1970年代に化合物半導体が登場し、80年代のV/UHF帯ローノイズアンプは正にGaAs MES全盛。90年代になると主役はHEMTに移り、私の所でも1.2GHz～10GHzのマイクロ波用LNAはすっかりHEMTオンパレードになりました。その一方で、VHF領域ではHEMTを過去試したことがありません。逼迫したニーズが無かったし、それに「X～Kuバンド用素子をVHFで使うと、1/f ゆらぎ雑音により思ったほどNFが下がらない。VHFはGaAs
MESで十分だ。」といつか聞いた事が影響したかもしれません。
それが本当かどうか？技術的興味だけが動機ですが、今頃になり両者を比較評価してみました。選んだのはGaAs MESからMGF1412とMGF4405それに2SK571。対するHEMTは10GHzでもVYFBだったFHX13LGとNE32984Dです。入力回路は同一のLC定数のままで、増幅素子を入れ替えトリマ調整したのち雑音指数とgainを記録しました。
結果としては144MHzでもHEMTが優り、定評あるMES FETのベスト値よりも低い値を記録しました。測定に供したサーキュレータのロスは実測0.50dBで、それを差し引くと単体NFは0.3dB以下。冷却無しで0.2dBオーダー、雑音温度に換算して20K以下というのはえらいことです。念のため別個体のHEMTに交換してみましたが、NF値に変動はありませんでした。
2mバンドでは天空の背景雑音レベルが高いから、月面反射通信でさえこの種のプリアンプを使用する意味はかなり薄いと思いますが、システムに雑音温度の余裕を持たせる意味で実用価値はあると思います。
JH3ERQ藤本さんによれば、1/f ゆらぎ雑音の影響云々はもっと低い周波数での話だそうです。それを実感できたし、今回のテストは十分エンジョイできました。
Compound Semiconductor (GaAs) device began to be used in 1970's. GaAs MES
got the leading position of V/UHF low noise amplifiers, and 1980's became
their best days. After that the main device shifted to HEMT in 1990's,
and LNAs for 1.2GHz-10GHz were completely occupied with HEMT even in my
shack. On the other hand、I had not tried HEMT on any VHF band so far. One
of the reasons is that there were no urgent necessity, and another reason
is some QSP that I heard before, "Because of 1/f fluctuations noise,
NF is not so good on VHF if a HEMT for X-Ku band was adopted. GaAs MES
is enough on VHF.".
Is that true? The motive was only technical interest, and both of them
were measured and compared on the same table this time. I chose MGF1412,
MGF4405, and 2SK571 from GaAs MES FET group. And FHX13LG and NE32984D which
worked on 10GHz very well were chosen from HEMT group. The same LC tank
was adopted as an input circuit, and the noise figure and gain were measured
for each transistor after tweaking for better condition.
The result was a win of HEMT also on 144MHz, and that recorded lower NF
than the best of GaAs MES FET. The insertion loss of circulator offered
to this measurement is 0.50dB, therefore the NF of LNA itself is less than
0.3dB. I think it's really wonderful to achieve less than 0.3dB of NF which
is equivalent to noise temperature below 20K without any cooling. (I was
anxious about individual difference and exchanged it for HEMT of other
individual, but NF didn't change.)
Because the background noise of the sky is high on 2m band, I think the
merit of HEMT preamplifier use is small even in EME activity. But there
is enough meaning to get margin of system noise temperature.
Recently, Masahiko Fujimoto /JH3ERQ told me in his mail that the influence
of 1/f fluctuations is matter of much lower frequency range. And that was
actually proved on my test bench. I could fully enjoy myself in this test.
