HIGH FIDELITY AM TRANSMITTER



TRANSMITTER::DETAILED DESIGN

The detailed design of a High Fidelity Low Power AM Transmitter is given in the following sections:-



top

  [1] OVERVIEW

  The low power transmitter consists of two sections:-
[A] An exciter which contains:-

[B] A power amplifier which contains:-



top

  [2] MODULATOR


  [1] Basic Design


  Modulation takes place at a low level in a Beam Deflection Tube.
The construction of the tube and its characteristics are shown on the right.
A rectangular electron beam is deflected by two deflection plates and intercepted by a split anode. The strength of the beam can be modulated by a grid near the cathode.
If the beam is modulated at RF and push-pull audio applied to the plates, an amplitude modulated signal will appear in both plates.
If only the middle portion of the deflection characteristic is used, the modulation distortion is very low.
As the beam current increases, mutual repulsion between the electrons causes the beam to spread, so a greater deflection voltage is required to totally deflect the beam at high beam currents.
If a small radio frequency sinusoidal signal modulates the beam current, it is assumed that it splits in the same ratio as the total beam current. Beam spreading can introduce a small second order error. This can be made insignificant by reducing the fractional beam modulation by the RF carrier applied to the control grid.
In all further work it will be assumed that the fractional split of the total beams is equal to fractional split of the small signal RF component.


  [2]Simplified Model

   A simplified model of the beam deflection tube is showm on the right.
I is the total beam current.
A fraction K is directed to the RH side giving a current = KI.
This leaves ( 1 - K )I on the LH side.
The transformer primary windings have a ratio of 1:n
The total magnetising current Im is then given by:-
  Im = KnI - ( 1 - K )I
∴ Im = [ K(n + 1) - 1 ]I
When used as a modulator, zero carrier condition occurs when Im = 0 at Ko
∴ 0 = [ Ko(n+ 1) - 1 ]I
∴ Ko = 1/( 1 + n )
So that :-   Im = [ K/Ko - 1 ]I
Unmodulated carrier occurs when K = 1/2
So that :   Imc = [ 1/2Ko - 1 ]I
100% modulation gives a peak carrier of 2Imc = Imp
So:-   Imp = 2[ 1/2Ko - 1 ]I
The ratio Kp for peak carrier is then given by:-
  [ Kp/Ko - 1 ]I = 2[ 1/2Ko - 1 ]I
∴ Kp = 1 - Ko
In the transmitter n = 2 so that:-
  Ko= 1/3
  Kc = 1/2
  Kp = 2/3
For full modulation the current division swings between 1/3 and 2/3 : that is only over the linear portion of the deflection characteristic.


  This modulator is capable of producing greater than 100% modulation in the negative direction.
An envelope detector produces horrendous distortion under these conditions, but the product demodulators in the Tuner demodulate the signal correctly.


  [3]Modulator Feedback

  A simplified circuit diagram of the modulator feedback is shown above.
The modulator distortion is inherently low since the deflection curves for the beam deflection tube are linear and only the central portion is used for full modulation.
As a further guard against distortion, 26db ( x20) overall feedback is applied to improve the deflection linearity of the beam tube.
The RF modulation is a linear function of the difference between the left hand plate current, Ilhp, and right hand plate current, Irhp, of the beam tube V3.
The variable ( Ilhp - Irhp ) must be used for feedback.
V4 and V5 form an operational amplifier with the gain increased by boot strapping through C10. R15 is the feed back resistor, so the voltage on the cathode of V4 is given by:-
V4c = -IrhpR15
The voltage on the left hand plate of V3 is then:-
Vlhp = IlhpR14 - IrhpR15
But R14 = R15
So: Vlhp = ( Ilhp - Irhp )R14    Here R14 = R15 = 1.8K
The currents Ilhp and Irhp contain an RF component at carrier frequency and this must be kept out of the modulator feedback loop.
R14 and R15 must therefore be bypassed.
This produces a pole in the feedback loop which can be used as the main stabilising pole.
Unforunately, the pole is in the feedback path not the forward path.
For a simple feedback system:-
Y(p) = A(p)/( 1 + A(p)β(p) )
where Y(p) is the closed loop response:: A(p) is the forward open loop response:: β(p) is the feedback transfer function.
Here A(p) = A the forward gain.
β(p) = 1/( 1 + RCp )
where R = R14 = R15 = 2.7k
and C = C8 = C9 = 1.8nF
The turnover frequency fp is given by:- fp = 1/( 2ΠRC ) = 32.75KHz.
This is chosen as low as possible while still keeping it above the audio band.
Subsituting β(p) in the above equation we get:-
Y(p) = [ A/( 1 + A ) ][ (1 + τp )/( 1 + τp/( 1 + A ) ) ]   where τ = RC
A zero has been introduced into the closed loop response.
This must be cancelled by a corresponding pole introduced into the signal path before the feedback modulator.
If we introduce the correction transfer function Yc(p) we get :
Yc(p) = 1/( 1 + τcp )
The total transfer function Ytot(p) is given by:-
Ytot(p) = [ A/( 1 + A ) ][ 1/( 1 + τcp ) ][ (1 + τp )/( 1 + τp/( 1 + A ) ) ]
For compensation: τ = τc so that :-
Ytot(p) = [ A/( 1 + A ) ][ 1/( 1 + τp/( 1 + A ) ) ] :: a single pole shifted out by ( 1 + A ).
Here A = 20, so the compensated modulator has a turnover frequency of 21x32.75KHz. = 687.75KHz.


