I first posted this blog entry in my personal blog. It was meant to be posted in Test and Measurement. It is now posted in both places.
In this post I compare the trend analysis feature on three instrument: The Agilent 34461A, the Tektronix DMM4050 and the Agilent 53230A universal frequency counter. My findings suggest to me that all three instruments provide useful trend features, but the 34461A does so at a lower price point.
I used each instrument to trend the stability of three signal sources: A 32.768 KHz real time clock (RTC) crystal oscillator, a 50.000 kHz output from an Analog Devices AD9835 numerically controlled oscillator and a 50.000 kHz sine wave generated by a Tektronix AFG3102 arbitrary function generator. In addition to using the graphical trend display available on all three instruments, I used each instrument to save data to a USB memory stick for transfer to a PC where I uploaded the saved data into Excel spread sheets. Note that I did not monitor or control for variation in ambient temperature, nor did I monitor or control for variation in DC power supply levels. These are important variables to consider if a serious assessment of oscillator stability is to be undertaken. The objective here is more so to compare the trending and logging functions of three instruments and less so to compare stability of oscillators.
The Agilent 53230A is the most accurate and precise instrument included in this comparison, and it should be; it has 20 ps resolution and 12 digits/s single shot frequency resolution. It is also the most expensive of the compared instruments with a base price of $CA 3773. Although the two multi meters are not in the same class as this fine instrument, both multi meters performed very well and both are capable of handling standard electronic measurement tasks, but at about ¼ the price of the 53230A.
Now, to the details. The three ≈10 s video clips below allow you to compare the trend display on each instrument. All three instruments are trending the output of an Analog Devices AD9835 numerically controlled oscillator programmed to produce a 50.000 kHz sine wave. The Agilent 53230A provides the most detailed trend chart. Next is the the Tektronix DMM4050, but with a different sort of representation. The DMM4050 presents a kind of error bar at each interval that marks the maximum and minimum values detected within the sample interval. Near the end of the DMM4050 video clip you can see the automatic compression of data as the screen fills and more data is added. Finally, the Agilent 34461A presents a traditional trend chart display, but even after auto scaling it does not show the fine detail visible in the 53230A.
All three of these instruments allow data to be logged to a file. I used this feature on all three instruments, then saved the files to a USB memory stick using the front panel USB sockets provided on each instrument. The log files were then loaded into Excel and graphed. I gathered a log for each oscillator on each instrument. By examining the Excel graphs below it is clear that all of the instruments examined here are quite capable of producing detailed data for investigating oscillator stability. The first three graphs below show the trend charts for the 32.768kHz RTC crystal.
I find it interesting that although the charts below were acquired from the same signal source and all have similarities, each instrument produces a unique view of the signal. Granted the time duration is different in each case (22 hours for the 34461A, 10 minutes for the 53230A and 1000 samples for the DMM4050). The large spikes near the right end in the Agilent 34461A trend chart were the result of:
a) petting our cat while it sat next to my bench,
b) turning off a vent fan in the adjacent room,
c) standing up from my chair.
Note also that in every case below only one measurement instrument was attached to the signal source during the tests. This method is especially important when measuring the signal from the crystal oscillator as capacitive loading imposed by the measuring instrument pulls the oscillator off frequency. I observed the effect of capacitive loading when I attempted to connect two multi-meters to the crystal oscillator at the same time. The frequency shifted by several Hz. When a third instrument was added, the frequency pulled even further and the amplitude dropped to a point where oscillation was barely sustained. So, rule of thumb: one low capacitance, high Z instrument at at time to minimize instrument loading effects on the collected data.
The Agilent 34461A shows about a 0.5 Hz positive offset compared to the other two instruments. I made note of this in my first Road Test review installment on this meter. One improvement I'd like to see in all three instruments is a time stamp value to plot on the horizontal axis. All three instruments simply log successive measurement values without a time reference, absolute or relative. The 53230A does include a starting time stamp and from that a column of relative time values could be created, but it would be convenient if a time stamp and interval were provided in the log file. Having these two pieces of information would make recreation of a time scale easier.
Next, the logs from each instrument when measuring the AD9835 at 50.000 kHz:
These logs were taken over approximately 10 minutes on each instrument. The two Agilent charts are very comparable. Tektronix uses an error bar over sample interval method that results in a different perspective on stability. I prefer the Agilent presentation, but all are valid and useful for interpreting stability. Interesting aside: I discovered through this experiment that the resolution on the Analog Devices AD9835 numerically controlled oscillator is remarkably fine. The video segment below (only 14 seconds!) shows the minimum step size for the AD9835 is about 11 milliHertz at 50 kHz! The left video shows the value sent to the programming register inside the AD9835 (FREG0) by the micro. The right video shows the step change in output frequency - in milliHertz. Wow.
Finally, here are the trend logs from all three instruments when measuring the output of a Tektronix AFG3102:
Notice on the log above acquired by the Agilent 34461A DMM there is a slope at the beginning of the trend chart. This chart shows the warm-up behavior of the AFG3102 for the first 12 minutes following a power up.
In summary, the Tektronix DMM4050 and the Agilent 34461A compare nicely when using the data logging features. The differences are in how the data is presented and in the steps used to set up the log. I'd say the on-screen presentation of trend data is better on the Tektronix as it auto scales to highlight even small variations. Conversely, I prefer the format Agilent uses for data logged to a file as it provides more of an oscilloscope like view of data points rather than the error bar presentation used by Tektronix. The Agilent meter is less expensive and has a cool colour display, so overall, I'm recommending the 34461A - at least based on comparison of data logging and trending on the frequency function. There are still plenty of other functions to compare.
Mark
-
DAB
-
Cancel
-
Vote Up
0
Vote Down
-
-
Sign in to reply
-
More
-
Cancel
Comment-
DAB
-
Cancel
-
Vote Up
0
Vote Down
-
-
Sign in to reply
-
More
-
Cancel
Children