TRF IV Stable RF Amplifier
2020/5/08
Mad RF amplifier
What I had did in Start From Crystal (II), was trying every wrong method to make a RF amplifier. That stupid still valuable, let know how pain if do it wrong.
RF amplifier for all frequency, undo tuner's selectivity
stable but un-efficiency
RFC could leading to resonant
Such a RF amplifier is crazy mad, stable is Illusion -- It possibly work standalone though, couple to next stage is almost impossible.
small Gimmick capacitors is tricky
Small gimmick capacitor couple is nothing wrong, apply to top of tuner tank is crazy, though. Most likely stable method is tap or another winding.
Deploy capacitor impendance transformation might be good idea, like double tuned filter. Brutal force couple like this is wired(not gonna too much dig it now).
TRF RF Stage
One obvious difficulty of TRF is achieving enough selectivity. RF amplifier must insert after tuner, classic design is servarl tracking LC tank--- tunable DTC/TTC or even more stages.
After tuner RF amplifier
AM band pass not good enough for TRF
Insert band pass filter, like FM receiver does, works, but LAN design is not for TRF, they works well for receiver with mixer.
Trying cascading un-tuned RF amplifier is tough enough. It gonna to oscillate everywhere at anytime.
Stable Means Everything
Base on TRF III receiver, It's fun to insert a RF amplifier before detector. Which named 1V1, I use LM386 as audio amplifer, so call 1V386.
Origin Lone Wire TRF, LM386 30db Gain
Inserted A RF Amplifier: around 30 dB Gain
A simple RF amplifier which use RF Choke is pretty common choice back to old AM receiver. With today's surplus small transistor, both Hfe and Ft is excessing for AM receiver. The RF choke is designed for power supply, I ever use it in Regen Receiver, and it's works amazingly well.
Amplifier to be evaluated
1mH RF Choke
LM386 input impendance is high enough. There is a input audio RC filter in my LM386 module, which make the input impendance around 6.8k. Such a high impendance is light loaded the RF amplifier, which means the Gain is almost unloaded Gain. Simulation indicate the Gain is 45dB around 1Mhz.
Of Course, It's gonna to be an oscillator, experiment confirm, nothing surprise. Low the bias current could low the gain, make it quite. But I won't do that.
Rules to Prevent Oscillating
There are tons of rule to prevent Amplifier to become an oscillator. Tese precaution had deep reason.
Group I
Cure Oscillating I --- lower the current: lower current might encounter some other limitation.
Cure Oscillating II -- - E degeneration, already deployed.
Cure Oscillating III --- B degeneration, a bit noisy.
Cure Oscillating IV ---- Lower the RFC Q, parallel a resistor to RF Choke.
Cure Oscillating V ----- put ferrite bead on transistor output lead(act as a resistor for RF but not affect DC)
Cure Oscillating VI ---- buffer signal for transmite a longer distance, keep low impendance on receiver end.
Group II
Cure Oscillating VII ---- Use Toroidal RF choke instead cylindery one.
Cure Oscillating VIII ---- Shield the RF choke or whole RF stage (2 stages 2 shields)
Cure Oscillating IX ---- Perfect decouple the power supply. 2 parallel 0.1uF disk, plus a 47uF electrolytic.
Cure Oscillating X ----Active power supply, capacitor amplifier, used to kill audio level boat motor sounds.
Cure Oscillating XI ---- For audio cascading: put a small capacitor to input, form a LPF with input impendance.
Group III
Choose different RF amplifier.
Use wide band RF transformer instead of RF choke, carefully match the input/output impendance.
Use feedback amplifier
use cascode amplifier which is isolated output and input well
Present low impendance to transistor, both input and output
Oscillating Rules Explained
The Group I (number I --- V) is trying to lower the parasitic oscillation frequency gain, so there is no strange high gain to excite oscillation.
The Group II (number VI-X) is trying to isolating inter-stages feeding back, which cause big loop oscillation.
And there is a deep rule: what exactly GAIN do you want? As high as possible is the most worst answer , ever.
Because for a given Amplifier and parts you use to built it, there is a mathematical condition, exceeding that fence, every amplifier became an oscillator. As a amateur maker, I'm not a fan for S parameters oscillating condition. Not because complex, the parts available to amateur always old, cheap, even random, there is no S parameters on parts datasheet.
Refer to <<Principles of T ransistor Circuits>> by S. W. Amos,BSc, CEng, MIEE , M. R. James, BSc, CEng, MIEE
Fig.9
while it's bigger than 2, oscillating started.
gm=Ie/26, Ie unit: mA c: Cob, unit Fala.
Back to write article why tap stabilize the IF Amp , I got some sense why IF amplifier oscillating. But speak of simple RF amplifier, what's wrong with it?
