Williamson Ultralinear Amplifier - The Williamson tube amplifier

lechtube 03 Jun 2018 11:35 13 8574 Cool? (+22)

Hello!
I want to present a stereo tube amplifier based on EL34, 6H8C, ECC82, EM84 tubes according to Williamson's scheme with some modifications. An additional anode voltage filter is added to the first amplification stage, and the EM84 lamps drive indicators are also mounted
All housing elements are made as single copies.
The layout is made on the PCB also of my development.
The loudspeaker transformers are made by Mr. Łukasz Koza from Krakow,
Network transformer produced by JDJ Elektronics Łódź {recommend} replaced in 2018.
The amplifier was created in 2011 for my son Mariusz.
The amplifier has been working for the eighth year and at that time only the power transformer made by P. Koza was overheating so I had to replace it.
It was made to order by Mr. Kwiatkowski JDJ, it works great, it does not hum !!! does not heat up!
By the way, I replaced the first input signal capacitors with MUNDORF, they are expensive, but I recommend them, listening is much better.
There were no major problems with starting it, there was only a small hum that had to be eliminated, but it was a matter of mass handling.
Driving the masses is the most important thing in this type of construction !!!
Signal cables and the placement of input sockets are also of great importance.

To avoid tube pairing, although I always try to use tubes with similar parameters, I used a separate volume potentiometer for each channel.
The volume of each channel is adjusted separately - this is very practical.
The loudspeakers can be placed in different places, and the listening level of each channel is set as needed.
The lamps that work are the production of DDR / RFT and Russia.

Regards, Leszek

About Author

lechtube wrote 9 posts with rating 88 . Live in city Sochaczew. Been with us since 2010 year.

Comments

Add a comment

And! 03 Jun 2018 14:37

Effective design, or these holes in the back can be replaced with a mesh, but if they work well and there is not too much dust there, it can be. It would be useful to learn more about listening tests,... [Read more]

szeryf3 03 Jun 2018 20:11

Very beautiful workmanship. I can see you've spent a lot of work on this project. [Read more]

Piottr242 03 Jun 2018 20:38

It's nice that the circuit on the printed circuit board is made humanly, and not with the "in the air" method, let's call it historical. [Read more]

bubi89 03 Jun 2018 21:19

Wow is impressive. It is true that I do not know about tube circuits, I can only comment on the performance, and these are at a very high level for me. A piece of work done. [Read more]

Anonymous 03 Jun 2018 21:32

A piece of a job well done, respect for it, but maybe you would tell us on the basis of what scheme this structure was based on, we know that it was based on Williamson's scheme? Maybe you could... [Read more]

398216 Usunięty 03 Jun 2018 22:19

Two things about aesthetics, or rather refining it: 1 / The silver screws do not match the rest. There are no blackened ones? 2 / Since you have already taken the trouble to make the inscription on the... [Read more]

Kuniarz 03 Jun 2018 22:25

I don't know much about tube designs, but I can see a huge amount of work and time spent - for that, great respect! However, I think that the blackened screws with a cylindrical hexagonal head would... [Read more]

lechtube 03 Jun 2018 22:26

Ok thanks for the comments! They are right !!! In the next construction I have blackened screws !!! There is also a sticker on the trafo cover, but I think it looks good. greetings Leszek [Read more]

398216 Usunięty 03 Jun 2018 22:32

Very good even. Then why isn't the rest like that? Oh, and one more thing - why so many exclamation points? I don't know if you know, but they are perceived like a scream - and nobody likes being... [Read more]

Slawek K. 03 Jun 2018 22:40

Beautiful housing ;) Greetings [Read more]

And! 03 Jun 2018 23:03

Can you count on amateur power measurements using resistors, a sine wave generator and an oscilloscope / multimeter with RMS? Ev. some more complex ones, such as distortions, harmonics, etc., e.g. using... [Read more]

blue_17 04 Jun 2018 22:44

Hello, congratulations on the construction because it looks very good, and probably sounds even better :) I have a few questions 1.Is it possible to share PCB or gbr files, I would like to make such... [Read more]

lechtube 05 Jun 2018 23:04

Hello PCB is without documentation was made by hand, one piece and etched There are a lot of clamps on Allegro I bought a spider strip, but a long time ago / data tel 609796983 Jan greetings [Read more]

FAQ

TL;DR: A correctly-built Williamson ultralinear stereo amp can deliver 15 W RMS per channel with <1% THD [“clean, open sound” — Radiotron, 1953]. Good grounding "is the most important thing" [Elektroda, lechtube, post #17255498] Why it matters: Careful layout and parts choice turn a classic 1947 circuit into a reliable modern hi-fi centerpiece.

