Graham's Blog Archive
Valves – on a Graham Slee product?
This post has been distilled from forum posts by Graham to make the journey of discovery easier reading . . . Enjoy!12AX7 (ECC83) and 12AU7 (ECC82) tubes suitable for preamp use
Yes, the V (or T) word has been uttered at ‘Slee towers’…
And four valves complete with bases arrived today! Does this mean I’ve gone crazy? Does this mean I’ve swallowed the marketing man’s pill?
I hope not, but I thought I’d try and hear what all the fuss is about and have a go at building a preamp line level stage to see just how good a modern performance can be gotten out of tubes. These are 12AX7 (ECC83) and 12AU7 (ECC82) tubes suitable for preamp use, and because they are very high output impedance (100k roughly) they’ll need the help of some solid-state electronics to make them usable.
Often these days they’re used in what’s called “plate voltage starvation circuits” — that’s what all these Audiophool things that run off 12V wall-warts are about. (see this forum post). But I’m not into making hi-fi fuzz boxes. I want to see how low I can make their distortion and how natural they can be made to sound . . . . . .
Well, 4 pages [on the forum] in less than 30 hours must be close to the record on here! Or it may have even beaten it. Such then is the popularity of valves! It shows most have a soft spot for the old beasties even if it’s a misplaced one . . .
Having used lots of valved gear early in my career I am glad solid-state done properly lifted the veil off all that musically interesting stuff I’d have never heard otherwise! But I also like the old world engineering which paved the way for (the few) better things today.
I will not be running valves outside their designed operating parameters. I will (if my experiments work…) be running the valves shown, at their lowest designed plate voltages, and in doing so there will be no lethal voltages. To do that will require solid-state techniques – for example: the plates (anodes) will not be aware that they’re not driving a resistive load connected to HT, because they are unable to think! However, let’s be simplistic about this and say if they were able to think, they’d think they were being utilised in a normal (for valves) way.
Valves produce their warmth because of even order harmonics. This was best described by Marcus Graham Scroggie under his pen name Cathode Ray in Wireless World some years ago, and I will attempt to reproduce the entire article in this thread at some point.
I will be amazed if the valves used in this experiment do the same! Why? Because if my experiment succeeds they will not produce the nice tubey flattened bottom (or top depending on phase) which causes those even-order harmonics. I am looking for neutrality.
So why muck about with valves when it can be done with solid-state (as I may have shown…)?
I suppose it’s because so many people are attracted to that warming glow of bottles stuck on the top of equipment. I call them sooners: they’d sooner see valves than hear how real music can sound.
And here I may have a problem. My research so far shows the nicest sounding valves I can obtain (not NOS – they need to be in production) have their filaments completely enclosed making it nigh on impossible to see their glow.
But, here’s the trick many employ: a LED nicely fits in the valve base, and I have an excess stock of orange LEDs that should outlast me . . .
So yes, it is mainly a commercial experiment.
The promised (38 year old) article by M G Scroggie aka Cathode Ray can be downloaded here: (each page, a PDF, is over 1 MB)
The article is from Wireless World October 1978, but is still as valid today.
Most will find the article heavy going because it is rather technical, and to fully understand it will require foundation level maths, but by skim-reading to the graphs which show asymmetrical sine waves, the reader will receive at least some education into what even-order harmonics is about – what valves without solid-state assistance do in abundance!
What I do with valves is in my spare time, and although it could have a commercial outcome, at present it is to satisfy my own curiosity.
Like many of you who might think valves offer something “special”, although my distant memories of valves suggest the opposite, I would like to find just how much can be achieved.
The current state of solid-state offerings (mine excepted) might be such that valves do in-fact sound better. And as the younger electronics design engineer tends to be digital-first — analogue last, it might be because of that.
Some of us however, are older, but the customer might not be able to differentiate solid state designed by the experienced against solid-state designed by the inexperienced.
Therefore the experienced (me) might have to offer tubes/valves to be accepted by the public at large.
The “working design” of the Proprius originally had a dome in the top which glowed and changed colour but I thought it too corny.
The phono preamps and headphone amps were the result of what I did for fun. Things I designed to obtain outright musical performance for myself, and so is the Proprius. I’m pleased as punch that so many have enjoyed listening with them and have served a good purpose.
The Lord sayeth whatever you make, make it for a fit and proper purpose, which means don’t fanny about!
I think placing valves on equipment, and some don’t actually do anything, is fannying about at the customers expense if they don’t serve a proper purpose. If they don’t bring anything better to the table then somebody is taking the piss and you’re paying to have it taken out of you. I also think that lighting up valves with LEDs to pretend they’re glowing is also taking the piss. But, it seems that is what people want, and to cater for people’s wants I therefore must also take the piss out of the customer??
Maybe I also need to start advertising amplifier power in PMPO? It might attract the younger generation? For those who don’t understand PMPO, it is a completely false assumption of power that is either made up by unscrupulous marketing departments, or still as falsely calculated as being the short circuit current of the power supply multiplied by its output voltage: the short circuit current can be tens if not hundreds of amps depending on the imagination of the company marketing department, and multiplied by even the most modest of voltages (3.3V or 5V on computational type devices) results in hundreds of watts! As Paul Daniels would say “That’s magic!” but the younger generation fall for it all the time.