Compensation Network
The compensation pole at -1/τc is produced by the input network R1,R2,C1.
The input terminal of A1 is a virtual earth, so the driving point impedance to C1 is R1 and R2 in parallel to give:-
τc = C1R1R2/( R1 + R2)
The change in steady state and step function response is shown as R1 changes to change τc.
The output from the modulator is shown in blue and the output from the RF tuned circuit in the plate is shown in red.
The transfer function for the modulation of the tuned circuit is given by:-
Ytc(p) = 1/( 1 + τtcp ) where:-
τtc = 1/2B where the total bandwidth, B, of the tuned circuit is given by:-
B = fc/Q
where fc is the carrier frequency and Q is the Q of the tuned circuit.
Here B ≈ 40KHz.

Carrier Trap
To reduce the level of carrier in the feedback loop a trap is placed between V1 and V2. This tends to decrease the stability margin in the loop. To increase the gain margin a shelf is introduced by the secondary fedback loop R5, R6, R7, C5. This shelf starts at 50KHz. and ends at 150KHz., thus improving the gain margin by 3.
The high and low frequency open loop transfer functions are shown.
The increase in slope between 50KHz. and 150KHz. can be seen.


    The closed loop steady state, step function and 1KHz. square wave responses are shown below.



top

  [3]AUDIO PHASE SPLITTER


  [1] Detailed Circuit


  The detailed circuit of the modulator is shown above.
V2,V3,V4 and V5,V6,V7 form two operational amplifiers.
Gain is increased by boot-strapping through V3 and V6. This increases the effective dynamic load on the EF184 pentodes and reduces the HF turnover frequency to about 8KHz. The low frequency turnover determined by the time constant (C6 R8) occurs at about 300Hz. The final low frequency gain is determined by screen and cathode degeneration.
The open loop gain of the amplifier is shown on the right.
Phase reversal occurs in the RHS operational amplifier.
R10-C9 is the front impedance and R11-C10 is the feedback network.


  [2] Closed Loop Response


High Frequency Closed Loop Response


The high frequency closed loop response ( both steady state and transient ) of the phase splitter with and without the shelf network is shown on the right.
The dominant and stabilising pole in the high frequency phase splitter loop is generated by the load on the EF184 pentodes and capacity to earth.
The 50KHz to 150KHz shelf to give extra gain margin in the overall loop is generated by R5-C5.

Analysis of Shelf Generation


Referring to the shelf generation in the phase splitter circuit:-
Let Ra = R6 :: Rb = R7 and R4 in parallel :: Rf = R5 :: Cf = C5 :: fb = bottom shelf frequency and ft = top shelf frequency. We get:-
Rf = Ra( 1 - (ft/fb)( Rb/( Ra + Rb ) ) )/( ft/fb - 1 )
Cf = 1/( 2Πfb( Ra + Rb ))
Here ft = 150KHz. :: fb = 50KHz. :: Ra = 47k :: Rb = 2.5533K so:-
Rf = 19.867k :: Cf = 64.23pF.

Low Frequency Closed Loop Response


The closed loop low frequency response of both sides of the phase splitter is shown opposite. They track down to about 0.1Hz where the RHS has more attenuation and phase shift because it has one more low frequency coupling network ( C8) than the LHS.
The open loop low frequency transfer function is fourth order:-
Y(p) = [(1 + Tkz p)/(1 + Tkp p)]
X[(1 + Tsz p)/(1 + Tsp p)]
X[(1 + Tbz p)/(1 + Tbp p)]
X[ Tcz p/(1 + Tcp p)]
Where Tkz, Tkp ;Tsz, Tsp ;Tbz, Tbp are zeros and poles on the real axis due to the cathode decoupling, the screen decoupling, and the anode bootstrapping.
There is a pole at the orogin, Tcz, due to the AC coupling : C7 and ( R6 + R7)
The low frequency open loop gain due to boot strapping turns over at about 300Hz. to ensure low frequency closed loop stability. The low frequency response has a 4.2 db peak at 18.5 Hz.. This is of no consequence, since it is covered by the modulator feedback, giving a peak of about 1.3db between 1 and 2 Hz.


top

  [4]AUDIO INPUT AMPLIFIER


    The circuit of the audio preamplifier is shown on the right.
The design uses the same bootstrapping technique used in the phasesplitter.
The input pentode is an EF86 - a tube designed for low noise and hum.


