Here is a quick project I made one weekend. I had just gotten my new toys, a Rohde & Schwarz FSIQ 7 GHz spectrum analyzer and two R&S SME 03 signal generators. I wanted to make something where I could employ my new test gear, so I decided to make a microwave noise source.

Noise source

Noise source

The noise source is quite simple. It has a reverse biased AT41486 RF-transistor and the noise due to the avalanche breakdown is very wideband. The transistor has its base grounded and the emitter is connected to a 330 pF decoupling capacitor via a 49.9 ohm resistor. The biasing voltage is connected to this capacitor via a 10k resistor. The noise source is the small board located on the left side of the enclosure.

After the noise source there is an amplifier stage which has three ERA-5 MMIC amplifier stages which are resistively biased (to make the response as wideband as possible). I made the first amplifier version using MSA-0386 MMIC amplifiers but they seemed to have too little gain. Here is a 1 MHz to 3 GHz sweep of the original amplifier with -50 dBm input power:

3x MSA-0386 amplifier response

3x MSA-0386 amplifier response

The peaks showing through are 900, 1800 and 2100 MHz cellphone signals which get coupled to the amplifier. I’m not sure about the dip near the 2 GHz if it is somehow related to the cellphone signals or not.

For the next version I used ERA-5 devices to get more gain. The first version seemed to be oscillating or something because there was a strong +10 dBm signal near 30 MHz or so which was not visible with the previous version:

3x ERA-5 instability

3x ERA-5 instability

It was easily fixed by adding some ferrite beads and more filtering to the biasing network. With the ERA-5 devices I got around 20 dB more gain:

3x ERA-5 response

3x ERA-5 response

I experimented by adding some resistive feedback over the amplifier stages (1 nF ceramic capacitor in series with a 240 ohm resistor) to get a more linear response which wasn’t so effective:

3x ERA-5 with 1n+240r feedback

3x ERA-5 with 1n+240r feedback

The response is flatter in the end portion but the lower end is still quite a bit higher, and I also lost some gain. I guess it is marginally better with the feedback.

Here is the noise profile of the noise source, measured with three different biasing voltages for the noise source (10, 12 and 15 V) and using the ERA-5 amplifier without the feedback:

Noise profile with three different bias voltages

Noise profile with three different bias voltages

The source has an ENR of close to 10 dB on the low side of the range and around 5 dB near 3 GHz with a biasing voltage of 15 V. As a next step I would need to make the enclosure more sealed, the cellphone signals seem to get through pretty well. I experimented with using ferrite beads and snap-on ferrites and different covers for the enclosure but none seemed that effective in blocking the cellphone signals. I think my cover is just too loosely fitting and badly grounded. But anyway, it was a fun quick weekend project!

Leave a Reply