Research – Page 7 – anderswallin.net

Clock Laser beat-note

Here's the beat-note, as seen on a spectrum analyzer, between a red laser at 445 THz (or 674 nm, if you prefer wavelengths instead of frequencies) and a femtosecond frequency comb. The frequency comb has evenly spaced (100 MHz in our case) 'teeth' at well-defined multiples of RF-frequencies that we can lock to a H-maser. This allows absolute frequency measurements of the optical frequency at 445 THz. Currently we are trying to improve the SNR of the beat-note so that a frequency-counter will give a stable output reading of the beat-note frequency. The video shows about 20 dB of SNR using 10 kHz RBW (if you are optmistic), but reliable counting requires around 25 dB SNR using a 100 kHz RBW.

This laser will be used as the 'clock-laser' in our ion-clock where it is used to drive a narrow clock-transition of a single laser cooled Sr+ ion. Earlier I blogged about measuring the thermal expansion of the optical cavity that is used to stabilize the laser.

Rack PC build

I assembled this new PC for use in the lab.

750W should be enough for anyone – right?

Note that the fan is connected to the wrong socket in this picture. The BIOS reported an error!

The case only has 5.25″ bays, and I didn’t have hardware to mount the SSD properly, so it is attached with double-stick tape.

Fusion splicing optical fiber

Update: Here's the other end of the 100m cable, spliced a few days later.

I spent most of the afternoon splicing this thick 12-strand single-mode cable to an LC patch-panel.

The fusion splicer and cleaver we have are quite similar to the ones shown here:

Pulse Stretcher - V2

Update: kicad files: 2014-07-23-pulse_stretcher_kicad

A development on pulse stretcher V1.

This circuit is used to stretch a short 10 ns pulse from a photon-counting module to a 100ns long pulse that can be more easily recorded or time-stamped e.g. with the white rabbit fine-delay FMC.

The new circuit is the same LT1711-based design as the old one, with an added buffer (BUF602) on the output. This improves output-load handling because the BUF602 can drive both 50 Ohm and 1 MOhm loads.

The PCB is made to fit a BNC-BNC enclosure by Pomona.

Some testing with an artificial input-pulse from a Keithley 3390 signal-generator..

.. and with the actual PMT-pulse. Note how the 100MHz scope produces nice round smooth signals while the 500 MHz Tektronix reveals more of the ugly truth.

ADF4350 PLL+VCO and AD9912 DDS power spectra

Update 2015-09-28: ADEV and Phase-noise measured with a 3120A:

Here's the 1 GHz output of an ADF4350 PLL+VCO evaluation board, when used with a 25 MHz reference.

The datasheet shows a phase noise of around -100 dBc/Hz @ 1-100 kHz, so this measurement may in fact be dominated by the Rigol DSA1030A phase noise which is quoted as -88 dBc/Hz @ 10 kHz.

The 1 GHz output from the ADF4350 is used as a SYCLK input for an AD9912 DDS. The spectrum below shows a 100 MHz output signal from the DDS with either a 660 MHz or 1 GHz SYSCLK. The 660 MHz SYSCLK is from a 10 MHz reference multiplied 66x by the AD9912 on-board PLL. The 1 GHz SYSCLK is from the ADF4350, with the AD9912 PLL disabled.

The AD9912 output is clearly improved when using an external 1 GHz SYSCLK. The noise-floor drops from -80 dBm to below -90 dBm @ 250 kHz from the carrier. The spurious peaks at +/- 50 kHz disappear. However this result is still far from the datasheet result where all noise is below -95 dBm just a few kHz from the carrier. It shouldn't matter much that the datasheet shows a 200MHz output while I measured a 100 MHz output.

Again I suspect the Rigol DSA1030A's phase-noise performance of -88dBc/Hz @ 10 kHz may in fact significantly determine the shape of the peak. Maybe the real DDS output is a clean delta-peak, we just see it like this with the spectrum analyzer?

Martein/PA3AKE has similar but much nicer results over here: 1 GHz refclock and 14 MHz output from AD9910 DDS. Amazingly both these spectra show a noise-floor below -90 dBm @ 50 Hz! Maybe it's because the spectrum analyzer used (Wandel & Goltermann SNA-62) is much better?

Photon-correlation test with modulated LED

Further testing of the time-stamping hardware. The idea was to generate a weak beam of light with an intensity modulation at 1-12 MHz, count and time-stamp the photons, and see if the modulation can be measured with a correlation histogram.

To generate a stream of photons intensity modulated at a frequency f_mod I used this simple LED circuit driven by an adjustable DC power-supply and a signal generator. I didn't test the bandwidth of the circuit and LED, but it seems to work well for this test at least up to 12 MHz.

If we are given such a stream of photons, with an average rate of say 10 kphotons/s, how do we detect that the modulation at MHz frequencies is there? Note that the average rate of photons is much much smaller than the modulation frequency. If we receive photons at 10 kcounts/s there is on average 1000 cycles of modulation between each photon-event, when f_mod=10 MHz.

One way is to count the photons with a photon-counter, and time-stamp each photon. We should now on average see more/less photons every 1/f_mod seconds. So if we histogram all time stamps modulo 1/f_mod, we should get a sine-shaped histogram. This assumes that the signal generator creating f_mod and our time-stamping hardware share a common time-base.

This works quite nicely!

At the start of the video we see only the dark counts of the PMT. A DC voltage is then applied to the LED and the histogram rises up, but remains flat. When the modulation is applied we immediately see a sine-shape of the histogram. If we adjust the the frequency, phase, or amplitude of the modulation we see a corresponding change in the histogram.

