Contacter l'auteur / Contact the author

Recherche dans ce site / Search in this site







Technical review

How to select a tube HF amplifier ? (II)


Knowing that the voltage can exceed four times the DC plate voltage, manufacturers use large "door-knobs" capacitors suited to handle high RF currents and should no more use ceramic disk capacitors. Usually we find 2 or 4 door-knobs capacitors working in parallel. When using several capacitors, each of them yield a low capacitance in order to carry higher current that a single capacitor of higher value. These capacitors are then connected in series with the plate tank capacitors to slow down the tuning rate of the plate tuning capacitor. 

Closeup inside the Wingfoot 813 amplifier. At left in dark red is the parasitic surpressor that prevents undesired oscillations in the amplifier. Just to its right is the yellow plate coupling capacitor, which keeps the 2 kV plate potential from reaching the plate tank circuit located just in front of it. Above the plate tank coil are two 40 mf "door knob" capacitors. Their function is to control the tuning rate of the plate tuning capacitor. In the image at right the antenna RF choke is the small device standing in the front and at right.

Antenna RF choke

Located between the output antenna connector and the ground it serves two purposes. First it should ground the plate loading capacitor for DC so that the loading capacitor handles only the output RF. Then it should short circuits the plate supply in the event of a short in the plate coupling capacitor. This will blow the security fuse and prevent the 2 kV plate potential from reaching the antenna system. How ? A leaky coupling capacitor lets current through from the preceeding stage and upsets the DC bias, usually turning the tube in such an hard way that no signal can pass through it. A power tube with a leaky or shorted coupling capacitor may blow fuses and cause low power or even damage transformers.

Pi-L network

The Pi-network is a symmetrical circuit that tunes the output tank circuit and transforms the load resistance in the nominal 50-ohm load. This is an excellent harmonic attenuator. To avoid harmonic distortions, the old Pi-network is usually replaced or complete with a Pi-L output network. Knowing that the quality factor Q measures losses (from conductor and dielectric) in a resonance circuit - the sharper the tuning curve, the higher the Q -, this factor defines its selectivity and sensitivity (Q = frequency / bandwidth). Applied to a Pi-L network a Q-factor 12 is enough to avoid any distortion. 

A Pi-L network.

At left, a Pi-L network schematic. At right, an input circuit made by SM2CEW constitued of one tuned circuit per band.

T-network attenuator

Also named PAD attenuator, this is a tuned network made of variable capacitors forming the arms of a " P" and two parallel inductors forming the top (also designed as a "T" network, hence its name). Its role is to attenuate or suppress harmonic distortions to present a 50-ohm load to the exciter and then to the tube, the tuning offering the advantage to get the lowest SWR and loss as possible. 

To work on all bands, including WARC, it is recommended of using individual tuned input T- networks for each band. This design will increase the Q factor and thus is far best than trying to use a single tuned network on several bands.

Direct feed of the tube is not recommended even if at first sight a tube supporting a 50-ohm input impedance could be directly fed by the exciter. In fact this nominal load is well present but only during half a wave of the input cycle, whilst during the other half of the cycle, the impedance load can yield any value. This effect distorts the input signal, increases losses and SWR during that time.

Instead of using a T-network we can also take advantage of a passive resistive input network. It can be found in many high gain triodes and larger multi-grids tubes, including pentodes.

Variable capacitors

To adjust a radio on a frequency we use variable capacitors which role is to equalize the inductive (from the inductor coil of the transformer) and capacitive (induced by the counter electromotive forces) reactance; this condition is called resonance. The particular frequency isolated by the equalized reactance is called the resonant frequency. 

At left, an Emtron DX-3 amplifier showing in the lower right compartment the tank oil in white and variables capacitors below it. At center, the variable (yellow) and loading capacitors (gray) of an Ameritron AL-572 with the four 3-500Z triodes to their right. At right, inside a Yaesu VL-1000 Quadra amplifier.

To tune an amplifier on the same resonant frequency as the radio we can use variable capacitors or switched fixed capacitors also called loading capacitors. The first are the most commonly used. They come in butterfly or differential model. Among usual problems, some unstable amplifiers or if you work on unappropriated high bands (i.e. on 10m where the capacitance can be too high) can lead to arcing the loading capacitors due to an extensive use. This problem will create paths of low resistance and in this case it is recommended to replace the plate. Therefore before buying an amplifier verify well its working frequencies to avoid such mistakes.

Butterfly capacitors support high voltage up to 5 kV, with plate spacing as small as .030" (0.8 mm). Quality floating stators and rotors should use silver plates and 1/4" shaft. Do not use a split stator as it yield a higher loss.

Differential capacitors plates should be tapered for uniform capacity distribution during rotation. Complementary tapers increase capacity in one section while decreasing in other. Rotor is usually grounded at shaft end and has a brush at rear.

At left, an air variable capacitor, also called butterfly capacitors due to the shape of its wings. At center, the author's Kenwood TL-922 amplifier showing the loading and plate tuning capacitors, the tank coil and the two new 3-500Z Svetlana tubes attached to their parasitic suppressor. At right, a model of differential loading capacitor. These accessories are sold by Surplus Sales.


