To sum up. Assuming I can add up

I did mention, oh….a year or two ago….. that I would like to design and build the Best Amplifier In the World. I didn’t? Well, I meant to. It’s your fault if you can’t follow simple smoke signals and temporal-time-warp-speke.

We have actually got some little way into our Best Amplifier….blah…blah…design.

Mr. Dumble based his designs on somebody else’s (Leo, no less), and a stratospheric pricing philosophy. There’s no doubt if you’re going to copy somebody (allegedly) you would be sensible to make sure that the ‘somebody’ knew what they were doing. There would be no point codging your neighbour in the next desk’s answers to your French exam if he happened to be doing the geography of Beijing.

The overiding piece of understanding we need (stratospheric pricing philosophies notwithstanding) is that we have so far loosely put together a number of parts.

Mains input transformer primary…goes to secondary…..goes to rectifier….goes to smoothing. IT IS MODULAR !!!!! And so is the rest of our Best Amp in the….blah, blah.

The power amp section of a solid state amp is less simple to section off into a modular form, because there are so many negative feedback loops (ac and dc), but it is still modular in essence.

So we have got as far as the smoothing section in our Best….blah, blah. What next?

Well, staying within our so far unmathematical approach, we get to the decoupling stages of the various dc rails, which supply the different preamps and power amp.

Onwards and upwards…………

So what is a ‘Phase Splitter’. Something to do with Flash Gordon, I bet.

And if you remember Flash Gordon, you must qualify for ‘Old Git’ status. Surely.

Let’s clarify this. There’s a first time for everything.

The ‘Flash Gordon’ I’m on about is not the one that Freddy Mercury waxed lyrical about. No. The Flash that I’m referring to was the one where the smoke from the rockets in outer space went upwards…..? I’m ashamed to admit that that is the only thing I can recall from my hundreds of Saturday morning visits to the local bug’s hut, the Coliseum.

Back on the topic. I think that might have been the shilling each way I bet on the dog with three legs called ‘Flash Gordon’. It lost.

A phase splitter in not a ‘phase inverter’. They are often called that these days (just as a thing you lose or break regularly is called a ‘phone’), but the phase inverter, is actually any valve that has its output from the anode. Whereas the phase splitter actually splits the phase of the input and is a specific circuit arrangement that is designed to drive two halves of a push-pull output stage. One side anode goes positive and the other side anode goes negative. Hence the phase splitting thing.

The point of this (there’s a POINT?), is that, although a lot of emphasis (and money) is put on output valve matching (and rightfully so), the phase splitter which feeds the grids of these output valves are rarely spoken about. But an internally matched ECC83 is basically a double triode valve, and if the outputs of those anodes are not well matched, the big bucks you’ve spent on some really nicely matched output valves wasn’t worth the dog I bet on called Flash Gordon.

The output valves can only reproduce whatever is going in, and if that is not symetrical, neither is your output signal.

It’s always hard to describe a sound, but the effect of this imbalance, to me anyway, is a loss of clarity and weakness of the sound. So next time you need to change output valves, get a nicely matched phase splitter to go with them.




Pi in the ski……erm….sky

In the great firmament of amp design, amp designers (Leo Fender, Jim Marshall, Dick Denny, et al) tend to be focused to the point of being obsessive. Whereas, anybody wandering his way through the vagaries of design issues who can chuck it for a couple of months at a time, (me), and then amble back into it as if nothing happened (because it didn’t), must be, at the very least, flippant.

Correct. So off we go again…….






The name ‘Pi type filter’ derives from the shape of the circuit and its resemblance to the mathematical symbol ‘Pi’.

Pi is a constant which relates the circumference of a circle to its diameter. Numerically it is 3.142 to three decimal places. However………………..

That has nothing to do with why the Pi type filter is so called. Sorry to bore you to death for no good reason. The circuit is called a ‘Pi-type filter’ because….erm… looks like it.

Below is a rather grand picture of the symbol ‘Pi’. If you compare the diagram at the top of the page to the symbol below, it becomes pretty obvious that the shape of the two capacitors and the symbol marked ‘L1 or R1′ is the same shape in essence as the Pi symbol below. If it was clever, I wouldn’t know about it, would I?



