. No idea why and obviously it measures slightly less well in class A/B (still pretty good though). I'll have to try all the various permutations again now of course:rolleyes:

Apropos of nothing but I found earlier that I can once again sign in to hi fi engine.... which is nice. Many people have lately been unable to access it so I wonder if they had technical issues...
 
"If Knowledge isn't shared, man would never advance, Libraries would be empty, and Genius would go unrecognised". "Share what you discover, that others may learn from it".

Marconi's 200Kw AM transmitters have a Transistor Ladder, to raise the drive level up to the output valve's grid voltage. Commissioned in 1984, they are still in service around the World and have been very, very reliable - despite the Hybrid Method employed in the final output stages. I am pleased you got your prototype up and running, but unfortunately it has all been done before:(.

Go - on, give us some pictures. I don't mind seeing a crow's - nest of wiring; all prototypes look the same:eek:.
 
Ah I see... somebody somewhere once used valves and transistors together for a completely different reason in an unrelated field so therefore ALL hybrid solutions have "all been done before":rolleyes: Actually it may well be unique (or not!) the way I've done it here and the output stage has a couple of twists to it that make it a potentially new topology.... depending on how much a known topology must be modified before it is considered new of course!

I'm sure that in this day and age absolutely any "advance" in pretty much anything analogue and especially the dying hobby of hi fi would be greeted by the wider world with the same enthusiasm as a newly invented and improved steam regulator for the likes of "The Flying Scotsman":eek:

Oh and if knowledge IS shared then it is no longer proprietary and can no longer be copyrighted etc. I do run a business and may at some time want to build some of these for sale... or at least something related to this incorporating things I've learned from this prototype.

My purpose in starting the thread (and I hope I will be forgiven a little hubris) is that after designing and building something like a hi fi amplifier and getting it working well I just wanted to share some of the excitement and sense of achievement over "my new baby":)
No doubt exactly as a man having just completed a scale model of the Eiffel Tower from matchsticks would likely feel like sharing this with other modellers on an appropriate forum:rolleyes: or like someone who has just built a diy variable loudness control for his Quad 909 may want to tell others how well it works on a hi fi forumo_O
 
I can sympathise with this view. I worked for a lovely man, Michael Cox, many years ago, one of the best electronics designers I've ever met, and he wouldn't patent anything. His view was that he couldn't afford to defend a patent against a Sony if they chose to copy his design, so he would publish everything to stop the Sonys of this world from patenting his inventions themselves. In any event, with technology moving so fast, patents are often outdated long lefore they expire, so not worth the cost of registering and defending.

I can understand that in the HiFi world, genuinely patentable inventions are rare, especially with anything involving valves, that keeping secret any clever configuration is sensible. Nevertheless, I for one would love to see the design details, even if I have to sign a Non-disclosure Agreement.

S
 
A few years back I designed a phono stage which works on the transconductance principal. It's complicated, all discrete, and I reckoned there was a potential 6-7 patentable features. It got mega rave reviews on a few forums from a few who heard it. I did nothing about it for the same reasons I gave above.

Now if I invented something in what I consider the utterly moronic fields of computer games or social media it would no doubt be worth $billions.... but for a phono stage for playing those old vinyl discs?? I doubt anyone with money to invest would even take a phone call on the subject!

OTOH I don't publish or share anything 'cos well... call me a mean spirited old bugger if you like but if I ain't going to make anything from it I'm damn sure no one else is!....even if it's another OMB operation making and selling 5 of them... or diyers who wouldn't pay me for one but who would jump at the chance to clone it and get a "free" one for their own use...

I still have to do it, ie design and build them, 'cos that's what I do... it's my "calling" if you like:)
 
I have a technical question you may be able to answer (outrageous, I know, on this forum)!
As I understand it, a transconductance phono amplifier has a near zero input impedance, so the cartridge drives current rather than provide voltage.
My question is what does this do to the cartridge's distortion? As cartridges are designed as voltage sources, do they behave differently when effectively shorted out.

Will this also work with a MM cartridge with much higher Inductance and DC resistance?

S
 
You're thinking of a transimpedance amplifier there. The Arkless GTI variant of the units I build reusing Cambridge Audio 640P casework is a transimpedance unit. I doubt that there is any distortion issue but have no data on this. It sounds great!

No it doesn't work with MM or even high output MC carts as a pole formed by the carts inductance with the low input Z of the phono stage comes down into the audio range and kills the HF response.

FWIW I have also designed and breadboarded a current in - current out MC stage which is balanced input and output (current > voltage in RIAA network impedance) and is a single stage (in effect) unit. Definitely unique! I only breadboarded a mono unit to prove the concept and haven't heard it.
 
