Bookmark and Share

Friday, 3 April 2009

Radio Telescope

Antenna for a Radio Telescope

For starters if you wanted something infinitely simple then I would erect, as the article suggests, a half wave dipole antenna for your radio telescope. At 30 Mhz (10 metres) the theoretical length of a wavelength is 285 / 30 = 9.5 metres. The derivation of the 285 is the speed of light (300,000,000 metres per second) X 95% (velocity factor) divided by 1,000,000 (because we are using Mhz or millions of cycles per second).

It follows of course a half wave dipole antenna would be 9.5 / 2 = 4.75 metres or as I prefer 4,750 mm. See figure 1 below

half wave dipole antenna for a radio telescope

Fig 1. Half wave dipole antenna for a radio telescope

Ideally your radio telescope's half wave antenna, which consists of two quarter wave sections of 2375 mm each, should be installed as high as possible above ground. Much will depend upon the available real estate, your locality and 101 other factors beyond your control. The width between the posts or trees or whatever supporting your antenna is largely immaterial as long as you can accommodate the full 4,750 mm length plus insulators. Try and buy egg shell type insulators but failing that use thick plastic, Plexiglas or whatever is non metallic. Wood would be fine excepting it will absorb water which will destroy its insulating properties. The end pieces can be nylon rope passing through pulleys so the antenna can be lowered for maintenance.

For the antenna wire itself I like to use what we call in Australia 2.5 mm2 earth wire. Originally it had green insulation (now green/yellow) which was less obtrusive visually than bare copper wire, resists corrosion and its diameter usually has sufficient mechanical strength for spans of this size.

Don't get too paranoid about precise millimetre dimensions because this is for a radio telescope receiving application NOT transmitting where ideal 100% efficiency counts. Similarly don't get paranoid about impedance. Listen to me - please read that again.

Unfortunately, at 30 Mhz the physical size of your antenna places some considerable constraints upon you.

Personally I'd like to construct a yagi antenna, complete with corner reflector, pointing directly at the Milky Way. Better yet would be two or more such antennas stacked horizontally and vertically (impressive). At 30 Mhz the physical dimensions definitely wouldn't suit your average domestic back yard. You would have an antenna occupying something like 14.25 metres ( 47' ) in height, width and length!

Ideally I'd look at frequencies around 300 - 400 Mhz or more so we can have practical dimensions but at the expense of more complicated electronics. That is beyond the immediate scope of this article. If sufficient interest is generated (read considerable) I'll investigate it further, but read my addendum of 22nd April, 2000 at the bottom of the topic- heh! heh!.

Radio Telescope Receiver

In the original article the radio telescope coaxial cable fed a typical short wave receiver of apparent reasonable quality. I have several problems with that concept.

Firstly our one goal is to detect and record noise emitted by the Milky Way. At 30 Mhz we are on the threshold of the VHF spectrum where the inherent noise of the receiver itself becomes a dominant and limiting factor.

Secondly our coaxial cable introduces some losses into the system. Thirdly our receiver, if sufficiently sensitive and stable, does not need to be particularly flash. In short I wouldn't part with large sums of money on a suitable receiver when items 1 and 2 are limiting factors.

Remember these statements are applicable at VHF and above, possibly even down to 15 Mhz on occasion. Below 15 Mhz it's totally irrelevant.

Anticipating an inevitable question which will arise - "can I use an FM receiver". Answer: NO! because an FM receiver's selling point is the exact opposite to our goal, it kills noise and I won't involve myself in a lengthy discussion here as to why or how.

Radio Telescope Antenna Preamplifier / Mixer

The way I would consider going and indeed the substance of this radio telescope project is the antenna preamplifier / mixer approach. Never heard of it? Consider this.

In the receiver mentioned in the original article (typical sw receiver for its time) if it was tuned to 30 Mhz (actually 29.98 Mhz but that's insignificant here) the local oscillator would perhaps be at 30.455 Mhz and when mixed with a 30 Mhz received signal a resulting standard IF signal of 455 Khz is amplified many, many times and then detected for AM. If this is all double dutch to you then later on go to my main page and a whole new education awaits you.

The same principle applies to a cheap AM transistor radio operating in the AM radio band 531 Khz to 1611 Khz or whatever.

My theory is this:

At the threshold of VHF a low noise preamplifier is obligatory to establish as low a noise figure as is possible for a radio telescope to overcome the limitations of inherent receiver noise. Ideally this preamplifier is situated AT the antenna, this philosophy happens to be a little quirk of mine.

To overcome the losses of the coaxial cable etc. why not put the first mixer and other bits and pieces directly at the antenna also? Another little quirk!