  The open loop and closed loop response of the preamplifier is shown on the right.
A single pole RC filter on the input attenuates any RF pickup by the audio chain.
This has a 3db frequeny of about 150KHz.



top

  [5]RF OUTPUT AMPLIFIER


  The buffer RF output amplifier drives about 1 M of unterminated cable connecting the exciter with the power output amplifier. The requirements are:-
(1)Very low distortion
(2)Flat amplitude response across any AM channel (here 80KHz. wide).
(3)Phase linearity across the channel.
V2 acts as a Miller integrator with Cm the Miller feedback capacitor.
V1 provides constant current drive for the integrator.
The steady state response of the amplifier across the broadcast band is shown on the left.



top

  [6]L-C OSCILLATOR


  The full circuit of the L-C oscillator is shown above.
A beam deflection tube, V1, is used as the L-C oscillator. Isolation is provided by the cathode follower V2a. The phase splitter, V2b, drives the full wave rectifier V3a, V3b. The output of this is used as feedback to control the beam current, and hence the oscillation level, in V1.
A beam deflection tube has the following attractive properties as an oscillator:-
[1] Positive feedback around the L-C tuned circuit is easily achieved because both positive and negative Gms ( mutual conductances ) are available.
[2] When the deflection plate drive is sufficient to fully deflect the beam, the plate current is a square wave. This greatly simplifies oscillator level control.

Level Control

The transfer function, Ya(p), between the grid of the V1 and the RF drive to the full wave rectifier is given by:-
Ya(p) = 1/( 1 + τap )
where τ = Q/( πfosc )
Here Q ≈ 80 : fosc = 739KHz. ; τa = 3.49x10-5
The response time of the full wave rectifier is smaller for increasing drive than decreasing drive.The time constant, τb, for decreasing drive is given approximately by:-
τb = (220pF)x(590k) = 1.3x10-4
Loop stabilisation can be obtained by making the beam tube grid time constant, τc much greater than either of these.
τc ≈ (0.1uF)x(100k) = 10-2 Secs.


top

  [7]RF POWER AMPLIFIER


  [1] Detailed Circuit


   The Circuit of the RF Power Amplifier is shown above.

  

Design Philosophy


The RF amplifier is broadband, so that the there is only one tuning control in the output matching network.
The plate current in the output tubes V4, V5, is sensed in the 18 ohm common cathode resistor and fed back into the cathode of V2 to give about 20db of current feedback. This ensures that V2, V3, V4, V5, act as constant current generator which is flat across the broadcast band with minimal distortion.
The internal and suppressor grid shielding of the EF80 removes any interaction between the output network and the current generator.
The output voltage on point L is divided down in a capacitive divider and fed back to the grid of V2 to give heavy voltage feedback. This reduces the overall distortion still further and ensures a wide bandwidth regardless of the loading of the matching network.

  

Loop Stability : High Frequency


The R-C network in the plate of V2 produces a pole on the real axis at 10.6MHz. and the R-C network in the plate of V3 produces a pole on the real axis at 2.26MHz. The network in the feedback path produces a zero at about 10.6MHz. and a pole at about 30MHz. This zero cancels the 10.6MHz. pole in the forward path to give two poles around the current loop at 2.26MHz. and 30MHz..
With 20db feedback this provides good current feedback loop stability.
The correction network is placed in the feedback path rather than the forward path to give a shelf which improves the voltage feedback loop stability.

  

Loop Stability : Low Frequency


The coupling network between V2 abd V3 ( 470pF/47K) has a 3db frequency of about 7.05KHz. This is more than 10 times higher than any other low frequency cutoff, so the forward path low frequency response falls at 6db per octave to well below the unity gain frequency giving good low frequency loop stability.


  

Overall Response


The overall step function response is shown on the right. Regardless of loading, the transmitter is flat to beyond 35 KHz.. At the time of the above tests the low frequency response had a 1.4db peak at about 10Hz. due to the audio preamp in the modulator.
This has been modified to give an overall response flat betwwen 10Hz. and 35KHz..


  [2 Matching / Output Network]


The RF matching network is essentially a parallel tuned circuit with taps.
Inductive and capacitive loads can be accommodated.


  [3]Tuning Indicator


The circuit of the RF level meter used as a tuning indicator is shown above.
V6 is an RF amplifier. The DC on the plate is applied to the grid of V7a via the 470k/0.01uF low pass filter.
DC plus the RF is applied to the grid of V7b which acts as a peak rectifier because of the capacitor Cf in its cathode. The cathode current is made large enough for the rectifier to follow the modulation envelope, so linear detection occurs. The potential difference between the the cathodes of V7b and V7a is then proportional to the level of the carrier.

TUNE

In the "tune" position, the level meter measures the RF voltage across the 18 ohm common cathode resistor in the output stage. This gives an indication of the RF current drive to the matching network. This is a parallel tuned circuit and presents the maximum impedance to the output stage at resonance. The heavy voltage feedback ensures that the output voltage remains constant as the circuit is tuned through resonance, and so the drive current has a minimum at resonance.
THE MATCHING NETWORK IS TUNED FOR A MINIMUM

RUN

In the "run" position the output voltage across the tuned circuit is monitored through a capacitive divider.
The output level can be set by the 1K helipot on the input to the RF anplifier.


top
back