(By the way, the short circuit current multiplied by the output voltage gives zero watts! Why? because there is no voltage if it is short circuited!!! Unfortunately most people don’t use the “free program” that came with their brain called common sense)
I can also remember how to make a light show. I had to be able to do so because music alone was too boring for the younger generation of my youth also. Instead of the gently changing colours some “hi-fi” emits I could do a sound to light??
Instead I have made the decision to stick with analogue! I know at least two people out there who would love me to do an analogue audio preamp of the “old school”, and I have discovered that some in my generation and above still want to spend on such sensible things.
As for the valves, I will find if they offer anything new. If I find they are just a diversion down a blind alley – which is what I suspect – then they’ll be going in the bin . . . Or maybe I just use them as decoration to grab more customers?? And qualify their use by lying?
One of the things I’ve realised about valves is the complete lack of support in my simulation software (Number One Systems Easy-SPICE). However, after much thought I realised the “amplifier sub-circuit component” could be married-up with the available sub-circuit SPICE models, and whilst they look nothing like a valve — and can’t be edited to look like them — it seems the simulations work.
This is important to see the actual operating characteristics, otherwise it’s just dabbling for dabbling’s sake. It is also hard to trust in such simulations until you actually build something and measure its performance to correlate simulation findings. Without being able to simulate it will be an absolute waste of effort.
Interestingly though the 12AX7 designed for a minimum HT of 90V shows useful gain operated on 48V. Another point of interest is that without grid stopper resistor the phase margin is vastly reduced confirming what people find reporting a very bright sound. So the model might be actually doing its job. We’ll see.
I am now going to see if I can utilise the double triode 12AX7 in a differential amplifier of some sort or another. The problem with valves in today’s world is their output impedance. Power valves required a matching transformer to be able to drive low impedance speakers because their plate impedances were so high. Preamp valves have even-higher output impedances as can be seen on this data sheet. The table on page 2 shows Rp values of 100k and higher. Obviously that isn’t going to drive any of today’s solid-state inputs without considerable insertion loss.
And this is why so many valve products are in fact hybrids which use solid-state output devices of one kind or another to be able to have the low output impedance required.
It should make us all laugh out loud at those who belittle solid-state whilst listening to their solid-state assisted valved equipment, completely unaware that it is partially solid-state.
Yes, even before putting them into a circuit I have come to my conclusion for what it’s worth. Having checked the simulation model against the data and the simulation returning the same information I think it is obvious that the SPICE simulation is telling it as it is. For the safe supply voltage of 48 volts the plate voltage will be obviously lower, and to obtain linearity the plate voltage would have to be approximately half supply voltage to obtain a symmetrical signal swing.
- The calculated cathode voltage required to bias the valve properly on this supply is quite low at 0.5VDC which tallies with the valve data. As such the ac voltage which can be accepted at the grid will be the inverse square root of 2 by half that voltage which is 175mV. This is the maximum allowable input voltage before clipping. Absolutely of no use for most signal levels.
- The plate resistance required to obtain the signal swing is approximately 100 kilohms and that is the output impedance. It is absolutely of no use at all in driving any other equipment.
- Without negative feedback the distortion is approximated at greater than 1% (possibly nearer 5%), so not even worth switching on the analyser.
- The open loop gain is approximately 28dB which for a gain goal of 8dB (for a preamp) allows just 20dB of negative feedback, which is 20dB short of Bell Laboratories Harold Black’s 40dB (1927) which still holds true today.
- Therefore applying negative feedback becomes empirical which is another word for messing about.
- The resulting reduction in distortion would be quite little and not reach my 0.1% target.
- The minimum HT voltage set by the valve manufacturers (90V) is obviously there for a reason, and that reason being that it allows for an “acceptable” distortion figure, but definitely not high fidelity.
- For the valve to be able to perform to near high fidelity standards the HT voltage would need to be near its maximum of 300 volts, which increases costs significantly.
- The necessary solid-state support circuitry required to make the valve “work”, which means lowering output impedance to be compatible with todays amplifiers, and to reduce distortion to acceptable levels, is just as complex as the sold-state circuitry which can do the same job without a valve being in sight!
Therefore why bother?
This experiment, cut short for the common sense resons set out above, conclusively shows (IMO) that all low voltage circuits utilising the common preamp valves of which there are many very similar types, will result in unacceptable levels of harmonic distortion.
Also that any of these valve circuits running at voltages below 48 volts, of which there are many, often running on much lower voltages such as 12 volt wall-warts (although many specify just the wall-wart mains supply voltage possibly to mislead the customer), will provide a considerable amount of distortion. That those devices are not high fidelity by nature of their distortion.
That those who find satisfaction using such devices must therefore prefer to hear the music with the addition of the effect known as fuzz.
That any device utilising such valves operating on low voltage supplies and claiming high fidelity distortion levels cannot be actually using the valves, and therefore they are there for decoration, and deceit.
Given the results of this research (albeit short lived) the customer reading this [post] has at least been informed.