432

(Kit of parts available. - See end of page for details) 3/Oct/2005 updated 2/04/06

The ATF54143 is a PHEMT device from Avago ( previously Agilent technologies ) that can provide low noise figures combined with high signal handling characteristics.

The prototype for this amplifier was originally designed for 70cm and was used on the 2004 '3B9C' Dx-pedition to Rodriguez Island for satellite and EME. It had a noise figure of 0.49dB with an associated gain of 20dB.

The intended application for this new version is reception of U band satellite signals where the receiver may have to perform in the presence of strong uplinks in the 2m band.

It was important to design an input matching network that provided a good match for best noise figure while offering a reasonable match to 50 Ohms. Additionally, it was important that whatever matching network was selected offer some rejection of signals on the 2m band.

The circuit includes active bias and has the ability to receive DC power via the center conductor of the coaxial cable.
Power requirement is 7 to 13.8V at 35mA. A surface mount 5V regulator is included on the PCB.

The PCB is a 31mil thick FR4 with nearly 80 plated through 0.7mm holes. It has shorting straps on the 2 source connections which allow a small amount of source inductance to be added to the design. This may allow the PCB to be used for the 2m band as a small amount of additional source inductance would increase stability
Overall size is 55.5mm x 32mm. This fits a standard waterproof diecast box available here in the UK. (Rapid Electronics 30-1035 )
The board can also be cut down to 45x 32mm for use with a different enclosure.
An extra line of plated through holes has been included to maintain good continuity when the board is cut to 45mm.
The PCB is shown below with the many vias omitted. For operation above 800MHz the best performance is achieved with the board fitted into an enclosure where it can be soldered around the edges on both sides of the board.

The exact circuit implemented on the board can be any of several published designs. I have used a circuit similar to that described in the ATF54143 application note but with the inductor and capacitor values scaled for 70cm and a damping resistor close to the drain.

Values of biasing components have been chosen to give a drain voltage of 3V with a current of 30mA

Below is the PCB made from 31mil (0.8mm) FR4 material. The board has over 70 plated through holes to ensure good grounding.

Since measuring the amplifier from 435 to 1500MHz I have decided that ability to add source inductance, which is printed on the PCB, has no real benefit. However, the board can be used for lower frequencies and on 145MHz the use of the source inductance would be necessary as the device has very high gain on that band.

Blank PCBs are available for £6.30 including post to the UK. 9 Euros to Europe inc post
or $10 to the USA inc post.

A kit of parts including the PCB and all the components is available for £20 in the UK 30 Euros in Europe or $35 dollars in the USA. Standard shipping by recorded delivery in the UK or standard airmail for Europe / USA is included in the cost.
Note that the enclosure and connectors (SMA or N-type recommended) are not included.

Hints from and for constructors:

1) The Tantalum capacitors have their positive terminal marked with a black line.
2) The correct orientation for the 78L05D regulator can be identified by either:
    a) Pin 1 marked with a dot molded into the plastic
    b) The edge along Pin 1 and 8 marked with a 'bar'
    c) The black plastic body has a chamfered edge along pins 1 to 4 while the body along pins 5 - 8 has a 90 degree angle.
3) When the board is completed, test and check DC voltages. When mounting in a box DO NOT FLEX THE PCB. If you do then use the DC voltages to     find the cracked SMD components !
4) The larger pin on the PHEMT is a source connection which is on the same side of the device as the drain. See overlay above for correct placement.

"The preamplifier arrived safely here and I already built it, succesfully I must say. Measured gain at 435 is around 20 dB! I am very satisfied with it. Thank you! "

"I have whittled out an aluminum case for the 54143 kit. I can also say that I have completed the kit and am pleased with the results. The unit measures .45 dB NF on 1296. I have the source inductance dialed in at the maximum that the board allows and measure an input return loss of 10 dB on 1296. The input return loss on 432 is slightly over 4 dB on 432 with the same source inductance. The input return loss on 432 with minimum source inductance is about 2dB. I have not measured the NF on 432.

I want to experiment with adding more source inductance to improve the input return loss on 432. I will let you know how the effort works."

Measured it. I cleaned up a bit removing some excess solder near the input connector, it made no difference. I also defluxed and gained 0.1 dB. Here are the results.. All this is unboxed.

This was done with a calibrated, temperature compensated, low ENR head, not a high ENR head like I have seen done at some NF measurement parties. All these are likely to be subject to measurement uncertainties, at least 0.1 dB maybe more. I stopped measuring above 3400 as the gain started to drop rapidly. Noise figure was not so bad but not enough gain to warrant it. Still, this is a 432/1296 amplifier so it has no business working at 3400! I swept it up to 11GHz. There is significant ripple above 3GHz presumably the matching not being good enough.

You had better start thinking of it as working at 13cms too. 0.7 dB NF and 14 dB gain is a useful if not stunning pre-amp. The NF rises rapidly below 432 and I had to modify the coil spacing to get it to work reasonably at 432MHz. I think the cut off is a bit too severe but I know what you were aiming for in suppressing 144MHz.

> Hello Mike.
>
> Thank you very much. That's a very interesting set of results.....
> I wonder if a little more compression of the coils may improve the 435MHz NF.
Not enough range as it turned out. I tried that as initially 432MHz was much worse. 435MHz is fine, 432MHz is not. It is that close to the cut
off point and the fall off below is very rapid. Maybe another turn on the coil or a slightly larger one? I might have made mine a little too
small as I think I only had a 2.5mm bit to hand. I did not wind it too tightly but that might be enough to make the difference. The capacitors
are probably only 10%.

.....I did it twice, (tested)
several times at 1296 and 432 and there was some variation between readings. There was also interference at 900 and 1900MHz, but the NF at 850MHz when quiet was under 0.4dB and with the device below, I saw about 0.37dB. That is getting too low to be accurate about without taking much more time and care.

An interesting thing we tried - the device gets quite hot. It has 30mA going through it at 3V. 90mW is not a lot but it is not very large
device. We put a Peltier cooler on the back of the PCB and cooled it to around 0C. The noise figure dropped by 0.05dB. Repeatedly as we cycled it. Now 0.05 dB is not a lot, but if you were to use this device in a narrow band design, it might give you say 0.3 dB NF which would drop to 0.25 dB.

I am not suggesting everyone uses a Peltier cooler, the EME guys can but nobody else would. What I suggest is you pay some attention to enlarging the source area so that the device can dissipate a little more heat via its leads. It might even be worth putting a dab of heat conductive glue under it.

Unfortunately the DC to daylight performance might also make it susceptible to interference from 1800MHz GSM and 3 GHz radars.