RF Amplifier

 The local garage asked me to improve an aerial in an old Bentley many years ago as the receiver was deaf. I made them a very simple amplifier for medium and long waves with a positive earth.

It had a car radio aerial plug and socket so could be fitted in the aerial lead. I think they discovered a wiring problem later so the thing was redundant. I found it the other day and being at a loose end decided to test it. First I rewired it so it had a negative earth. It's a very simple design, just a single NPN transistor with no emitter resistor, an automatic bias circuit with a 1kohm to ground and a 3.3kohm to the collector. There's a collector load of 270 ohms. I'd marked on the box that it had a gain of 85.

The transistor is a 2N5109 and at 12volts it consumes about 26mA, and the collector sits at around 4.5 volts.

See the 2N5109 datasheet I chose this transistor because it had a very high value for fT which is the product of gain and frequency. It's rated for example to provide a power gain of 6 at 200MHz so, all things being equal, should provide a power gain of 40 at 30MHz.

The amplifier is built in a small plastic box using pieces of tinplate cut from a biscuit box. Components are soldered directly to the tinplate, obviously constructed in a hurry.

 There are lots of features on my spectrum analyser about which I've no idea how to use, so I thought I'd experiment with the simple amplifier. It took me ages to work out how to use the more esoteric features of my digital frequency meter and I reckon I've just about worked out how to use the main features. Initially to get the feel of the amplifier I used my oscilloscope.

I fed in a 10mV RMS signal and measured the output on my oscilloscope. This is supposed to be rated up to 100MHz so should be pretty flat up to 10MHz or so, however I'm using a 10:1 probe of uncertain origin. The scope is one of these new fangled types using a flat screen display and is described as "digital". After fiddling around looking at the output of the amplifier for ten minutes wondering how to relate the sinusoidal peak-to-peak display to RMS I happened to push one of the buttons and would you believe it the thing can indicate RMS. I selected this and pressed "Measure". Now I can see the amplitude and frequency of the display in volts RMS and frequency in KHz or MHz written in text at the bottom of the screen. Very different from my old Tektronix scope where you need to squint at the graticule and ponder over the switch for timebase frequency and attempt to figure out exactly what you're seeing.

Here are the results of initial tests.

At 10mV input the output was fairly flat at around 600mV from 1MHz down to 200KHz.

This represents a voltage gain of 60 or about 35dB. Below 200KHz the gain dropped off so I changed the input capacitor from 0.1uF to 0.22uF which helped a bit.

The low end now drops off to 270mV output at 10KHz, a voltage gain of 27, or about 28dB.

Looking at the HF end I was getting about 360mV at 10MHz, 250mV at 30MHz and 260mV at 50MHz. As I don't know how the 10:1 probe affects measurements just yet I'll need to carry out further tests, but as it stands the test rig says a gain at 50MHz of 26, or about 28dB.

I have in mind using the amplifier to beef up a receiver working on long waves as most have a pretty poor performance below 150KHz.

Well, I tried the amplifier using the Rigol DSA815TG. I fiddled with the settings and eventually got a display which sort of made sense, however I turned it off and looked for inspiration on the Internet. A discussion about the tracking generator suddenly triggered something in my mind. The TG output level is adjustable down to -20dBm. I did a quick calculation and the penny dropped. The transistor collector runs at about 4 volts with the auto bias I've set it at. It's what I'd describe as a Class A amplifier. Undistorted output therefore cannot exceed 2 volts peak to peak and its likely to be less than this. This voltage represents 1.4 volts RMS. Converting this figure to dBm gives 16dBm across 50 ohms. If the amplifier has a gain of say 60 or 35dB then you cannot drive it with anything greater than 23.5mV before clipping occurs. This means the TG cannot have an amplitude greater than -20dBm and I wasn't particularly taking notice of this fundamental fact.

The answer was simple. I added a pair of 20dB attenuators in series with the TG output, taking it down to about 220uV. This is much more representative of typical signal strengths met on broadcast bands.

Continuing with tests. I found, with the amplifier disconnected, the SA noise level measured -60dBm and with the amplifier in circuit it gave me a level of about -21dBm from about 500KHz to about 1.5MHz dropping to -27dBm at 15KHz in the LF direction and -30dBm at 30MHz in the HF direction. At 69KHz the level was -23dBm so in its design purpose as a long and medium wave amplifier it's not bad, giving a voltage gain of 39dB.

The next step was to see what the amplifier did to the latest rig in for repair, as this Trio TS-940S has a general coverage receiver tuning down to 30KHz. Connecting it up and plugging in my long wire produced a huge noise background, lots of second channel or image reception and really strong long wave stations. Radio 4 read off scale or a lot greater than S9 plus 60dB. The next step is to decide what I'd like to do with the amplifier. A first step is to add an input filter to limit reception to say 15KHz to 300KHz, thus eliminating break-through of higher frequency stations and the very high noise levels which appear to be at the low frequency end of the medium waveband. This exercise will give me a little more experience with the DSA815.

The first thing was to fit BNC connectors in place of the car radio aerial connectors making it easier to test the amplifier.

I had in mind building a simple low pass filter and connect this between the amplifier and its output socket. It might be better to fit this at its input though in order to reduce cross modulation effects, but I chose the former. Looking in my box marked "coils" I found a chunky little ferrite cored coil and measured its inductance with my Peak tester. It read 997.6uH. A quick calculation showed a pi filter would need around 1000pF to make it work in the long wave band. I connected the coil between the 0.22uF output capacitor and the output connector and wired a pair of 2.2nF capacitors from the ends of the coil to ground. At this point let me outline a potential problem. Most test equipment uses 50 ohm terminations and this is fine for a transmitter but not ideal when it comes to receiver aerial inputs which are often high impedance. This being so I used a probe to measure the output of the amplifier. The 10:1 probe for the scope and a home made probe for the spectrum analyser.