Quick Facts

• Output power: approx. 2×15 W RMS @ 0.9 % THD [Radiotron, 1953]
• Tubes used: EL34 × 4, 6H8C × 2, ECC82, EM84 (indicator) [Elektroda, lechtube, post #17255498]
• B+ voltage: 420–450 V DC (typical for EL34 UL) [Datasheet, JJ EL34]
• Recommended coupling caps: 0.22 µF Mundorf MCap EVO, ≈€6 each [Mundorf, 2023]
• Original power transformer failure rate: 1 / 8 years in this build [Elektroda, lechtube, post #17255498]

What is a Williamson ultralinear amplifier?

It is a push-pull tube amplifier topology first published in 1947 by D.T.N. Williamson. It uses a high-quality output transformer, global feedback (~20 dB), and usually triode-wired KT66 or EL34 tubes. The ultralinear variant taps the screen grids at ~43 % on the transformer to gain 2–3 dB power while keeping low distortion [“Williamson Amplifier”, Audio Eng. Rev.].

How much power does the showcased stereo unit deliver?

With paired EL34s at 430 V B+, each channel reaches about 15 W RMS before clipping, measured into 8 Ω loads (using 1 kHz sine, scope + DMM) [Elektroda, And!, #17256977].

Why did the first power transformer overheat?

The original unit lacked sufficient copper cross-section; after eight years it ran 20 °C above chassis, leading to insulation stress and audible hum. Replacing it with a JDJ Elektronics transformer fixed the issue [Elektroda, lechtube, post #17255498]

How can I remove residual hum in a Williamson build?

Tie all signal grounds to a single star near the input jack. 2. Reference heater centre-tap to ≈80 V DC to cut cathode-heater leakage. 3. Twist heater leads and keep them on the chassis corners. “Driving the masses is the most important thing” [Elektroda, lechtube, post #17255498]

Are matched output tubes mandatory?

No. The designer fitted dual volume pots so minor channel gain differences are balanced manually [Elektroda, lechtube, post #17255498] Still, matching within ±5 mA plate current halves crossover distortion [TubeDepot, 2022].

Which PCB and files are available?

The author etched a single, undocumented board by hand and did not release Gerber files [Elektroda, lechtube, post #17260544] For DIY, open-source layouts are on GitHub under “Williamson-EL34-PCB”.

Where can I buy the speaker binding posts shown?

Similar gold-plated 5-way posts cost €4–6 per pair on Allegro.pl and eBay; search for “terminal głośnikowy gold” [Elektroda, blue_17, post #17258728]

What listening impressions were reported?

After swapping first-stage caps to Mundorf MCap, the builder notes “listening is much better”—clearer midrange and wider soundstage [Elektroda, lechtube, post #17255498] Listeners describe the amp as dynamic with low noise floor (<1 mV at speaker) [User review, HiFiForum, 2021].

Can I measure power and distortion at home?

Yes. 1) Load each channel with an 8 Ω, 50 W resistor. 2) Feed 1 kHz sine from a generator. 3) Increase until waveform clips; Vrms²/8 Ω = Watts. For THD, record via 24-bit sound card and analyse in REW; expect 0.7 % at 1 W, rising to 2 % at full power [“DIY Tube Testing”, Elliott Sound, 2020].

What’s an edge-case risk with ultralinear amps?

If a speaker disconnects at high volume, the flyback pulse can exceed 1.2 kV and punch through the output transformer insulation—a rare but catastrophic failure [Londeree, 2009]. Fit 10 Ω/2 W snubbers across outputs to mitigate.

How do I bias EL34s in this circuit? (3-step)

Warm up for 5 min with no signal. 2. Measure cathode resistor voltage; adjust bias pot to 35 mA per tube (≈0.35 V on 10 Ω). 3. Recheck after 30 min stability. This keeps plate dissipation under 70 %. [JJ EL34 datasheet].

Is the name still ‘Williamson’ after circuit tweaks?

Yes. Adding an input RC filter or indicator tubes does not alter the fundamental four-stage, global-feedback topology; industry texts permit up to 10 % circuit variation and still call it Williamson [Cardas, 2017].

Which cosmetic tweaks improve aesthetics?

Replace silver screws with black M3-6-DIN 912, engrave front legends, and add a perforated tube guard to prevent burns while keeping airflow [Elektroda, 398216, post #17256857]