Re: [time-nuts] minimalist sine to square
An npn emitter follower with collector connected to the 10V supply will produce a 4.2V swing in a 50 ohm load. However it only works well if reflections from the load are small. Otherwise, assuming that you want a 0-5V signal at the 50 ohm load its possible to build a back matched driver using 3 BJTs that will switch 200mA into a 25 ohm load with transition times around 5ns or so. Faster transition times require using transistors with ft's somewhat greater than 300MHz. However the 10V rail current increases to 200mA when the output is 5V at the 50 ohm load. Bruce > > On 21 January 2018 at 08:43 Jerry Hancock wrote: > > Tom might have started this as I was playing around with PICDIV and had > asked him the best conditioning circuit. Turned out I had all the parts to > copy the TADD-2 including the mini circuits transformer so that’s what I did. > It works well, pretty sensitive, etc. I’ve also used the bias trick with a > TTL or CMOS buffer when I needed to convert SPIDF signals to baseband for > driving an optical connection. > > Now that I had the input conditioned, I need to drive a 50ohm load with > the signal coming from the PICDIV. Can someone point me at a circuit using > transistors and 10V if possible? > > I am trying to duplicate one channel of the TADD2 so I can bring 10Mhz > down to 10Khz. > > Thanks > > Jerry > > > > > > On Jan 20, 2018, at 11:21 AM, Poul-Henning Kamp > > wrote: > > > > > > In message , > > jimlux writes: > > > > > > > > > > > > > > > > > > I played with that, I used a small transformer to > > > > balance the signal > > > > and then into LVDS receiver through a voltage divider. > > > > Worked well, > > > > but I didn't measure the jitter, it was just for a > > > > micro-controller. > > > > > > > > > > > > > > You can also do it with capacitive dc block to one side, and > > > some > > > resistors - the ap notes describe it. The receivers are a > > > fairly high Z > > > input, so you pick the voltage divider resistors to make the > > > termination > > > resistance right for the incoming signal. > > > > > > > > > > Yes, but that doesn't give you galvanic isolation, which I think is > > almost > > mandatory unless it is a metrology situation. > > > > -- > > Poul-Henning Kamp | UNIX since Zilog Zeus 3.20 > > p...@freebsd.org | TCP/IP since RFC 956 > > FreeBSD committer | BSD since 4.3-tahoe > > Never attribute to malice what can adequately be explained by > > incompetence. > > > > ___ > > time-nuts mailing list -- time-nuts@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@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@febo.com To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts and follow the instructions there.
Re: [time-nuts] CSAC purchase
On 1/20/18 5:05 PM, Ronald Held wrote: I am thinking of buying a CSAC plus evaluation board. Eventually I might want to make it portable. Any suggestions including where to buy it? MicroSemi is the manufacturer - Find a distributor and order it. For instance, Digikey has the CSAC ($5312.50) and the eval board ($928.75) portable isn't a problem - just run it off batteries. It's pretty easy to hook up - power, 1pps in and out and 10 MHz out (they have versions that put out 5, 10.24, and 16.384 MHz too). A serial port to control the device The eval board has SMA connectors, a sub-d for the serial port, and comes with a wall wart to run it. Download the Microsemi CSAC UserGuide for more info you can fool with the disciplining algorithm, etc. It's a low power device compared to a OCXO (<120mW) They're a pretty nifty device, even if the price more than tripled in the last couple years. Jackson Labs makes some integrated systems using CSAC, I believe. ___ time-nuts mailing list -- time-nuts@febo.com To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts and follow the instructions there.
[time-nuts] CSAC purchase
I am thinking of buying a CSAC plus evaluation board. Eventually I might want to make it portable. Any suggestions including where to buy it? Ronald ___ time-nuts mailing list -- time-nuts@febo.com To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts and follow the instructions there.
[time-nuts] minimalist sine to square
If you have a TADD2-mini you can just replace the divider chip with the 10 KHz version. That doesn't get you a 10 V output, though. -- > I am trying to duplicate one channel of the TADD2 so I can bring 10Mhz down > to 10Khz. ___ time-nuts mailing list -- time-nuts@febo.com To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts and follow the instructions there.
[time-nuts] Time interval measurement vs dual mixer method
Luciano, I don't know of a commercial version. As you have seen as the two signals move away from close phase coincidence the system noise level will increase. This is because the common offset oscillator noise will only cancel when the phases are closely matching. This is mainly important for the lower Tau (like 1 to 10 seconds). I will normally adjust the phase of my reference or DUT so that myTIC is reading 0.0XX (close to phase match) and very slowly rising. (with my setup most DUT will age downward). Then I start logging the data. A typical run with a 1 week aged quartz will show the TIC count slowly increase (and may wrap) and then as the aging continues the count will reverse and eventually go "below" zero and wrap. A phase plot will show a nice parabolic looking curve going up and then as the aging crosses it will curve down. If there are wraps the plotter program can take them out. Any long plot of Quartz WILL wrap! Now if you plot the AD you will see a normal type plot for a good Quartz. If the TIC count increased well past coincidence only the Higher Taus will be accurate. For the lower Tau you need to cut off the data past where it climbed too high. I typically will setup as described and run a short log of say 5 minutes. I might only keep the first 100 Seconds and plot the AD against that. This will give you an accurate plot for the lower Tau. These two plots can then be combined to give you the whole range. The attached plot of an FE405B illustrates this. The Red plot is against a very good FTS 1200 (2X10-13th at 1 thru just past 10 Sec) The blue continuation is against a very good HP 5065A (1.5X10-13th at 100 Sec) So if the lower Tau are important only use logged data that are in near phase match! Hope this helps! Cheers, Corby___ time-nuts mailing list -- time-nuts@febo.com To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts and follow the instructions there.
