Counters

Counters / Time Interval Analyzers

Under elaboration....

Philips / Fluke PM6680B

Details TBD

HP 5370A / 5070B Universal Time Interval Counter

These oldtimers are still popular in the time nut community thanks to the 20 ps one-shot resolution. It can be tricky to set the trigger thresholds to get the lowest possible noise floor, and the input level should not be too low. Note that the manual states that the input peak level should be between 150 mV and 700 mV above the trigger level.
TimeLab may conveniently retrieve the data using GPIB, and thanks to the Talk Only mode the ADEV remains unchanged for different sample rate settings. The graph shows the noise floor for three different settings of the sampling rate, when letting TimeLab set the sampling interval based on the readings (0.070 s CCW, 0.46 s for the middle position, and 4.07 s CW).

HP 5334A Universal Counter

Details TBD

HP 53310A Modulation Domain Analyzer

Details TBD

Wavecrest DTS-2077 Communication Signal Analyzer

The DTS-207x family features 800 fs one-shot resolution, which provides a very low ADEV floor, comparable to what you get from DMTD or cross-correlation techniques. These devices do not seem to be very widespred in the time-nut community, perhaps due to the bulky form factor, or the not so straightforward user interface. Also, you need to go through an alignment procedure after each power-up. The inputs prefer steep edges rather than sine waves, and the level must be limited to a maximum of 1.3 V peak. I use a signal shaper in front of each input when I have to measure 10 MHz signals. As a side effect, this also protects the sensitive inputs.

Data may be retrieved by means of TimeLab but GPIB latency must be taken into consideration (more to follow...)

4-channel mixer design

This is a prototype of a 4-channel version of the Dual Mixer Time Difference (DMTD) technique. This front-end contains a local source and four down-converted channels to be digitized by a synchronous-sampling DAQ from eg. National Instruments. The application written in LabVIEW provides time-difference data for all of the channel combinations, allowing three-cornered hat analyses to be performed, or to have real-time monitoring of multiple sources. The huge advantage of the system is the high resolution and the low ADEV noise floor. In order to increase the measurement gain, the difference frequency is often kept low in DMTD and multichannel versions, like in the range of 1 Hz to 100 Hz. As a consequence, the maximum update rate is typically not high, so the method targets analysis of slowly changing oscillators of nominally the same frequency. The prototype here allows the user to set the offset frequency in order to experiment and to find the best setting for a specific application.

High-speed digitizer as TIA

The advancements of high-speed, synchronous sampling digitizers have made it viable to digitize clock signals and process the data to extract timing differences. By using fitting and interpolation techniques you can get a standard deviation in the order of a few ps when analyzing two 10 MHz clock references. However, the results depend vastly on the digitizer settings, the slew rate vs. the sampling rate, the signal levels, and whether you analyze sine or square wave signals. For sine wave signals you are like going to introduce signal conditioning, or to overdrive the inputs to increase the slew rate. It's very easy to spoil the standard deviation by an order of magnitude!

Counter comparison; Noise floor with 10 MHz oscillator signal split into two and dumping the time interval data.