-----Original Message----- From: moon-bounces@moonbounce.info [mailto:moon-bounces@moonbounce.info] On Behalf Of Sam Sent: Sunday, June 13, 2010 11:31 PM To: moon; moon-net Subject: [Moon] 23cm preamp Folks, Some updated results from implementing a few of the RW3BP mods. Sergie has certainly put in a lot of work to extract better performance from the 23cm VLNA. My thanks to him for publicising his efforts and results. Some of the modifications described by Sergie require considerable effort. I'm thinking here of the input connectors. The surplus connectors I supply are gold plated and although used, seem to be of fairly high quality. Sadly, that source is drying up and I will need to find an alternative. I'll bet they are not as cheap! Anyway, onto the results. I thought it would be useful to try increasing the source lead inductance beyond that already introduced by the existing pad arrangement and FET leads. Using MGF4919 FETS has already been observed to provide slightly better results than the NE32584C. However, the 4919 has much wider source leads and hence introduces less inductance than the leads of the NE325. By bending the source leads of the MGF4919 down by about 30 degrees they will just about make contact with the first vias on the source grounding pads. This causes the FET to sit some distance above the PCB. The drain lead can easily be bent down to contact the pad with the drain resistor. R2 is now reduced from 22R to 10R. I have yet to try replacing it with a 0R link. Air-supporting this resistor, as a few people have tried, surprisingly provided no noticeable improvement in performance in this test preamp or in others where I have tried it. The gate lead is bent up slightly. L1 is 3.75 turns such that the input and gate ends effectively cross over to contact C1 and TR1 gate respectively. L2 remains as 2.5 turns. With TR1 and L1 gate-end well above the PCB this noticeably improves performance. There is still a tendency for the amplifier to oscillate around 16GHz until L1 is correctly adjusted. The small, supplied, strip of absorber, when stuck to the end wall near L1, effectively suppresses the oscillation. Sergie suggests placing absorber near L5 track. Both work. One further point needs to be emphasised. The quality of the construction around the input connector, including solder, cleanliness and integrity of the connector to box contact are VERY important. The input of a low noise preamplifier us no place for sloppy construction or substitution of parts (connectors!) unless you know their quality. The supplied C1 is a low ESR Microwave part that costs more than the usual 0805 chip cap!!! With these simple changes the prototype preamp now provides a noise figure of between 0.17 and 0.2dB with around 35 to 36dB gain. Also, the noise figure remains sensibly constant across the range from 1250 to 1420MHz, the variation is about 0.03dB. Of course these results are for one sample and several more preamps are being assembled to confirm these numbers. I tried several of the other mods suggested by Sergie without any further noticeable improvements, although I want to repeat these tests on other preamplifiers. I am sure they will provide that improvement towards 0.14dB that Sergie has set as the target! I just wanted to share these results with the community to show that the further improvements suggested by Sergie can, largely, be duplicated. I will supply MGF4919 FETs in new 23cm VLNA kits, where requested, for EME users, as well as the 0603 size 10 and (if I can see an improvment, as expected) 0R. I need to try similar mods to the 13 and 9cm preamps. I have alread seen improvements to around 0.3dB for the 13cm VLNA without the lead bending. I'll publish details on my web page in due course. 73 de Sam, G4DDK BTW. There was a small error in the July NL. I worked more than 8 stations on 23cm last year....... The actual words were that I worked 1 more station than the 8 stations in the DUBUS CW contest last year!