Re: [time-nuts] minimalist sine to square
Tom might have started this as I was playing around with PICDIV and had asked him the best conditioning circuit. Turned out I had all the parts to copy the TADD-2 including the mini circuits transformer so that’s what I did. It works well, pretty sensitive, etc. I’ve also used the bias trick with a TTL or CMOS buffer when I needed to convert SPIDF signals to baseband for driving an optical connection. Now that I had the input conditioned, I need to drive a 50ohm load with the signal coming from the PICDIV. Can someone point me at a circuit using transistors and 10V if possible? I am trying to duplicate one channel of the TADD2 so I can bring 10Mhz down to 10Khz. Thanks Jerry > On Jan 20, 2018, at 11:21 AM, Poul-Henning Kamp wrote: > > > In message , jimlux > writes: > >>> I played with that, I used a small transformer to balance the signal >>> and then into LVDS receiver through a voltage divider. Worked well, >>> but I didn't measure the jitter, it was just for a micro-controller. >> >> You can also do it with capacitive dc block to one side, and some >> resistors - the ap notes describe it. The receivers are a fairly high Z >> input, so you pick the voltage divider resistors to make the termination >> resistance right for the incoming signal. > > Yes, but that doesn't give you galvanic isolation, which I think is almost > mandatory unless it is a metrology situation. > > -- > Poul-Henning Kamp | UNIX since Zilog Zeus 3.20 > p...@freebsd.org | TCP/IP since RFC 956 > FreeBSD committer | BSD since 4.3-tahoe > Never attribute to malice what can adequately be explained by incompetence. > ___ > time-nuts mailing list -- time-nuts@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@febo.com To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts and follow the instructions there.
Re: [time-nuts] minimalist sine to square
In message , jimlux writes: >> I played with that, I used a small transformer to balance the signal >> and then into LVDS receiver through a voltage divider. Worked well, >> but I didn't measure the jitter, it was just for a micro-controller. > >You can also do it with capacitive dc block to one side, and some >resistors - the ap notes describe it. The receivers are a fairly high Z >input, so you pick the voltage divider resistors to make the termination >resistance right for the incoming signal. Yes, but that doesn't give you galvanic isolation, which I think is almost mandatory unless it is a metrology situation. -- Poul-Henning Kamp | UNIX since Zilog Zeus 3.20 p...@freebsd.org | TCP/IP since RFC 956 FreeBSD committer | BSD since 4.3-tahoe Never attribute to malice what can adequately be explained by incompetence. ___ time-nuts mailing list -- time-nuts@febo.com To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts and follow the instructions there.
Re: [time-nuts] trimble Thunderbolt, how to get 25 or 27 mHz from it??
I am using Leo Bodnar's GPSDO that is based on the 5328 and what looks to be a good TCXO as an external reference clock for a Perseus SDR. Using that, I measured phase noise and Allan Deviation of the best sources I have available and found the phase noise and Allan Deviation to be close to what has been measured on more expensive equipment. From a quick look at the 5351's data sheet it's jitter appears to be on the order of 100 times more than the 5328, and would result in unacceptable phase noise for my application. I would expect that the 5328 with the Thunderbolt as an input source would be even better than Leo's GPSDO. Note that the two outputs from the 5328 can't be two arbitrary frequencies, and I have found that different values of the various dividers can produce the same output frequency by varying levels of close in spurs. See my sites listed below for all the details: https://sites.google.com/site/perseusmods/ https://sites.google.com/site/perseusmods/home/performance https://sites.google.com/site/spectrumlabtesting/ 73, Mark W7MLG On Sat, Jan 20, 2018 at 6:59 AM, John Ackermann N8UR wrote: > Hi Mark -- > > Thanks! To clarify, when you say you've found "it" acceptable, you're > referring to the 5328? > > What caught my eye about the 5351 was the three (or eight) outputs. My > idea was to build a board that would provide independent LO oscillators for > multiple VHF/UHF transverters. It looks like the 5328 has two outputs, > which is still useful but would require putting two or three of them on the > board. Which isn't the end of the world. > > Thanks, > John > > > > On 1/19/2018 8:56 AM, Mark Goldberg wrote: > >> On Jan 19, 2018 6:01 AM, "John Ackermann N8UR" wrote: >> >> Sorry to hijack the thread, but the Si5351 looks interesting for another >>> project I'm working on. I know it specifies "low jitter" but has anyone >>> looked at the phase noise? Is it usable for RF applications? >>> >>> >> >> Datasheet states jitter in the 40-70 ps range. That is not very good. If >> you want something suitable for RF applications, look at the Si5328. It >> has >> jitter on the order of 300fs. >> >> I believe Leo Bodnar uses it in his GPSDOs and I can confirm the phase >> noise and Allan Deviation others have found and it is suitable for me for >> RF applications. It is not nearly as good as the Thunderbolt though. It >> will generate almost any frequency you want. >> >> Mark >> >> >> >>> ___ >> time-nuts mailing list -- time-nuts@febo.com >> To unsubscribe, go to https://www.febo.com/cgi-bin/m >> ailman/listinfo/time-nuts >> and follow the instructions there. >> >> ___ > time-nuts mailing list -- time-nuts@febo.com > To unsubscribe, go to https://www.febo.com/cgi-bin/m > ailman/listinfo/time-nuts > and follow the instructions there. > ___ time-nuts mailing list -- time-nuts@febo.com To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts and follow the instructions there.
