|
Guys,
I've heard people say that having a totally accurate external frequency reference available and distributed to all the test devices on one's test bench which can accept it is an optimal solution. But what is it an optimal solution to? What's the advantage of such a system? I've heard various absolute standards mentioned over the years. Currently GPS satellites seem to be in, but I've heard such-and-such a radio station being recommended as well. What's it all about?
J.
|
I'm not a time nut, so take this with a bit of salt.
You're dealing with two things here in measurement, stability and accuracy.
Accuracy you can get easily enough by a GPS, or even Rubidium, and in the short term, OCXO.? This answers the "how close is that to 10 Mhz" question.? I suspect that it means most with uncoupled test equipment not locked to a lab standard, or portable equipment where locking is not possible.
Most portable stuff will be OCXO if that.
Stability is another matter.? Various sources can be made on frequency, but then how long do they stay there?? Depends on the reference.
Do you need this?? Depends on the measurement, IMHO.? If you have multiple counters in the lab, then the best thing is to lock them to a source, if you can.? Not all will.
If you do a lot of portable work, you pays your money and you takes your chances.
If you are specifically interested in long term stability, or accuracy, then you've answered your own question.
Then again, there's the pursuit of accuracy for the sake of accuracy
Harvey
toggle quoted message
Show quoted text
On 6/18/2024 3:25 PM, Jinxie via groups.io wrote: Guys,
I've heard people say that having a totally accurate external frequency reference available and distributed to all the test devices on one's test bench which can accept it is an optimal solution. But what is it an optimal solution to? What's the advantage of such a system? I've heard various absolute standards mentioned over the years. Currently GPS satellites seem to be in, but I've heard such-and-such a radio station being recommended as well. What's it all about?
J.
|
In my case, I don't bother with a common reference piped around all my test gear. I usually rely on the various OCXO references in my test gear. I do check and adjust them every year or two.
However, I do sometimes connect the reference from one (vector) sig gen to another if I want to vary the phase of one output compared to the other.
There is also a risk that an external reference can add spurious terms or it can add jitter and degrade the noise performance. It can also be quite a complicated thing to set up because you would want the reference to distribute reliably even when some bits of test gear are powered off. I just don't see much advantage to having a common frequency reference.
However, some people will run a GPSDO 24/7 and pipe it around their workroom using a suitable distribution amplifier. I like to turn everything off when I finish in my workroom so this isn't a realistic option for me anyway.
|
I have one of these inexpensive units running to send 10MHZ to all my devices.
Since the GPSDO runs on 12v it can be used with portable gear. Takes about 3 mins to lock from cold.
https://brisbaneradiosociety.net/10-mhz-reference-by-vk4amg/
|
There are two or three parts to "optimal" in the lab frequency reference context. The first is that if all gear uses the same reference, then they all will agree within +/- one count of whatever units are measured, typically to 1 Hz resolution for counters, generators, and spectrum analyzers and such. This is regardless of absolute frequency precision. If each piece has its own internal reference, any differences in measurements will include the differences in references. A common reference for all conveniently eliminates these differences. Most "regular" gear won't notice these details. It shows when making high resolution measurements, or when frequencies are multiplied up by large factors, say, in communications gear.
The second is the absolute precision at any point in time, depending on the reference source quality. The third is the stability, over various time frames. These usually go hand in hand - the best references have very high precision and stability, and the never-ending pursuit of improvement and perfection.
Ed
|
Add to this a fourth metric; phase noise.
You won't be doing yourself any favor if the frequency reference you use in your shop is noisy.? Of course, you'll need to have some equipment capable of measuring said noise, as well as a less noisy reference to use in those comparisons.
There's a reason one who pursues the most favorable characteristics in all of these metrics is called a "Time Nut"... it is never-ending; the deeper one dives into learning and striving for accuracy and precision, the farther down one finds the bottom of the hole.
?
|
Hi J:
Some decades ago I wanted to measure mixer spurs in an automated test system and speed was very important since a lot of spur frequencies needed to be checked.
In order to see low power levels the resolution bandwidth needed to be narrow.? That means the spectrum analyzer needs to tune to the exact frequency of the spur.? A couple of synthesizers supplied the LO and RF inputs to the mixer.? All that was needed to get this to work was to connect the 10 MHz reference output from the instrument with the highest quality internal reference to the Reference Input of the other instruments in a daisy chain.
--
Have Fun,
Brooke Clarke
axioms:
1. The extent to which you can fix or improve something will be limited by how well you understand how it works.
2. Everybody, with no exceptions, holds false beliefs.
|
I second that about spur tests taking large amounts of time.? We have several test benches at work and each one has a daisy chain for the 10 MHz reference, one instrument used as the source output, and the others are inputs.? We do sometimes see Reference Unlocked error messages on some of the instruments, though.? It's surprisingly easy to get suckouts at 10 MHz on the line.? Video equipment has switchable 75 ohm terminations, so that only the piece at the far end from the source has the termination connected to the line.? IDK why the test equipment manufacturers don't do the same with the 50 ohm internal terminations.? My garage lab has a distribution amplifier between the GPSDO and the instruments, each instrument with a dedicated RG-58 cable to it.? I'm not going for those suckouts.? The Leo Bodnar GPSDO cost me about $150, and the DA was less than $100.? Cheap way to get atomic clock stability.? ? HTH.? ? ? ? ? ?Jim Ford?
On Wed, Jun 19, 2024 at 1:07 PM, Brooke Clarke via groups.io <brooke@...> wrote: Hi J:
Some decades ago I wanted to measure mixer spurs in an automated test system and speed was very important since a lot of
spur frequencies needed to be checked.