We need the radio telescope preamplifier to establish a low noise figure and provide sufficient gain to overcome the insertion losses of the input filtering.

A double balanced diode mixer serves us well in these circumstances (see reference later)

A post mixer amplifier accomplishes a number of goals (again see later)

The only downside is near lightning strikes often play havoc with the type circuitry in such a vulnerable position. You can't have everything and remember this project is a lot like life, a matter of compromises. I have toyed with the idea of installing a NE-2 neon bulb at the input between signal and ground to bleed off static build up but I still remain in two minds.

So what have we decided? In place of the centre insulator in Fig 1 we will now instead have a shielded box (preferably diecast variety). In it we will construct, ugly style, in various compartments, a low noise preamplifier, various passive pre-selection circuitry, a passive mixer, a broad band post mixer amplifier and some output circuits. Our D.C. power supply will be fed through the same coaxial cable that the signals come down. Real Flash eh!

Here is a block diagram. The whole radio telescope circuitry may be called a "converter".

Radio Telescope converter block diagram

Fig 2 - Radio Telescope converter block diagram

Here we have a balun at the input because our half wave antenna presents a balanced 50 ohm (nominal) input whereas the input low pass filter and the succeeding circuitry is unbalanced. Similarly a coaxial cable presents an unbalanced load.

Radio Telescope Low pass filter

In this case we present our radio telescope antenna input after the balun to a 30 Mhz (nominal) two stage low pass filter. This is followed by a low noise dual gate fet amplifier. After that we have several stages of band pass filtering which feeds a double balanced passive diode mixer. The other ports of this mixer receive our 36 Mhz crystal oscillator frequency and the remaining port is the 6 Mhz band pass (36 - 30 = 6) filter and post IF amplifier. My choice of L.O. and I.F. frequencies was to an extent influenced by the ready availability of cheap microprocessor crystals at around 36 Mhz plus a desire to keep the I.F. at about 20% of the received input signal for image rejection purposes. Another quirk!

The input is a two stage low pass filter, why? Well stage one is a bog standard butterworth 3 pole filter with a Q of 1 (nominal 50 ohms both ports in / out) BUT modified to present a trap at the image frequency (30 + 36 = 66). No big deal however it doesn't hurt because in some areas that frequency falls within certain active TV channels. ABC-TV Channel 2 in Sydney is 63 - 70 Mhz and quite strong. The second stage is again a 3 pole butterworth filter with a Q of 10 and designed to transform the nominal 50 ohms to 2000 ohms input of the dual gate mosfet amplifier.

Here it is (notice I have included relevant reactance's and values in a table below) - a further quirk:

Radio Telescope input low pass filter

Fig 3 - Radio Telescope input low pass filter

The resistors in Fig 3 above are depicted solely to indicate input / output impedance's. These resistors are NOT actual resistors used in the circuit. Component values for capacitors are nearest standard 5% value or a suitable parallel combination.

Radio Telescope Low Pass Filter Component
100 pF
27 pf
100 pF
130 pf
27 pF
0.2 uH
1.18 uH

Having used standard capacitors and a nominal 30 Mhz for our Radio Telescope input low pass filter we can run into a few problems. As one example take capacitor C1 with a required reactance of 50 ohms. At 30 Mhz the actual value calculates out to 106 pF or just marginally higher than a 5% tolerance 100 pF standard capacitor. On the other hand 100 pF EXACTLY at 50 ohms reactance is a frequency of 31.83 Mhz.

Generalized Rule with low pass filters - use the next lowest standard value capacitor DOWN from those actually calculated if necessary. With inductors use the next LOWEST number of whole turns e.g. a calculated 10.6 turns - use 10 turns NOT 11.

In practice with the circuit above in Fig 3, C3 and C4 would often be replaced by one single capacitor (it has the same effect). In this case there is a 5% tolerance standard close to (100 + 130 = 230 pF) the required value. A capacitor of 220 pF fits our generalized rule. Also be aware that although capacitors are 5% tolerance they can also vary considerably with temperature variations.

If you don't entirely understand this then perhaps you should see my tutorial on low pass filters, hopefully you will come away with some better understanding.

Radio Telescope Low noise amplifier

For our radio telescope this is going to be the first critical part. Purely on whim I'm going to use a dual gate mosfet. Another reason is I have on hand quite a number of these devices so if I can ever build the unit I have whatever I need. I think my choice of a 40673 should still be readily available - I will provide sources of all parts later on if I see sufficient interest is generated. Other devices may be substituted. I make no particular claim that this circuit or the devices selected are necessarily optimum.

