I played around with this today, recording 5 minute runs: freerunning, P/10 gain, P gain, 10x P gain (where P is some unknown value...) I then added either a 1M/100n or 500k/100n integrator which really only shows in the TDEV-plot, not in the PN-plot. https://photos.app.goo.gl/V9xL5MkVsL6RLRXJ7
maybe "P/10, with integrator" is the best one so far? Seems any increase in P-gain always creates an ugly PN-bump? Increasing the I-gain is maybe the next thing to try, either with a shorter RC-constant, or by increasing the I-weight in the summing op-amp.. Didn't have time to read the references much - so this is mostly trial and error for now... Anders On Sun, Feb 3, 2019 at 7:01 PM Tom Van Baak <[email protected]> wrote: > Bert, > > That looks like a very nice board. Thanks for posting. Let us know how to > obtain one; as board, kit or product. > > /tvb > > N.B. The original posting was aol / html. It has been reformatted below to > work with the time-nuts server. > > > ----- Original Message ----- > From: ew > To: [email protected] > Cc: [email protected] > Sent: Sunday, February 03, 2019 5:26 AM > Subject: Re: [time-nuts] OCXO PLL gains? > > There is a need for clean up loops across time-nuts applications. Right > now Corby has one to test clean up for a HP 5065A and a HP 5061B. How ever > there is room for improvement. > > In my simple mind there may be an answer for most applications and enough > brains among time nuts to define and develop a loop. Many of the telecom > GPSDO's use frequency to correct for time, FE5680 a candidate even other > Rb's depending what you have and try to accomplish. Morion MV89A, HP10811, > Austron 1150, FRK, M100, FTS1200, M1000, OSA8600 all candidates for the > other side. > > I recently did two OCXO boards one based on many time nuts inputs. Did not > find some one to pick up the designs and make it available. I than decided > to post Gerber files and based on off list questions and comments boards > have been made. > > I propose something similar. Our experts come up with an answer, I do a > board and add it to my next order, projects are waiting to be ordered after > the Chinese holiday. As always Corby will along with Juerg get the first > boards and once passed Corby's test I will release information. I do make > layout mistakes and I do not want mistakes out there. Cause confusion with > corrected boards. > > I keep looking at Microsemi 4145C datasheet and I am convinced with the > time nuts brains out there one could get close. There is a digital loop > between an the HP5071A and a OSA8600 and an analog loop between 8600 and a > Wenzel Super Low Noise OCXO. Not now, walk before we run, but the brains > and parts are out there. > > Bert Kehren > > Attached a picture of our first board, we now have later versions, one > based on Wenzel 600 sec. design and one similar to the dither filter used > in the HP GPSDO. We are not experts. > > _______________________________________________ > time-nuts mailing list -- [email protected] > To unsubscribe, go to > http://lists.febo.com/mailman/listinfo/time-nuts_lists.febo.com > and follow the instructions there. > _______________________________________________ time-nuts mailing list -- [email protected] To unsubscribe, go to http://lists.febo.com/mailman/listinfo/time-nuts_lists.febo.com and follow the instructions there.
