RF Experiments...has to be a better way:)

I’ve been spending a fair bit of time these days characterizing the AD8307 and AD9850 module.  The 8307 experiments are pretty much completed and I started in on the AD9850 module that I originally sourced from eBay.  Both these components are to be used in the up and coming SNA project that will be the fall buildathon put on by the Peel Amateur Radio Club.

I’ve always found that when being part of the design team for these projects, they are nothing less than an ideal learning opportunity on many levels.  This project has not disappointed.

For years I used the standard prototype board for my projects. While I know that they are not ideal for RF experimentation, I always went ahead and used them.  Not sure what was different this time, but I had all kinds of issues when trying to characterize the AD9850. Between some suspect continuity between the different connections on the prototype board and placement of wiring, the results were not consistent between the different runs of experiments.  In fact just a hand placement or a probe connection swung readings that could have contributed to at least a .5dBm error.  In one case a wire placed one pin over where it should have been did not rear its ugly head until scope readings on the output signal were not what was expected by a long shot. 

The 8307 board was sensing the rf and giving solid readings so it took a little head scratching and a few runs through the experiments before digging into the scope reading problems revealed the issue.

I was finally convinced that a proper rf testing jig was needed and forget the prototype board. When precision is needed as in this case, you need to have not only the test equipment but the proper testing setups as well.

My 2^nd kick at the can was to try and minimize the wiring that involved rf.  The Arduino, and the power supply were easy to take care of. The 8307 board was not much of an issue due to it been isolated in an Altoids tin with coaxial cable connecting to the 9850.  However I now connected the 8307 to the dummy load and the 9850 thru a series of sma bnc connectors that eliminated most of the un-shielded cabling I used for some of the rf.  See picture below.  Results, better but not good enough.  It will have to do for now as I’ve ordered some boards to make a more permanent rf testing environment.

In the meantime I’m now putting together a lpf and amplifier board.  The lpf will replace the lpf that comes with the 9850 module as most of us know is a pretty crappy filter and not the right impedance to boot.

Board is pretty much done and hopefully I can populate it in the next day or two.

Ad8307 Testing up to 500mHz

Ran another linearity test this time from 60mHz to 500mHz. Power level was set at -70dbm and -30dBm.  

Two conditions for each power level. Compensation components in circuit and components bypassed.  Results were pretty interesting for the bypassed assessment. It was about a 160mv drop, which would represent about a 6dBm loss, over a 50mHz span and 205mv over the entire range.

Since I soldered in a jumper in order to bypass the compensation components, I wonder if it is having the effect that is responsible for the rapid falloff.  I suppose I could put in another jumper of a different size/shape and see if there is any change in the curve. Even though I’m not all that interested in in frequencies this high, the curiosity factor alone is enough to try it.  Besides the whole idea of these experiments is to learn all I can about this device first hand.

The compensated response shows a 94mv drop over the range from 60mHz to 500mHz.  That would be close to a 4dBm drop. Not too bad and certainly can be easily taken into account either by software or a “cheat sheet”. 

As well I injected various hf frequencies into the Arduino input along with the 8307 output. I used a .1 cap to isolate the signal generator. Long story short here was, unless you inject a signal at the level of 0 to 5 dBm, there was no effect on the readings of the dvm.  Even then it was only 1 or 2mv. However the important thing here was not how the dc output was performing but how the Arduino behaved to any harmonics that may bleed through and get into the adc. The readings being displayed by the Arduino were steady and never once changed values no matter the frequency or power level. So for me the debate begins on how necessary is the lpf in that path? 

 

Linearity Test for ver 3 Board

Ran a linearity test from 400kHz to 50mHz. Power level was set at -70dbm.  I’ll do another run at a higher power level later, but previous testing showed an even better response curve.

Surprisingly, the circuit over this range of frequencies without the frontend compensation components had better results. 32mv compared to 40mv.  Over the hf range the variance was 12mv for no compensation and 20mv for the circuit with the compensation components added in. Maybe the circuit has better results on frequencies between 100 to 500 mHz.   Since a change of 1dBm results in a change of 25mv, this represents about .5dBm difference when not using the compensation components within the hf frequencies.