SWR-meters display various accuracies, some are digitals others analogs, and even using the same circuit different models of SWR-meters will give different readings. Changing the length of the coaxial can also change the SWR reading.

Modern transistor-output transceivers always use a set of switched broadband output-filters, offering about 1.5-octave per filter, in respect to the national regulation requirements on spectral purity.

When using such a filter at the limit of its bandpass, 29.500 MHz, the filter can introduce a reactance into the transmission line. This reactance might avoid any optimization of the SWR in the amplifier tuned input circuits.

The best way to avoid this problem is still using a tube-radio (e.g. Kenwood TS-830S) when optimizing the tuned-input circuits. The radio must be first tuned for maximum power into a 50-ohm termination, and then it must not be corrected during the adjustment of the tuned-input circuits. Indeed, if the transmitter is tuned again, it may introduce a reactance that will affect the SWR.


Of course in purchasing or building a multi-band amplifier, a quality band-switcher is a must. Here also due to the high current, avoid the low cost phenolic/bakelite switch and select a switch made of ceramic. Similar to the electro switch, this last is up to 20 times more expensive than phenolic, but it is much more resistant and sometimes explosion proof, confirming its high efficiency.

Knowing that a kW amplifier displays a peak voltage of about 10 kV DC, the band-switch should be able to carry at least 20 amperes DC, the contineous component rating 60% of the DC level. If you select a smaller switch the lesser failure of this device will switch-off your amp immediately and in all cases you will have to replace it. So the biggest the best and never skimp on its quality.

At last each year contacts of the band-switcher should be clean using a dedicated cleaner to ensure it a top condition.

At left, inside this QRO HF-2500DX amplifier we recognize in the middle near the Svetlana 4CX800A/GU74B metal/ceramic tubes, the orange plate choke on which are grounded blue parasitic suppressors. To its right is the white tank coil with the gray variable capacitors in the bakground. Far right is the band-switcher fixed to the chassis. At right, a closeup of the QRO HF-2500DX Mark III RF tank showing the band-switcher at foreground. An excellent quality !

Plate choke

After tubes and the tank coil, this is the largest component. This is a cylinder displaying a diameter over 25 mm, 150 mm high and that yield no less than 200 ÁH. The wire forming this choke is between 18-26 ga (0.4-1 mm) in size and insulated.

It must be placed away from the heat radiated by the tubes to avoid that it burns up. Therefore used with glowing glass tubes (vaccum tubes containing gaz) chokes of small diameter request to be placed in contact with the direct air stream coming out of the blower. This problem does not occurs with metal/ceramic tubes. The plate choke is usually not placed in a direct air stream but to avoid overheat the plate choke displays a larger diameter.

Most amplifiers working on all bands including WARC. To prevent parasitic resonances on WARC bands a capacitors set named parasitic suppressors is grounded in the plate choke. However, before using it on WARC bands verify with the manufacturers that the plate choke is modified consequently, otherwise you risk to blow up the plate choke at the first emission.

T/R switching

PIN diodes.

A PIN diode.

In many amplifiers, the T/R (Transmit/Receive) switch is an open frame mechanical relay (20 A). Although reliable, even after 20 years or more of use, this kind of switch is loud by design and can not switch fast enough to allow amateurs to operate fast break-in (QSK) in modes like CW, SSB-VOX or AMTOR. Indeed in these applications is it requested that the switching reaches or is below 3 ms. To get this speed some modern amplifiers can be equipped at extra cost with vacuum relays or use from factory high power PIN diodes

A vacuum relay is sensible to failure when used in high RF applications and is a bit louder than a PIN diode. This latter on the contrary is an active semiconductor that operates as a variable resistance at RF and microwave frequences. 

The PIN diode resistance is determined by the forward biased DC current. Better, in switch applications the PIN diode is able to control the RF signal without introducing distortion.

PIN diodes should be your best T/R switches because they are much quieter and faster than vacuum relays.Drawbacks,  PIN diodes must be associated to a complex circuit constitued of dozen components; they are subject to damage by electrostatic discharges (i.e. lightning), and when SWR is high,they don't have a great reliability record and tend to fail due to high RF voltages and currents. Of course they are not the sole components to fail in such conditions.

Two worlds dealers are recommended, Jennings and Kilovac. Both manufacturers provide high-speed relays (3 ms) rated up to 30 kV peak and that handle over 30 A if necessary.

At left two Kilovac vacuum relays installed in an amateur gear. At right a closeup on the open frame vacuum relay used in a QRO HF-2500DX amplifier that can be equipped with a vacuum relay QSK at extra cost.

Note that due to the slowness of  some T/R switches many amplifiers are often unable to operate fast QSK. For these gears some manufactuers like Ameritron provide an external unit called a "PIN diode QSK switch". This small backbox is inserted in your RTX chain in front of your amplifier and handles 2.5 kW PEP.

Next chapter

Power supply

Page 1 - 2 - 3 - 4 -

Back to:


Copyright & FAQ