Author Topic: OCXO's and friends  (Read 7363 times)

0 Members and 1 Guest are viewing this topic.

Offline Lemonizer

  • Regular Contributor
  • *
  • Posts: 55
  • Country: fr
Re: OCXO's and friends
« Reply #75 on: April 09, 2020, 01:56:14 pm »
Yes I did read your answer, you explained how to do it but I'm asking how to do to amplify the signal, as it is indeed too low ! :)
Or if it is possible at all ! I know there's other ways to measure the internal HP OCXO's, I just wanted to know, if the amplitude of the distribution amp output is too low, if there's a workaround.
 

Offline Dr. Frank

  • Super Contributor
  • ***
  • Posts: 2053
  • Country: de
Re: OCXO's and friends
« Reply #76 on: April 09, 2020, 04:15:26 pm »
 well then , did you test whether the level is really too low?
What's the direct output level from your GPSDO, and the one from your new counter ?
Maybe that's sufficient, just to check..
 

Offline Lemonizer

  • Regular Contributor
  • *
  • Posts: 55
  • Country: fr
Re: OCXO's and friends
« Reply #77 on: April 09, 2020, 04:57:44 pm »
Ok so I measured it with my tek 2245, from the distribution amp I get 6V p-p, from the ouput of the FA-2 I get 2.24V p-p, and 3,56V p-p from the GPSDO. I don't really understand ! Was it too high ? Is there a limit ? I can't find much infos in the manual :/
 

Offline Dr. Frank

  • Super Contributor
  • ***
  • Posts: 2053
  • Country: de
Re: OCXO's and friends
« Reply #78 on: April 09, 2020, 05:57:13 pm »
The FA_2 is a bit low with 0.8Vrms, but the other level should directly drive the EXT REF IN of your 5345A.
Again no, all these levels are not too high at all.
Maybe you actuate the int/ext switch several times, as this may be oxydized.
You can also test with other sub harmonic frequencies of 10MHz.

Frank
 

Offline Lemonizer

  • Regular Contributor
  • *
  • Posts: 55
  • Country: fr
Re: OCXO's and friends
« Reply #79 on: April 09, 2020, 06:50:50 pm »
Well I only tried (yesterday) with the distribution amp, which is the highest, that's why I find it strange, do you think it's still not enough ?

I'm still reading on harmonics. I don't fully understand everything, can you give me a one or two line explanation about your last suggestion please ? :)

And yes, the switch felt a bit wonky. But I could clearly see two states, either the normal internal ref, or that 'all segment lighted up' state. Maybe I need to give it more time ? Usually, when there's something that I don't expect, I quickly shut it off and research more, to be sure that I'm not damaging stuff !
 

Offline Dr. Frank

  • Super Contributor
  • ***
  • Posts: 2053
  • Country: de
Re: OCXO's and friends
« Reply #80 on: April 09, 2020, 08:17:36 pm »
Well I only tried (yesterday) with the distribution amp, which is the highest, that's why I find it strange, do you think it's still not enough ?

I'm still reading on harmonics. I don't fully understand everything, can you give me a one or two line explanation about your last suggestion please ? :)

And yes, the switch felt a bit wonky. But I could clearly see two states, either the normal internal ref, or that 'all segment lighted up' state. Maybe I need to give it more time ? Usually, when there's something that I don't expect, I quickly shut it off and research more, to be sure that I'm not damaging stuff !
Ok. first thing, the 5345A ext ref input accepts not only 10MHz, but sub harmonics like 1MHz, 2, 2.5, and 5MHz. The PLL of the internal OCXO can lock in to all of these frequencies.
Please allow your instrument more time to lock-in on the external signal, obviously the PLL needs some time, but I have no clue how long, as this is not described in a detailed manner in the manual. To shut off when something odd happens, in this case is not necessary, as you can't damage anything.

You're seeing a working state, obviously when on internal, the display is updated and everything is running well.
The other state with all segments lit up (that is the LAMP TEST mode) is a non-functional state, very obviously, so the OCXO can not properly lock to the external frequency, either because the PLL is not working properly, or because the external signal is too weak, or simply is not connected properly via switch S9. The detection and behavior of the display is described in paragraphs 4-127 .. 4-131.

My 5370B has some similarities with the 5345A, they both share the same input front end, for example.
The external switch on my unit does not work properly any more, so I have to push a bit harder to the external position, until the ext. 10MHz is fed through.
That might be similar on your unit, so also try to wiggle this switch.

