dom's software
what do you mean it doesn't work?
ECL82 valve amplifier - PP UL
ECL82 Amplifier - Push-Pull Ultralinear


This is my first valve amplifier (or tube amplifier if you must).

Before I start please let me assert that:

  • I am no expert
  • The voltages in valve amplifiers are dangerous - do not copy anything here unless you are competent to do so - I accept no responsibility as to the safety or fitness of the circuits presented below
  • There are other sources of information on these subjects especially on the net. Do not rely on me!
  • This is part of a learning process for not rely on the veracity of anything below...check it yourself, or ask someone else!


I would greatly value any feedback you may have on this article, the amplifier it describes or anything else that is on your mind. Please send all feedback positive or negative to , be gentle though!


For years, I've always been fascinated by radio valves. Ever since poking around in my beloved Ultra reel-to-reel tape recorder, at the age of three - and finally breaking it after several years. I've always loved the sound, the smell, the look, the mystery of the mysterious glowing blobs. As time went by I lost my tape recorder, my Dynatron Mazurka record player, the kitchen radio, the old TV that used to periodically burst into flames...all valve equipment. All consigned by my parents to the tip, classified variously as either "too dangerous", "too big" or "too heavy".

But recently, whilst travelling in Germany in my old bus (see here ) I visited an antique shop and bought (porbably for far too much money) and old "Saba Freudenstadt" (working well) and a Phillips Saturn (not working). These Teutonic monsters rekindled my old passion for valves and set me on a course to build up a small collection of radios, radiograms and one old telly, all in various states of repair.

Over the past year I've re-learnt lots of things I once new, finally understood some things I never fully understood and discovered just how little I actually know about electronics - a degree in Physics not withstanding! Now I am finally getting to grips with the theory and design of radios, I've decided to have a stab at some Hi-Fi using radio valves, encouraged by the fact that a friend has kindly donated a pair of excellent GEC "FR" speakers - 1950s vintage but beats any modern so called HiFi speakers I've ever owned.

The starting points

I bought, some time ago an excellent book by John Lynsley Hood, "Valve and Transistor Amplifiers", ISBN: 0-7506-3356-5. This book is a brilliant buy for anybody from amateur to expert. It goes into some detail about basic circuits, how valves and transistors work, building up into very detailed analyses of some of the best vintage and contemporary Hi-Fi amplifiers. One of the circuits in this book is JLHs own Class-A transistor amplifier, this is in itself a design classic and has been built by many amateur constructors (including myself). I'm not going to go into any details except to say that it is an excellent amplifier and a brilliant project for anyone to undertake. I've had one sitting in my living room, running off an old UPS lead-acid battery and a battery charger powering the GEC speakers. It sounded great but after a year of daily use its finally died (inadequate heatsinks!). If you are interested in building one of these I'd not only recommend the book above but also the Class A Amplifier site. These are, to my mind, the next best thing to valve-amps.

However I decided mine was underpowered (12V doesn't give much power!), and I couldn't be bothered getting new heatsinks made or bought.

I decided I'd go for gold and build a valve amp, and if possible, match it as closely as possible to my speakers. This article describes, how I went about it and the trials and errors I've gone through! I hope you find it useful or interesting or at least amusing!

Component Selection

This probably isn't the right way to start, but its how it was with me!

I had already, on the spur of a drunken moment, bought some mains and output valve transformers from Maplin (Ultra-linear 43% tapped Danbury output transformers, 100mA max 240V-240/6.3V mains transformers), some tin boxes and valve holders and decided that these would all have to be used.

Also, another drunken buy from ebay, was a box of assorted radio and telly valves from somebody's loft. After going through this box the best set I could come up with was a number of PCL82s - these are 1950s valves designed for used in telly audio stages. They contain a low-power triode and a pentode power output valve, similar but not quite as powerful as something like an EL84 inside a single envelope.

At 7W maximum dissipation these would, appear to many, as either being too low powered, or two "modern" but for me they are more or less perfect. The speakers they will be matched to are pretty sensitive and in nice big cabinets, anything over about 15W would melt the voice coils! I also have a matched pair of KT88s/6G6s (can't remember which exactly) these though are expensive and I didn't want to go breaking these on my first attempt...I'll save them for when I'm a bit more confident.

