Pros and cons of valve amps

[anthonyTD]

Hi John,
Yes, its the one in the avatar, Its not a very good pick, I will try and put up a better one.
A...

Any chance of a photo or two of *your* copper amp Anthony?

Seen Marco's and Joe's

Edit: Is that it in your avatar? 8 x KT88?

[Arkless Electronics]

The Mullard 5-20 circuit is still doing the rounds on a lot of amplifiers today in various guises, and for good reason. I've been following with interest, and some points made RE damping factor, I did in fact made some posts back (lost in the debate). I'm following the feedback V's no feedback arguments for design, and am of the opinion that both can work equally well, but one of the primary reasons to use it for valve amplifiers is to get the output impedance low. Again, I made the point that few use it because some can't be had with the design needed to handle the instability if time compensation isn't properly applied. Designs such as Toms and my own STA25 do use a lot of it, 26dB compared with the 5-20's 30dB, and remain perfectly stable due to a well designed compensation circuit and indeed the STA design shares some similarity with the 5-20. I remember Will providing the specifications and far from dropping as low as Jez suggests, the design has plenty power reserve at 20Hz because it has been thought about and is designed well, and has low THD throughout the operating bandwidth. The specs are (-1dB)25 Watts from 20Hz to 40KHz; 36watts from 25Hz - 25KHz with just 0.1% THD at 25W output. That's nothing like as low powered into 20W as was suggested earlier, so with a decent damping factor, will compare very favourably to any equivalent SS amp into something as sensitive as the Tannoys, which have a minimum impedance of just 7.2 Ohms. (average of measurements taken from MGs and HPDs)

It is also worth remembering that some passive 'speaker designs don't have the losses quoted earlier of a typical 0.6R. For a typical 3 to 4m speaker cable run using a 16AWG cable has a cable loop resistance of about 0.11 Ohms. My own RFC Reference crossovers have losses in the bass circuit of just 0.13 Ohms at DC (compared with the 0.6 quoted earlier) giving a total load of 0.24. Ohms. If we round that up to 0.25 and add to say a Radford output impedance of 0.18 ohms, we have a total of .43 Ohms. Minimum speaker load is 7.2 ohms which gives a damping factor of 16.7 which is more than enough for vice like bass grip. Compare with a SS device of say 0.05 ohms output impedance. Total impedance will be .3 ohms and damping factor 24. The reality is that once past a DF of 12 to 15 or so, any more is not adding a great deal to the grip. Even of an amp had zero output impedance (the ideal), maximum DF would still only be 29 or so, not the 200 or 300 that some manufacturers claim because they neglect to include cable and crossover losses. So the summary is a healthy DF and plenty of power into 25Hz with speakers having around a 92dB/1w sensitivity. In such cases, it is not game, set or match, but more what you prefer to listen with.

Like Tom, I'm open to any design, be it SS or valve, and have heard a lot of really quite decent SS amps with the Tannoys. Most though have been in the same price ball-park as the valve amps, save one or two, and there's something which just seems to give more life to the music with a really good valve amp, so for the time being, that's where my preference lays. As with Tom, I'm watching with interest where Class D is going, but consider it still somewhat in it's infancy with regard to availability of some really worthy designs for now.

I repeat for the third time that Radfords are not normal valve amplifiers and so picking probably the best valve amp ever made and using it as representative of valve amplifiers is hardly erm... representative! Also IIRC it has 36dB of NFB. Also I repeat that there is no such thing as time compensation.
Oh and most valve amps do use feedback!
The only factor that matters in maintaining the output power down to 20Hz is the physical size of the output transformer...
I stick by what I said on game set and match to SS on bass performance 100%. and stress again that although, on average, SS amps will be 10 X or more better than valve ones at the bottom end, valve amps can easily be good enough for it to not be particularly noticeable subjectively. Whilst it can be amazing what the human ear is capable of in some areas it is also surprising how tolerant of gross errors it is in others!

[RothwellAudio]

I'll just chip in here with some general comments which (I think) address the OP's question.
Firstly, I think it's a mistake to assume that there's a "valve sound" and all valve amps have it, or that there's a "transistor sound" and all transistor amps have it. The cliches about "warm and lush" for valves and "hard and aggressive" for transistors are just gross generalisations.