The results on the scope were as follows:-

Using 10mV input I measured the outputs as shown below in the column marked 1000uH + 2.2nF. Being voltage, the 6dB points are roughly 170KHz and something like 15KHz (the latter is a guess as I didn't check that low).

In summary then. The amplifier and filter would be good for the low frequency end of the long wave band.

I tried it on the Trio TS-940S and the results were very good. The huge number of stations from cross modulation and images were gone and the normal long wave broadcasts were amplified by a decent amount. For example Radio 4 had moved from around S9+18dB to S9+30dB on the S-meter.

Tuning down to 60KHz the timing signals from the UK transmitter were clearer and with less background noise.

I think a coil of perhaps 500uH would be better than the 1mH coil I used, significantly shifting the gain towards the Irish station at 252KHz.

I removed about 40 turns from the coil and measured its value at 492uH, then refitted it.

Below is a table showing results:-
10 1000uH + 2.2nF Input 10mV 492uH + 2.2nF Input 10mV 492uH + 1nF Input 10mV
Frequency KHz Output mV RMS gain dBmV Output mV RMS gain dBmV Output mV RMS gain dBmV
300 34 11 228 27 756 38
290 41 12 252 28 898 39
270 92 19 310 30 1290 42
250 142 23 385 32 1900 46
230 198 26 503 34 2320 47
210 255 28 699 37 1920 46
190 310 30 1030 40 1440 43
170 443 33 1620 44 1130 41
150 703 37 2000 46 944 39
130 1200 42 1540 44 816 38
110 2260 47 1120 41 733 37
90 1560 44 886 39 670 37
70 992 40 746 37 620 36
50 742 37 648 36 576 35
30 597 36 561 35 510 34

 The general response has a lowish attenuation to the LF of a peak then initially a slow fall to the HF followed by a more rapid attenuation.

As you may be able to interpret, the third result will be fine for listening to the long wave band. This set of readings were made after changing the filter output capacitor from 2.2nF to 1nF.

I tried an input filter also but the results were very disappointing so abandoned the idea. I puzzled over the effect of increasing the input voltage from 10mV. What happened was a gradual increase in output up to over 12 volts RMS whilst maintaining what looks like a perfect sinewave on the scope. I checked the transistor collector DC voltage. It stood at 4 volts at all inputs, very odd. I moved the scope probe to the collector and found there was a sinewave at low inputs but this doubled in frequency and stayed very low in amplitude whilst the output from the amplifier via the filter rose to 12 volts RMS. In fact this voltage was established quite quickly as the input was increased and didn't change at all beyond 70mV input.

To read more about this see below...

I might consider adding a small multi-way switch and use this to change the filter output capacitance so that reception can be peaked on specific stations. Probably a range of capacitors from 470pF to 3.9nF judging from the above table.

The 2.2nF filter input capacitance has very little effect.

 Above is the amplifier after modifications. The diode is fitted to protect against reverse connection of the supply. The coil at the bottom right is from an old TV circuit board, initially 1mH but 40 turns were unwound to give about 500uH. The orange capacitors are 0.22uF and were just handy. Physically much smaller types can be used. The input resistor is not essential. The 1kohm base resistor is underneath the capacitor. The auto-bias resistor is 3.3Kohm and the collector load 270 ohms and there's a ceramic and a 10uF decoupler...

I tried the final configuration and it improved long wave broadcast reception by around 20dB on the receiver S-meter and lifted the VLF transmissions from zero to good volume.

In view of several new LF/MF amateur bands now being available, I need to peak the amplifier on some spot frequencies, viz 136KHz, 472-479KHz and 500-515KHz.

 

Overdriving the amplifier

 A few lines ago I mentioned increasing the input voltage above 10mVolts.

Well, I investigated what happened using a spectrum analyser and the results shouldn't have been surprising. Initially I connected Channel 1 of my scope to the output of the pi-filter and Channel 2 to its input (the collector of the transistor). As my standard test input had been 10mV I started with this and surprisingly I had to reduce this down to 1mV before the second and third harmonics of the input disappeared into the noise. At 10mV input, harmonics were very strong and as the input was raised the level and number of harmonics increased, even though the signal at the transistor collector looked like a sinewave. Once the output from the pi-network had reached 12 volts, and the signal at the collector looked very distorted, the harmonics at the transistor collector were very large in amplitude and stretched up to VHF. Interestingly the supply voltage to the amplifier was 12 Volts and this reflected in the maximum RMS voltage of 12 observed and, as the supply voltage was increased the pi-network output voltage increased in step.

Here are four pictures representing:-

(a) Amplifier with 100mV input, Signal Input; scope traces; spectrum at filter input

(b) Amplifier with 20mV input, Signal Input; scope traces; spectrum at filter input

(c) Amplifier with 3mV input, spectrum at filter input, reduced bandwidth to enhance signals

(d) Amplifier with 1mV input at 100KHz, spectrum at filter output, bandwidth reduced further

 Fortunately the problems re harmonics are due to the test parameters (much higher voltages than met in practice).

The amplifier is OK if used with a long wire where typical voltages are not very high.

I'm not the first to design such an amplifier.. click the valve to see Mr G.L.Morrow's amplifier which built in 1923!

Wireless Telephony Explained was co-written by G.L.Morrow and published in 1924, other than that who was he?

I discovered a George L Clare-Morrow 1895-1965 was it he?

Return to Reception.