Valve preamp feedback defeats the purpose

I cut my teeth on valve amplifiers and have been repairing them for musicians ever since. Despite that, I totally agree with the last line of your Publisher’s Letter in the November issue – if you compare the LM833 with any valve, it’s no contest. If you want fidelity, you don’t use valves. But I’ll leave that debate to others.

What I’m really wondering is who will use your valve preamp design and for what? Instead of a normal opening stating the need or inspiration we have a hint in the editorial about interested experimenters, and at the tail of the article that "guitarists" might be interested in valve sound.

You do concede that valves can still be legitimate electronic components, and you cover valve theory quite well, but why the big back-flip? Valves still have some very useful advantages as input amplifiers such as being almost indestructible by input overvoltage, combined with almost infinite input resistance.

For guitarists, there is a moderate (3-5%) second harmonic distortion caused by the inherent non-linearity of the transfer characteristic (your Fig.3 p26). This is progressive over the signal swing and not to be confused with clipping, soft or otherwise. But your valve preamplifier design very effectively irons that out flat! (Figs.5, 6 & 7).

A Fender "Stratocaster" guitar fed into a Fender "Twin-Reverb" is considered a classic and practical situation. The input stage of valve guitar amplifiers like the "Twin" are, almost without exception, are as shown on the left of your Fig.2 (page 25, November 2003). The only feedback applied is DC from the cathode resistor; no AC feedback at all.

They normally lack the 100pF top-coupling capacitor in the input on the simple reasoning that neither the speakers nor the guitar have much output above about 5kHz. Despite this apparently seriously "nasal" bandwidth, this rig can produce ear-splitting highs.

I’m no "feedback purist". I’m just pointing out that the "classic" (ie, distorted) sound comes from NOT having feedback for linearity.

Roly Roper,

via email.

Valve guitar sound is desirable

I’m writing to congratulate SILICON CHIP for publishing the Valve Preamp project in the November 2003 issue.

I play guitar in a band and have designed and built two all-valve guitar amplifiers, a 2-channel 50W rack-mount monster and a small 4W "combo" amplifier, which I am happy to say were either built using easy to obtain (within Australia) components or salvaged parts from old radios.

I assume that most of the people who will build the valve preamp kit will either fall into the category of musician or hifi enthusiast, and I would like to offer one or two comments.

The idea of "Valve Sound" (at least as far as guitar amplifiers are concerned) is something made up from a number of contributing factors, one of the most important of which is the output transformer. They are big and heavy and a prime source of high harmonic distortion but that’s the whole point of that sound! The user wants high levels of harmonic distortion and the perceived warmth of valves is due to high levels of second and third harmonic distortion, which in a tube guitar amplifier is desirable.

Coupled with a high gain preamp (three or more triode stages) and a pentode or triode output stage (push-pull configuration is the most common), this creates the overall valve sound that many guitarists will shell out megabucks to obtain.

Also, in regard to Leo Simpson’s comments about the prohibitive cost of
a 60W all valve guitar amp/speaker kit, a price of around $1000 is peanuts compared to something similar available commercially. A quick scan through some of my mail order music catalogs reveals prices floating around $1800 - $3500 for a 60W guitar amplifier.

I would hazard to guess that there would be quite a few readers willing to part with $1000 to obtain a valve guitar amplifier that they could build themselves.

Andrew Curtis,

via email.

Digital TV dropouts

I would like to add a few comments to the current debate about Digital Television. I have owned one of the early Thomson set top box (STB) decoders for quite a while and have recently been given a PC/STB combination unit. Both perform very well with very good picture quality but they both will drop reception when the signal is not quite up to scratch. By not up to scratch, I mean an equivalent analog signal from the same antenna, that can only be described as perfectly adequate, can still produce intermittent dropouts.

Recently I spent a few days at a house in Pacific Palms near Forster, NSW. As I knew the TV reception at that location was marginal and a check on the DVB website indicated the ABC digital transmission was available, I decided to take the small decoder to try it out. The ABC analog transmission is on a VHF-Hi channel and is a little noisy and has some ghosting.