I can sympathise with this view. I worked for a lovely man, Michael Cox, many years ago, one of the best electronics designers I've ever met, and he wouldn't patent anything. His view was that he couldn't afford to defend a patent against a Sony if they chose to copy his design, so he would publish everything to stop the Sonys of this world from patenting his inventions themselves. In any event, with technology moving so fast, patents are often outdated long lefore they expire, so not worth the cost of registering and defending.

I can understand that in the HiFi world, genuinely patentable inventions are rare, especially with anything involving valves, that keeping secret any clever configuration is sensible. Nevertheless, I for one would love to see the design details, even if I have to sign a Non-disclosure Agreement.

S
Me too! Another point, with regard to Patent rights is, by publishing, it puts a "Date - Stamp" on the design. Thus, it is unlikely that Sony or anyone else would be granted Rights, given it would turn up in searches and probably be deemed as common knowledge. Also, if for some reason Sony were granted Rights, the original author would still retain the right to manufacture, without infringing Sony's Patent and no Royalties would be payable as it pre-dates the granting of P.R.

If Sony added a "Twist" to their design, asserting it is different from Jez's, just as others claimed when effectively "Copying" Peter Walker's Current Dumping, they could not come back and claim it to be the same and therefore, again, Jez would be free to manufacture.
 
Ah I see... somebody somewhere once used valves and transistors together for a completely different reason in an unrelated field so therefore ALL hybrid solutions have "all been done before":rolleyes: Actually it may well be unique (or not!) the way I've done it here and the output stage has a couple of twists to it that make it a potentially new topology.... depending on how much a known topology must be modified before it is considered new of course!

I'm sure that in this day and age absolutely any "advance" in pretty much anything analogue and especially the dying hobby of hi fi would be greeted by the wider world with the same enthusiasm as a newly invented and improved steam regulator for the likes of "The Flying Scotsman":eek:

Oh and if knowledge IS shared then it is no longer proprietary and can no longer be copyrighted etc. I do run a business and may at some time want to build some of these for sale... or at least something related to this incorporating things I've learned from this prototype.

My purpose in starting the thread (and I hope I will be forgiven a little hubris) is that after designing and building something like a hi fi amplifier and getting it working well I just wanted to share some of the excitement and sense of achievement over "my new baby":)
No doubt exactly as a man having just completed a scale model of the Eiffel Tower from matchsticks would likely feel like sharing this with other modellers on an appropriate forum:rolleyes: or like someone who has just built a diy variable loudness control for his Quad 909 may want to tell others how well it works on a hi fi forumo_O

I had thought of designing my own Loudness Control - well modified to suit what I saw as being a problem. The enthusiasm is still there, but Arthritis in my hands means, any soldering I do looks like to has been done by a Welder:D. Plus, my eyes are shot to bits, so I would probably solder something else lying on the bench, rather than the bit I am supposed to:mad::D.

Plus, what I did, hardly qualifies as a design - a small mod to an pre-designed Band-stop Filter Network - by removing capacitors from it, which resulted in treble frequencies suffering the same attenuation as the mid-range stop - band. It works, but without a signal generator or my Spectrum Analyzer, I have to rely upon "Subjective Listening Tests". In this regard, I don't have the same confidence as Jez. But it seems to be effective and after all, at normal listening levels, it it "Out - of - Circuit".

Further to the above, my 909, Sugden and Spendors are stock items - none of which are currently in production. So the pleasure I get is from exploiting the performance of the Spendors, which although very good, I never appreciated "Just how good". This is by way of what are now two geriatric amplifiers.

Whilst my Sugden is currently unused; and in truth I bought it purely on reputation and experience of my first Sugden, all those years ago, the 909 has been subjected to some fair listening tests. It too however, was bought on the speculative argument that it would operate wholly in Class A at the levels I wanted to use it - low level.

That aside, it is proving to be an astoundingly good purchase. I won't use the term muscle although it most certainly can provide it. The term "Weight" is more appropriate as it gives a solidity to the sound-stage, which I have rarely heard outside of a concert hall - despite the low level it is operating at. Whilst having enjoyed my 303 for some 50 years and my 405 for 20 years - the 405 being a brilliant design, I did not appreciate just what a step - up the 909 is. Truly, wonderfully precise - arrogant even, in the way it deals with complex music. Well done Quad.

