Why is my RF circuit behaving like this?

[thomaskuhn]

I am not an RF engineer but play one at work. I am working in uhf SatCom. In the lab, In order to test two radios communicating over a satellite, I built a small RF circuit out of cots components. I basically put a mixer between the two radios and inject the delta of the uplink and downlink frequency ( one radio also has discrete inputs and outputs so I use a circulator to split that out). Finally there are some attenuators so I'd do not blow anything up.

Here is the strange thing. I can use this circuit for most of my testing when I operate in SBPSK 2400bps. Also in CPM 2400 - 56kbps. BUT, if I operate in FSK 16k, the receiver will not get the transmission. No RX shows up.

Then I noticed that if I unscrew the N connector JUST enough ( most likely when the center connector is no longer engaged) but the shell is still on, reception all of the sudden works perfectly fine. My theory is that it is now tiny antennas vs. a solid connection) What is causing this? And is there another RF component that I should put in to properly replicate this condition? If so what is it called?

 

[itsmoked]

I'd expect you were over-driving the receiver front-end causing all sorts of non-linear results that breaks the detection.

But, I'm sure V1BLL will be by soon and give you the straight answer.

Keith Cress
kcress -

http://www.flaminsystems.com

 

[thomaskuhn]

I am currently TX at 24dBm and I have 33dB of attenuators. The receiver is -10dBm to -80dBm. I "think" I tried it earlier with a different setup with 50dB attenuation with the same results.

 

[VEBill]

Hi.

ViaSat offers the DOCCT/S UHF MilSatCom satellite (entire path) simulator, but they're 6-figures expensive. These systems do everything, far more than you might need for simple testing.

In general, I've seen quite a few examples where the 'over the air' (last foot) RF link for lab use was implemented using a batch of rubber duck antennas installed into the lid of a metal can (like a one-gallon paint can). The can lid would have bulkhead BNCs installed, and a bunch of rubber duck antennas inserted into the can, and then the metal paint can sealed up in the usual manner tapping with a hammer. It's a nice, sealed-up and isolated, RF link.

Of course, you still need inline power attenuators (often -30dB) for the transmitters (e.g. 100 watt dummy loads with -30dB sample ports), as well as quite a bit of additional low power attenuation. The impedance match of the rubber duck antennas doesn't matter in the least, due to the vast attenuation (50-ohms) required in the circuit.

A basic frequency mixer should work exactly as you're doing. But you might have to adjust the attenuation to account for the additional loss of your mixer circuit (although the dynamic range is typically huge, if you're in the middle).

Based on your description, it sounds like something odd is happening (difficult to see over the Internet). In other words, you'd have to poke around with a Spectrum Analyzer and otherwise troubleshoot.

-Double check for odd ground loops at power supply.
-Try plain voice comms, to make sure that the basic radio link sounds good to your ear.
-Try listening to the bad signal to see if you can hear something (for example, loud 60Hz noise). Use your ears(*).

(* Many decades ago, I went as far as reformatting C: and reinstalling the OS because I couldn't log into the BBS via my dial-up modem. After trying everything, I finally *listened* to the modem audio on the telephone handset. What I heard was a telephone system recording: "The Number You Are Calling Has Changed..." That's when I learned to listen to "bad signals" with my ears; don't rely on digital systems to tell you what's going on.)

Your FSK 16 should be low enough bandwidth to be human listenable. Unfortunately this old lesson doesn't work with modern Mbps (wide bandwidth) signalling.

Good luck.

 

[VEBill]

"...TX at 24dBm and I have 33dB of attenuators. The receiver is -10dBm to -80dBm."

+24dBm - 33dB = -9dBm. So maybe Keith was on the right track.

If the receiver is intended for -10dBm to -80dBm, then you should aim for the mid-point, on this log scale. IN other words, around -45dBm.

It's not the slightest bit critical when you're in the middle. But it can be critical when you're at the edges.

So aim for the middle. One less thing to worry about.