To my surprise, the decoder worked and produced a first-class picture until the dropout problem struck again. The decoder would stop working, sometimes every few seconds. Sometimes it would go for many minutes without dropping out; working consistently it was definitely not! The point I am trying to make is that when analog is switched off, a lot of people will have to invest in satellite TV or have nothing usable. It is thoroughly annoying to watch a perfect digital picture and have the sound drop out frequently.

All this early talk about being able to watch digital signals in marginal areas, even in a moving car, is absolute rubbish. The signal quality needs to be better than good in my experience. Recently, I walked into an electronics store at Warringah Mall, in Sydney. There are a number of STBs on display there, all working, and guess what – all dropping out every now and then. Things will need to work better when analog is turned off!

Horst Leykam,

via email.

Happy with 12AX7 valve preamp

I’ve recently completed the 12AX7 valve preamp from the November issue and I am quite delighted with it. I am using it with my wideband audio, double-tuned crystal set tuner for the AM broadcast band. The high input impedance of the valve preamp is ideal for the application and the valve preamp drives the line level input of my transistor power amplifier and speakers.

As suggested in the SILICON CHIP article, this preamp doesn’t really have the "warmth" of a typical valve preamp (second harmonic distortion?). However, in comparison with my standard line preamp, using an NE5532 low noise op amp, the valve preamp has a much more "open" sound that is an absolute pleasure to listen to.

Possibly this is due to the typically heavy use of negative feedback in op amp circuitry, compared to the somewhat lesser use of negative feedback in the valve design. Whatever the reason, the preamp sounds very nice.

By the way, in my version, the ferrite-cored transformer hisses quite audibly during operation. Is this normal?

Felix Scerri,

Ingham, Qld.

Comment: the line version of the preamp featured in the February issue would be more appropriate to your application. It is normal to hear some "frizzle" from the step-up transformer.

UHF to VHF converter

In the December 2003 issue (p91), W. B. asked about a UHF-to-VHF converter. I made such a converter some years ago when SBS first started transmitting and my old TV did not have a UHF tuner. I purchased a scrap UHF tuner with a mechanical capacitor tuning circuit. I reduced the capacitance by carefully separating the vanes of the variable capacitor until the output of the mixer was in the VHF range instead of the original UHF tuner IF output. I then connected this new output to the antenna input on my VHF TV and presto – a down-converter which produced a very good picture.

By fiddling with my UHF tuning capacitor modification I was able to get fairly good tracking, allowing me to tune in other UHF community stations and watch them on a spare VHF channel (I think I used Ch. 0 because it would have been the lowest frequency). The whole job was done with no test equipment at all; just a good background in electronic theory, some youthful enthusiasm and some luck. The whole thing (TV, tuner, etc) went in a garage sale many years ago. I wish I had kept it, for sentimental reasons.

Brad Fuller,

via email.

PCB design tutorial

I have read David Jones’ tutorial articles on PC board design and I enjoyed them. As someone with a reasonable amount of design experience, I found them a "refresher" course. I do have some comments though.

For information on the silkscreen, component values and/or designators are fine if there is room for them but I find that with tight layout and tiny components, there quite often isn’t room to put anything. So one has to prepare separate manufacturing drawings of the PC board showing at least the designators. Even on manufacturing drawings there may not be space to put the component values.

I do disagree with the comment that back annotation is rarely needed. When preparing the net list, all the components have to have a designator. In all but the simplest of boards, you can’t know where the components are going to be on the board. So when the board is laid out, the components in terms of an ordered placement against designator are all over the place, which on any reasonable sized board makes hand assembly difficult, as a lot of time is wasted locating the component.

So the re-annotation feature of the PCB package is used to re-assign the component designators in an orderly arrangement, in one of a number of arrangements of increasing rows and columns. Then this new annotation is passed back to the schematics.

Note that the re-annotation may not be perfect. On the program I use, an occasional component will be out of sequence. This happens particularly with resistors and capacitors.

Regarding the choice of PCB file format, anything Protel is certainly fine for local manufacturers. But for overseas manufacturers, it’s been my experience that Gerber is the preferred format. At my work, when we get prototype PCBs made locally, we send Gerber files. This acts as a cross-check that the files are correct, as these files will be sent to the overseas manufacturer. It is also an additional protection of intellectual property, as none of the component information appears in the files. This is not to denigrate in any way the integrity of the local manufacturers; it is the old story of not supplying more information than you have to.