According to my RCA data book the 40673 offers these performance features:

  • superior cross-modulation performance and greater dynamic range than bipolar or single-gate Fets (this may be somewhat negated by newer devices with which I have no experience)

  • wide dynamic range permits large-signal handling before overload

  • excellent thermal stability

Additionally the device features

  • high unneutralized RF power gain

  • low VHF noise figure - typically 3.5 dB at 200 Mhz
Radio Telescope low noise amplifier

Fig 4 - Radio Telescope low noise amplifier

Radio Telescope low noise amplifier components
100 pF
47 pF
47 pf
82 pF
15 pF

Without going into a lot of technical discussion about the 2 pole band pass filter ( C7 to C11 and L4 and L5 ) let me say for those who like to follow these things, the centre design frequency, Fo is 30 Mhz. The theoretical design bandwidth is about 1 Mhz and using the recommended toroids the unloaded Q of the inductors should be about 165. The filter equivalent impedance calculates out to about 2530 ohms. C7 is calculated to match back to the 2200 ohm source and C11 is calculated to match the 2530 filter to the 50 ohm input of the mixer. C9 the principal coupling capacitor which influences the bandwidth was calculated at 2.72 pf.

You will note two possible oddities in the table above. Firstly C7 and C9 are possibly funny values to some people, tell me which you would find the easier to discern without possible mistake, 6.8 pF or 6P8 and 2.7 pF or 2P7. Could not 6.8 pf to some tired old eyes be taken as 68 pF? The capital "P" merely indicates a decimal point placer as well as pico-farads. Similarly I have used 82R, 220R, 2K2, 33K and 100K for resistors

The second oddity to some people might be the two identical capacitors for C8. At my peril, I did not want to introduce trimmer capacitors and as the resonator capacitance Co at 30 Mhz with 0.27 uH inductors you need 106 pF. C8 needs to be Co - C7 - C9 or 106 - 6.8 - 2.7 = 96.5 so I elected to split it in two. You could equally use one 100 pf capacitor. C10 was arrived at similarly ( Co - C9 - C11 ). There is also stray capacitance to take into account.

If you prefer changes C8 could be 82 pF fixed and C10 could be 68 pF fixed, each with a 20 pF trimmer in parallel. This is fine if you have available a 30 Mhz signal source, know how to peak simple filters and have some means of detection. I'm punting on the fact that for most readers this is sufficiently difficult as it is with very limited test resources they have available.

Double balanced mixer

Again there is already a tutorial on this if you need to know. In this instance I'm going to use a literal "black box" approach by recommending we use a Mini-Circuits SBL-1 mixer. Knowing the insides of it won't necessarily help you and to build it from four hot carrier diodes plus two baluns would probably cost you more anyway. Here are the connections. Note this is the bottom view, pins facing toward you

double balanced mixer

Fig 5 Double balanced mixer

Now that is pretty straight forward isn't it? In actual practice we will mount the SBL-1 on its back on a piece of blank circuit board. Note pins 1 - 7 and 2 - 8 are standard 0.1" (2.54 mm) pitch and pins 1 and 2 are standard 0.3" apart. The device is the same size as an 8 pin I.C. Mini Dip.

Crystal Oscillator

Again we are going to use something pretty bog standard here. Surplus microprocessor crystals are readily available and quite cheaply too.

crystal oscillator

Fig 6 Crystal oscillator

All of these circuits are just plain text book designs used in pretty well every manufacturers' application notes e.g. RCA, Motorola, National, Amidon, Mini-Circuits and others. Of course I have used classic butterworth filter design techniques for the filters.

Post mixer amplifier and output circuits

Here is where the fun starts

post mixer amplifier

Fig 7 Post mixer amplifier and filtering

Looking from the left hand side of the schematic where the coaxial socket is, here is the output which goes via coaxial cable to your receiver.

Going down the schematic you will note the 10 uF/ 25v electrolytic capacitor, 220 uH choke, a LM-340T-12 voltage regulator and associated by-pass capacitors. At the receiver end we are going to have circuitry which will inject 15V DC up the coaxial cable and here we are tapping off the 15V DC, passing it through a choke etc. to clean it up and then regulating it a second time on board down to 12V DC. Here is where our power supply comes from!. It came UP the coaxial cable.

Back to the left of the schematic I have taken the IF output from pins 3 and 4 of the SBL-1 mixer which needs to see a 50 ohm termination (quite important!). I think from memory I learnt my next trick from Dr. Ulrich Rhode DJ2LR in one of his many excellent contributions to the now defunct Ham Radio magazine in the 1970's.

