The unit, which I christened Anasi, was entrusted to me by its owner due to a series of faults. The problems included tube sockets that did a poor job of retaining the tubes, crosstalk and interference between channels: a disruptive listening experience in which inactive channels could be heard in the background even when switching channels. In addition, the balanced input turned out to be a simulation with the phase shorted, essentially operating as a standard unbalanced channel.
Opening the unit was eye-opening… The customer insisted on a complete rebuild, practically turning the device into a brand-new one, with very few elements of the original retained. Personally, I would not have taken on this job if it hadn’t been for the client’s strong request. Don’t get me wrong: the customer is always right, and designing a new circuit and reusing the framework were still fun and stimulating experiences. I carried out this work in my spare time, putting in passion and effort to achieve a satisfying result. When the unit was delivered to me, I immediately took a series of detailed photos and recorded short videos to accurately document its original condition.
The birth of Anansi
After completely disassembling the original amplifier down to the last screw, the idea of what Anansi should be began to take shape. Anansi retains the basic concept of the unit that was scrapped, namely a Parallel Single-Ended with KT150 tubes wired as triodes, 25 watts RMS per channel. The original unit used an ECC82 with sections in parallel to drive the two KT150s. However, preferring to avoid tubes with parallel sections and wanting a robust drive for the KT150s, I opted to use the 6CL6 tube wired as a triode. Below are the curves of the 6CL6 in triode mode; it has a mu of 20 and a decidedly higher current-delivery capability than an ECC82.
This tube was implemented as the driver stage, ensuring adequate drive capability for the KT150s. For the input stage, I used the ECC82 configured as a long-tail phase splitter, despite the circuit being single-ended. This solution was adopted to avoid excessive circuit gain.
I also planned to handle the balanced input with coupling transformers, specifically the Split-TL082. These are 3-winding 1:1:1 transformers of 600 ohms, suitable for accepting a balanced input signal and converting it to unbalanced. This solution preserves the benefits of a balanced cable, such as noise rejection, which does not occur if only one phase of the signal is used.
I then decided to eliminate the Chinese remote control board and design a custom Arduino-based board. This not only manages volume control, but also channel switching and driving the Nixie indicator for the active channel. I completely replaced the original transformers, saving only the covers, and designed a new set of power and output transformers. In addition, I calculated a new filter choke with an air gap. The switch originally used for channel selection was converted into a 2-position switch to enable or disable the feedback signal.
Next, I designed in CAD and had a workshop make an essential steel cover, meant to be firmly fixed to the amplifier body with screws; I continued with the creation of a mounting support for all the wiring, using insulating materials. I then began assembling Anansi’s circuit. During this process, I managed to salvage some parts from the original circuit, including capacitors, anchors, the output tube sockets, the board for the VU meters, the motorized potentiometer, the infrared sensor for the remote control, the speaker connectors and the input RCAs, as well as various screws, washers and nuts, in addition to the original chassis and the copper plate that supported everything.
Assembly in practice
In the photos below, you can see the design process of the sturdy steel cover, the circuit mounting plane made of insulating material, as well as the bottom and mounting supports:
In the following photos I show the construction of the two audio transformers dedicated to the balanced input, together with the relay board that manages the 3 total inputs with signal-grade relays, equipped with silver contacts sealed in an inert atmosphere. The goal of this approach is to minimize the length of audio connections to avoid interference between signals, while always ensuring a perfect contact. This is in stark contrast to bronze switches exposed to air and humidity which, over time, can oxidize, and to long, unshielded wires running in parallel suspended from a piece of iron wire.
In the next set of photos and videos, I show the prototyping of the Arduino-based control board, and the source code of the control program is available for anyone who wants to explore it.
/*
* Firmware 1.0 per Ricevitore Remoto Tektron - Progetto PSE-KT150
* Autore: Stefano Bianchini
* Azienda: SB-LAB (www.sb-lab.eu)
* Licenza: GNU General Public License version 2 (GPL-2.0)
*
* Descrizione:
* Questo firmware controlla le funzioni di un amplificatore a valvole KT150 single-ended parallelo,
* ricostruito utilizzando il telaio di un amplificatore Tektron demolito. Il programma accetta input dal
* telecomando infrarossi originale Tektron. Gestisce la fase di accensione dell'amplificatore impostando
* un periodo di muto pre-riscaldamento, controlla il cambio dei canali e l'aggiustamento del volume tramite
* gli input del telecomando e visualizza il canale attualmente selezionato su un tubo Nixie originariamente
* integrato nell'amplificatore Tektron. Il programma include una funzione per prevenire l'avvelenamento del
* catodo del tubo Nixie.
