Electronics > RF, Microwave, Ham Radio

How far can a spectrum analyzer go in terms of transistor fmax?

(1/2) > >>

I know the title doesn't make a lot of sense. What I am trying to ask is; let's say that I have a spectrum analyzer, a tinysa ultra, with 800MHz bandwidth, which claims to be calibrated to 5.3GHz in ultra mode. It uses some kind of harmonic mixing and spur removal algorithms but I am not quite sure how it works yet.

I am going to buy some transistors and schottky diodes to build radio frequency receivers and some testing equipment like a return loss bridge and signal sources.

What kind of fmax should I aim for? Should I limit myself to transistors with fmax<800MHz? Maybe 5.3GHz? I know there is no hard rule about this kind of thing and also that it is pretty much impossible to actually get any gain at fmax from a transistor. Most of them seem to be specified for about a third of their transition frequency.

What kind of scares me is that I will buy these transistors, build some circuits and then they will oscillate at some very high frequency that I can't detect, though I suppose it is not very likely that a transistor can be made to oscillate at frequencies near fmax especially considering that the rest of the circuit will have some loss even if I were to try and make it oscillate.

I am thinking of getting some BFP183 for amplification/oscillators and some cheap medium barrier schottkies for mixers. I've already got lots of lower frequency transistors and I've been building RF circuits as a hobby for about 10 years or so. It's just that seeing "GHz" in a datasheet kinda scares me :)

BFP183 seems to have an fmax of about 10.3GHz based on its fT, Rbb and Ccb. I figure I can just run it at a lower than ideal current to lower its fT, although I will need to run it at a higher current to drive those schottky diodes for mixers. Maybe I can buy a more powerful but slower transistor to do that. The reason I am considering it is because it seems to be quite cheap. Another reason is that I need a low noise figure for some weak signal projects.

I am open to transistor-diode suggestions and thanks in advance for your answers.

What is the frequency you are going to receive. Ft should/must be larger than that. 10GHz transistor will oscillate, it is your job to stop it. So, with 8GHz SA you are in trouble. A better tool would a nanoVna. Use it to see the impedances. There are many models of those. See https://nanorfe.com/nanovna-v2.html

Besides, I think you should study some receiver schemas, Ft is a wrong end to start building a receiver.


The tinySA ultra will suffice for the BFP183's 10 GHz fmax. The former can detect oscillation beyond its 6 GHz specified upper limit, but with amplitude error. The firmware can support up to 12 GHz. https://groups.io/g/tinysa/topic/95556944#8485

Other low cost ways to detect unwanted oscillation:

1. In the future, when your transistors' fmax is >>10 GHz, probably the cheapest way to detect unwanted oscillation into the mmW region is a power meter (e.g. HP436/437) and a thermocouple power sensor. The power sensor can be used many times above its specified freq limit although the reading will be inaccurate.

2. The collector / drain current can be monitored for oscillation. When oscillating, the device is driven into its P1dB, hence the Icc/Idd is higher than the stable state.

3. If the noise figure (the tinySA can measure NF) is higher than normal, it could indicate oscillation


--- Quote from: LM21 on June 06, 2023, 06:55:50 pm ---What is the frequency you are going to receive. Ft should/must be larger than that.

--- End quote ---

For a receiver (and honestly even a generator) this is not true. We use harmonic receivers all the time to sense up to the 100's of GHz with spectrum analyzers, and these don't have gain at any frequency. You just pump them at one port with an LO, and listen on another port with a diplexer, and then use some clever frequency jiggling and math to get only the signals from the harmonic you want.


Thanks, that's nice to know. I will be using the BFP183W BJTs for low noise amplifiers and maybe low phase noise oscillators. It seems that the optimal current for the best noise figure (5mA) is actually much lower than the current for optimum power gain (15mA) or peak fT (25mA) so that should decrease the fmax even further. One other problem was the fact that at higher currents the impedance seen looking into the emitter can actually be as little as 1 ohms at 25mA, and when combined with the frequencies involved, makes it very difficult to provide a ground point with low impedance for the emitter. Even a 1mm wire has 1 nanohenry of inductance, which is 6.3 ohms at 1GHz. Operating at lower currents should help with that too.

I am guessing that's also how the "ultra" mode for the tinySA works?


[0] Message Index

[#] Next page

There was an error while thanking
Go to full version
Powered by SMFPacks Advanced Attachments Uploader Mod