The benefits of valves are that you can have something which looks cool (ironically, since it's actually hot) and sets you apart from your non-audiophile mates, keeps your house warm in winter and can sound extremely good - if it's a good amp.
The cons are that some maintenance will be required, the amp will cost a lot more than an SS amp of similar power (probably) and your electricity bills will go up.
I wouldn't want to generalise beyond that.

[anthonyTD]

I'll just chip in here with some general comments which (I think) address the OP's question.
Firstly, I think it's a mistake to assume that there's a "valve sound" and all valve amps have it, or that there's a "transistor sound" and all transistor amps have it. The cliches about "warm and lush" for valves and "hard and aggressive" for transistors are just gross generalisations.

The benefits of valves are that you can have something which looks cool (ironically, since it's actually hot) and sets you apart from your non-audiophile mates, keeps your house warm in winter and can sound extremely good - if it's a good amp.
The cons are that some maintenance will be required, the amp will cost a lot more than an SS amp of similar power (probably) and your electricity bills will go up.
I wouldn't want to generalise beyond that.

[JohnJo]

Another Copper valve amp John?...top plate's copper?

Very nice, that looks proper!

[Marco]

I repeat for the third time that Radfords are not normal valve amplifiers and so picking probably the best valve amp ever made and using it as representative of valve amplifiers is hardly erm... representative!

Yes, but surely all that matters is that the *actual topology* in question (i.e. valve-based electronics) is capable of said performance, regardless of whether Radfords are representative of your average valve amp?

If the topology itself wasn't capable of producing the measured performance in the first place, as cited by Paul, then he wouldn't have used it as an example to back up his argument. Furthermore, if the majority of other valve amps fail to measure up to the Radford, then that's not its fault (or the designer's) - it's up to other manufacturers to up their game accordingly!

It's a bit like quoting and marvelling over the technical specs of a Bugatti Veyron, and citing how it's a shining example of automotive technology, and then a dissenter saying: 'Ah yes, but it's hardly representative of your average car...!'

Marco.

[Macca]

Can't add to the technical stuff but will say

1) worst expensive systems I have ever heard all used valve amps. May not have always been the amps fault but there is deffo some really expensive rubbish in the valve world. Not so much with SS, you seem to get what you pay for there.

2) Did not know about the technical superiority of Radfords until recently, but I had already concluded several years ago that it is one of the best amps of any topology I have ever heard. An STA 25 sold here not long ago for £1200 so not much money for something so good, if you can get away with 25 watts of course.

3) At the top levels there is little to choose between them. Although single ended valve amps have their own flavour that no other topology seems to be able to mimic. Seriously low power is the drawback.

4) For me personally valves are too much faff. But I am pretty lazy. I also don't like that they slowly degrade, so slowly you can't hear it. That would get my audiophilia nervosa going a treat.

[Reffc]

I repeat for the third time that Radfords are not normal valve amplifiers and so picking probably the best valve amp ever made and using it as representative of valve amplifiers is hardly erm... representative! Also IIRC it has 36dB of NFB. Also I repeat that there is no such thing as time compensation.
Oh and most valve amps do use feedback!
The only factor that matters in maintaining the output power down to 20Hz is the physical size of the output transformer...
I stick by what I said on game set and match to SS on bass performance 100%. and stress again that although, on average, SS amps will be 10 X or more better than valve ones at the bottom end, valve amps can easily be good enough for it to not be particularly noticeable subjectively. Whilst it can be amazing what the human ear is capable of in some areas it is also surprising how tolerant of gross errors it is in others!

Radfords are an example of a competently designed valve amplifier Jez, and they are not alone in that, though I agree, there is a plethora of poorly designed ones doing the rounds (and not all at bargain basement prices!).

The feedback figure is correct and is taken from Radford's own literature which states in section 2.5 (below) that "...A low inherent output stage resistance is achieved by using anode/screen feedback.....and then applying 26dB overall amplifier feedback..."



As for most valve amps using feedback, many do, but many do not. I have owned a few myself which used zero negative feedback (driving very efficient loudspeakers). OK, lets re-phrase the time compensation. As distortion reduction can be proportional to amount of feedback used, then in theory, the more that is used the better. However, as stated within the Radford design literature, the amount used is a function of the phase shift at low and high frequencies. As oscillation can occur at the frequency extremes, then it becomes necessary to reduce gain at these extremes to zero, where the phase shift becomes 180 degrees. That is all that I meant by "time" compensation, ie, that it is derived because at specific frequencies where phase shift becomes 180 degrees, oscillation and instability would occur otherwise.

