What's sets lower frequency limit of LNA devices?

[gbugh]

Hi all, newbie here, I hope this is an appropriate question/topic.

I'm wanting to make a broadband LNA with an upper frequency limit of 30MHz, to be used with a varactor tuned magloop antenna.

FETs will be used to sense the voltage oscillations across the varactor stack. I need very high input impedance rather than the typical 50ohm inputs.

When I look at the lowest noise devices available like these:



they all have their lower frequency limit higher than what I need.

What is it about these devices that make them unsuitable for lower frequencies?

When I look at the noise figure charts, they look to have even lower NF values below their specified lower frequency limit so I don't know why they have those lower limits.

Can someone please explain?

What would be a more suitable FET for the lowest noise at frequencies below MHz?

Thanks for your patience with me and your help,

George

 

[drkirkby]

Microwave devices tend to be hard to stabilize at low frequencies. I've not looked at any of the data sheets your link points to, but I can see they are PHEMPT devices. Getting them stable at low frequencies would be a challenge.

With a loop antenna at low frequencies, there is no point in having a low noise figure. There is galactic noise and other sources of noise which make it pointless in having a low noise figure at low frequencies.

Dave

 

[mtwieg]

Using fancy HEMT or SiGe devices for such a low frequency won't give good results. A while back I was designing LNAs in the 10-15MHz range and found that the BF998 works quite well (I got NF<1dB). If you want high impedance, then a J310 JFET might work well too. Regardless, you will likely need some impedance transformation network to get good results.

 

[BigBoss]

GaAs FETs and its derivatives' are very noisy (1/f Noise) below 50-100MHz, don't use it.

 

[RCinFLA]

Generally, less then 30 MHz RF communications is limited by environmental noise and lots of signals. This makes noise figure of less importance and signal handling capabilty very important.

Tuning over the range will yield better noise figure, limit signal passband signal quantity, but depending on what your desired tuning range, can very difficult to accomplish. Be careful of the varactor Q over the tuning range. If you don't need continuous tuning consider pin-diode controlled inductor increments to keep the varactors tuning range requirement low and tank Q's high.

 

[gbugh]

Is a BF862 a good choice? I have some of those.

 

[vfone]

BF862 is a JFET and is very good for your application. JFET's have much lower noise than any other transistor type (BJT, MOS, GaAs) at low frequencies.

 

[vfone]

The maximum frequency specified in the initial post is LESS than 30 MHz. At this frequency an Avago Pseudomorphic HEMT will have more noise than a cheap BJT.

 

[drkirkby]

The maximum frequency specified in the initial post is LESS than 30 MHz. At this frequency an Avago Pseudomorphic HEMT will have more noise than a cheap BJT.

When I did my MSc (microwaves and optoelectronics) about 20 years ago, I covered these devices in some detail. The old brain cells have died a bit since then, and I have forgotten much of what I knew then. But can you explain why the PHEMPT should have such a high noise figure at low frequencies. Do you have an references to peer reviewed journals which indicate this? I'd like to know what in the semiconductor physics makes this so.

I've designed and built preams at 144 MHz from FETs deisgned to work up to 20 GHz or so. The problem as far as I can see is that the gain high at such a wide range of frequncies, there is a high probability of the things being an oscillator rather than an amplifier.

Dave

 

[gbugh]

If I use a source follower circuit I should have less problem with oscillation, yes?

I'm letting the high Q magloop provide voltage gain when sensing across the varactor diodes used to tune the magloop.

... The problem as far as I can see is that the gain high at such a wide range of frequencies, there is a high probability of the things being an oscillator rather than an amplifier.

Dave

 

[drkirkby]

If I use a source follower circuit I should have less problem with oscillation, yes?

I'm letting the high Q magloop provide voltage gain when sensing across the varactor diodes used to tune the magloop.

The problem is that components deisgned for 30 MHz will usually have behaviour at microwave frequencies which are hard or impossible to predict.