[time-nuts] Measuring ADEV for a beginner
kb8tq at n1k.org
Sat Nov 1 16:17:50 EDT 2014
The problem with setting up to measure any of this stuff is that it’s *very* dependent on the gear you have. There’s no big surprises below. It’s all “spend more money and have fewer things to figure out”.
First you need a way to measure frequency out of your mixer (there are LOTS of ways …):
If you grab a HP 5334 or HP 5335 to measure the frequency or time, you need to get the data out via GPIB. That’s a project by it’s self.
If you grab a HP 53131 or CNT 90 to measure the frequency, you can use a serial port. One less project, way more money.
If you grab a HP 5371, you just push a button and it pops up an ADEV plot. Less money than a 53131 (usually). It’s a bit limited on the range it will do ADEV over.
If you grab a brand new Symmetricom Time Pod, you spend (gulp!!) a bit more. You can now do measurements easily over it’s frequency range. (If you can afford that option - can I come play at your house ?)
Yes each of these is it’s own little rabbit hole to wander down and each has it’s own issues.
Now you need a reference:
GPSDO’s have their fans, they often have spurs and other “crud” on the output. They do have very good long term stability.
OCXO’s are often pretty quiet and spur free. You need some way to calibrate them for really long term stuff.
Rb’s come in a wide range of sizes and costs. Some are better than others. Their ADEV at 1 second often not as good as an OCXO.
Cs standards and Hydrogen masers are (to me) in the same cost category as exotic test gear above. Fine for a company, not quite so easy for a basement guy.
Again each of these is a bit of a project by it’s self. Often it turns into a “a couple of these and a couple of those” sort of solution. You only know something is right if you have another one to compare it to.
Now you have some gear. Let’s ignore building the mixer board and assume that’s taken care of already.
Now you need software:
TimeLab is free and it will interface with many of the counters you might be using. It does all the math for you and puts up pretty plots. Highly recommended. There are other free software packages out there.
Stable32 is another commonly used program. You need to get the data out of the counter before you can use it. There are other paid programs out there, Stable 32 is the best of the bunch.
An Excel spread sheet is indeed another option. The ADEV math is *not* very complicated. Excel can do it all very easily.
Getting data out of the counter could be a terminal program sort of thing (serial port) or something more complex (GPIB). In both cases there are a number of programs to pick between.
With software, things like what computer and operating system you have will influence your choices. I’m a Mac person, but I also run Linux, Windows, FreeBSD and a few other things as needed. Most people have a favorite ….
So, that’s the quick and dirty start to the “quick ADEV setup” process. There are:
4 counter like things X 4 reference ideas X (at least 4 mixer approaches) X (way more than) 4 software options = > 256 paths you could follow.
That’s quite a few, and many of them would be utter nonsense. Even pruning out the ones are unlikely to ever be followed, you still have lots.
1) Drive your “known good” reference OCXO into a Minicircuits RPD-1 mixer.
2) Drive your “device under test” into the other side of the RPD-1
3) Amplify and limit the output with a couple of OP-37 op amps running off of +/-15V.
first stage is an amp with high pass and low pass sections
second stage is an inverter / limiter (back to back diodes in a normal op amp inverter).
third stage is same as the second
you might (or might not) want a fourth stage
4) Feed the +/- 0.7 V limited output into a 5335 (DC couple the input) in talk only mode
5) Hook up the 5335 to a National Instruments GPIB card on a PC
6) Run a Visual Basic routine to grab the outputs and write them to a text file.
7) Process the text file with Excel. Use the ADEV formula from the original NIST papers.
There have been a *lot* of OCXO’s tested that way in many different factories over a couple of decades. (Yes, it’s probably easier these days to substitute Time Lab for steps 6 and 7).
Is it the ideal or perfect way to do it? Certainly not. It is one of many simple ways it can be done. Even this simple way has two forks in it (TimeLab or not). It also has a few features like a know good OCXO and the undocumented circuit on the output of the RPD-1.
Quick, simple and pretty much useless to you. It’s my favorite way to do it, but that really should not make it your favorite.
What to do?
It’s really a multi part process. Break each section down and address it separately. You *do* have a goal in mind, but get each chunk running by it’s self. Do simple verifications each step of the way. Tie it all together at the end. Accept that you *will* spend some money on this. If you have a $10 budget, do not expect $10,000 or $100,000 results. If time spent matters, factor that in as well (is 4 months worth $300?).
How long should it all take?
Well, I’ve been poking at this and that part of the list above for 50 years or so. I’d say I’m still in the “working on it” phase as far as the basement is concerned. There are a number of people on the list who have a *much* better setup than I do downstairs. I doubt that there is anybody who has exactly the same piles of this and that lying around that I have.
> On Nov 1, 2014, at 9:29 AM, Anthony Roby <aroby at antamy.com> wrote:
> I've been reading a lot about ADEV and following the threads on the list, particularly Karen's in-flight thread. What I haven't come across is a simple explanation of the basic setup required to go about collecting the data. John Miles referenced this page http://www.ke5fx.com/tpll.htm, and the simple setup at the bottom of the page looks like a reasonable place to start. Seems that I'd need to acquire a phase detector and build or buy some filters and the amp. I can probably figure that out, but how do I get the data into a PC? Is there a basic hardware and software setup that someone could point me to or recommend?
> time-nuts mailing list -- time-nuts at febo.com
> To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
> and follow the instructions there.
More information about the time-nuts