Frank 

Edit: I just successfully tested the TALK ONLY mode on my 5370B, this probably works on the 5345A identically. Flip the switch on the rear to TALK ONLY.
On the counter, set up the measurement mode like frequency, T.I., the gate time/ resolution.
Set the parameters in TimeLab accordingly, then press MONITOR, and you may see the incoming samples like in the picture.
If you press START, TimeLab will acquire the samples, and inside MEASUREMENT you can chose which mode you want to get analysed, i.e. T.I. ADEV, or frequency difference
 
« Last Edit: April 10, 2020, 02:37:47 pm by Dr. Frank »
 
The following users thanked this post: edavid

Offline Lemonizer

  • Regular Contributor
  • *
  • Posts: 55
  • Country: fr
Re: OCXO's and friends
« Reply #81 on: April 09, 2020, 08:32:19 pm »
Ok ! Thanks, you gave me confidence about it, tomorrow I'll re-try and tells here what happens ! I can indeed try other sub-harmonic frequencies with my siglent 1032x, with the GPSDO as its external ref.

And yes I will give it more time and play a bit with the switch, maybe it's simply that, it needs more time and the switch needs to be pushed ! :)

The 5370B is a really nice instrument !  8)

By being 'phase locked' to the external ref, does it means that it acquire it's stability ? If I understand correctly, the offset will still be there, I will still need to adjust it, right ?
 

Offline Dr. Frank

  • Super Contributor
  • ***
  • Posts: 2053
  • Country: de
Re: OCXO's and friends
« Reply #82 on: April 09, 2020, 08:37:53 pm »
Well, on the 5370B, the clock is simply switched between the internal and the external signals.
On your unit, they really phase-lock the internal OCXO to the external signal, so that the internal OCXO takes over the stability, but maybe keeps the good phase noise of the 10811 (depending on the realisation of this PLL).

You need to adjust this offset only without an external clock, of course.

If the 10811 locks in to an external clock, like a GPSDO, it also takes over its accuracy, and you will not see an offset any more.

Frank
 

Offline texaspyro

  • Super Contributor
  • ***
  • Posts: 1406
Re: OCXO's and friends
« Reply #83 on: April 09, 2020, 10:58:26 pm »
Well, on the 5370B, the clock is simply switched between the internal and the external signals.

Also, the selected 10 MHz clock drives the processor.   I once bought an auction lot of "dead" 5370's.  Their only problem was they were set for external clock.
 

Offline ocw

  • Regular Contributor
  • *
  • Posts: 240
  • Country: us
Re: OCXO's and friends
« Reply #84 on: April 11, 2020, 04:09:52 pm »
I have three GPSDO's
  • One built into my Anritsu MT8222A analyser
  • Trimble GPSDO that I just received with three sine wave outputs and three square wave outputs and two 1PPS outputs
  • What might be a BG7TBL unit that I have had for several years with one sine wave and one 1PPS output
Since I like to calibrate my spectrum analyzers and other equipment close to 1 GHz, I have a sine to square wave converter to use with that older GPSDO.  I thought that the Trimble unit would make that both easier to do and permit connection to multiple equipment.

I thought that I would cross compare the three in the following fashion:
  • Connect two Trimble GPSDO square wave outputs to the SA input of my Anritsu MT8222A with its internal GPSDO active and to a Rohde & Schwarz ETL broadcast TV/Radio analyzer's SA input
  • Connect the BG7TBL GPSDO to serve as an external 10 MHz reference for the ETL
I measured the frequency of the 10 MHz square waves 99th harmonic at 990 MHz through either a 470 or 700 MHz high pass filter.  Attached are views of the measurements of the MT8222A and the ETL.  The ETL had a very stable reading 0.6 Hz low.  That translates into about 0.006 Hz low for 10 MHz.  The MT8222A attached display shows a nearly perfect frequency.  However, while it was centered there its reading drifted about +/-2 Hz (0.02 Hz at 10 MHz).

Both analyzers were adjusted for their minimum RBW, which was 1 Hz for the MT8222A and 10 Hz for the ETL.  Because of those differences the span of the MT8222A was set for 10 Hz as compared to 100 Hz for the ETL.  That made it easier to see any drift on the MT8222A and hurt the accuracy ability of the ETL.

1 Hz accuracy at 1 GHz meets my current accuracy needs.  And having a 1 Hz/cm display on my MT8222A is nice.  That makes 0.1 Hz resolution easy at 990 MHz.
 

Offline ocw

  • Regular Contributor
  • *
  • Posts: 240
  • Country: us
Re: OCXO's and friends
« Reply #85 on: May 09, 2021, 03:28:35 pm »
Having made too many frequency reference oscillators, I decided to take it to the next level.  Look at the first attached picture—ten different distribution amplifiers, each having eight outputs.  If eight outputs are not adequate, the feedthroughs can be used to create a single eighty output DA.

