Polski (PL) 
English (United Kingdom) 
Started by Piotr, finished and edited by Grzegorz "gsmok" Makarewicz,
Instead of an introduction
Like many enthusiasts before me and many after me, I decided to build a tube amplifier myself. I admit that I do not have much experience when it comes to tube designs. However, I am a practitioner who has "made" several systems and a soldering iron is no stranger to me. I decided to describe the process of building the amplifier in quite detail. I decided to risk making mistakes that will be visible to all visitors to this page.
I think it's worth a try though. In this way I would like to break with the tradition of many websites on the Internet that show the final results of construction work omitting all the often painful details that accompanied the achieved results. Such websites are very interesting, but they bring little practical value to "experts" like me, who sometimes feel stupid to ask questions that bother them without fear of being exposed to a contemptuous shrug of the shoulders of "experienced tube enthusiasts". Well, enough of this talk. I take a deep breath and begin the report.
Piotr
Report 1 (March 5, 2002)
What amplifier to build? First of all, it has to be a proven design. There are plenty of amplifier schematics. From experience with other systems, I know how easy it is to get into a dead end by starting work on an unproven schematic. I have already had unsuccessful attempts even with systems from the famous Elektor. In electronic magazines, in the "Editors' Answers" or similar sections, there are many reports from desperate readers who cannot start up published devices that, as it turns out, only work in the editorial laboratory. I have reviewed a lot of materials on the Internet, in magazines and books. I think that the "gentle pressure" from my colleague from Fonar not to go crazy and choose the design of the so-called Williamson amplifier is completely justified. Why? For two reasons:
- The design is proven and works not only on paper. There are plenty of schematics that can be verified for correctness by simple comparison.
- I can use (this gentle pressure) the embryos of the design lying in the Fonar company's workshop.
I did some research in the literature and based on it I drew (or, to be more precise, simply copied) a schematic diagram of the amplifier, which is presented in Fig. 1.
Fig. 1. Schematic diagram of the amplifier
I deliberately did not enter the values of the components, because in the diagrams from different sources they differ slightly or sometimes even greatly. I do not know how significant the differences in these values will be for the amplifier's operation, but due to the upcoming search for them in order to complete the components, I have to create a list of parts before Sunday. Why before Sunday? Because on that day I intend to go to the famous Wolumen in Warsaw, one of the largest electronics markets in the open air. So I will present the list of components with commentary at the end of the week. In the meantime, I am taking up theory (it is sad but it cannot be completely avoided). In the next report I will write what I managed to dig up "on this topic".
Report 2 (March 8, 2002)
In order to determine the values of the components for building the amplifier, I analyzed the following sources of information:
- Article "Wzmacniacz lampowy audio", Elektronika Praktyczna 2/1998,
- Article "Wzmacniacz lampowy Amplifon WL36, Radioelektronik 8/1998,
- Article "Wzmacniacz lampowy Amplifon WL36 dla koneserów muzyki", Radioelektronik 8/1998,
- Williamson D.T.N., The Williamson Amplifier, Audio Amateur Publications Inc., 1994,
- Materials available on the Internet (e.g. materials on the ECC website).
The list of amplifier components is presented below:
- R1 - 470K
- R2 - 470/0,5W
- R3 - 47K
- R4, R5 - 22K
- R6, R7 - 470K
- R8, R9 - 24K/2,5W (4W)
- R10, R11 - 470K
- R12, R13 - 2K2
- R14, R15 - 470/4W (8W)
- R16, R17 - 1K/1W
- R18 - 220/4W
- R19 - 33K
- R20 - 5K6/2W (3W)
- R21 - 5K6/4W (5W)
- R22 - 220/2W
- C1 - 0,1µF/630V
- C2 - 0,01µF/630V (0,47µF/630V)
- C3 - 10µF/450V
- C4, C5 - 0,1µF/630V (0,22µF/630V)
- C6, C7 - 0,22µF/630V (0,47µF/630V)
- C8 - 0,01µF/630V (0,47µF/630V)
- C9 - 10µF/450V
- C10, C11 - 47µF/100V
- C12 - 0,22µF/630V
In brackets I have provided alternative values given in the source materials, in which the elements differ despite the same schematic diagram. As for the capacitors, I will try to use both variants of the values and measure their effect on the amplifier parameters. As for the given power losses of the resistors, for safety reasons I will initially assume the variant with a higher value. I have not yet made a final decision on the final power supply system. I think that in order to simplify the system I will give up the automatic delayed switching of the anode voltage for now in favor of a manual switch. I also do not have a specific opinion on the values of the filter capacitors in the power supply. The values found in the source materials are so divergent that, unfortunately, having no substantive basis for performing the calculations, I have adopted the following solution: I will use capacitors with a capacity of at least 220µF. The exact value of the capacitance will be determined simply by the possibility of purchasing such capacitors. Considering that the anode voltage of the system is about 430V, this may cause problems. When making a shopping list, when I was looking at the "anode voltage switch" item, I started to wonder if there were any switches for such voltages available on the market. I started to investigate and now it's time for one of the "stupid" questions I have to ask.