I appreciate my Stax much more now too, although listening to my son's Hifiman's "Planar's", may well super-sede them. I didn't get much time to listen to them when we were visiting him, but as he will be down in August, a decent session should provide some answers. Even so, driven by the 909, the Stax seem to come alive as they can handle more drive than my Spendors and being single Electrostatic Elements, the transition across the musical spectrum is seamless, a function of both the 909's frequency response, sustained power level, relative to where the music is (bass, mid-range or high frequency) - and zero phase delay due to absence of crossover network, transition points.

Jez makes the point about well designed amplifiers, with good figures not sounding too good. Well, the parameter which was most important when matching amplifiers to loads is always "Return - Loss". An amplifier with an excellent Damping Factor is only one half of the equation. If the return loss of a loudspeaker is poor, it will have a great impact on the output stages of an amplifier and "That" would certainly show up in listening tests. The original Quad Electrostatic loudspeaker was "full - range". That in itself is enough to set it greatly apart from any loudspeaker of its day and is probably still true of many modern loudspeakers. To design a Phase - linear Cross-over Network is both expensive, a fundamental compromise and a pain in the arse. And they will always be load - conscious. "The Loudspeaker is the load for the Amplifier". But equally true, "The Amplifier is the load for the Loudspeaker". Mis-match a valve amplifier into a load and you will burn out the output valves. But, before it goes "Pop", you will hear the consequences of that mis-match.
 
A few years back I designed a phono stage which works on the transconductance principal. It's complicated, all discrete, and I reckoned there was a potential 6-7 patentable features. It got mega rave reviews on a few forums from a few who heard it. I did nothing about it for the same reasons I gave above.

Now if I invented something in what I consider the utterly moronic fields of computer games or social media it would no doubt be worth $billions.... but for a phono stage for playing those old vinyl discs?? I doubt anyone with money to invest would even take a phone call on the subject!

OTOH I don't publish or share anything 'cos well... call me a mean spirited old bugger if you like but if I ain't going to make anything from it I'm damn sure no one else is!....even if it's another OMB operation making and selling 5 of them... or diyers who wouldn't pay me for one but who would jump at the chance to clone it and get a "free" one for their own use...

I still have to do it, ie design and build them, 'cos that's what I do... it's my "calling" if you like:)

Forget about what others may do. "You did it" and no one can take that away from you. Whatever anyone else says or does, you know the truth and that is what is important. As an employee, the company owns the rights to any design an employee comes up with. That is something one has to come to terms with. But it pretty much guarantees one will be the last out of the door when business turns sour. And word gets around, so there will always be companies looking for individuals that bring more to the table than what their job description states.

"Hiding your light under a Bushel" is not the answer. Demand for what you design is small - relatively speaking. However, making known just what you do, do, will benefit your core business as trust in your expertise will grow - exponentially.
 
A true plethora of technical errors there I'm afraid... you're still thinking in terms of RF!

I do want to correct the assertion that I have said amplifiers that measure well do not sound good. I am not saying anything of the sort!

What I am saying is that, as you found with the 303, 405 and 909, amplifiers that measure well enough that they should be "perfect", and therefore indistinguishable from one another, often sound very different from one another.

Two amplifiers can share a measured spec of <0.01% THD, 10Hz - 40KHz response, 100WPC, damping factor >50 and yet sound completely different. One may sound great and the other may sound awful even. Science fails to explain why this should be so at the present time. There has been all sorts of fads over the years that have come and gone and which were put forward as possible reasons for deficient sound quality.... often with an associated "cure"! There's been TIM, SID, excessive feedback, capacitors causing it, inability to deliver 200 Amps (even if driving LS3/5A's at 1W!) and loads of other guff and most has proved to be completely baseless.

Obviously an amp with 3% distortion, audible mains hum and hiss, frequency response falling off beyond 8KHz etc etc will sound bad.
When things are this bad then classical measurement techniques do adequately predict this. The designer and electronics author Doug Self came up with the concept of the "blameless" amplifier and I interpret this to mean that in all those parameters that can be measured, an exemplary result is returned such that one could expect the amplifier to sound perfect. ie there is no measurable problem with the amplifier that one could expect to cause it to sound poor.

Now in practice of course all this matters not as amplifiers have an annoying habit of sounding good/bad/ugly in spite of measuring good. However it is both good engineering practice and common sense to make sure it measures well enough that nothing can be blamed on the measurable parameters. An amplifier may sound somewhat "shut in" and "lacking in air" at the top end even though it measures as flat to 50KHz... hence here the amp is at least "blameless". However it could also sound that way because it is -3dB by 12KHz. It is then not blameless and a simple classical frequency response measurement reveals the problem.
 
