27 November 2016

Si5351 output waveforms

I finally managed to have myself some shack time this weekend and one of the curiosities I had was to check the output waveform of the Silicon Labs Si5351 I2C programmable VCXO. The chip is quite popular among homebrewers thanks to its low price, ease of use, the fact that it generates 3 independent outputs across a very wide bandwidth (8kHz to 160MHz) and that it offers lower phase noise than the cheap & cheerful AD9850/1 DDS modules.

Adafruit Si5351 module

By the way, there are also some versions of the Si5351 that have 8 independent outputs instead of 3, if you really need that many.

The test setup included an Arduino Duemilanove with the NT7S library, an Adafruit Si5351 module and a Tektronix 2225 oscilloscope. I used the middle output for the test (CLK1), which was terminated with a 75ohm load. Oscilloscope vertical scale is 0.1V/div and horizontal scale is adjusted depending on frequency.

First test, 100kHz output:

Si5351 output at 100kHz

Output looks like a clean square wave, with a nice rise time. Peak to peak amplitude is about 650mV into 75ohm, which puts the power output at 0.7mW (-1.5dBm).

Si5351 output at 1MHz
No significant change observed at 1MHz output.

Si5351 output at 10MHz

At 10MHz we start seeing some distorsion in the square wave. Part of this may also be due to stray capacitance or inductance in the probe ground lead.

Si5351 output at 20Mhz

At 20Mhz the output still looks like a square wave, but the distorsion is more significant. The Tektronix 2225 is a 50MHz bandwidth oscilloscope so waveforms above ~10MHz are not displayed very accurately.

Si5351 output at 40MHz

At 40Mhz the shape seems to get closer to a sine wave. Displayed output is still constant at 650mV peak-peak.

Si5351 output at 50MHz

50MHz looks like a very clean sine wave, almost ideal for RF work. Unfortunately without a spectrum analyser we can't know how clean it really is.

Si5351 at 60MHz

Si5351 at 80MHz

In the last two screen captures we are already above the oscilloscope's bandwidth limit and at 80MHz the wave looks already smaller on the vertical scale.

The chip has a rated output impedance of 85ohm, so 75ohm is a bit out of spec but not enough to worry about. However, the output level is too low to drive a mixer directly, so an amplifier will be required at the output.

Square waves like Si5351 produces are great for driving mixers, the problem is they also contain harmonics of the desired signal which we generally don't want in our receivers. Depending on application, a (switchable) low-pass filter might also be required between the Si5351 and the output amplifier.

No comments:

Post a Comment