Question number 1
How is it that (I have checked it organoleptically in a few factory amplifiers) that ordinary elements for much lower permissible voltages are used to build delayed anode voltage switching systems (relays in automatic systems and manual switching switches). And this is done not only in amateur radio systems but also in factory systems. I understand that the value of voltage that can be applied to the relay or switch terminals is the higher the lower the value of the current flowing through it. The problem is that this only happens up to a certain point, i.e. until the maximum voltage is reached, which is usually around 250V. I looked in various catalogs and I admit that I did not find a switch for a voltage of, for example, 450V. I think that in my system I will use a series connection of contacts in a double switch for a voltage of 250V. Just in case. Unless I am making some general error in my reasoning?
Next Sunday I will also do some research on the possibility of buying a mains transformer and speaker transformers. I have learned that it probably won't work, but it's worth a try. I will write about it in the next report, after a visit to the electronics exchange. Fortunately for me, I have original Amplifon transformers at my disposal just in case. In the meantime, I am studying the descriptions of the operation of the circuit shown in Fig.1 and in the near future I will try to create my own study on this subject. I am also looking through Mr. Williamson's original articles from the mid-1940s. In the articles section of the Fonar website, a "free" translation of these materials should appear in the future. I have read them for the first time and I can already say that I am impressed by the requirements that Williamson imposed on speaker transformers, for example. I think I will return to this topic.
Report 3 (March 15, 2002)
I have just visited the Warsaw "Wolumen". As I expected, I had no major problems with acquiring passive components for building an amplifier. The problems started when I had to buy electrolytic capacitors for the power supply and transformers. I did not come across any electrolytic capacitors for voltages exceeding 450V (except for one seller who had "not very well identified electrolytic objects" for which he wanted some completely astronomical money. In this situation, all that was left for me was to buy a handful of capacitors for lower voltages, from which I will try to put something suitable together. I will write about it in the report, in which I will describe the concept of the power supply system. I have not written about it yet, because the system will be created with the purchased capacitors in mind.
A stone has been lifted from my heart when it comes to the switch for delayed switching on of the anode voltage (Question number 1). I am lucky that at the right moment (even before visiting Wolumen) I received from my friend Olek, who has the appropriate practical knowledge, several important comments on this subject. As I suspected in the case of relays and switches, the matter is based on the use of the fact that these elements operate at relatively small currents. This issue is perfectly illustrated by a fragment of Olek's statement in our internet discussion: "...Relays - in our case it is primarily about current. "Normal" relays operating in automation systems in household appliances are a 12 or 24 V coil supply and a voltage of 220 V, but they carry large currents - up to 10 A. In the case of the amplifier's output stages, the total current reaches values of, say, 200 mA. A voltage of 300 or 350 V is of no particular importance here. I have installed relays from a Sharp microwave oven: 12 V supply, 250 V, 10 A working contacts. They fulfill their role 100%...". So the matter is clear.
I will now return to the problem of transformers. First: a mains transformer. It turns out that unfortunately this is an exceptionally difficult element to obtain. I was unable to buy it at Wolumen. What is more, at the "transformer stands" I received information that "I do not have and will not have such transformers". In only one place did the seller express interest in winding a transformer, but on condition that I give him the exact winding parameters and order a sufficiently large number of pieces. I was unable to determine how much and how much it would cost.
Secondly: speaker transformers. To my surprise, I BOUGHT A SPEAKER TRANSFORMER!!! I was convinced that it would be easier for me to buy a mains transformer than a speaker transformer, but here's a surprise. After searching the entire Wolumen twice, I found a stand where speaker transformers were offered. Judging from the visual inspection, these transformers may be of good quality. Typical EI shapes are used as cores. There is a sticker on the transformer informing that it transfers frequencies from 20Hz to 20kHz. The upper frequency of 20kHz is not too much according to the requirements that Mr. Williamson set for himself in the 1940s, but one should not prejudge anything before taking measurements. I think that first I will use original toroidal transformers from Amplifon as transformers, which have proven themselves in the circuit according to Fig.1. After building the amplifier, I will try to connect the speaker transformer I mentioned and do comparative tests with the "Amplifon" transformer.
