
A little history, way back we used and developed various noise sources, some custom, utilizing the usual techniques such as Noise Diodes, PN Sequences, Amplifier (ed) noise and so on. We even utilized noise sources with Time Domain Simulations to study the Non-Linear effects with noise modulation, frequency translation, and limiting, this was well before Cadence offered special Non-Linear Time Domain Noise Analysis Modes. So we are somewhat familiar with noise sources and such for both simulations and measurements.
On another thread the OP requested a Noise Reference and we suggested maybe using an AWG to generate a noise reference waveform. This suggestion got us thinking about these AWG as Quality Uniform Noise Waveform Generators, something that hadn't occurred to us prior.
https://www.eevblog.com/forum/metrology/white-noise-standard/Seems that having a separate thread about this topic might be useful regarding AWG Generated Noise Waveforms & Measurements, so here we go!!
We have a SDG2000X and 6000X, unfortunately the 2000X limits the Noise waveform Bandwidth minimum to 20MHz (maybe Siglent is listening

), while the 6000X goes to very low bandwidths, like 1mHz, but you'll need to wait until next month to find out!! Since the OP in the reference thread above was interested in a lower frequency noise source (0.1-10Hz), we took a look at the spectral output of the 6000X AWG.
With a bandwidth setting of 100Hz the output was flat from 0.1Hz to 100Hz, this took a long time (~hour) to stabilize with spectral averaging in the FFT mode (SDS2000X+ and Pico 4262), so many averages. This showed the quality of the AWG output waveform uniform spectral density across the band as set by the bandwidth parameter. We also did the same measurements with bandwidth set to 10Hz which took hours to stabilize but showed extremely flat response with no hint of 1/f characteristics. We also observed the usual 6~8 peak to peak to rms (std dev) ratio common to Bandlimited Gaussian Noise. All in all very impressed with this AWG as a source for low frequency noise.
Then we decided to "see" how well the True RMS features of our DMMs behaved, especially at the lower frequency ranges. We connected a pair of KS34465As with a DMM6500 to the AWG and set the noise bandwidth parameter to 10Hz.
One 34465 was set to AC Mode Filter >3Hz, with input at 10V Range and Stats On, the other set to DC at 1KSPS, 10V Range and Stats On. The DMM6500 was set to 1KSPS, 10V Range and Stats On. All were viewed in Historgram Mode and displayed the beautiful classic Bell-Shaped Curve!!
So this begs the question why would we not just use the AC RMS like the first 34465A. The answer is these are AC coupled modes and thus produce the RMS of the AC input as "seen from the point of digitation". These have a lower corner "High Pass Filter" to remove any DC from the RMS computations, so a waveform with very low frequency content, such as these Noise Waveforms, will be attenuated by the High Pass Filter Characteristic and introduce a measurement error by underestimating the waveform.
Sure enough this is what we see, the 34465A with the AC RMS shows an Average RMS of 0.9444 Vrms, while DC coupled 34465 shows 1.0031 Volts StDev and the DMM shows 1.0035 StDev. The AC RMS is ~5.9 % low. Should note that the AWG StDev setting is ~9% low, as we needed to set 1.1V on the AWG to measure ~1V StDev on the DMM.
Now the AWG is changed to 100Hz bandwidth, which should reduce the High Pass Filter effects as they are much less % of the total measured bandwidth. Now the AC RMS agrees with the other Digitized StDev DMMs, as 0.9963 Vrms vs. 0.9966 V StDev and 0.9966 V StDev for the two 34465A and DMM6500 respectively. Note the Digitized measurements showed a VPP/RMS ratio of over 8/1 at 10 and 100Hz Bandwidths, indicative of a good Band-limited Gaussian Noise representation!
Edit: Added FFT Plots from SDS2000X and PicoScope 4262 for 100 and 10Hz. Note the Pico FFT doesn't go below 100Hz, and you can see the noise roll-off after 10Hz in this 100Hz span plot.
Anyway, others may want to consider an AWG as a Uniform Low Frequency Noise Source Generator without spectral artifacts, and make some measurements! We certainly did not expect the AWG to be this effective as a noise source

Best,