The video first has testing at f_mod=1 MHz, with a histogram modulo 1/f_mod = 1000 ns, and later with f_mod=12 MHz and the histogram modulo 83333 ps. The later part of the video also has a least-squares sin() fit to the data.

This technique is very sensitive to mismatch between the applied frequency f_mod, and the histogram mod-time 1/f_mod. I first wanted to try this at 12 MHz, so I set the histogram mod-time to 83333 ps - and saw no sine-histogram at all! This was because I had rounded 1/f_mod to the nearest picosecond, and 1/83333 ps is actually 12 000 048 Hz - not 12 MHz!
At 12 MHz a deviation of 48 Hz is a few ppm in fractional frequency, and I later tested that changing f_mod by a fraction of ca 1e-8 makes the histogram slowly wander to the left or right. Any larger deviation and the correlation is lost.
All of this is similar to a lock-in technique, so the same principles should apply.

White Rabbit Fine-Delay time-stamp testing

I'm testing the White Rabbit Fine-Delay FMC. It has an ACAM TDC-GXP time-to-digital converter that time-stamps the leading edge of an input trigger signal with ~30 ps resolution.

Recent work by Alessandro Rubini introduced a raw_tdc=1 driver mode which on my computer is able to collect time-stamps at a maximum rate of ca 150 kHz. Each time-stamp is 24-bytes, so this corresponds to a data-stream of roughly 4 Mb/s.

The video shows two graphs that update in real-time on the machine that collects time-stamps. The first is simply a (reciprocal) frequency counter where we count how many time-stamps arrive within a certain gate/window.

The second part of the video shows a modulo(tau) histogram where we bin time-stamps modulo tau into a histogram. The histogram was calculated for tau=100 us and an input frequency of 1 kHz was used. This results in the central peak in the histogram. I then slightly increased the frequency to 1.000010 kHz which makes the peak wander to the right. The peak on the right was produced by again tuning the input frequency to exactly 1 kHz. Similarly a lower input frequency of 999.970 Hz makes the peak wander to the left, and the peak around 20us was produced after tuning back to 1 kHz.

This hardware/software combination will be useful for collecting statistics and correlations in any experiments where a pulse type detector is used to measure something - provided the pulse-rate is below 150 kHz or so.

Pulse Stretcher - v1

A first try at this pulse stretcher circuit based on the LT1711 comparator. I need it for stretching a short 10ns pulse from a PMT.

The idea is to use the output latch of the LT1711. Once the output goes high, the combination C4 R4 keeps the latch pin (and thus the output) high for a time R*C. The Schottky diode is there to prevent the latch pin from swinging to far negative once the output goes low.

The PCB is made to fit into a BNC-BNC enclosure such as the ones from Pomona.

Messing up the pin-order of voltage regulators is becoming a habit! Note how the regulator is mounted the wrong way round compared to the PCB design - because I had the pin order wrong in my schematic.

I used a Keithley 50 MHz function generator to generate a 20ns long input pulse (the shortest possible from the Keithley) and the pulse-stretcher outputs a ca 483 ns output pulse. The prototype used a 1 nF capacitor with a 500 Ohm resistor which gives a nominal time-constant of 500 ns. The output pulse duration is far from constant and varies quite a bit from pulse to pulse.

This verifies that the propagation delay of the LT1711 in this circuit is within specifications, ca 4.5 ns. In addition to the comparator there is also maybe 70 mm of BNC-connectors, wires, and PCB-traces in the signal path, but that would add only ~350 ps to the propagation delay (assuming 2e8 m/s signal velocity).

One problem with this design is that it is sensitive to the load impedance connected to the output. With a 1 MOhm setting on the oscilloscope the pulse-length is correct, but switching to a 50 Ohm load impedance allows the capacitor to discharge significantly through the load impedance.

Version 2 of this circuit should thus add an output buffer (fast, low-jitter!) that can drive both 1 MOhm and 50 Ohm loads. An adjustable trigger level for the -Input of the LT1711 comparator could also be useful.

allantools

Python code for calculating Allan deviation and related statistics: https://github.com/aewallin/allantools

Sample output:

Cheat-sheet (see http://en.wikipedia.org/wiki/Colors_of_noise).

Phase PSD	Frequency PSD	ADEV
1/f^4, "Black?"	1/f^2, Brownian, random walk	sqrt(tau)
1/f^3 ?, "Black?"	1/f Pink	constant
1/f^2, Brownian, random walk	f^0, White	1/sqrt(tau)
f^0, White	f^2 Violet	1/tau

Phase PSD is frequency dependence of the the power-spectral-density of the phase noise.
Frequency PSD is the frequency dependence of the power-spectral-density of the frequency noise.
ADEV is the tau-dependence of the Allan deviation.

These can get confusing! Please comment below if I made an error!

White Rabbit DIO vs. FDELAY PPS output test

I ran a test to compare the quality of PPS outputs from a DIO (used as GM) and an FDELAY (used as slave) White Rabbit SPEC/FMC combinations. Here is the phase data measured with a Time-interval-counter. The nice clean traces are derived from a H-maser. We then lock a BVA (internally 2x multiplied to 10MHz) to this PPS signal. The GM node is locked to this 10MHz signal.

The average of each trace was removed, and the traces are offset for clarity.

The Allan deviations look like this. Both the DIO and FDELAY PPS outputs have allan deviations about 5x worse than the BVA used as input clock for the GM.