Circuit selection

The next step for me was to decide on a circuit topography. The output transformer and the choice of pentodes for output valves had me more or less decided on an Ultralinear push-pull setup.

As this is my first attempt I needed a starting point, so I had a bit of a look on the internet for a circuit to start with. I could then just mess with various component values and decide on a layout. I did a few searches and in the end came up with this site. After looking at the various circuits on offer I settled for the Supertone S9 circuit, on the basis that:

  • it looks simple - I undertand it!
  • it looks nice - I like the symmetry of it
  • it looks cost effective - not too many parts
So once again, a nice subjective set of reasons to good as any when you're experimenting!

I will of course be fiddling a little with this circuit straight off. Firstly I'll make my own power supply, using solid state devices for now...more on this later. But also I'll be messing with overall NFB (negative feedback) and adding Ultralinear mode.

First-cut circuit

Firstly I printed out the ciruit from the internet and labelled up all the components (R1,R2 etc).

First cut circuit diagram

I also scribbled a few extra components for UL connection on here, as you can see I started off just connecting these up through 100R resistors. I had decided to do this after a bit of reading around on the net that this can help with oscillations etc. More on the screens later.

Next I checked that the values for the cathode bias resistor on the output valves was roughly right. I did this by plotting a load line on a rough graph I found in the brimar valve data for the PCL82. An excellent article on load lines and amplifier construction in general can be found at Boozhound.

I calculated the load on each anode as follows:

Firstly find out the turns ration of the transformer. The simplest way of doing this was to apply 2Vac to the secondary of the transformer, then measuring the voltage across one half of the primary, this came out at roughly 34Vac. Giving a ratio of 17:1.

The load of the speaker at the primary is given as the square of the ratio multiplied by the impedance of the speaker = 17*17*15 =~ 4300ohms

First go at load lines

I have made several assumptions here, most of which turned out to be wrong! more on this later:

  • B+ will be 300Vdc
  • therefore I want the anode to be 150Vdc - wrong!
  • the load (as seen by the valve) will be about 4300
  • The screen voltage will be about 200Vdc - wrong

I then picked a point on the graph where I reckoned roughly that I wanted to be. This was with Vg=-12, Va=150, Ia=40ma. This would have each valve drawing 40ma, so there will be 80mA going through R9. We want the voltage across R9 to be 12V, using ohms law this comes out as R9 needing to be roughly 150ohms. Rather than the 120ohms given on the diagram.

Next I started up my trusty RISC OS computer so that I could use the free, yes that is free, vector drawing program "!Draw" to allow me to do a quick sketch of a layout. It still amazes me after all these years that I still haven't found anything to better the combination of simplicity and flexibility in !Draw. I can usually lay something out in here fairly quickly (although I don't have any stock-part symbols for the stuff I'll need here. Anyway after a couple of hours I have something like this:

First cut layout

This layout is a 1:1 scale view of what I actually laid out. As you can see I'm being fairly neat (for me anyway) and laying all the components out on tag boards rather than hanging them off the valve pins. I do this not only out of a sense of neatness, but also because I may - indeed will, probably want to make lots of changes. Otherwise there probably would be more components close round the valves - to keep lead lengths to a minimum. The dotted line indicates the NFB line. Though I'm not going to include this...for now no NFB in my amp!

Build it!

I am an impatient bugger at the best of times, and now I really wanted to get going with the soldering iron. The only thing I need now is a way of cutting out the holes in the boxes. For this project I'm using some soft aluminium two-part boxes. In the past I've tries using my hand-drill and a hole cutter to do this, and also Dremels hack-saws, files all sorts of faffing around. Finally I bought the bullet and got some Q-cutters. Lawson's deliver them a couple of days later and within 2 minutes I've got two beautiful, perfect round 7/8" dia holes cut for my valve bases. About one ear-splitting hour later I've finished filing away the rough hole for the transformer! Mount the tag boards, and fire up the soldering iron! About 2 hours later I'm finished. A few more scribbles on the circuit diagram where I've substituted values and voila I'm ready to test.