So, what is it for? What does it do?

Obviously not to make me look clever because I’ve already failed miserably on that one. If you look back at a blog of February ‘Valve rectifier and how it works’ the output of the circuit is unsmoothed dc. Which is ok for charging batteries but no good if you were charging folks to go into a show full of amps with no smoothing circuits.

Great to accompany a buzz-saw convention though. The purpose of the circuit is to convert the raw (unfiltered) dc, from the rectifier (valve or diodes, makes no difference) to constant and stable dc that the electronics it supplies can usefully work with.

Many thanks to Wikipedia for the diagram. I don’t feel too bad about nicking it because I paid ‘em a fiver on their last trawl.


C1 is called the reservoir capacitor because it stores voltage. Strictly speaking it stores charge, in Joules. It charges up as the voltage increases (on the dotted line above) and discharges as the input voltage falls.

The red line is the ripple voltage, and although it is much smoother than the raw rectified dc as depicted by the dotted line, it is unlikely to be much use for supplying circuits that have to supply audio output. That ripple voltage is what happens across C1.The rest of the Pi type circuit smooths the ripple further by the action of the smoothing capacitor C2 dropping the ripple voltage across the circuit element L1 (or R1).

The physical difference between L1 and R1 is that the L signifies an inductance which is usually an iron cored choke (a coil of wire round an iron core), or a resistance. The significant difference in design is that an iron cored choke (inductance) is a lot more expensive than a resistor, and the difference in operation is that the choke is far more effective as a smoothing element in the circuit, and, unlike the resistor, doesn’t tend to drop much dc voltage across it.

So, harking back to the ’Things in my favourite amp’ subject; which afterall is what this is all about, the smoothing / Pi type filter circuit will definitely involve the series inductance, and not the resistor. However, most amps will have a number gain stages, which are often arranged in cascade, (each stage feeding the next), and this is where we can use the dropper resistor effectively, because we often need to arrange different voltages for different stages.

There is also another issue. That of ‘decoupling’. But more of that in due course……

Tea. And I did get a chuckle or two out of this . So, also, macaroon!!


Sound City Concord….it was quite a buzzzzz.

There can’t be that many folks about who remember Sound City. And even less who want to. I hold up my hand. I actually used to sell them, when I worked in the venerable establishment of C.E.Hudson and Sons of Chesterfield (that’s Derbyshire, England.) That would have been in the ’70′s.

Some of the best, and certainly most useful, times of my life were spent during my six year sojourn there. At that time Dallas Arbiter were a big name in musical wholesale, and it was that company who was responsible for the introduction of the Sound City name to the market. The amps were designed by Dave Reeves (later of Hiwatt fame), and some big names used them; Pete Townshend being amongst them. But this all sounds too much like a boring old git carping on. Let’s just make the point that these amps were (when the design settled down) very well made and well thought of.

The one that turned up at the workshop was a Concord, which was a fifty watt 2x 12 combo, and now not far off fifty years old. It had a quite distinctive look to it, having faders for the channel volume, treble and bass controls. This one did buzzsaw impressions at a level that would have completely drowned a buzzsaw next to it. The main reservoir capacitors had dried out and that seemed to be the extent of it. After replacing the two 100uF capacitors……it stopped buzzing….and started humming. This was not quite at the level of the buzz (which had disappeared) but it wasn’t messing about, and certainly wasn’t usable.

This was a mystery. It sounded (and looked on the scope) like heater wiring hum. If the heater wiring has been moved around it can pick up. So you get a pair of longnosed pliers and (very cautiously) move the heater wiring around. The effect of that can be extreme. In this case it didn’t make any difference at all. Sometimes the heater winding on the transformer can develop shorts, so that the cancelling effect of the ac heater wiring is upset, and you get hum. One thing you can do to get round that (but not if the heater winding has an internal centre tap) is to fit a hum cancelling preset pot, which can trim the out the hum. That was not possible on this amp because it had a grounded internal centre tap on the heater winding.