In order to see low power levels the resolution bandwidth needed to be narrow.? That means the spectrum analyzer needs
to tune to the exact frequency of the spur.? A couple of synthesizers supplied the LO and RF inputs to the mixer.? All
that was needed to get this to work was to connect the 10 MHz reference output from the instrument with the highest
quality internal reference to the Reference Input of the other instruments in a daisy chain.
--
Have Fun,
Brooke Clarke
axioms:
1. The extent to which you can fix or improve something will be limited by how well you understand how it works.
2. Everybody, with no exceptions, holds false beliefs.
|
On Tue, Jun 18, 2024 at 01:19 PM, jmr wrote:
I just don't see much advantage to having a common frequency reference.
If it really is a reference, and not a 'reference' then it will help eliminate systemic measurement errors, but, the combination of the reference, and the synced instrument, will always increases uncertainty...? due to random errors in the synced instrument. However you can use statistics to reduce the random errors. Statistics can't reduce systemic errors.
|
On Wed, Jun 19, 2024 at 06:08 PM, Jim Ford wrote:
Cheap way to get atomic clock stability
AFAIK, Cesium atomic clocks are the most accurate timekeepers... 100 to 1000 times better than a GPSDO, with a rubidium oscillator. Anyway, a rubidium oscillator, or OCXO, will drift as it ages... beyond the capability of the GPS system to correct that... if I understand it correctlly?
|
On Wed, Jun 19, 2024 at 09:03 AM, Matt Harris wrote:
Time Nut
'time nut' ... 'volt nut'... or 'audiophile'... IMO... they sometimes make it harder for other 'nuts' to get the resources they need to actually 'do' stuff.
|
On June 19, 2024 9:59:33 PM "Roy Thistle" <roy.thistle@...> wrote: Time Nut 'time nut' ... 'volt nut'... or 'audiophile'... IMO... they sometimes make it harder for other 'nuts' to get the resources they need to actually 'do' stuff.
The difference bei6, of course, that the former two actually know something about technology. -Dave -- Dave McGuire, AK4HZ New Kensington, PA
|
On June 19, 2024 10:10:57 PM "Dave McGuire" <mcguire@...> wrote: On June 19, 2024 9:59:33 PM "Roy Thistle" <roy.thistle@...> wrote:
Time Nut 'time nut' ... 'volt nut'... or 'audiophile'... IMO... they sometimes make it harder for other 'nuts' to get the resources they need to actually 'do' stuff.
The difference bei6, of course, that the former two actually know something about technology. (err, "being") -- Dave McGuire, AK4HZ New Kensington, PA
|
Hi there!
I'm not a time nut, but almost. I like my equipment to go at the same rythm, so I spent about €200 on two boxes: a GPSDO and an 8-output distribution amplifier and I am very happy with the result. My frequency measurements are precise and synchronized enough. Those were bought at eBay, from China, you know, that green and aluminum small boxes.
Regards,
Ruben
|
We used the excellent Leo Bodnar GPS for years,
USB and Windows interface allow selection of F and V, 800 hZ..800 MHz, out to 32 dbm
We set the F to 10.000 MHz and output to MAX.
We distribute the ref to all our TEK/HP/other via Mini circuits splitters, one Leo B GPS drives 5..10 50 Ohm ref inputs.
Enjoy
Jon
|
I assume that you distribute the reference through a resistive splitter to each device.
The output of the LEO is a square wave and long cable lengths can cause waveform distortion due to reflections, especially for those units tapped off along the line.
ed
|
Ed: The Leo B GPS (and most GPS/TCXO etc) have Zo 50 Ohms.
We have a stock of vintage BNC 2X, 4X, 6X 50 Ohm wideband splitters.
These are used often for video , so perfect BW for 5..10 MHz.
Running the GPS at max output, the losses in the splitters (use transformers) still give > 500 Mv at the loads.
Our setup: Leo B 10 MHz>>Mini Circuits 2X>>Mini Circuits 4X, 6x etc.
We drive 6 to 10 devices: Counters, spec/network analyzers, scopes.
Min-Circuits are good cost/availablilty.
Jon
|
So having read all the replies so far - for which many thanks - is it the same reason that all the PLLs within my 8566B spectrum analyzer are not only locked themselves but inter-locked with each other as well? So to ensure everything's in phase with everything else?
|
I'll attach some block diagrams of the various setups when I get home tonight.? Can't draw stuff up in any reasonable amount of time on my phone here at work.? I'll show the 50 ohm 10 MHz distribution method we have at work, the method I use in my home lab, and the 75 ohm baseband analog video distribution method.? ? ?Jim
On Thu, Jun 20, 2024 at 9:16 AM, Jinxie via groups.io <paul666@...> wrote: So having read all the replies so far - for which many thanks - is it the same reason that all the PLLs within my 8566B spectrum analyzer are not only locked themselves but inter-locked with each other as well? So to ensure everything's in phase with everything else?
|
Roy, with one difference... Time and volt nuts chase quantifiable metrics with measureable and definable absolute values using standardized nomenclature and accepted methodology. Audiophools measure nebulous metrics with values such as "warmth", "brightness", and my favorite, "presence", while throwing out the factual formulas and ideas like ohms law and human ear response because they love brand X better than Y. Also, in the audiophool world, quality is always directly proportional to the cost if the device. ?
|
Jinxie
Jack
Dave McGuire
RubenRubio
Froggie the Gremlin