Analog (especially RF) circuit designers like to characterize their product’s distortion performance by means of the classic dual tone intermodulation test. Since this test requires two signal sources operating at very similar frequencies and identical amplitude, two expensive signal generators would usually be required. Up to 500 MHz, a dual channel SDG6052X can serve that task and provide perfect results – if used correctly.
The main requirement for this test is a clean dual tone signal and the challenge is to combine two signals in a way that the output stages of the generators do not produce intermodulation distortion products by themselves.
If we want to utilize the full available frequency range, we can consider three different setups suitable for the full frequency range of the SDG6052X. Consequently, the following test compares three configurations:
1. Resistive power combiner with two attenuators in the source paths from the generators.
2. Resistive power combiner with a single attenuator at its output.
3. Internal “Wave Combine” function with external attenuator.
The attenuators in this test are 20dB. The big advantage of this arrangement is that we need not touch the generator settings (except for frequency) and get the exact same output levels. As a consequence, the spectrum analyzer always sees the same signal level. All in all, the three tests are perfectly comparable over the entire frequency range.
I’ve tested both 4 MHz (low frequency) and 400 MHz (high frequency), as we can expect the generator performance to degrade at higher frequencies. The 2nd tone is just 10 kHz above the first one. An automatic Intermodulation measurement has been used to get the results quickly and accurately.
Let’s start with the textbook approach: resistive power combiner with a 20 dB attenuator in each source path. As expected, the resulting signal is near perfect and we can measure the intermodulation performance of the spectrum analyzer, which is better than -87 dBc when both input signals are at -30 dBm level at the analyzer input.
No point in showing the 4 MHz result, the first screenshot is for 400 MHz:
SDG6052X_Ext_2x20dB_400MHz_-4dBm
Now we remove the attenuator in the source links and place a single one at the output. This way, the levels remain unchanged, but the isolation between the two generator outputs is now only 6 dB, whereas it was 46 dB before.
Not a problem at 4 MHz:
SDG6052X_Ext_1x20dB_4MHz_-4dBm
But look at 400 MHz!
SDG6052X_Ext_1x20dB_400MHz_-4dBm
The Intermodulation products are very significant at only -58 dBc now.
Finally, we can try the internal wave combine function, 4 MHz at first:
SDG6052X_Int_1x20dB_4MHz_-4dBm
Well, it works, but distortion is already significantly higher than with the external splitter. -82 dBc might be good enough for many tasks, yet it is about 6 dB worse than the worst result with external power combiner, which in turn quite likely marked the limits of the SA.
The true disaster becomes apparent if we try to use the “Wave Combine” at high frequencies – no wonder, now we have both frequencies in a single amplifier without any isolation at all. That means, we are measuring the linearity of the output buffer in the SDG6052X now, and we cannot expect it to be highly linear at 400 MHz anymore.…
SDG6052X_Int_1x20dB_400MHz_-6dBm
With only -38 dBc, this is unusable even for the most undemanding tasks. I had to reduce the output level to -6 dBm, because the SDG6052X cannot maintain the -4 dBm per channel in wave combine mode at frequencies that high. The result is valid and fully comparable nevertheless.
Verdict: As tempting the internal “Wave Combine” feature might look, its use has to be restricted to low frequencies.