For those who would like to get an idea how the performance of the Bode plotter can benefit from the 12 bit acquisition in the SDS2000X HD, here you go:
Depending on the signal levels, ADC resolution doesn’t have to make that much of a difference for Bode plot in Auto Gain mode. At low signal levels, the frontend noise might be the limiting factor despite the frequency selective detector used by Siglent’s Bode Plot implementation. At higher levels, the automatic gain control keeps the signal near full scale anyway.
I’ve used the highest test signal levels that I had available, which also happen to approach the limit of the lowest 50 ohms sensitivity of 1 V/div (I had to fit a 10 dB attenuator for the reference channel already).
The test object was a ceramic 455 kHz IF-filter (with the familiar imperfections). Because of the complexity of this filter the phase plot doesn’t look particularly pretty and obscures the picture, hence I’ve disabled it. For this test it is irrelevant anyway.
The input level in this test was 24 dBm = 3.54 Vrms = 10.02 Vpp.
The first screen shot shows the measurement in Auto Gain mode. This plot is quite clear, the display range is 120 dB and the usable dynamic range in this scenario certainly exceeds 115 dB.
SDS2504X HD_FRA_455kHz_A
In Gain Hold mode we enjoy a considerably faster operation at the cost of a drastically reduced dynamic range. This is where the 12 bits of an SDS2000X HD might help to improve things by up to 24 dB in theory.
The second screenshot shows the same measurement as before, but this time in Gain Hold mode. For higher levels, there is no visible difference, but below about -55 dB the increasing noise gets quite obvious. It’s up to you to judge what you still accept as usable dynamic range in this scenario, but I would call it about 60 dB before the noise seriously starts obscuring signal details too much.
SDS2504X HD_FRA_455kHz_H
This result is still not too bad. 60 dB is the relation of 3.54 Vrms to 3.54 mVrms = 10.02 mVpp. In theory, we would expect about 60 dB improvement with the Auto Gain feature and this test suggests that it is at least 115 – 60 = 55 dB. This cannot be totally exact, because neither of the two tests clearly defines a distinct dynamic range:
• With Auto Gain, we didn’t see any significant noise yet, so the true dynamic range might be a bit higher, like 120 dB.
• In Gain Hold mode, it’s up to us to define the threshold where the noise becomes too obvious, and this is of course subjective.
Now let’s see how the SDS2354X Plus compares in this scenario when using Gain Hold. There is not much difference up to 400 kHz between SDS2000X Plus and HD, just the noise kicks in about 10 dB earlier on the 8-bit instrument (-55 dB instead of -65 dB). But we do see a substantial amplitude error above the second resonance at about 660 kHz; The level is shown up to 10 dB low. See the third screenshot:
SDS2354X Plus_FRA_455kHz_8Bit_H
All in all it seems that the results don’t look too different at a first glance, but the accuracy and reliability at low signal levels is indeed much better on the 12-bit instrument. It shows consistent results all the time, whereas the SDS2000X+ gets a bit random for levels below the maximum dynamic range of an 8-bit system, which is <50 dB.
BTW, the 10 bit mode of the SDS2354X Plus doesn’t make much of a difference either, see the fourth screenshot:
SDS2354X Plus_FRA_455kHz_10Bit_H
At lower signal levels, the analog noise from the frontend will ultimately be the limiting factor, at least partially killing the advantages of the high resolution acquisition. Again, we should stop and think for a moment, what the previously measured 115 dB dynamic actually means: 3.54 Vrms to 6.30 µVrms = 17.83 µVpp! A sensitivity that high would be totally impossible without the frequency selective detector, which also ensures that the measurement result is not falsified by spurious signals.