[time-nuts] Building a DMTD/phase noise set in the 21st century (was: Keysight N5511A - phase noise measurements down to theoretical-177 dBm/Hz)

[Attila Kinali]

On Thu, 22 Aug 2019 07:35:06 -0700
"Richard (Rick) Karlquist" <richard at karlquist.com> wrote:

> It is easy to homebrew a phase detector using a high LO drive mixer
> (eg JMS-5H+ $12.95) followed by an LT1028/LT1128.
> Then all you need is a low frequency spectrum analyzer.
> If you don't need to go below 20 Hz offset, you can use
> software that makes your PC sound card into a spectrum analyzer.
> Depending on what you are doing, you may need to add an op amp
> after the LT1X28 to make a phase locked loop.

If you are going the soundcard route, I would not recommend to mix
down as low as with DMTD systems. The ADCs used in soundcards
have very poor DC performance. They don't need to have one anyways,
as the lowest most people can hear is in the 20-50Hz range. 
Due to that, I would suggest to have the IF frequency between 100Hz
and 1kHz and then perform the rest of the analysis in software,
maybe even including another down-mixing step down to DC. As you
have always two channels, I would recommend to use two mixers fed with
an LO that is 90°C out of phase to get I and Q components. This will
further reduce noise, at least to some extend. While the isolation
between the ADC inputs in soundcards is pretty darn good, the noise
isolation isn't. So there is some correlation there that cannot be
removed (easily).

Another important thing to know with audio ADCs is, that they are
almost always some sigma-delta variant. This gives them a high linearity
and low noise with moderate cost. But that also means that the higher
the bandwidth you are using, the higher the noise floor becomes, as
you are "averaging" over less and less samples. I.e. while an audio
ADC might be spec'ed to be sampling at 192kHz, you will be only able
to use up to 20kHz with the nice low noise floor that is specified.
If you go beyond that, you will have a corresponding increase in
noise.

In a mail on this ML pretty much three years ago, I wrote the
following:

Subject: [time-nuts] ADCs for phase noise measurement (was: windows for FFT measurements of phase noise)
Date: Sat, 11 Jun 2016 15:33:30 +0200
---schnipp---
Alternatively, I would suggest using one of the modern sigma-delta
or SAR ADCs, which can deliver increadibly high ENOB and SNR at 
astonishing sampling rates. Good candidates might be:

AD7982, 18bit 1Msps
AD7984, 18bit 1.33Msps
LTC2378-20, 20bit 1Msps
LTC2368-24, 24bit 1Msps

Eval boards for these are available (between 100 and 200€) and interface
with SPI. You can either use an USB SPI dongle (between 5 and 50€) or
use a small uC board to interface with the PC. Saving the samples in
a wav file and using one of the many FFT tools shouldn't be a problem.
---schnapp---

And in another mail:

Subject: [time-nuts] measuring noise of power supplies (was: For those that insist on using switching power supplies)
Date: Sat, 15 Oct 2016 00:53:25 +0200
---schnipp---
The mid frequency range is mostly influenced by the telecom
noise requirements, which for historical reasons cover the 10Hz
to 20MHz range. It is probably the easiest region to measure
with homebrewn instruments. A decently fast ADC with a low
noise voltage reference (like the LTC6655) are all you need.
Depending on how accurately you want to measure the noise, it makes
sense to further split this range into a lower range up to ~500kHz
and an upper range above 500kHz. The reason is that there are today
several high resolution ADCs available that support sampling rates of
up to 1Msps (and some beyond),eg:
AD7982, 18bit 1Msps
AD7984, 18bit 1.33Msps
AD7960, 18bit 5Msps
LTC2386-18, 18bit 10Msps
LTC2378-20, 20bit 1Msps
LTC2368-24, 24bit 1Msps
These would allow to accurately measure the noise range that is
IMHO most interesting for most applications. Interesting because
a lot of applications are insensitive to noise below 1Hz or even
below 10Hz and noise above several 100kHz becomes easy to filter
out using inductors, ferrit beads and ceramic capacitors. When
choosing an ADC for this range make sure you check the actual
SNR/SFDR performance as it a higher output resolution not necessarily
corresponds to the actual performance delivered. This becomes
especially pronounced when going higher with the sampling rate
to cover the higher noise frequency ranges. Beyond 5-10Msps 16bit
is the best you can get and conversly the SNR is limited to
something around 90dB-95dB.
---schnapp---

Especially the AD7960 and the LTC2386-18 are interesting in this application,
due to their high sampling rate, which allows to keep the IF frequency
far out of the flicker-noise region and at the same time does not require
aggresive anti-aliasing filters, making the measurment more stable.
An important part in this setup is to have two (or four) low noise mixers
that are driven coherently with the same LO signal, and more importantly,
the same LO noise. Designing the mixers such they can cope with the wide
range of input power, input frequency and still maintain a low noise
figure (and low flicker noise) is an art in itself. Unfortunately, I have
very limited experience in this and thus have to refer you to more
knowledgeable people. 


As most of you know, the state of the art is using direct sampling
of the input signal and doing all the processing in the digital domain
à la Timepod and Timepod v2.0 (aka Jackson Labs PhaseStation 53100A).
For this you need higher sampling frequency ADCs. But, even though
you can get 16bit ADCs with 210MHz sampling range, their SNR is
limited (the LTC2107 has ~80dB). Another problem with this approach
is to have the aperture jitter to be correlated among all ADCs. While
the sampling clock's noise is the largest contributor to aperture jitter
in high-speed ADC applications, it is not the only one. And at the
performance we want to achieve, the jitter induced due to voltage
fluctuations inside the ADC due to its own circuitry switching is 
significant. Unfortunately, the best I can tell you here is that
(some?) dual-ADCs have a quite correlated aperture jitter, which 
might help in lowering the instrument noise if used properly. But
why exactly the aperture jitter is correlated is not exactly known
(people have been tossing theories around, but nobody has actually
proven any of them).


			Attila Kinali
-- 
<JaberWorky>	The bad part of Zurich is where the degenerates
                throw DARK chocolate at you.