[Moon-net] LNA 144 ATF54143? Phemts *Earl (Jim) Shaffer, WB9UWA* wb9uwa at verizon.net /Sat May 27 08:31:09 CDT 2006/ * Previous message: [Moon-net] LNA 144 ATF54143? <005305.html> * Next message: [Moon-net] LNA 144 ATF54143? <005303.html> * *Messages sorted by:* [ date ] [ thread ] [ subject ] [ author ] ------------------------------------------------------------------------ Hi All. I did some pretty extensive simulation and testing of Phemt LNAs and concluded that a practical design for a Phemt does not currently exist for 144Mhz.. There are several factors badly at odds with each other when the desire is to use a Phemt device on 144Mhz. The first problem is that it is virtually impossible to assure unconditional stability. The thing just wants to oscillate at some frequency. In simulation, it is possible to "gimmic" good looking stability, but parsitic capacitance in a real inductor tends to shoot down a good "gimmic". It is possible and even likely that the thing is oscillating at microwave frequencies and you won't even know it until you test it for IMD. The second problem is IP3. It is common to swamp the output with resistance to help control stability. This clearly has the effect of reducing power output and therefore IP3. Much of the "advantage" is lost for using this device for better IP3. Some designers use higher bias levels for better IP3, but noise figure is sacrificed. A third, more serious issue for IP3 is the low input impedance of the device at 144 Mhz (more specifically the "noise match" impedance. It is between 150 and 200 ohms! This is good news for noise figure because little impedance change is required in the input circuit to bring the preamp to an excellant noise figure. The result is that a very low loss input network is very easy. This good news for noise figure is very bad news for IP3. Not that an input circuit will affect the power output of a device, but simply that a practical design with a very low noise figure will have a very wide input selectivity curve. The thing is broad as a barn on its input. This is because you have an input cavity that is very heavily loaded (about 150 -200 ohms). A popular preamp made in W3 land uses a 2 pole filter on its OUTPUT to hide the poor input selectivity. This does NOTHING to help with the problem of overloading the preamp with signals. Some of the power gaAsfets on the other hand have around a thousand ohms input impedance, so selectivity is easy in that case, however stability is still generally an issue with them. I did do some simulation with a two pole input cavity configuration to restore a sane amount of input selectivity to the circuit, but it becomes difficult to keep the low noise figure. Designing the preamp for best return loss does little or nothing to solve the problems here that I mentioned. Too much source inductance tends to quickly further reduce stability. Input selectivity is a very important attribute of a preamp. The noise figure should be a order of magnitude lower then your site noise level. Generally around .4 db nf is sufficient for this. In my opinion, a preamp with very good input selectivity should be chosen over a preamp with the lowest possible noise figure. Input selectivity works in combination with gain and power output to control IMD in a preamp. By the way, I am NOT an RF engineer, but I have had extensive conversations on this subject with some good ones. I take a little bit of knowlege from everyone. I have also done a lot of preamp building and rf simulation using Serenade SV. If I missed the mark somewhere along the line, I would be happy to hear what and why. I'm still looking for an optimal design. What do I use? I still use the old MGF1302 with a two capacitor input optmised for good input selectivity and around .4 to .5db noise figure. A good two pole helical filter AFTER the preamp does help when using the Yaesu FT736r. 73, Jim Shaffer, WB9UWA. P.S., How many QSO points for an AM QSO (hi, hi)? ----- Original Message ----- From: "Conrad_G0RUZ" > To: > Sent: Friday, May 26, 2006 6:41 PM Subject: Re: [Moon-net] LNA 144 ATF54143? >/ 18dB RL on the input of a noise matched PHEMT LNA, oh dear that is very />/ much />/ not optimal for noise figure, it is utterly impossible to have a RL like />/ that for the ATF54143 on 144 MHz because S11* does not equal Gamma opt, in />/ fact it is quite a lot different at 144. So the amplifier is gain matched />/ and not noise matched, this is physics and cannot be disputed. In saying />/ that quite why anyone needs a 0.3 dB noise figure for a typical 2m array />/ in />/ a typical environment is beyond me; also how exactly does one measure that />/ kind of NF with any degree of certainty? />/ />/ I am quite prepared to stick my neck out and say that you cannot achieve a />/ correct noise match with an input return loss that low with the ATF54143, />/ the numbers don't add up. There is something