It didn’t make any sense to me at all, so I shelved it, and sat in a corner for a month sucking my thumb. Then I realised that there was a distortion if a signal was put through it. Difficult to make anything of, when you have a monumental hum going on. But it was not far off half wave on the scope, which meant that half of the output stage wasn’t doing anything. THAT WAS IT!!!!!!! Eureka. Half of the phase splitter wasn’t putting a signal to one of the EL34′s.

On most phase splitters (not Ampeg) an ECC83 or ECC82 valve is arranged so that each anode (pin1 and pin6) puts out an identical (but out of phase) signal to the grids of the output valves. So that explained the distortion. But what about the hum?

There was an open circuit anode resistor, which was responsible for the lack of half the signal. But this meant that there was an open circuit at the anode of the ECC83 and that was connected via a coupling capacitor to the EL34 control grid………and picked up the hum from the heater wiring!!!!!!!   Yes!!!!!!!!

Although there were few laughs in this job, I award myself a congratulatory macaroon. And tea.




Valve rectifier circuit and how it works.

This looks horribly intelligent.

Which means (by inference) that my coconut macaroon is fading into the sunset.

On the plus side, it’s not impossible I might get a laugh out of this. Not likely, just not impossible.

Onwards and upwards, bull by the horns, goldfish by the goolies……maybe not that…..

The first thing we need to supply our amp is something to plug into the socket on the wall. On the end of this cable, we go into an IEC (kettle) connector, and this has a line voltage, a neutral, and an earth. The line voltage and neutral connect to the mains transformer ‘primary winding’. This ‘primary winding’ is quite sensibly named. In a world of electronics bullshit, you should hang onto anything that might be sensible. Because there’s not much of it about.

This primary winding is the first thing that the mains voltage sees (so it is primary, or first) and it’s a winding, because it’s a long length of copper wire, wound around some metal former. In the drawing, that primary winding is represented by the block marked 240 volt. Also wound onto this same former is another (sometimes many) more lengths of wire. This is the secondary side of the transformer, and in this case we have two HT (stands for ‘high tension’ or ‘high voltage’) windings which supply the two anodes of the rectifier valve. American for ‘anode’ is ‘plate’. But this doesn’t make all that much sense because the cathode is still ‘the cathode’ in American. Don’t ask me??? We also have a heater winding, which is much lower voltage (5volts or 6.3 volts usually, depending on the valve. This heater winding supplies the rectifier valve heater (and possibly other valve heaters.) The property of ‘induction’ enables the transformer to function. The ac voltage applied to the primary winding has some fraction of that voltage ‘induced’ in the secondary winding, which is why a voltage appears at the secondary output without there being any direct connection between the windings.

The heater of the valve often (not always though) is completely separate electrically from the operation of the valve. All it does is heat the cathode.

Here we get to actually how the valve rectifier (and nearly all valves actually) works. By heating the cathode (usually made of a tube of copper with the heater inside it) we excite the free negative electrons on the cathode. I knew we would get to something exciting eventually. The copper cathode is often coated with something like boron, which gives off a lot more free electrons than copper, so making the valve more efficient.

This does next to nothing at all, other than to create a ‘space cloud’ which is a cloud of electrons that surrounds the cathode. However, if then we switch on mains voltage, the secondary of the transformer supplies a high voltage to the anodes, which is positive in one direction. Positive attracts negative, which means that the electrons from the space cloud are drawn towards the anode when it is positive. This has now begun an electron flow from cathode to anode (but only when the anode is on its positive cycle).  Conventional current flow is actually the opposite (positive to negative), because electron flow is negatively charged.

So what we have now is positive voltage appearing at the cathode (marked HT dc out).

On the end of this terminal, we have to provide smoothing for the HT output. That voltage is not usable as it is. So we will look at the choices of smoothing circuits next.

Tea. And not many laughs. Sadly.


Valve rectifier?…..solid state rectifier?????

Further to the amp design series. This is part three. If I can get to three without suffering brain fade there must be a slim chance I might know something useful. Eh?

a) So what’s the difference? b) Is there a difference? c) Will I notice a difference?

a) One uses a valve. Duh.

b) Not if you’re deaf.

c) See above…..and…..some people wouldn’t notice a difference if a herd of wildebeest stampeded through their kitchen.