*
* Description:
* This firmware controls the functions of a KT150 single-ended parallel valve amplifier,
* reconstructed using the chassis of a Tektron remolito amplifier. The program accepts input from the
* original Tektron infrared remote control. It manages the amplifier's power-up phase by setting a
* pre-heating mute period, controls channel changes and volume adjustments via remote control inputs, and
* displays the currently selected channel on a Nixie tube originally integrated into the Tektron amplifier.
* The program includes a function to prevent cathode poisoning of the Nixie tube.
*
* Nota sulla Licenza / License Notice:
* Questo software è distribuito con la Licenza Pubblica Generale GNU versione 2 (GPL-2.0).
* Ciò significa che sei libero di utilizzare, modificare e distribuire questo software in base ai termini
* della GPL-2.0. Ogni modifica apportata al codice deve essere resa pubblica e deve includere una nota
* indicante le modifiche effettuate. Ogni distribuzione del codice deve includere questa nota sulla licenza
* e preservare i riferimenti all'autore originale.
*
* This software is distributed under the GNU General Public License version 2 (GPL-2.0).
* This means that you are free to use, modify, and distribute this software under the terms of the GPL-2.0.
* Any modifications made to the code must be made public and include a note indicating the changes made.
* Any distribution of the code must include this license notice and preserve the references to the original author.
*
* Contattami / Contact Me:
* Per eventuali domande o miglioramenti al codice, puoi contattarmi all'indirizzo:
* https://www.sb-lab.eu/mandami-una-e-mail/
*/
#include <Arduino.h>
/* compilare in presenza della libreria IRRemote 4.20 */
#include <IRremote.hpp>
/* pin associato al ricevitore infrarosso */
#define IR_RECEIVE_PIN 2
/* Adatta il ricevitore IR, un valore di 20 è consigliato per il ricevitore recuperato dal tektron, valore diversi causano ricezione di dati casuali */
#define MARK_EXCESS_MICROS 20
/* definisco le stringhe che rappresentano i tasti */
#define tasto_meno 0xF6093A00
#define tasto_piu 0xEA153A00
#define tasto_on_off 0xF20D3A00
#define ripeti 0
/* definisco i pin dei rele' degli ingressi */
#define rel1 3
#define rel2 4
#define rel3 5
/* pin che controllano il demultiplexer MC14067b */
#define nix_A 9 /* pin 10 MC14067b */
#define nix_B 10 /* pin 11 MC14067b */
#define nix_C 11 /* pin 14 MC14067b */
#define nix_D 12 /* pin 13 MC14067b */
#define nix_EN 14 /* pin 15 MC14067b */
/* pin che controllano il driver motore L293D */
#define motor_up 8
#define motor_down 7
/* led integrato nel pulsante di accensione */
#define power_led 6
/* codici telecomando */
unsigned long value;
unsigned long old_value;
/* memoria canali*/
unsigned char canale = 1;
/* variabili per timer non bloccante usato per il controllo fluido del motore */
unsigned char vol = 0;
unsigned int timer = 0;
/* costante per fludita' motore del volume */
#define nbdelay 12288
/* impostazioni del timer per la funzione disossida */
unsigned long ultimoRichiamo = 0;
unsigned long intervallo = 60000; /* Intervallo di 1 minuti */
/* impostazione del timer mitigazione */
unsigned long ultimoRichiamo2 = 0;
unsigned long intervallo2 = 3000; /* Intervallo di 3 secondi */
void setup()
{
pinMode(LED_BUILTIN, OUTPUT);
pinMode(power_led, OUTPUT);
pinMode(rel1, OUTPUT);
pinMode(rel2, OUTPUT);
pinMode(rel3, OUTPUT);
pinMode(motor_up, OUTPUT);
pinMode(motor_down, OUTPUT);
pinMode(nix_A, OUTPUT);
pinMode(nix_B, OUTPUT);
pinMode(nix_C, OUTPUT);
pinMode(nix_D, OUTPUT);
pinMode(nix_EN, OUTPUT);
/* Abilita il lampeggio del led integrato che indica la ricezione di qualcosa, utile per debug */