inherently wrong with the />/ design. I am sure that there are much cleverer people than me out there />/ who />/ will back me up. I am not saying that you did not measure that return loss />/ Peter, I am sure that you did but this fact points to errors elsewhere in />/ the design. This LNA must be conditionally stable, in fact the original />/ article in the chapter on stability issues shows this to be the case. Also />/ the series drain resistor realised as it is in this circuit will limit the />/ IMD performance of the amplifier. It should be shunted so as to be />/ resonant />/ at 144 MHz which then makes the resistor 'invisible' at the operating />/ frequency, then the IMD performance will not be compromised at the />/ operating />/ frequency. There are more elaborate but perfectly achievable techniques />/ that />/ are even better than this. />/ />/ />/ 73 />/ />/ Conrad G0RUZ />/ />>/ -----Original Message----- />>/ From: Peter Frenning [OZ1PIF] [mailto:oz1pif at privat.dk ] />>/ Sent: 26 May 2006 23:52 />>/ To: conrad at g0ruz.com ; moon-net at list-serv.davidv.net />>/ Subject: Re: [Moon-net] LNA 144 ATF54143? />>/ />>/ I don't quite agree with you there Conrad, I've built more />>/ than a doze PA3BIY LNA's for 144 and I've never seen less />>/ than 18dB RL, best was 26 dB! />>/ I've also built four 432 MHz versions! never worse than 15dB RL />>/ />>/ In my book that's excellent input matching for an LNA />>/ />>/ Vy 73 de OZ1PIF, Peter />>/ />>/ ** CW: Who? Me? You must be joking!! ** />>/ email: oz1pif(no-spam-filler)@privat.dk />>/ http://home24.inet.tele.dk/oz1pif/ />>/ Ph. +45 4619 3239 />>/ Snailmail: />>/ Peter Frenning />>/ Ternevej 23 />>/ DK-4130 Viby Sj. />>/ Denmark />>/ *********************************** />>/ />>/ ----- Original Message ----- />>/ From: "Conrad_G0RUZ" > />>/ To: > />>/ Sent: Saturday, May 27, 2006 12:41 AM />>/ Subject: Re: [Moon-net] LNA 144 ATF54143? />>/ />>/ />>/ >I seem to recall that there were some problems with the PA3BIY design />>/ > highlighted in Dubus, the input match is not optimal and />>/ also there is too />>/ > much source inductance. I know what the best solution is />>/ but unfortunately />>/ > it is a well known French commercial design and the design />>/ data is not />>/ > freely available, it is a great LNA though! />>/ > />>/ > I think that the key to a good LNA for 144 MHz using a />>/ ATF54143 or 55143 />>/ > is />>/ > to use both resistive gain tailoring in the output network (shunt />>/ > resonators />>/ > across drain resistors with spread resonances) and more importantly />>/ > controlled source feedback. The value of source inductance />>/ is extremely />>/ > critical, these devices are not easy to make />>/ unconditionally stable at low />>/ > frequencies, adding excessive amounts of source inductance />>/ shifts the />>/ > instabilities into the GHz region. There is plenty of />>/ information in Dubus />>/ > Technik VI about how to best design the input circuit in />>/ DC8NR's article />>/ > which starts on page 337. />>/ > />>/ > 73 />>/ > />>/ > Conrad G0RUZ />>/ > />>/ > PS this is a classic example of learning from other's />>/ mistakes, reading, />>/ > some reverse engineering and some very recent lectures on />>/ LNA design. I am />>/ > not trying to be clever, I really need my hand holding for />>/ this stuff, it />>/ > makes my head hurt! />>/ > />>/ >> -----Original Message----- />>/ >> From: Moon-net-bounces at list-serv.davidv.net />>/ >> [mailto:Moon-net-bounces at list-serv.davidv.net ] On Behalf />>/ Of DW Harms />>/ >> Sent: 26 May 2006 23:03 />>/ >> To: 'ur5lx Sergey'; moon-net at list-serv.davidv.net />>/ >> Subject: Re: [Moon-net] LNA 144 ATF54143? />>/ >> />>/ >> Hi Sergeij, />>/ >> />>/ >> I think PA3BIY's design is the best. I must say though, that />>/ >> the ATF54143 is extremely sensitive for statics etc. I lost a />>/ >> few of them, not knowing what is the cause and I do have a />>/ >> very decent sequencer.. />>/ >> />>/ >> 73, Dick PA2DW />>/ >> />>/ >> -----Oorspronkelijk bericht----- />>/ >> Van: Moon-net-bounces at list-serv.davidv.net />>/ >> [mailto:Moon-net-bounces at list-serv.davidv.net ] Namens ur5lx Sergey />>/ >> Verzonden: vrijdag 26 mei 2006 22:32 />>/ >> Aan: moon-net at list-serv.davidv.net />>/ >> Onderwerp: [Moon-net] LNA 144 ATF54143? />>/ >> />>/ >> Hello moon-net, />>/ >> />>/ >> Pse tell me best LNA for 144 mhz for ATF54143, YU1AW????? />>/ >> TNX!Sergej.UR5LX />>/ >> />>/ >> -- />>/ >> Best regards, />>/ >> ur5lx mailto:ur5lx at vhf-dx.net / ------------------------------------------------------------------------ * Previous message: [Moon-net] LNA 144 ATF54143? <005305.html> * Next message: [Moon-net] LNA 144 ATF54143? <005303.html> * *Messages sorted by:* [ date ] [ thread ] [ subject ] [ author ] ------------------------------------------------------------------------ More information about the Moon-net mailing list