[time-nuts] MT3339 chipset 10ns jitter?

Azelio Boriani azelio.boriani at screen.it
Mon Nov 26 22:45:34 UTC 2012


OK, try that setup, measuring the period every 2s. You should end up with a
file (use the GPIB) with the PPS periods and so you can apply the usual
statistic computations. You should, at that point, be able confirm or not
the strange datasheet reporting 1uS first, then 100nS and finally 10nS.

On Mon, Nov 26, 2012 at 11:15 PM, Fabio Eboli <fabioeb at quipo.it> wrote:

> Azelio,
>
> I dont have a GPSDO and this will be may first attempt
> to make one.
> I have a pair of working FE5680 with their 10MHz and PPS
> outputs, at 10s integration time the counter measures them
> at 10MHz+ 15mHz and 10MHz +45mHz. The counter oscillator
> is the opt 04E "high stability" OCXO.
> So if I understand correctly my option could be the
> "Second choice": the FE5680A 10MHz output as ext timebase
> for the counter, and measure the PPS under test period.
> Using 1s integrqation I will have 1nS resolution, with
> one result per 2s.
> Using 10s integraqtion time the counter returns 10digits,
> showing 100nS resolution. I will collect one reading every
> 20s, a slow process. Will this setup be enough?
>
> Fabio
>
> Azelio Boriani <azelio.boriani at screen.it> ha scritto:
>
>  You need a reference to test the PPS output from a GPS receiver. That is,
>> you need an already running GPSDO so that its stable PPS output can act as
>> the reference for your test. Then you need a time interval counter and use
>> the GPS under test PPS as start and the reference PPS as stop (or the
>> other
>> way, it is irrelevant which is the start and the stop). It helps if the
>> GPSDO's PPS can be offset slightly (say, 200nS, 300nS, ...) so that the
>> PPS
>> under test doesn't "cross" the reference. Second choice: you can use the
>> counter period measurement, feeding the counter with an external stable
>> reference. Last (really the last to use): the counter has a very good
>> internal OCXO.
>>
>> On Mon, Nov 26, 2012 at 10:29 PM, Fabio Eboli <fabioeb at quipo.it> wrote:
>>
>>  Hello Jacques, and the List.
>>>
>>> Thank you for sharing these docs.
>>> Indeed the PPS specs of the MT3339 are confusing.
>>> Maybe measurement on a living specimen can help?
>>>
>>> I have a R-D 1992 counter, and a GPIB interface.
>>> When I will have the GPS I will be able to log
>>> some data on the PC, although I never tried to use
>>> the Racal Counter remotely I will try.
>>> Also I hope somebody can suggest the best way to test
>>> the PPS from the gps using the Racal 1992, I'm not a
>>> time nut, so I could make gross errors.
>>> I was thinking simply to measure and log the PPS
>>> connected to input A with the counter set to
>>> "Period A", and 10s integration time to gain
>>> more resolution, all this for some (many?) hours.
>>>
>>> Fabio.
>>>
>>> Jacques Tiete <jacques at tiete.org> ha scritto:
>>>
>>>
>>>  Hello TNutters,
>>>
>>>>
>>>> maybe new thread?
>>>>
>>>> you can have a look at two specsheets of the MT3339 chipset on my
>>>> FTP-site
>>>> 1) http://nixie.ramdac.be/TNutstuff/MT_IT530_Data_Sheet_rev_1_0.pdf
>>>>                 on page 8 "PPS output 100 ms high pulse, rising edge
>>>> +/-1
>>>> μs @ full second GPS epoch..." so it states +/-1 μs
>>>> 2) http://nixie.ramdac.be/TNutstuff/gtop module application note -a00
>>>> _mt3339 series_.pdf
>>>>                 on page 32 "The accuracy level of 1PPS for GlobalTop GPS
>>>> module is within 100ns (same for majority of MTK GPS modules).
>>>>                                         The reference design below can
>>>> help improving this accuracy to within 80ns."
>>>>                 but on page 14 "1PPS signal is an output pulse signal
>>>> used for timing application, its electronic characteristic are listed
>>>> below:
>>>>                                                 1PPS signal and its
>>>> pulse
>>>> width with 100ms duration
>>>>                                                 Low Voltage level:
>>>> 0~0.4V