A rectifier of any sort has one purpose, and that is to convert (rectify) an ac (alternating) current/voltage to a dc (direct) current/ voltage. But the valve rectifier has certain properties that the solid state rectifier doesn’t. From a point of view of how it affects the sound, the valve rectifier has the property of ‘sag’. I know how it feels. The valve itself has an internal impedance, and what that means is that the current path from anode to cathode (the conducting direction) isn’t a short circuit, it has a certain resistance. (Impedance, strictly speaking.)

This is a very complex characteristic and varies from valve to valve in imponderable ways. This will not endear the rectifier valve to the digital fraternity. They like to put these numbers into this black box and it does this. Every time. No valve will do that, because the valve is a physical device and as such every one is different (within certain tolerances). Not only does it vary from valve to valve, but also depending on the current it is drawing, how hot it is, what is the smoothing arrangement on the dc side. And what day of the week it is, for all I know?

The characteristic of ‘sag’ in the rectifier, means that as the valve takes more current, the output voltage tends to drop, which tends to limit the output of the amp it is supplying, so providing compression. This is not a simple compression. It is affected by frequency, amplitude, phase angle, power amp loading, even speaker impedances. So far as I know, there is no compression circuit that can simulate this form of compression, it is not very predictable and varies (as we’ve already noted) from valve to valve.

A solid state rectifier, on the other hand, has very little impedance in the forward direction. It will run itself without varying the output voltage, to destruction. So it does not compress, and has no sag factor. This is a big difference, as any guitarist will tell you who has wound up a valve amp with a valve rectifier.

But there is a downside to the valve rectifier. It’s expensive and complicated. It needs, for a start, an extra valve base. It also needs a transformer that will give a heater supply voltage for the valve, which is often a different voltage to rest of the amp heaters. In the case of a GZ32 or GZ34 or 5Y4….etc….5 volts. This means that the mains transformer has to have an extra winding to supply this voltage.

You don’t need any of those things for a solid state rectifier circuit. Just two (or four) diodes, or a bridge rectifier. Nevertheless, our amp (should it ever get to that state) will have a valve rectifier. If we keep it small, say 15 watts  rms, we can do the valve rectification with an EZ80 or EZ81. ….So what?…..ah well….these valves have 6.3 volt heaters, so the mains transformer is much simpler; we don’t need a separate 5 volt supply.

So we’ve actually made a decision. Which is quite something in a life of doing my best to avoid them. We use an EZ rectifier valve and design the output to be around 15 watts rms. And that solves another decision. If the amp is going to be some variation on class AB (we’ll look into this a bit further into the proceedings), it is likely to be using a pair of EL84′s. Not unlike the Vox AC 15, or the Watkins Dominator.

The thick plottens………..

Tea! Macaroon! Oh yessssss!

Puctec ZD-987 solder/ desolder station

This is a bit off the beaten track, but might be useful. There’s a first time for everything……

I’ve used a desoldering hand pump device forever, and they work ok. For about a week. You really can’t expect too much from something that costs a couple of quid, can you? And also, if you have a lot of desoldering to do you can get a serious case of ‘pumper’s thumb (?)’ We will not go into that; but it hurts.

The big plus for this soldering/ desoldering station was that it was cheap. I may as well be brutally honest. Most of the other soldering stations (Weller, Antex, etc.) would need me to sell my grandma to drum up the cash. As I don’t have a grandma any more, my style is somewhat cramped in that investment area. It was actually surprisingly good (I’m off grandma’s now) but for one piece of missing information. It was like building a flat pack bookcase from Korea (or somewhere else foreign). On following the instructions to the letter (usually an interpretation from the early Sanskrit) it looks remarkably like a food mixer. A food mixer with a lot of screws missing.

There is a glass tube with a spring in it that should fit behind the sucky-pipe-thing. I found that a jack for a Transit van was helpful in installing this. You really have to have powerful thumbs to put this in. It is probably worth an investment in a thumb-building-up course as you have to take the tube out on a regular basis to clean out the solder waste. Except there isn’t any waste if the temperature isn’t effectively set, because it won’t desolder.