IrReceiver.begin(IR_RECEIVE_PIN, ENABLE_LED_FEEDBACK);
/* accende il led nel pulsante power */
digitalWrite(power_led, HIGH);
/* nixia spenta */
digitalWrite(nix_EN, HIGH);
}
/* funzione che abbassa il volume quando si esce da un canale */
int fade_out()
{
digitalWrite(motor_down, HIGH);
/* durante l'uscita dal canale lampeggia il numero sulla nixia */
for (char i = 0; i < 10; i++)
{
digitalWrite(nix_EN, HIGH);
delay(500);
digitalWrite(nix_EN, LOW);
delay(500);
}
digitalWrite(motor_down, LOW);
}
/* funzione che rialza il volume quando si entra in un nuovo canale */
int fade_in()
{
digitalWrite(motor_up, HIGH);
delay(2000);
digitalWrite(motor_up, LOW);
}
int cambia_canale()
{
if(canale > 3) { canale = 1; }
if(canale == 1) { fade_out(); digitalWrite(rel1, HIGH); digitalWrite(rel2, LOW); digitalWrite(rel3, LOW); scrivi_nixia(canale); fade_in(); }
if(canale == 2) { fade_out(); digitalWrite(rel1, LOW); digitalWrite(rel2, HIGH); digitalWrite(rel3, LOW); scrivi_nixia(canale); fade_in(); }
if(canale == 3) { fade_out(); digitalWrite(rel1, LOW); digitalWrite(rel2, LOW); digitalWrite(rel3, HIGH); scrivi_nixia(canale); fade_in(); }
canale ++;
ir_mon();
delay(250);
}
/* Slampeggia il led power */
int ir_mon()
{
digitalWrite(power_led, LOW);
delay(50);
digitalWrite(power_led, HIGH);
}
int motore(char tasto)
{
if (tasto == '+' && vol == 0) /* Tasto Vol + */
{
ir_mon();
vol = 1;
}
if (tasto == '+' && vol == 1) /* Tasto Vol + con pot in marcia */
{
ir_mon();
timer = 0;
}
if (tasto == '-' && vol == 0) /* Tasto Vol - */
{
ir_mon();
vol = 2;
}
if (tasto == '-' && vol == 2) /* Tasto Vol - con pot in marcia */
{
ir_mon();
timer = 0;
}
}
/* questa funzione accetta in ingresso un valore da 0 a 16 e imposta i 4 bit all'ingresso del demultiplexer */
int scrivi_nixia(unsigned char cifra)
{
digitalWrite(nix_A, (cifra & 0b0001) ? HIGH : LOW);
digitalWrite(nix_B, (cifra & 0b0010) ? HIGH : LOW);
digitalWrite(nix_C, (cifra & 0b0100) ? HIGH : LOW);
digitalWrite(nix_D, (cifra & 0b1000) ? HIGH : LOW);
}
/* visto che la nixia per la maggiorparte del tempo visualzzera' sempre e solo i numeri 1-2-3 e questo potrebbe causare avvelenamento dei catodi
* con la compromissione della nixia stessa, questa funzione richiamata ciclicamente fa una scansione
* di tutte le cifre della nixia per evitarne il deterioramento precoce. */
int disossida()
{
for (unsigned char i = 0; i < 10; i++) { scrivi_nixia(i); delay(125); }
/* alla fine della scansione ripristina la cifra corretta che corrisponde al canale correntemente attivato */
scrivi_nixia(canale - 1);
}
void loop()
{
/* procedure di partenza */
/* all'accensione gli ingressi sono tutti e 3 disattivati faccio lampeggiare lo zero sulla nixia per 40 secondi in attesa del riscaldamento
* mentre lampeggia abbasso anche il volume a zero per evitare uragani alla partenza */
scrivi_nixia(0);
for (char i = 0; i < 40; i++)
{
digitalWrite(nix_EN, HIGH);
delay(500);
digitalWrite(nix_EN, LOW);
delay(500);
}
digitalWrite(nix_EN, LOW);
canale = 1;
cambia_canale();
/* l'amplificatore e' avviato, inizio un ciclo infinito dove leggo gli input del telecomando */
for(;;)
{
/* Questa e' la funzione non bloccante che attiva ciclicamente la funzione per disossidare la nixia */
unsigned long tempoCorrente = millis();
/* Verifica se è passato un minuto dall'ultimo richiamo */
if (tempoCorrente - ultimoRichiamo >= intervallo)
{
/* Richiama la funzione disossida */
disossida();
/* Aggiorna il tempo dell'ultimo richiamo */
ultimoRichiamo = tempoCorrente;
}
/* Nota sul telecomando Tektron:
* Il telecomando invia ripetutamente comandi zero ("0" ossia ripetizione) dopo l'invio del comando effettivo.