Ian Johns,

via email.

Amplifier design philosophy questioned

I have a few unrelated comments about audio amplifiers. I note with some interest your 350W amplifier in the January 2004 issue. One design detail in complementary symmetry amplifiers of this sort has puzzled me for some time. Is it really necessary to have a 10Ω resistor between signal ground and the real DC return earth?

This must surely assume the input to such an amplifier always has a floating ground, which will not necessarily be so. Thus, the output from an external preamp or equaliser of some sort feeding into an amplifier like the Studio 350 is very likely to have a signal earth that is also at DC earth return (usually the chassis or equipment housing). If both pieces of equipment have a mains earth, this will effectively short out your 10Ω resistor. What then?

Your article on the Studio 350 could say more about the choice of output transistors and the considerations that drove this choice. To know this would be of some interest.

I do not understand why three key performance measures are not included in the performance data for amplifiers like the Studio 350. They are: (a) power bandwidth, which is rather more meaningful than small signal (1W) frequency response. After all, a small signal bandwidth from DC to daylight is of little use if the slew rate is poor; (b) intermodulation distortion and transient intermodulation distortion, both of which are perceptually intrusive; and (c) power supply performance under full power tone bursts.

Yes, I’ve seen the very impressive scope picture but what happens on longer tone bursts at full power? Does it sag? Is the regulation of the supply well damped?

I do not understand the emphasis you place on ultra low harmonic distortion performance. Sure, getting this down to 0.1% or a bit better is important but beyond that, given the distortion levels in signal sources and transducers, could anyone hear the difference? Would it not be better to minimise intermodulation distortion (static and transient)?

On a completely different topic, I am constantly amazed at the extraordinary science fiction written about valve amplifiers. I am forced to conclude that much of it is more about being part of a cult than about pursuing serious engineering science.

Over 40 years ago, I was involved in designing hifi preamps and power amplifiers (see Miniwatt Digest for 1962 or thereabouts) and I still have a very clear view of the real limitations of valve amplifiers – not least the horrible things that happen in output transformers. My colleagues of that era would be astonished at some of the claims being made today! I have even read apparently serious discussions on the web about the big (?) differences in the sound produced by different brands of the same output tube! Give me a break!

Emeritus Professor J. E. Clark,

via email.

Comment: the 10Ω resistor is included in the signal earth return in order to improve the separation between channels when the module is used in a stereo set-up – it reduces the circulating currents which inevitably occur when a stereo source is connected and when both modules are powered from a common power supply. It can be omitted in a mono set-up.

We did not dwell on power transistor choice as it would have made the article a lot longer. But for power output versus cost, the ones we chose are pretty good. There are some bigger power transistors used in Japanese amplifiers but they are harder to get and more costly.

Power bandwidth is more meaningful than frequency response at 1W but far more difficult to measure, particularly if you don’t want to overload components in the output filter, not to mention the increased power dissipation in the output transistors themselves. Of course, we also do distortion runs versus frequency at high power but only up to 20kHz and since distortion is always low, slew rate limiting does not come into the picture.

We can easily do intermodulation testing but we have always found that if THD is very low, then so is intermodulation. TID is more difficult to measure but if there is bandwidth limiting at the input, TID should never be a problem. If we were to publish all these tests it would make already large articles much larger.

Longer tone bursts would inevitably lead to sagging as the simple power supply is not particularly well regulated. We do not believe there is much value in really "stiff" power supplies – there is no audible benefit and the cost is much higher.

Partly the emphasis on extremely low THD is because we can! If the THD of bog-standard CD players is as low as .001% (or lower) then ideally the THD of the amplifier should be much less. Yes, the distortion in speakers is going to be much higher but that will not necessarily mask the distortion from amplifiers. Indeed, high order harmonic products from amplifiers can be heard at extremely low levels in quite ordinary loudspeakers.

Our 15W class A design from July 1998 has the lowest THD we have ever measured. It is also the best sounding amplifier we have ever heard – the clarity has to be heard to be appreciated. On that basis alone, we feel justified in always striving for the very best THD performance.