Here is the first real crunch. In the instruction sheet, it says that leaded solder melts at 180 degrees C, and leadless solder 220 degrees C. Which it does. However after some experimentation, we find that after setting the temperature to around that recommended level, we pull off the tracks quite well, but don’t actually desolder anything.

You need to desolder at around 335 degrees C. At that level it does work very well.

However, another problem (which is described in the instructions) is that at high heat levels (see workable operating temperatures) the solder tends to clog the intake pipe with a lump of lead. This lump of lead (also according to the operating instructions) you can’t get out. That’s because all the tin content of the solder alloy has been burnt off, which leaves you with a lump of lead that you can’t shift with the poking tool. Fair do’s though; Puctec do inform you of this. And also that your nice desoldering machine is unusable. Decent of them.

There is a way round this. You might have to do this a lot if you are soldering valvebase joints where there is a lot of solder to get rid of. You bring the temperature up above 400 degrees C and thump the cleaning stick in and out repeatedly. As the temperature gets up to around 380+ the solder will shift and you will have a clean tube. Sounds a bit brutal, but it works and is a lot better than chucking your nice new machine away.

Cuppa tea! And I certainly deserve a macaroon for this one.

The Second One in the amp design series.

The general idea in designing/ inventing anything at all, is to get together some wonderful ideas, build them, and then find out why it don’t work. Unless you use a Computer Aided Design piece of soft brain (sorry, software), in which case you can get it to virtually design it, virtually build it, virtually test it, and come up with something you can then virtually poke around with until you get something new. Except it won’t be, because the intuitive human input  has been relegated to accidental poking about and the imagination relegated to accountancy and marketing.

I am not, then, very likely to be a CAD person. In which case I shall have to think a bit. Oh dear.

We’ll start with the power supply. “Why would you do that?” you might enquire. Because if we get that wrong I am not going to get a macaroon. If that part of it is less than it should be, the rest of the amp is unlikely to be worth plugging in. So what do we want from a power supply? At its most basic, some device that takes the ac mains (240-250 volts in UK) and converts it to a useable dc voltage for the electronics. There is little difference in that requirement whether you are supplying a semiconductor (solid state, transistor; call it what you like) or valve circuit. The difference is that semiconductors tend to be higher current, lower voltage, and valves tend to want the opposite. So a power valve will be likely to have several hundred volts at the anode, whereas a semiconductor amp will need voltages of less than hundred volts. The actual voltages depend on the power output of the amp. The power transistors will put out amps, and the power valves, tens (or maybe a few hundred) milliamps.

Clearly an over simplification, and the actual figures will be dependent on the sort of power rating of the amp. But it at least illustrates the general principle.

There are two different possibilities for the power supply, and the main difference between those alternatives is in the way the incoming ac voltage is rectified.



This is a basic arrangement for a full wave rectifier circuit using solid state diodes. It could also be a bridge rectifier arrangement, but the result is the same, so we will stick with this one.

A diode will only conduct in one direction, and the transformer’s job is change the 240 volt input to a useful voltage for the amp to work with. The primary side is the 240 volt input and the secondary is centre tapped, which enables the ends of that winding to act in opposition. That means that when the voltage increases at one end, (with respect to the grounded centre tap) the other end decreases. This enables the diodes to conduct alternately in their forward direction, so producing a dc voltage from the ac input.

All this sounds fine and dandy, but the output is like a rowboat on a ski jump. Not smooth in other words. And therefore not useable for the amp, as it stands, because what you would hear most is a 100hz buzz. Most tunes really need a bit more than that.

So far there is no difference between that which the valve amp requires from a power supply and that of a transistor amp. The difference comes in with voltage/current requirement, and that brings in a big possibility with the valve amp. It can use valve rectification, and the solid state amp does not have that option.

In the next blog we shall consider the why’s of that, and what the differences are.

And just in case you might think  I’ve turned sensible, I shall, especially for you, dear reader, come up with some really, really daft ones.

Have a good macaroon…..I mean…



This might be a bit of a shock

…..and surely will be if you were silly enough to trudge through the last, extremely daft, post.