The DA ten-pack is a two-rack unit Imagine-Leitch FR-684.  I am using it with ten UDA-683 DA boards installed.  That uses an AD8010ARZ-16 current feedback op amp at its output.  The boards have a typical measured frequency response up to 180 MHz and are made for unbalanced inputs.

I also tested the VDA-683 and VEA-683 DA boards.  They both have differential inputs.  The VDA-683 uses a MAX4145 input a with board response up to 146 MHz.  THE VEA-683 uses a MAX4144 input with a board response up to 68 MHz.  Due to the lower frequency response, and my differential input need is more at higher frequencies, I have stayed with the UDA boards for now.  However, a quick board change can alter what is installed.  A look at the closed and open front door is the second attachment.

Due to the many frequency standard oscillators which I have built lately, I used one with each FR-684 board.  I attached the circuit boards of the smaller oscillators to the UDA-683 boards.  Fine frequency adjustment trimmer controls can be adjusted with the front door down.  The UDA-683’s have fine level adjustment controls already on them along with an equalization control.  The EQ control can be used to clean up a square wave oscillator output.

While my FR-684 has two power supplies for both redundancy and double the current capability, powering all of the oscillators having heaters would be taxing them.  Plus, I did not want all that heat inside of the FR-684.  And with some of the oscillators already mounted in metal boxes, there was not an easy way of mounting those inside.  So, I have several oscillators mounted externally.  The next attached picture shows them sitting on top of the FR-684.  Remember, there are five more oscillators which have been attached to UDA-683 boards.

The next attachment shows the outputs of all the DA’s.  Because of the limitations of my four-channel scope, I have combined three scope views.  That permits my changing the time base for each, so please note those differences.

The UDA-683 frequency response turns some oscillator square wave outputs to something closer to a sinewave on the higher frequencies.  I am not concerned with that for those oscillators.  The Extron MDA DA’s which use OPA2691’s generally have response up to 400 MHz.  I have those and other DA’s to address the higher frequencies.  I am modifying a THS3202EVM for DA use.  It has measured frequency response up to 810 MHz—see the attachment.  I will be using that with a 100 MHz square wave oscillator.  Still greater frequency response would be nice and I have yet another option.

I also have a CDCLVP1212EVM Clock DA.  It has the choice of two separate differential inputs while the EVM board has four differential (or eight unbalanced) outputs.  I have one of those inputs changed to unbalanced so that both options are available.  Points are available on the EVM board to increase those outputs to 12 differential (or 24 unbalanced) outputs.  I have measured the frequency response of the EVM to nearly 2600 MHz—see the attachment.  My 0.5 and 1 GHz LVDS oscillators to it are obviously below what TI says is “a maximum clock frequency up to 2 GHz.”

Since TI has upgraded the CDCLVP1212 to a similar 1216 having 16 differential outputs, the CDCLVP1212EVM is no longer made (although a few are still available).  The 1216EVM and other CDCLVP****EVM’s are also available.
 

Offline hgl

  • Contributor
  • Posts: 27
  • Country: de
Re: OCXO's and friends
« Reply #86 on: May 15, 2021, 04:45:50 pm »
I'm wondering what people think about that FA-2 counter, compared to high-end counters like Agilent, Pendulum's ones, just accuracy wise (once used with a well stabilized GPSDO, with a good signal ofc). I mean, it seems to be pretty nice, it got a huge resolution, and its mostly the external reference that does all the work accuracy wise, right ?
I thought it would be nice to later replace its OCXO with a better one (once I got my hands on a few that were measured during days, prior to that) and power it with a linear supply, instead of a wall-wart. Am I right ?

The FA2 isn't that bad. The short-term stability is limited by the noise of the OCXO and the CPLD interpolator and the long-term by the drift. With an external reference, you get additional noise from the input stage. If you don't have a stable and low-noise reference, it might be better to lead the control voltage to the outside and regulate the internal OCXO via GPS.

Measurements of my rubidium standards with the FA2.



 

Offline FriedLogic

  • Regular Contributor
  • *
  • Posts: 108
  • Country: gb
Re: OCXO's and friends
« Reply #87 on: May 15, 2021, 05:50:10 pm »
With an external reference, you get additional noise from the input stage.

Is that actually the case? I just don't recall seeing a specific test of that, but I may have missed it.
It should be easy to test if someone was replacing the oscillator anyway.
 


Share me

Digg  Facebook  SlashDot  Delicious  Technorati  Twitter  Google  Yahoo
Smf