* Questo comportamento, sebbene non ottimale, è una caratteristica intrinseca del telecomando Tektron.
* Il problema principale sorge quando si preme un tasto e, per qualsiasi motivo, il primo impulso
* non viene ricevuto correttamente, ma successivamente vengono ricevuti i comandi di ripetizione.
* Questo può causare un comportamento indesiderato, poiché il programma interpreta erroneamente
* gli zeri come la ripetizione dell'ultimo comando ricevuto correttamente anche se diverso dal tasto effettivamente premuto.
* Per mitigare questo problema, è stato implementato un timeout di 10 secondi nel programma
* che interrompe qualsiasi azione in corso se non viene ricevuto alcun comando durante tale periodo.
* È importante notare che questa mitigazione potrebbe non essere completamente efficace
* se vengono premuti tasti diversi molto velocemente, perche' finche' non scadono i 3 secondi la variabile old_value non viene azzerata
* e quindi il problema potrebbe presentarsi, questa mitigazione interrompe anche la corsa del potenziometro del volume in corsa */
/* codice della mitigazione*/
unsigned long tempoCorrente2 = millis();
/* Verifica se è passato un minuto dall'ultimo richiamo */
if (tempoCorrente2 - ultimoRichiamo2 >= intervallo2)
{
old_value = 0;
/* Aggiorna il tempo dell'ultimo richiamo */
ultimoRichiamo2 = tempoCorrente2;
}
/* fine del codice della mitigazione */
if (IrReceiver.available())
{
if (IrReceiver.decode())
{
/* memorizza il valore RAW nel buffer value */
value = IrReceiver.decodedIRData.decodedRawData;
/* il telecomando manda il codice pulsante solo al primo impulso, dopo continua a inviare degli zero che significano "ripetizione dello stesso
* tasto", questa funzione che ricopia il registro value in old_value serve per richiamare l'ultima routines richiamata in caso il codice ricevuto
* sia zero */
if(value == tasto_meno)
{
/* memorizza value in old_value */
old_value = value;
}
if(value == tasto_piu)
{
/* memorizza value in old_value */
old_value = value;
}
/* utilizzo il tasto on/off del telecomando per cambiare i canali */
if(value == tasto_on_off)
{
/* memorizza value in old_value */
old_value = value;
}
/* richiamo la funzione legata al tasto premuto memorizzato, gestisco i segnali di ripetizione e discrimino eventuali segnali errati */
if ((old_value == tasto_meno && value == ripeti) || (old_value == tasto_meno && value == tasto_meno))
{
motore('-');
}
if((old_value == tasto_piu && value == ripeti) || (old_value == tasto_piu && value == tasto_piu))
{
motore('+');
}
if(old_value == tasto_on_off && value == tasto_on_off)
{
cambia_canale();
}
/* prepara il ricevitore per un nuovo ciclo */
IrReceiver.resume();
}
}
/* funzione per rendere fluido l'andamento del volume con timer non bloccante */
if(vol == 1 && timer < nbdelay)
{
digitalWrite(motor_up, HIGH);
timer ++;
}
if(vol == 2 && timer < nbdelay)
{
digitalWrite(motor_down, HIGH);
timer ++;
}
if(timer >= nbdelay)
{
digitalWrite(motor_up, LOW);
digitalWrite(motor_down, LOW);
vol = 0;
timer = 0;
}
/* Fine funzione timer non bloccante */
}
} You can download the “.ino” file by clicking here: Telecomando.zip
What is Nixie tube poisoning and how to prevent it.
Nixie tubes have more than one cathode, each representing a digit. When a cathode is lit, the material it’s made of can vaporize and deposit on both the glass and the other unlit cathodes. Some of you have probably seen dark Nixie tubes with a gray or silvery coating inside the glass. This is the extreme case of material vapor deposited on the glass, which occurs toward the end of the tube’s life. If a cathode is not used for a long time while the others are on, the deposit on the inactive cathode becomes thicker and thicker. This coating is highly resistive, so if it becomes thick enough, it prevents the inactive cathode from lighting. When it does light, parts of the cathode may be dimmer or completely dark. Different types of tubes have different sensitivities to cathode poisoning, and it usually takes thousands of hours before it becomes a significant problem. You can learn more on this site: https://www.tube-tester.com/sites/nixie/different/cathode%20poisoning/cathode-poisoning.htm
To prevent Nixie cathode poisoning, it is sufficient that all digits are lit at least once in a while. In the case of Anansi, where an Arduino manages the Nixie control, I included a routine in the program that scans through all numbers every 5 minutes; in the 2 videos below we see preliminary tests for software tuning.