To preserve this 50 ohm termination we pass the output to a "diplexer". A diplexer simply means our output from the mixer (which contains myriad's of signals) is going to take two different paths, one high pass ( 2 X 220 pf and L6) and the other low pass (L7 and 68 pF).

This diplexer consists firstly of a high pass filter (L6 and two 220 pF capacitors). L6 and each capacitor exhibits 50 ohms reactance at the cut off frequency of about 14.5 Mhz. Some would question my choice of cut off frequency here because traditionally three times the IF frequency is used i.e. 3 X 6 Mhz = 18 Mhz. Personally I wouldn't slash my wrists over it either way. If you prefer to stick with tradition then make L6 50 ohms at 18 Mhz (0.44 uH or 10 turns on T-50-6) and change the two capacitors to 180 pF each.

An important note on the toroids! The AL factor for T-50-6 is 40 while the T-37-10 is 25. In some instances I have deliberately used a reduced number of turns than those calculated, this is to allow for some stray capacitance which does exist between windings. Also be aware some toroids, even from the same batch will offer varying AL factors. Nothing is perfect!

The other part of the diplexer is a simple L network comprising L7 and the 68 pF capacitor (normally includes a trimmer) which forms a low pass filter at about 6.5 Mhz and transforms the impedance from 50 ohms to 2K2. This filter is terminated by the 2K2 resistor at the gate of the fet although I suspect it doesn't greatly help our noise figure.

The high pass filter is terminated in an absorptive termination of two parallel 100 ohm resistors which of course equals 50 ohms. I elected once again not to include the trimmer because I doubt many readers will have suitable test equipment. If you do have test equipment replace the 68 pF fixed with a 20 pF trimmer in parallel with a 56 pF fixed capacitor. Ideally I would have left the 68 pF fixed and made L7 variable (another quirk of mine) but that's a lot more difficult to achieve in this day and age. Boy it's exciting to see a test signal grow on an oscilloscope when tuning an inductor.

The amplifier is a re-run of the earlier preamplifier we used and comes from RCA literature. The output transformer with the 2K2 resistor presents a load of 2K2 to the amplifier and the 40:6 turns ratio ( 40 / 6 = 6.667) which is pretty close to the square root of 2200 / 50 which is 6.63. The reactance of the transformer using a FT-50-77 ferrite (NOT iron powder) toroid is about 130,000 ohms at 6 Mhz. The inductance is in the region of 3.5 mH (milli-henries) and it's a plain vanilla RF transformer.

Review so far (for the technically minded)

  • We have a dipole antenna cut for close to 30 Mhz. Minor variations in dimension would not be significant.

  • This feeds our initial 50 ohm low Q (1) low pass filter (LPF) at slightly above 30 Mhz which also contains a trap at 66 Mhz

  • The signal then goes to a further higher Q (10) LPF which also transforms the 50 ohms up to a nominal 2K2 input to our preamplifier.

  • The input termination for these input low pass filters is our 50 ohm Antenna.

  • The output of the preamplifier also sees a load of 2K2 thanks to the loading resistor across L3 and passes to a two pole butterworth band pass filter of about 1 Mhz wide centred around 30 Mhz. (29.5 - 30.5 Mhz). To achieve a narrower bandwidth would require a higher design Q and once the design Q becomes an appreciable fraction of the unloaded Q (Qu) of the inductors, L3 to L5, then the filter insertion loss becomes substantial. Design Q is about 33 and the inductor Qu (from the Amidon data book) is about 165.

Tables indicate at a ratio of about 5 for a two resonator filter, the insertion loss is about 3 dB (if I wasn't lazy I'd calculate it but what the hell). In any event the output should represent a 50 ohm source to the double balanced mixer.

On reviewing this section I decided to become "un-lazy" and present the insertion loss calculations for you:

IL = - 10 log ( 1 - QL/QU)2

= - 10 log ( 1 - 33 / 165)2 = - 10 log ( 0.8) 2 = - 1.938 dB or 2 dB loss
  • I deliberately refrained from designing any narrow band circuitry in the output of the post mixer amplifier because I am unsure what bandwidth was required. A filter with a design bandwidth of 200 Khz could easily be inserted here.

  • The input filtering (both sides of the preamplifier) probably accounts for a total insertion loss of about 5-6 dB. The preamplifier gain is probably about 13 dB and the succeeding double balanced mixer is another loss of about 7 dB, so our gains and losses are about square after coming out of the mixer.

  • The diplexer probably accounts for a couple of dB loss and assuming a similar gain of 13 dB from the post amplifier I reckon overall we have a net gain from antenna to coaxial socket of about 10 dB. I'm guessing a noise figure of between 3 - 5 dB (as I said earlier NOT an optimum design)

  • For the non-technical, an approximate net 10 dB gain means, in signal voltage terms, that a one micro volt signal on the antenna becomes 3.162 micro volts at the coaxial socket. Not exactly heart stopping eh? However we should have established a significantly improved receiver noise figure and here that is the name of the game or object of the whole exercise.