Sorry,

"A true plethora of technical errors there I'm afraid... you're still thinking in terms of RF!" - not true, I'm afraid. Matching is fundamental to all electronic engineering. And, if test equipment can't detect failings, to say an amplifier sounds bad, solely on the basis of audible tests, then the results must be as inconsistent as the hearing differences between individuals. Of course, that provides great opportunity to those that sell "Snake Oil" who peddle their nonsense - an issue you go to great length on.

It is taken as a given, we are talking about well designed pieces of kit, so resorting to variances in response are irrelevant, given they are outside the remit of what we are talking about. Classically, damping factor is taken to mean Z load/Z source. So, I fail to see how a combination of loudspeaker impedance/amplifier output impedance - across the specified audio range of an amplifier could sound anything but dreadful if it is 1 - if there is nothing to arrest a driver, it would be bursting out of its housing and into the lounge!

There are fundamental differences in the design of the 303, 405 and 909, but none result in them sounding dreadful. What is clear however, is the damping factor of the 303 is inferior to the 405, which is inferior to the 909. If return loss is irrelevant, then so too is phase response and group delay, which are both a function of return loss - "Match". What that provides is scientifically proven. Any other explanation is just hocus-pocus.
 
I think you're using the term 'return loss' incorrectly, as that term is used in RF to mean the amount if reflection from a misterminated transmission line. It doesn't apply to audio, where lines are never terminated in their characteristic impedance. (except perhaps analogue telephony)

If you mean that a loudspeaker driver's back-emf is more or less attenuated by the amplifier's output impedance, that's a different thing, not return loss.

S
 
Absolute rubbish from start to finish!

There is no such thing as "return loss" in audio electronics.

A damping factor of 1 can sound fine with the right speakers. (clue: there is damping other than electrical damping at play)

Damping factors over around 20 are largely irrelevant.

While I'm at it....

The Quad ESL is not full range but has bass and treble panels and a crossover.

There is no such thing as an (analogue) phase linear crossover. A minimum phase crossover is the closest we can get but this means 1st order only.

A valved audio power amp will not be damaged by "mismatch" eg using 4 Ohm speakers on the 8 Ohm tap etc. Operating into an open load may cause damage in some cases though.
 
A true plethora of technical errors there I'm afraid... you're still thinking in terms of RF!

I do want to correct the assertion that I have said amplifiers that measure well do not sound good. I am not saying anything of the sort!

What I am saying is that, as you found with the 303, 405 and 909, amplifiers that measure well enough that they should be "perfect", and therefore indistinguishable from one another, often sound very different from one another.

Two amplifiers can share a measured spec of <0.01% THD, 10Hz - 40KHz response, 100WPC, damping factor >50 and yet sound completely different. One may sound great and the other may sound awful even. Science fails to explain why this should be so at the present time. There has been all sorts of fads over the years that have come and gone and which were put forward as possible reasons for deficient sound quality.... often with an associated "cure"! There's been TIM, SID, excessive feedback, capacitors causing it, inability to deliver 200 Amps (even if driving LS3/5A's at 1W!) and loads of other guff and most has proved to be completely baseless.

Obviously an amp with 3% distortion, audible mains hum and hiss, frequency response falling off beyond 8KHz etc etc will sound bad.
When things are this bad then classical measurement techniques do adequately predict this. The designer and electronics author Doug Self came up with the concept of the "blameless" amplifier and I interpret this to mean that in all those parameters that can be measured, an exemplary result is returned such that one could expect the amplifier to sound perfect. ie there is no measurable problem with the amplifier that one could expect to cause it to sound poor.

Now in practice of course all this matters not as amplifiers have an annoying habit of sounding good/bad/ugly in spite of measuring good. However it is both good engineering practice and common sense to make sure it measures well enough that nothing can be blamed on the measurable parameters. An amplifier may sound somewhat "shut in" and "lacking in air" at the top end even though it measures as flat to 50KHz... hence here the amp is at least "blameless". However it could also sound that way because it is -3dB by 12KHz. It is then not blameless and a simple classical frequency response measurement reveals the problem.

This is where we disagree fundamentally, as I have never ever heard amplifier differences that weren't obvious from the measurements.
I also remind you of the blind tests carried out by James Moir and Associates that showed that the Quad 303, 405 and bridged IIs were indistinguishable, a set of tests replicated by Martin Colloms using different amplifiers, including Naim, Quad and Michaelson & Austin amps.

You might also be aware of the Spanish comparison between a Behringer A500 and a Levinson with the same result.

I've never seen any results from any blind tests that showed up amplifier differences when level matched and kept out of clipping.

If you know of any, other than anecdote, I would be interested in knowing about them.

S
 
Back
Top