Now that I have made a decision about the amplifier's components, it remains to tackle its assembly. The amplifier will be assembled on an aluminum base in the shape of the letter U. In the next report, I will address the problem of the power supply and the tube filament. I already feel that I will have problems with hum, because the mains transformer that I have has only one filament winding.
Report 4 (April 9, 2002)
On a simple U-shaped base I mounted the basic components of the amplifier. The amplifier base measures 360mmx300mm. Photos of the chassis in three views are shown below.
Fig. 2. Top view
Fig. 3. Rear view
Fig. 4. Bottom view
I have already started the initial wiring of the amplifier (switches, speaker sockets, etc.). In the nearest report I will present a drawing with a detailed arrangement of the components. As you can see, I have tried to assemble it spatially. I am wondering whether the power supply itself should not be assembled on a small printed circuit board screwed to the chassis.
By the way, I would like to thank Mr. Jarosław, who sent me some very useful information via e-mail. Probably thanks to them I will try to make a second version of the amplifier on slightly different elements in the future.
Report 5 (18 May 2004)
Since I had the opportunity to look at the dusty construction of the amplifier described above for the last two years, I decided that its designer would probably never finish it and with his consent I decided to continue building the amplifier as the third!!! person (some fate hangs over this amplifier). I will try to describe my struggles with this device on this page and, in case of any problems, report them on the discussion forum (http://www.trioda.com/php/forum/index.php), where I speak as gsmok.
So, to the point. I inherited the mechanical construction with attached tube bases, transformers, switches and a set of most elements (photos above). I started by developing a general work plan. I decided:
- not to change anything from the mechanical side, although the chassis seems a bit too low to me,
- to assemble the power supply system and one channel of the amplifier first using the collected elements. The elements of this amplifier channel will be assembled using the spatial assembly technique,
- depending on the achieved effects, assemble the second channel using the same technique or using connecting strips.
I started with the power supply part. The schematic diagram of the anode power supply (one channel) is shown in Fig. 5. The anode voltage of the amplifier exceeds and is equal to approximately 430V. Obtaining electrolytic capacitors for a voltage of around 500V is very difficult. Hence, the system uses electrolytic capacitors connected in series. In order for the voltage to be distributed evenly between the capacitors connected in series, they were bypassed with resistors.
Fig. 5. Schematic diagram of the power supply
Since the no-load voltages of my mains transformer alarmingly exceed the values intended for the amplifier in relation to the original design, which assumed six electrolytic capacitors connected in series and parallel, I made a small modification consisting in inserting resistor R7, which I decided to use to reduce excessive voltage from the transformer if necessary. I will probably have a similar problem with the filament voltage. Unlike the anode winding, I plan to unwind the appropriate number of turns if necessary. In order not to worry about the values of these voltages for now, I will start the amplifier by powering it via an autotransformer. The criterion for selecting the supply voltage will be maintaining the correct filament voltage value.
It turned out that the electrolytic capacitors I have are slightly lower than the height of the chassis. To avoid any potential problems related to isolating the high-voltage components from the chassis (safety considerations first and foremost!!!), I decided to place the components on a glass-epoxy laminate board and then glue the laminate board to the chassis from the bottom.
Here is the board with the anode power supply components installed. The capacitors were glued and used as carriers for the remaining components.
Fig. 6.
Here you can see the details of one power supply channel. I decided to make independent anode power supplies for each channel (there is only one mains transformer). Before gluing the components to the board, I drew a grid allowing for their even distribution.
The holes in the board were cut out so that the nuts of the screws securing the mains transformer and the speaker transformers could fit in them.
Fig. 7.
The photo below shows the amplifier chassis from below and the board prepared for mounting inside the chassis.
On the chassis (at the bottom) you can see three large nuts securing the transformers.
Fig. 8.
This is what the amplifier looks like after inserting the board containing the anode power supply elements.
Fig. 9.
Another shot from a different angle. Here you can see how low the chassis is. Fortunately, thanks to the mounting method used, the most protruding elements of the power supply are 1 cm below the bottom surface of the chassis, so that even after turning the amplifier upside down, they are not in danger from the ground.
Fig. 10.
