TECHNICAL

Pre-amplifier for 70 MHz
Published on 25 May 2011 by David G3PTU

This unit uses a dual gate MOSFET. You can use virtually any of this variety of semiconductor; the ‘Philips’ radial type e.g. BF981 appears to be just as good as the top hat types. The Classic is the 40673 but supplies are somewhat difficult and costly. There is an ‘equivalent’ available on eBay, these are not quite as good and have more gain; this is what was used in this unit. By luck (for this argument) however on 4 m the background noise level is quite high so absolute perfection in a preamplifier would be wasted at most locations. Gain was more then I needed at about 15 dB max.

A source of FETs is old Analogue TV tuners although some of these are bi-polar transistors so take care; some old Analogue satellite receivers also have them in the IF Chain. Rallies also provide good pickings at times where two 40673’s where bought on a board for a few pence. Like EF80 Valves all MOSFETs appear to be masters of most circumstances placed on them.
This amplifier was unconditionally stable, although it must be stressed that a metal box, VHF techniques and short component leads are mandatory to achieve this.

A word about location, location, location; if you have lousy coax and put this at the Coax foot it will do no good at all, you add the coax loss to the noise factor. This amplifier was devised to make up for a very long run of buried coax across a field and was located at the Antenna end of this cable in a waterproof IP44 Box. Without a good reason i.e. a long feeder, it is very debatable whether a preamp will do any good at all; indeed it could be a considerable hindrance. If the receiver is that deaf, find out why first.

Construction, the circuit is somewhat flexible, although the input coil needs to be the best that you can make, the output is not so fussy. The original was powered up the coax, if this not possible a separate Cable / Local PSU / Battery will be required to supply approximately +12 V. A suitable alternative feed point is shown. If powered coax wise, keep your eye on the connecter centre connections, as these will be at +ve and will suffer corrosion, like car batteries and tram rails do. As illustrated there is no RX/TX switching as this was used at a location is where 4 m TX is forbidden.

The input coil is tuned to resonance with C1; C2 provides matching for the best results. L1 was positioned at right angles to the out put coil L2 even though a screen of copper clad board across the box is used as well. Regeneration is the death knell of the noise figure and although can be problem with some layouts, lots of decoupling, inductors at right angles and a screen across the box are the wise and usual starting points. 
The source resistance used was 270 Ω, now opinion dictates that for some FETs this needs to be different, if you really want to optimise things then make this a 500 Ω Pot. The BF981 type FET prefer about 33 Ω, 40673 maybe 100 Ω. Philips claim low noise Fet amplifiers require a predetermined current through the device for the best possible noise factor, “for the BF981 this is 10 mA”. This current is a factor of both the G2 voltage (which should be about 4 V to source - again quote Philips) and the source resistance, so to be pertantic look on the WWW for the best quoted for your FET if it’s available. But as said, on 4 m generally the noise factor is not the main concern, QRM and QRM are so dominant and excessive gain can push the S meter to S9 with just the background noise, adding to the risk of cross modulation in the following amplifier stage, besides being most inconvenient. In short the contribution of noise from the preamp is insignificant.
It’s this background crud that usually sets the overhaul possible lowest signal readable on this band. So unlike other bands its not the lowest noise factor that is top priority. Even without any research with values shown, the noise will be low.
Besides and ignoring the above points, G2 is used in this design to vary the gain via a tweak pot; this is done as being “very convenient” because of the up the coax power was used, thus making attenuators within the coax line difficult. The gain finally used needs only to make up for any losses after this amplifier, therefore sufficient without making the background noise level too high.

Coil dimensions are as follows; -
L1 = 0.3 mm (not critical) 12 turns; tap 5 turns from earthy end 6 mm diameter, original on nylon former no tuning slug
L2 = 0.3 mm (not critical) 12 turns 6 mm on nylon former with tuning slug. Output 2 turns over cold end, same direction winding, insulated wire. On this coil, the tuning was quite flat compared to L1. Alternatively do not use slug tuning and make the 7 pF capacitor a 15 pF variable.

C1 was the best the spares box could provide and was a tubular type of trimmer. The 10KR pot was a skeleton device. The lead through capacitors are probably about 1Kpfd, they where used as they provided a nice supporting pegs and have decouplers across them in any case. On my FET the metal case was believed to be connected to the Source, take care as use of an ohmmeter might damage the device and as most are connected to one of the leads, in principle just do not let the FET case touch the earth plane / box or anything else.

Construction requires to be as compact as possible around the FET. The location of this device is next to a (say) 3 mm hole in the centre screen; only the Drain lead passed through this hole so as to provide a high degree of isolation. The majority of the bits including the FET where therefore in the Input compartment. The author has built other versions with gate lead through the hole and the bits in the output compartment; as long as the input and output are segregated it appears to make no difference. 
The centre screen is made of Copper clad board soldered around its edge into the box (cut this hole and possibly another hole for a lead through capacitor - the g2 supply, before soldering the board in place!). Leads should be cropped where necessary and everything is built dead bug style. The box I used was a little tin plate cheapy bought at a rally.
I used BNC for the input (SO259’s are not very good) and the output was B&L TV (as it had to interface with an existing lossy cable run) this is not critical and can be what you desire. The decouplers where all small disc units off the shelf ‘what was available’, they are not critical except for short end leads, I used quite heavy decoupling on the supplies etc as on long runs of coax MF radio can sneak in and cause problems. D1 is there because suspected lightening up the coax damaged the Mk1 and it seemed a good idea; when the Antenna and its Mast are remote from the shack and both ends of the feeder are ‘earthed’, during lightening storms there can be a potential difference between the two ends. Fortunately on that occasion the 10 µF sacrificed itself and saved the rest of the bits. Likewise the LED is only a help point and plays no part in the function, if the amplifier is battery powered; forget this luxury. The two 1N4148’s protect against low-level lighting discharges. Connect the metal box to the main mast and if the antenna has a rotator, bridge a copper braid across this mechanical box to connect the stub mast to the main section.

The result appears to be slightly better than the Spectrum front end. The Beacons from the south of England at 350 km while it was on test in Yorkshire are about 1/2 S-point better than with receiver barefoot. This may be due the higher Q in L1.
Tuning is best firstly done on the background noise. On 4 m there is lots of this so simply adjusting everything for maximum can be done rapidly. Then find a beacon and with careful adjustment of C1, C2 and source value (if fitted and you can be bothered) etc tweak for the best difference between the background and the beacon signal. Finally set the gain and repeat loop set up with a final tweak on the input C1. In operation there should be a marked increase in noise when the antenna is connected and if you beam down the road through the row of houses the noise usually at most locations can be heard increasing markedly. Bandwidth appears to be adequate to cover the band, although final tweak is probably best done on 70.25 MHz (mid band, UK) or whatever part of the band you are interested in.

The whole thing cost only a few pounds and appears to fulfil the requirements demanded of it. These dual gate Mos Fet amplifiers are very easy to get going and are very tolerant of component differences, there is no need to look for example for a 120 Ω. A 100 Ω or 150 Ω will just as good. You need a small soldering iron, a pair of tweezers, magnifying glass and be able to count to 12. 
If the box was made out of Copper board and spares box is provident it could cost nothing! 
To use on 6m simply adjust the Inductors / Tuning a bit!

Download the schematic in high resolution.

D J Long G3PTU