  • Note: - for the true "NON-BELIEVERS" - the main gain ALWAYS comes from the final IF stages later in the receiver, not here!. Repeat that after me! Therefore don't be a "non-believer" and be the first to tempt fate and my wrath by sending me an email asking "How can I beef up the front end gain?". YOU DON'T.

  • The first amplifier is only designed to establish an acceptable "noise figure" and its net gain in conjunction with the post mixer amplifier gain are simply expected to recoup the net losses elsewhere in our converter system including some anticipated losses in the coaxial cable.

Reservations (mine)

1. As I said earlier this is a far from optimum design, however I am hazy at best as to what the actual goal is (technically). I have no experience in this type of reception and what is to be expected. Obviously we are attempting to receive "noise" and measure it.

All my technical life in receiving I've been doing the exact opposite, trying to avoid or minimize noise. In this case I guess we have to differentiate between "good noise" that emanates from the Milky Way and "bad noise" which comes from terrestrial signals (local radio signals and interference etc.) and the inherent noise in our receiver set up.

Certainly I would expect the set up I have suggested above to be a significant marked improvement over a coaxial cable of indefinite length, attached to a dipole at one end and to a short wave receiver at the other end.

2. Also the chosen frequency of 30 Mhz being so close in proximity to both the CB band and the Amateur Radio 10 metre band also causes me considerable unease.

On reflection I have a fairly shrewd hunch this frequency was selected solely because it was the upper limit of the original author's short wave receiver and hopefully beyond terrestrial signals. In fact I am quite confident about that statement.

3. The chosen frequency does not lend itself to easily physically constructing a good directional antenna with high gain, excellent front-to-back ratio and good side lobe rejection. That's high-tech-speak for "it'd be nice to point it at the Milky Way and pick up nothing else apart from our expected noise from that source and perhaps E.T.'s phone calls being returned!"

Hey, this exercise was brought to you by a guy well versed in the technology of DX-TV reception. Used to pull in usable TV signals in Sydney from as far away as New Zealand and Brisbane and that's no small technical feat, especially as I lived in a valley. For me this exercise is like trying to answer the age-old question "how long is a piece of string".

Answer: Twice the distance from one end to the centre

On balance with these reservations I think I could make a case for a good project in the UHF region say 300 - 400 Mhz (or higher) instead of the limitations 30 Mhz brings to a radio telescope.

On the positive side

If you have got this far then your knowledge should have been somewhat improved. If so, then my time spent was worthwhile.


At the receiver end we need some circuitry to strip off the 6 Mhz signals, pass them to your receiver and inject our 15V DC up the coaxial cable to power the converter. Here is my suggestion.

Receiver input and power supply

Fig 8 Receiver input and power supply

Presumably this part of the circuitry will be installed indoors and connected to the converter at the antenna by a suitable length of 50 ohm coaxial cable (not 75 ohm TV type).

You will note to the bottom left of Fig 8 a 15V A.C. plug pack (wall wart) socket depicted. To avoid the use of transformers and dangerous exposed 120V or 240V A.C. wiring I always recommend using plug packs or wall warts as they are called in the USA and elsewhere. Please NOTE it is a 15V (or slightly higher) A.C. NOT D.C. The current rating will be whatever minimum size you can obtain. The power consumed here is absolutely negligible.

Next follows a bridge rectifier. I specified 1N4004 diodes only because they are cheap in Australia. Any silicon rectifying diodes will do or if you prefer, buy a cheap small bridge rectifier instead of the individual diodes. The diode depicted above the 15V regulator is there to protect the regulator. The 2200 uF capacitor is not sacred in value either but ensure the voltage rating does not go below 25V nor its value below say 1000 uF. The 10 uF capacitor value is not sacred but the comments about voltage rating still apply. The 0.1 uF capacitor should be as close as possible to the output of the regulator. Another one on the input wouldn't hurt.

The fuse depicted is COMPULSORY. Voltage rating is irrelevant and current rating is about as small as you can get, I'd suggest 100 mA maximum. The red led and associated 1K5 resistor are optional extras with the LED so mounted as to indicate power is on.

The 0.1 uF capacitor shown in the signal line between the two signal sockets (IN / OUT) is also COMPULSORY and its exact placement as depicted is CRITICAL. It blocks D.C. power from the receiver but allows radio signals to pass through.


Post a Comment

There was an error in this gadget