Re: [time-nuts] trimble Thunderbolt, how to get 25 or 27 mHz from it??
Hi Mark -- Thanks! To clarify, when you say you've found "it" acceptable, you're referring to the 5328? What caught my eye about the 5351 was the three (or eight) outputs. My idea was to build a board that would provide independent LO oscillators for multiple VHF/UHF transverters. It looks like the 5328 has two outputs, which is still useful but would require putting two or three of them on the board. Which isn't the end of the world. Thanks, John On 1/19/2018 8:56 AM, Mark Goldberg wrote: On Jan 19, 2018 6:01 AM, "John Ackermann N8UR" wrote: Sorry to hijack the thread, but the Si5351 looks interesting for another project I'm working on. I know it specifies "low jitter" but has anyone looked at the phase noise? Is it usable for RF applications? Datasheet states jitter in the 40-70 ps range. That is not very good. If you want something suitable for RF applications, look at the Si5328. It has jitter on the order of 300fs. I believe Leo Bodnar uses it in his GPSDOs and I can confirm the phase noise and Allan Deviation others have found and it is suitable for me for RF applications. It is not nearly as good as the Thunderbolt though. It will generate almost any frequency you want. Mark ___ time-nuts mailing list -- time-nuts@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@febo.com To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts and follow the instructions there.
Re: [time-nuts] Time interval measurement vs dual mixer method
Hi The quality of the offset signal source matters in a DMTD system. If you are using a “noisy” source you will have trouble with the data. You also will have trouble if the offset signal is correlated in some way to one of the DUT signals. Bob > On Jan 20, 2018, at 3:43 AM, tim...@timeok.it wrote: > > > Hi Corby, > I did several tests to verify the system noise floor using the same source > for both input channels. I have seen that the result varies a lot as the > difference in time between the two channels varies. > Moreover, if I do two successive tests without changing the parameters and > the duts, the result can not be superimposed. > It may be that I make a mistake or do not take any parameter into account. > I also wonder if there is any commercial realization of the DMTD system. > Cheers, > Luciano > > > Da "time-nuts" time-nuts-boun...@febo.com > A time-nuts@febo.com > Cc > Data Thu, 18 Jan 2018 11:13:54 -0800 > Oggetto [time-nuts] Time interval measurement vs dual mixer method > Luciano, > > Not sure what you mean by "two measures made on the same DUT never > coincide". > > The plotter program takes out wraps which since your phase gain is so > large are due to happen. > > For lower Tau measurements just adjust the phase for close to equal and > run shorter plots. > > These can be spliced into the longer Tau plot. > > I routinely get repeatable plots on the same DUT with my DMTD setup. > > Careful setup and understanding what you can and cannot accomplish is the > key. > > Cheers, > > Corby > ___ > time-nuts mailing list -- time-nuts@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@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@febo.com To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts and follow the instructions there.
Re: [time-nuts] Time interval measurement vs dual mixer method
Hi Corby, I did several tests to verify the system noise floor using the same source for both input channels. I have seen that the result varies a lot as the difference in time between the two channels varies. Moreover, if I do two successive tests without changing the parameters and the duts, the result can not be superimposed. It may be that I make a mistake or do not take any parameter into account. I also wonder if there is any commercial realization of the DMTD system. Cheers, Luciano Da "time-nuts" time-nuts-boun...@febo.com A time-nuts@febo.com Cc Data Thu, 18 Jan 2018 11:13:54 -0800 Oggetto [time-nuts] Time interval measurement vs dual mixer method Luciano, Not sure what you mean by "two measures made on the same DUT never coincide". The plotter program takes out wraps which since your phase gain is so large are due to happen. For lower Tau measurements just adjust the phase for close to equal and run shorter plots. These can be spliced into the longer Tau plot. I routinely get repeatable plots on the same DUT with my DMTD setup. Careful setup and understanding what you can and cannot accomplish is the key. Cheers, Corby ___ time-nuts mailing list -- time-nuts@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@febo.com To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts and follow the instructions there.