I have seen examples of valve amps using negative feedback without any compensation, and they are probably not very stable as a result.

As for the ONLY aspect of power output being the size of the transformer, that's not entirely correct either! That may have been the case 40 years ago, but don't forget that, for example, the advent of modern grain oriented Silicon steel cores has allowed more compact output transformers to be used so they don't have to be the size of a small bus to get the power gains at modest sizes....mind you, the Emille I had did use output transformers that probably were the size of a small bus!

[Arkless Electronics]

Radfords are an example of a competently designed valve amplifier Jez, and they are not alone in that, though I agree, there is a plethora of poorly designed ones doing the rounds (and not all at bargain basement prices!).

The feedback figure is correct and is taken from Radford's own literature which states in section 2.5 (below) that "...A low inherent output stage resistance is achieved by using anode/screen feedback.....and then applying 26dB overall amplifier feedback..."



As for most valve amps using feedback, many do, but many do not. I have owned a few myself which used zero negative feedback (driving very efficient loudspeakers). OK, lets re-phrase the time compensation. As distortion reduction can be proportional to amount of feedback used, then in theory, the more that is used the better. However, as stated within the Radford design literature, the amount used is a function of the phase shift at low and high frequencies. As oscillation can occur at the frequency extremes, then it becomes necessary to reduce gain at these extremes to zero, where the phase shift becomes 180 degrees. That is all that I meant by "time" compensation, ie, that it is derived because at specific frequencies where phase shift becomes 180 degrees, oscillation and instability would occur otherwise.

I have seen examples of valve amps using negative feedback without any compensation, and they are probably not very stable as a result.

As for the ONLY aspect of power output being the size of the transformer, that's not entirely correct either! That may have been the case 40 years ago, but don't forget that, for example, the advent of modern grain oriented Silicon steel cores has allowed more compact output transformers to be used so they don't have to be the size of a small bus to get the power gains at modest sizes....mind you, the Emille I had did use output transformers that probably were the size of a small bus!

Other than my falsely remembering the NFB as 36dB when it is in fact 26dB I stand by every word I said. Oh and having this time read up the specs it seems the output impedance is actually 0.45R and so right in the middle of the range I gave as typical.... I've still got something in the old grey cells saying 36dB so maybe that is for the STA100.. I can't find a figure for this after a quick search though. Certainly 36dB is very high for a valve amp if so....

The vast majority of valve amps for hi fi use have NFB. It would also be very unusual to find one without NFB that doesn't feature big triodes, usually directly heated, as output valves as triodes have inherently low output impedance for a valve and so if you don't mind an output impedance from the amp as a whole of as high as 3R or more (I do!) they can be used without feedback. I'll say as a guesstimate that of all hi fi valve amps ever built at least 90% use feedback.

Grain orientated silicon steel laminations have been around for donkeys years! Also size is indeed everything here so yep, I am entirely correct.

No point in re-phrasing time compensation as there is no such thing. I gathered you meant phase compensation but for the sake of accuracy....

I also disagree with both you and Marco over the choice of the Radford as an example. Bizarrely Marco gives the same ish example I was going to myself so yes lets stick to the automotive analogy. To me it would seem very silly to name a Bugatti Veyron as a typical example of a motor car and to then say that Mondeo's, Uno's, Xantia's, BMW 3 series etc were all absolutely shite cars cos they fall so far short of the performance of the Veyron!!

[Radford Revival]

(Posting just to take my mind off unfortunate events in our personal lives right now)

Other than my falsely remembering the NFB as 36dB when it is in fact 26dB I stand by every word I said. Oh and having this time read up the specs it seems the output impedance is actually 0.45R and so right in the middle of the range I gave as typical....

Hi Jez

For once for a specsheet, the Radfords actually better what is quoted. On the 8 ohm tap with good valves it is as low as 0.18 ohm in the midband (measured), on the 16 ohm tap it's around about 0.3ish from memory, and around 0.10 on the 4ohm setting. It does rise slightly at either frequency extreme as per circuitry limitations. Pretty good though!