The keypoint, amplifier with RFC is a wide band amplifier, from khz to tens Mhz. We want AM band signal being amplifier, but that amplifier have gain under frequency far beyond this range. At some frequency the gain could be too much, cause oscillation.
So, Fist of all, what gain do I expected? 50 dB? 30 dB or 10 dB? Evaluate the gain for signal we are interested is first thing. The signal gain should be feasible, that's mean, the stable margin given by :
2*pi*f*c*gm*Rb*Rc < 2 better condition is 2*pi*f*c*gm*Rb*Rc < 4
Obviously, the load present to transistor should be known, stale under such a load is the thing we care about. There is possible configuration, under all kind of load the amplifier always stable. That's overkill for me.
Stable the Simle Amplifer
Now It's not that fuss for me to got a RF amplifier stable. First of all, I want >30 dB gain available for AM band. 30 dB is very high gain, already. Practically, it's possible, I had made it. But still need answer many "Why".
Why 30 dB Gain?
Because this gain could be achieved stably, by experiment.
What LM386 high input impendance means to Amplifier?
High impendance, means the RF amplifier is light loaded. And I set it by a RC filter which roughly 6k. Light load RF amplifier means the Gain is not set by load, the LM386 and the gain is set by merely the 1mH RF choke. This gonna mad wild. So the 2.7k* resistor is not only lower the Q of RF choke, the most important function of this resistor is set gain for in-band signal. At same time, This resistor also set the maximum gain for all unwanted signal.
So Then why bother use a RF choke?
Good question. This is always bother me. The RF choke here, had a side effect, the choke provide a method to let bias the transistor more easy and freedom. use merely a resistor is works, but you need much higher power supply to generating same bias current. And the voltage swing is also limited. Overall effect, the RF choke make bias easy, make voltage swing large, make high gain by feasible low voltage supply.
Is it possible cascading another 30 dB gain RF amplifier?
Not likely if not deploy well shielding. Even with 30dB gain, prevent inter-stage feedback is very important. The toroidal ring is much better than cylinder one, both them is 1 mH.
Why 1 mH?
This actually a random value. The inductance should provide 4 times or higher impendance than load. 1mH provide 3.3k reactance at 500 kHz, not high enough, but acceptable. Old MW radio use 2.5mH even higher value RF choke. Those high value RF choke, usually have a very low SRF( self resonant frequency), and low Q, the RF choke works in a frequency region which beyond their SRF frequency.
High beta means what?
High beta most likely cause instability on low frequency, ie, motor boating, audio frequency oscillation. The Re degeneration resistor is prevent audio level instability (also works for RF).
What's Difference Choke
Toroids RF choke ( similar AL as FT50-43), how better it gonna be?
40 turns got 1.4 mH on a 15x13x7 mm Toroidal
Evaluation base on same radio, and 2 steps testing: signal generator gain evaluation and actually hear the stations from the radio.
signal generator testing result
500khz:
input: 5mV output: 140 mV(loaded by LM386), 180mV(unloaded) Gain: 29dB
1 Mhz:
input: 5mV output: 170 mV(loaded by LM386), 210mV(unloaded) Gain: 30.6dB
500khz:
input: 5mV output: 250 mV(loaded by LM386), 310mV(unloaded) Gain: 29dB
1 Mhz:
input: 5mV output: 250mV(loaded by LM386), 300mV(unloaded) Gain: 30.6dB
Outdoor testing
The cylindery RF choke seems works better than Toroids, the gain is high and flat. But obviously on high gain the unshield RF choke is very easy causing oscillating.
Most surprising, while compare the RF choke on the TRF radio, they are so different. The cylinder choke only got one station, the most strongest station, around 600 kHz. But Toroids TRF choke received 3-5 more stations, all on higher band, all above 1 Mhz. Received a station, here means could be heard comfortably.
This strongly suggesting the cylinder choke is lossy on high frequency. Signal source testing use strong signal (-30dBm), overwhelm this effect. I suspect it at first, but confirmed this several time.
This is a lossy RF choke, this not good for RF amplifier, but could use to suppress EMI. And loss choke use in regen receiver, might could bring some positive effect: high frequency, might be smoothed controlled, I don't know.
Why Shiled
For RF gain 30dB, plus 33 dB audio gain (50x), this receiver is pretty fragile for audio oscillation. Overall layout and wiring is pretty bad. Hand touch some wire, joint easily leading to audio oscillation.
Long antenna could magnetical couple to receiver in a very long distance, depending on signal level, 5 cm to 25 cm couple distance is normal, strongest station need 30 cm - 40 cm to avoid overloading the receiver.
weak signal overgo 30 cm couple to TRF
This is a strong proof, why shield for RF amplifier is so important, imagine how strong signal the RF amplifier could deliver. This also the reason the toroid RF choke is far more better than cylinder choke.