Report 6 (19 May 2004 - updated 25/05/2004)
I checked the anode power supply. It works OK. The voltage on the capacitors when the mains transformer is powered by 230V AC voltage without load is 470V. This is definitely too much in relation to the project. I hope that after connecting the system this voltage will decrease, and I will reduce the rest using a resistor in the anode power supply. By the way, I modified the anode voltage switch (this is the blue switch visible in the photos). In order for it to be illuminated by the neon lamp installed inside, the value of the series resistor had to be increased. Originally, a resistor of 150K was installed there, which I increased to 240k. The neon lamp shines slightly dimmer than in the original switch powered by 230V, but the difference is small. I must admit that I had a hard time assembling the disassembled switch (oh, those springs!!!). After "burning in" the power supply (the electrolytic capacitors had been lying around for a few years), I started the final wiring of the power supply and assembly of the elements of one channel of the amplifier. In Fig. 11. I show again the schematic diagram of the amplifier (I reworked it for my purposes, compared to the original I added the numbers of the tube terminals and the basic voltages that I will be guided by when starting the amplifier.
Fig.11. Schematic diagram of the amplifier (one channel) with the numbers of the tube socket pins and the most important voltages that determine the resting operating points of the tubes.
Below are photos showing the next stages of assembly.
This is what the fully assembled anode power supply looks like. The ceramic "cube" visible on the right was used to attach the thick glow wires.
Fig. 12.
Directly on the L1 and L2 vacuum tube socket I mounted all the elements of the voltage amplifier and phase inverter. In the lower part you can see the elements filtering the voltage supplying these stages (R16, C8, C9 and R8, C2, C3 - fig.11). The filters were assembled in the form of a module shown in fig.14. and fig.15, after which this module was glued to the chassis and served as an important element supporting other electronic elements. In this way I avoided using additional supports.
Fig. 13.
Fig. 14.
Fig. 15.
And here's a little "back to the past". While assembling the output stage elements, it turned out that the anode power supply was a bit of a nuisance. And here the design using the base with the power supply elements came in handy. With a quick movement, I removed the entire anode power supply to assemble the elements on the output tube sockets.
Fig. 16.
And here is a view of the amplifier after inserting the anode power supply board (of course I also did it with a "quick move").
Fig. 17.
Front view of the assembled interior.
Fig. 18.
View of the assembled interior from the back.
Fig. 19.
And here is the "spiderweb" in all its glory, i.e. a view from a "bird's eye view".
Fig. 20.
And that's it. After checking the correctness of the assembly, I placed the amplifier on the measuring stand (Fig.21). The measuring stand. The amplifier was loaded with the equivalent resistance, which can be seen on the right. Tomorrow is the day of truth.
Fig. 21.
Report 7 (20 May 2004)
I started to start the amplifier. I did it in stages, in such a way as to minimize the risk of its damage in the event of assembly errors. In order to avoid boring description of the subsequent start-up steps, I am posting it together with several photos of the amplifier on the measuring stand.
In the first step, I removed all the tubes, disconnected the feedback (resistor R21 according to the second diagram), connected the dummy load and turned on the supply voltage, observing whether nothing disturbing was happening, i.e. whether there were no short circuits, "jumping sparks", smoke rising above the heating element, etc. It turned out that everything was fine. I turned off the amplifier and waited for the capacitors in the anode power supply to discharge. You don't have to wait too long for this due to the fact that bypass capacitors are resistors through which they discharge after the voltage supplying the amplifier is turned off.
Fig. 22.
In the second step, I inserted the preamplifier stage tube (L1) and the phase inverter (L2). I shorted the amplifier input so that no signal would appear on it. After switching on the amplifier, I measured the voltages at the most important points of the circuit (the operating points of the L1 and L2 tubes). I found that everything was fine. I switched off the amplifier again and waited for the anode power supply capacitors to discharge.
Fig. 23.
In the third step, I inserted the push-pull output stage tubes and turned the amplifier back on. Then I measured the voltages on the electrodes of the L3 and L4 tubes. I checked the signal at the amplifier output with an oscilloscope. Unfortunately, it turned out that the amplifier had a large mains hum with a peak-to-peak value of 180mV. I accepted this fact calmly, preparing for the later tedious search for the causes of this phenomenon.
In the fourth step, I connected a sinusoidal waveform generator to the amplifier input and used an oscilloscope to measure the signals at various points in the system, including the signal at its output.
Fig. 24.
In the fifth step I turned on the feedback resistor. I was hoping that with a 50% chance that the feedback would be negative and not positive and that I would not have to swap the ends of the secondary winding of the speaker transformer. After turning on the amplifier it turned out that I was lucky and the amplifier did not excite. While tuning the generator I checked whether there were any visible signal distortions in the acoustic band. It turned out that there were none. However, the signal was superimposed with the previously mentioned hum. Adjusting the symmetrizing resistor placed in the filament circuit did not provide any significant improvement. I decided to stop the measurements and, as a relaxation, develop this description. While preparing the photos I looked at the details of the amplifier shown in Fig.24. and I was surprised to find that the mass of the filter module (details in Fig.14. and Fig.15) was not connected anywhere, so capacitors C2, C3, C8 and C9 were "hanging in the air". In the next step I will correct this error and re-measure the amplifier parameters.