>>>>                                                 High Voltage level:
>>>> 2.8~3.1V
>>>>                                                 Duration: 100ms
>>>> (Firmware
>>>> customization for duration is available)
>>>>                                                 Period: 1s Accuracy
>>>> (jitter):10ns"
>>>>
>>>> So what shall we believe?
>>>> The rest of the PDF's make a good read.
>>>>
>>>> 73's
>>>>
>>>> Jacques
>>>>
>>>>
>>>>
>>>>
>>>> Message: 4
>>>> Date: Fri, 23 Nov 2012 16:44:27 +0100
>>>> From: Jacques Tiete <jacques at tiete.org>
>>>> To: "time-nuts at febo.com" <time-nuts at febo.com>
>>>> Subject: Re: [time-nuts] time-nuts Digest, Vol 100, Issue 95
>>>> Message-ID: <7C466C42-C05A-46E2-B3C4-793672874AC8 at tiete.org>
>>>> Content-Type: text/plain;       charset=us-ascii
>>>>
>>>> About:
>>>> {
>>>> Date: Thu, 22 Nov 2012 22:26:24 +0100
>>>> From: Fabio Eboli <FabioEb at quipo.it>
>>>> To: time-nuts at febo.com
>>>> Subject: [time-nuts] MT3339 chipset in Globaltop's FGPMMOPA6H
>>>> Message-ID: <20121122222624.71854hb9oslingcg at webmail.micso.net>
>>>> Content-Type: text/plain; charset=ISO-8859-1; DelSp="Yes";
>>>>   format="flowed"
>>>>
>>>> Hello, while asking info about the FE5682,
>>>> I realized that for another project
>>>> (model plane data logging) I will buy
>>>> some cheap gps modules built around an
>>>> MT3339 chipset.
>>>> In my mind somewhere I stored the fact
>>>> that in the datasheet of these module
>>>> is mentioned the pps output, so I checked
>>>> and the manufacturer is claiming that it
>>>> has "High accuracy 1-PPS timing support
>>>> for Timing Applications (10ns jitter)"
>>>>
>>>> This is the module I'm talking about:
>>>> http://www.mirifica.it/store/gps/342-globaltop-fgpmmopa6h-pa6h.html
>>>>
>>>>
>>>> http://www.alcom.be/binarydata.aspx?type=doc/GTop-FGPMMOPA6H-Datasheet-V0A.pdf
>>>>
>>>> Has anybody used this chipset or similar for GPSDO?
>>>> I have no experience with discipled oscillators
>>>> but if there is any hope that the pps out
>>>> of these modules is good for the task maybe
>>>> it's time for me to start reading more about
>>>> the subject.
>>>> The final goal will be to check the timing
>>>> of the Rb I have around.
>>>>
>>>> Thanks,
>>>> Fabio.
>>>> }
>>>>
>>>> Hello TNutters,
>>>>
>>>> I did some test with this module from Adafruit (
>>>> http://www.adafruit.com/products/746) and it seems not a bad unit.
>>>> Tested it together
>>>> with my TBolt and the PPS seems quite stable most of the time, I did not
>>>> do any logging and math on the result,
>>>> just letting the TBolt trigger the scope and apart from the cable offset
>>>> I guess 80% of the time the module was
>>>> within 10 to 15 ns.
>>>> However the specs are not very clear and it is a typical chip for mobile
>>>> use in eg. carnav stuff. I also saw some
>>>> other specs where the jitter was specified as 100ns. and even one with a
>>>> 1 microsec. so I'm a little confused.
>>>> As soon as I have some time I plan to connect this module to a
>>>> BeagleBone
>>>> and do some testing for an NTP server
>>>> I think the Beagle is better for this than the RaspBerry because of the
>>>> "real" ethernet port.
>>>>
>>>> Greetz,
>>>>
>>>> Jacques Tiete
>>>> 0499 99 83 78
>>>> jacques at tiete.org
>>>>
>>>> Op 23-nov.-2012 om 00:19 heeft time-nuts-request at febo.com het volgende
>>>> geschreven
>>>>
>>>>
>>>>
>>>>
>>>
>>> ----------------------------------------------------------------
>>> This message was sent using IMP, the Internet Messaging Program.
>>>
>>>
>>>
>>>
>>> _______________________________________________
>>> 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.
>>>
>>>  _______________________________________________
>> 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.
>>
>
>
> ----------------------------------------------------------------
> This message was sent using IMP, the Internet Messaging Program.
>
>
>
> _______________________________________________
> 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 mailing list