Test it!

I'm now ready to give it a test. First I need some power-supplies. For the HT I've got a small power-supply I've had lying around that will give ~300VDC from a solid-state rectifier. For the heater supply I'll knock up a small 15V supply, PCL82s are meant to be 16V but it'll have to do! Plug in a speaker from an old radio. Connect to portable CD player via a 1000V/0.1uF capacitor - for safety?!?!

Now the moment of truth, and, nothing....Wiggle a few wires with insulated scew-driver, nothing...hit power supply, aha yes we have something, turn up the CD player, aha not bad. Quite a lot of hum, but not bad.

Now try the big speakers, - less hum funnily enough. Will have to investigate, but I think this is due to a difference in impedance? That sounds pretty bloody good though - and first time, big grin...listen for 20 minutes...time for the pub to celebrate!

Fine-tuning and fault finding

Now for a more serious bit of investiagation, first with NO input. The observations are:

  1. Hum - lots of it
  2. Buzzing - 50Hz clicking, shows up on scope as a "click" just before the peak of the hum. This does not show up on the mains input though!
  3. Microphony - tapping the chassis makes a "bong" noise and affects the humming???
  4. Oh dear, the PSU is getting hot
  5. Oh dear, oh dear, the plate of one of the valves is glowing!
  6. A strange buzzing can be heard coming from the valves!

Switch off, turn over, switch on and investigate....

The hum, can be left until later...firstly lets do some measurements on whats really going on...Measure the voltage across the cathode bias resistor, this will give us an idea of how much current is being drawn about 20V - hmm, voltage at anode, about 270V, hmmm, voltage at screens, about 270V also....Obviously I've got something very wrong and need to go back to the drawing board....First I think I ought to get it all running a bit cooler!

Load lines - do it right this time!

First I measured the actual impedance of the speakers (using a 1Khz sinewave)... these are more like 17ohms! So the anode load we will use will be 5000ohms

Next I need to rethink, the screen voltages. I'd not take into account how much the screen voltage affects the Ia/Vg characteristics. Going back to the valve characteristics I found this graph.

Another set of Ia/Vg from Brimar

As can be seen, by comparing this graph to the previous one, the higher screen voltage has affected the anode currents....that'll be where I've gone wrong and why everything is getting so hot!

Now for some re-thinking!: Firstly some new assumptions:

  • The grid bias will need to be higher roughly 20V
  • The screen voltages will be (measured from Cathode) about 225V, I will change the screen resistors on the amp to achieve this.
    I've no reason for choosing this point, I just did, I don't know whether it is right or wrong!
  • The anode voltage will be 275V

This still leaves me with a problem...I could use the Brimar data sheet, but it has a lower 200V screen setting so the measurements could still be wrong! Luckily though a while back I bought an AVO valve characteristic meter. This machine should let me plot my own Ia/Vg graph. And so it does!

Loadlines, Vg2=225V, using an AVO Valve Characteristic Meter

This whole fixed screen thing concerns me though - as this is a UL amp it won't stay fixed, so what should I do instead. Anyway for my static point calculations this will all have to do!

Here I've decided on a new point giving 275V @ 36mA, Vg=-19. As can be seen this is still running the valves a bit too hot, but not stupidly so like before. This will reduce the life of the valves a little, but hopefully not by too much!

This requires a cathode resistor of R9=19/(36mA * 2)=263ohms, the nearest handy size is 270ohms, this will give us slightly more grid bias, slightly lower anode voltages and currents and so will bring us closer to the 7W maximum dissipation

I'm also assuming that the screens will draw about 10mA. So to get my 50V drop I'll put in some 4k7 resistors.

Out with the soldering iron, in with the new resistors, turn it on and voila, everything measures within a few volts of where I was expecting, the valves don't get too hot after half an hour continuous play (although the PSU is still struggling a little).

Humming, Buzzing and other noises

First the microphony. I am a bit disappointed about this, especially as the valves seem to actually make a noise! A kind of annoying 50Hz rasp! From the speakers is a fairly normal 50Hz hum, but with a bit of a strange ringing sound to it! Tapping the valves and moving them in their seats affects this noise. Swapping valves improves the situation a bit but does not eradicate the problem.