Report 8 (22 May 2004)
I took the measurements after correcting the error related to the mass (Fig.25). It is very good. The hum has decreased to the level of single millivolts!!! Below I present photos allowing you to familiarize yourself with the details of the assembly. Some photos can (by comparing with previously presented photos) be used to see and avoid errors that I made during the assembly and which I had to correct.
The amplifier is back on the measuring stand. This time with a wire connecting the filter capacitors (those red blocks in the photo) to the mass.
Fig. 25.
Here you can see all the elements of one amplifier channel. The only unconnected element is the input capacitor - the signal is fed to the input not through the input socket but by means of a "tweezer" (the brown one) connected to the input capacitor leg. The second "tweezer" (green-yellow) was used to measure the anode voltage.
Fig. 26.
Here is an enlarged version of Fig.26. You can see more details here.
Fig. 27.
Here you can see how the anode power supply was connected in the final version.
Fig. 28.
Here are the details including the routing of the glow wires and the potentiometer used to minimize hum.
Fig. 29.
I couldn't resist showing this view. The oscilloscope shows the output signal with the input shorted - the oscilloscope sensitivity is 10mV/div.
Rys. 30.
To make sure of the hum level, I connected an 8-ohm woofer to the amplifier output. The hum is barely perceptible despite "sticking your ear to the speaker".
Fig. 31.
And this is what the amplifier looks like in all its glory from the top.
Fig. 32.
This is a side view. You can see that the mounting technique used has allowed the amplifier to maintain a relatively large clearance from the bottom despite the low chassis height (at least like in an off-road vehicle).
Fig. 33.
And finally, the last photo before continuing with further work.
Fig. 34.
I am now starting to assemble the second channel of the amplifier. Considering that I will be modeling myself on the channel that has already been started, I should not make any mistakes. In the next step, I will add detailed photos and a short description of these activities. I am also wondering whether to move the input sockets of the amplifier from the rear to the front panel. I could then connect the input capacitors directly to these sockets, thus shortening the path that the signal must travel from the input socket to the grid of the first tube.
Report 9 (29 May 2004)
I have finished assembling the second channel and thus the entire amplifier. Following the example of the first channel, I have assembled the second one without making the mistakes mentioned earlier. Below are photos allowing you to familiarize yourself with the subsequent stages of assembly.
Of course, I started with the filament wires. In accordance with the assembly rules, the wires were twisted before soldering. I used two wire thicknesses - a higher current obviously means a higher thickness. The thick wire is used to supply voltage from the transformer, which is then "distributed" to the voltage stage and phase inverter tubes and to the output stage tubes.
Fig. 35.
After routing the filament wires, I glued in the module containing the voltage stage filter capacitors and the phase inverter.
Fig. 36.
Now I have mounted resistors and capacitors on the ECC82 vacuum tube socket. The free terminals are waiting for connections.
Fig. 37.
Next, I soldered the components to the pins of the 6N8S tube socket and the output tube sockets (EL34).
Fig. 38.
Below is an enlarged fragment of the photo from Fig. 38.
Fig. 39.
Most of the components are already fully connected. The terminals connected to the amplifier's ground are not yet soldered.
Fig. 40.
And this is what the second channel of the amplifier looks like after assembly. The only "free end" is the feedback resistor leg.
Fig. 41.
Here are some more detailed photos. Here you can clearly see the feedback resistor waiting to be connected.
Fig. 42.
The cathode resistor of one of the output tubes (EL34) is clearly visible.
Fig. 43.
Some more details...
Fig. 44.
... and more...
Fig. 45.
... and more.
Fig. 46.
This is what the insides of the amplifier look like after assembly.
Fig. 47.
Report 10 (June 2, 2004)
The amplifier is finished and running. I only had to correct one mistake in assembly. The amplifier passed listening tests successfully. I will post the results of parameter measurements soon, in the meantime here are photos of the final form of the amplifier (in this version, because I am preparing a second version).
Front view.
Fib. 48.
Rear view.
Fig. 49.
"Amplifier guts", i.e. the upside-down construction.
Fig. 50.
Prepared by: Grzegorz "gsmok" Makarewicz, (formerly www.trioda.com)