In this video I show how the channel-change routine works: it lowers the volume with the Nixie flashing on the current channel, followed by changing to the new channel number and a fade-in as the volume rises again.
Main body assembly
When assembling an object that requires so much effort and time, and therefore involves significant costs, it is crucial to use high-quality materials. In two photos I show the contrast between the plastic switch of the original unit and the stainless-steel anti-vandal switch I chose to install. The latter alone costs almost €60, but it’s a deliberate choice to ensure durability and long-term resistance, and the same applies to everything else.
 |
 |
In the next series of photos, I show the progress of the amplifier assembly. In the first image you can see the bare chassis of the amplifier, without transformers or any other components. This initial stage represents the starting point of the assembly process. In the second photo, the transformers have been mounted on the chassis, highlighting the base on which the rest of the circuit will be built.
As I proceed with the assembly, the following images show the addition of the mounting plate and the start of laying out the electronic components. We see capacitors, resistors and other essential parts being positioned precisely on the chassis. Each step illustrates the meticulous work and attention to detail required to complete the amplifier. The process continues until the amplifier takes shape, with all components soldered and installed correctly.
After the first power-up, the unit was not yet perfect in every detail. It was necessary to carry out a series of measurements and small tweaks, which involved changes to the circuit as I had initially built it. For example, I had to change the type of diodes and revise the very configuration of the high-voltage supply. I also reworked the supply for the current sink under the input phase splitters, making it quieter thanks to the addition of a small inductor I had on hand.
The tuning phase is crucial in a project like this. Even the smallest adjustments can make a big difference in the performance and reliability of the final device. This process requires patience and precision, but it is essential to ensure that the amplifier operates at its full potential. Each modification, however small, helps refine the circuit and optimize its overall operation.
Measurements
Maximum undistorted power: 25 Watts RMS per channel
Damping factor with zero feedback: DF 5
Damping factor with feedback engaged: DF 7
Total power consumption: 440 Watts
Bandwidth, zero feedback: 10 Hz / 20 kHz ?1 dB
THD no NFB @ 1 watt: 0.55%
THD with NFB @ 1 watt: 0.38%
The bandwidth figure of the output transformers is not as extended as that of other transformers I have produced over the years. However, in this project I was constrained by the original chassis and had to use a core identical to the original. This forced me to apply a current density more than double my usual practice and to settle for a smaller core than I would have liked. Despite these limitations, the performance achieved is acceptable, also considering that the amount of feedback applied is really minimal.
The transformers I built are based on the design of the originals, which I unwound and to which I made some small changes. I kept the primary impedance, wire gauge and number of turns, but I changed the arrangement of the windings and the insulating sections. This compromise made it possible to achieve a satisfactory performance, given the conditions and the limitations imposed by the original chassis. Now let’s look at the classic plots…
Frequency response
THD without NFB
THD with NFB
Square waves at 100 Hz – 1 kHz – 10 kHz
Presentation of the completed Anansi
In the following photos, you can admire the Anansi amplifier fully assembled and ready for use. Those who read and see these photos might think that a project like this is quick to build, but in reality the process was extremely complex and detailed. From demolishing the original unit to SPICE design, from transformer design to salvaging all the components, from designing the various internal parts, to writing the software, fabricating the metalwork, assembly, testing and tuning, it took a full 82 hours of labor!
In this video, you can see Anansi in action and observe how it responds to the remote control. We start by showing the volume control: you’ll see how the volume level can be adjusted smoothly and precisely. Next, we demonstrate channel switching, with the Nixie clearly indicating the current channel. Finally, we show the automatic scanning function I implemented to prevent Nixie cathode poisoning.
Conclusion
The Anansi project represents a complex and rewarding journey, from dismantling a malfunctioning amplifier to creating a completely new and optimized device. Every phase of the process, from initial design to final tuning, required attention to detail, precision, and considerable commitment.
Anansi is not just an amplifier; it is a demonstration of craftsmanship, engineering, and dedication. Every component has been carefully chosen to ensure excellent performance and long-term durability. Advanced features, such as volume and channel control via remote and the prevention of Nixie cathode poisoning, add a touch of modernity and practicality to a classic design.
I hope this article and the related images and videos have provided a clear view of the work carried out and the attention to detail that led to the creation of Anansi. This project proves that, with dedication and skill, it is possible to turn an idea into a working, high-quality reality. Thank you for following this journey and for your interest in my work.