A brain wave...I remembered reading recently about the heaters of amps actually interacting with the cathode and drawing current that way! The heaters in this set so far are floating around, lets measure whats going on with the heaters with an oscilloscope. I'm afraid I don't have a picture of this but its a very scary spiky messy picture! First I'll try grounding one side of the heaters. Da da! An instant vast improvement...however I remember somewhere reading about this being "bad". I'll bung in two resistors and have the heater chain at 0V. I'm guessing this will a) hold the heaters steady. b) keep the +/-8V haveform away from the +ve 20V cathode. So long as the heater is more negative than the cathode it should repel the electrons? Is this true?.

This simple (not so obvious to me except in hindsight!) solution has got rid of a lot of the problems above, even reducing the microphony! I guess the filament moving about inside the cathode was responsible! I'd be very interested to hear from anyone with ideas on this!.

More hum

There is still a lot of hum...Measuring the output from the PSU with an oscilloscope shows why...+/-15V ripple! Quick swap of the two 8uF smoothing capacitors for 47uF versions makes a fair improvement in the hum, but a worsening in volume and character of the rasping...I think I know what this is...yes its the noise caused by the sudden turn on of the current through the solid state rectifiers. I discover this firstly by putting a lid on the PSU! Secondly be reading this excellent article on ground-loops and star-earthing. My first move is to re-wire the PSU as advised....this more or less eliminates the problem. However there's still some pickup - I guess I will have to re-wire the whole amplifier, with a star grounding layout!

New layout with star earthing and tied heaters

This new layout gives much improved sound! There is still some hum. But I find it acceptable at the current level.

For completeness I decided I'd do a full circuit diagram. Here it is: (If you want this in !Draw or SVG format email me and I'll send it on).

New layout with star earthing and tied heaters
The finished insides

Heat, the heat!

My PSU is still over heating (especially with the lid on the box!). I didn't expect this to be such a problem. My next step will be to cut some vents, and possibly fit a fan. I'm guessing that this is due to:

Anyway here for completeness is the final PSU layout and a picture! I would come up with something better if I were you though!

PSU layout
Finished PSU innards
  • An inadequate transformer (Danbury DB342) - though it is rated at 100mA, I guess the capacitors may be drawing a lot more instantaneous current, leading to core heating?
  • Inadequate ventillation

Any advice from anybody? I'd like to not have to switch it off every hour to cool down!

Finish construction

As can be seen on the layout I've now included a simple volume control, a phone jack input etc. I finished off the one amp, then ordered some more parts for a second amp. Also I've decided to fork out on NEW VALVES! And instead of PCL82s I'm going to plump for ECL82s. The very nice chap at Crowthorne Tubes can do me 40 for &156;20, (about $40) so not exactly breaking the bank. These tubes should hopefully improve matters in several ways:

  • They run on 6.3V heaters, so there may be less cathode interference?
  • They run on 6.3V heaters, I can use the 6.3V output of my B+ PSU instead of having a separate supply
  • They are new!

The new valves certainly do improve matters! The hum is less and is far less intrusive sounding (it is now a nicer shaped sine-wave instead of slightly jagged). I'm not sure if this is due to running from a bigger transformer, the interaction between the cathode and heater, or that it is not from a separate supply.

Things I may try in the future

If you have any suggestions or requests I'd be happy to hear from you!

Over the next few weeks I'm going to try and build a signal generator (I'll write that up too if the response to this article is positive!). When I've done this I'll publish some updates here with some measurements of bandwidth, bode-plots, attempts at applying more or less NFB, and some rough THD figures if I can measure them reliably

  • A bigger mains transformer - less heat more power, and I can fit bigger/more smothing caps
  • A smoothed DC heater supply - less heater hum?
  • Negative feedback - more bass?
  • Bigger valued capacitors in signal path more bass?
  • Fiddle with the input stages, better symmetry, more gain?


All in all this has been a very enjoyable project. The amp sounds great (to me anyway, and even my girlfriend was fairly impressed). I'll try and get some of my mates round for a listen...they are usually fairly critical!

The finished article playing some Queen!

Useful Links