I have mentioned (probably many times) my ambition to design an amp. After a year or so I’ve got a bit further with this ambition. It is called ‘A Guitar Amplifier’. Slick, eh? At that rate of progress the second great flood will have swept the planet before I get down to soldering something. At least you will know who it is paddling the ‘Guitar Amplifier’ across the frozen wastes  of Biggleswade.

My second idea (on a roll, here) is to entertain you, oh erudite blog readers, with a sequence of blogs that will take you from the very first stages of amp design, though to the…erm….very second stages, etc, etc.

Alright, let’s make start. Questions first, answers second.

What level of marketing bullshit are we going for? Zero, because I don’t care if I sell any, because it’s for me, and because it’s just a fascinating thing to do, to build something that skirts around all the stuff that doesn’t work (or doesn’t work very well) and takes the stuff that does work to another level. Hopefully.

Another question. What colour is it? Don’t care. Next question?

Valve or semiconductor or hybrid of some sort? This sounds a bit serious, and I can see my macaroon dwindling. To come up with an answer to this one, we have to consider what a Guitar Amplifier actually does. The text book answer for this is that an amplifier makes whatever it is going in, bigger when it goes out. It has the property of gain, in other words. But in my Guitar Amplifier, this is only part of the story. A semiconductor of more or less any sort will produce (all things being equal) an output very close to the original input, but bigger, because its characteristics are close to being linear. A valve will rarely do this, it will add things; harmonics for instance, because its characteristics are rarely linear; and also certain sustain capabilities, because it will tend to ring in some degree. A guitar player might find this very attractive, or obnoxious. Two guitar players might have both opinions, it is a subjective thing. As this is for me, I can poke around with it until it sounds as I would like it. Bearing that in mind, I am very likely to produce, at the end of this process, something based on valves.

And CERTAINLY NOT a ‘digital modelling amp’. This is something that can sound like everything but not much like anything.

There we are, a start. Zen saying “A thousand mile journey starts with one step”. But so does a journey to your front door. Doesn’t it?

Tea. Macaroon. But only because it’s nearly New Year. I preferred the ‘Charter Flight’ blog. I got a laugh (and a legitimate macaroon) out of that.


Fender Hot Rod Deluxe. You really don’t want this kind of fault.

Of any fault that can beset an amplifier, oscillation is the most damaging and also the most obscure to fault find, unless your happen to have an oscilloscope about your person. Not everybody has those. Even the symptoms are not easy to interpret.

So what is it, and why is it? The ‘what’ is easy. The amp puts out its full power, but you can’t hear it. That’s because the oscillation frequency (often) is way above any Earthlings’ hearing. I have heard that folks on Calisto whistle at frequencies in excess of a MHz, but you shouldn’t believe that because I just made it up.

The ‘why’ of this can be tricky. Without the benefit of a ‘scope you would be able to hear a fairly heavy buzz (this is the power supply putting out everything it’s got), and distortion on anything you plug into the input. That is the signal fighting with the huge oscillation signal.

But the heavy buzz could also be reservoir/ smoothing capacitor problems (which could also cause the oscillation), or possibly an output bias fault, (check bias volts, around minus 55 or so is a safe one to go for 6L6 valves), or possibly a heater/ cathode short on any valve (interaction with the volume/ tone controls will narrow it down to the first stages) before the phase-splitter.

The causes are often reservoir/ smoothing caps drying out, especially on older reissue amps. There are three 22uF and a 47uF in the Deluxe/ Deville amps, and the best plan is to change them all. Another cause is shorted windings on the output transformer. The centre tap becomes not centre because the lacquer sealing the windings has melted. You can check for that by measuring the impedance (ohms) of either output valve anode to the positive on the 47uF. They should be the same but anything up to 10% or so discrepancy should be workable.

A high voltage (1kv working) poly capacitor (say, around 0.0022 uf) across the anodes of the of the 6L6′s can work wonders in stabilising an oscillating output stage. Not a cure for all ills, though.

There we are; an almost sensible blog. I’ll need a cold shower before my macaroon.