Re: [time-nuts] Effects of noise on EFC line? - Resolved
Hi The 10811 is very unusual in that the EFC is not bypassed. It’s still got a modulation bandwidth. The important part is that the EFC has an inherent “lowpass" due to FM to PM conversion. The EFC line on the 10811 should be bypassed to ground at the socket with at least a 1000 pf cap. Yes that’s not in the little manual. It’s the only way to knock out all of the RF tune effect on the EFC line. Of course if you are driving it with an op-amp that will terminate. Bob On Aug 7, 2014, at 10:08 PM, Bob Stewart wrote: > Hi Bob, > > I hadn't even considered a filter in the OCXO. This isn't a 10811, but > that's the OCXO I have a schematic of, so I'll assume that's the benchmark. > Following the EFC in, it looks like it goes to a 100K resistor and then tees > to the 100pf varicap and a 15pf to the xtal. Other caps are attached as > well, but it doesn't look like it's bypassed to ground anywhere along the EFC > line. I see that there's another 100K to a 6.4V reference with a 6.8uf cap. > So, that means that the EFC line ranges from +6.4 to -6.4? I haven't worked > out the time constant, but that wouldn't seem to apply for a 10MHz signal > riding on the EFC voltage. > > > Like I said, I don't have a 10811 on my GPSDO. It's my faithful Trimble > 34310-T. Still, I would imagine that they at least looked at HP's design. > > As to Hal's comment about probe pickup. I was careful to specify the X10 > position of the probe. In the X1 position there was a signal that wasn't > visible in X10, but should have been. So, I assumed that was some sort of > induced signal. I'm using a generic cheap Chinese probe available on ebay. > > > Bob > > > > ________________ > From: Bob Camp > To: Discussion of precise time and frequency measurement > Sent: Thursday, August 7, 2014 6:11 PM > Subject: Re: [time-nuts] Effects of noise on EFC line? - Resolved > > > Hi > > Your EFC line is probably bypassed internally to the OCXO. A 3db modulation > bandwidth beyond 1 KHz is unlikely. A modulation bandwidth below 100 Hz is > quite possible. > > Next thing to consider is that the EFC does FM on the OCXO. Phase noise is PM > modulation. FM is 1/Fmod relative to PM. If I go up a decade in frequency > with constant FM, my PM sideband will go down by 20 db. Yes that’s for small > modulation indexes. That’s very likely the case if we are dealing with noise. > > You can calculate exactly what PM sideband in dbc you will get from a 1 Hz > tone at 1 mV p-p on your EFC. From that you can pretty quickly work out what > this or that number of microvolts will do at this or that frequency. The > answer is normally that the noise you have from a reasonable regulator or > op-amp isn’t a big deal. > > Bob > ___ > 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] Effects of noise on EFC line? - Resolved
Hi Bob, I hadn't even considered a filter in the OCXO. This isn't a 10811, but that's the OCXO I have a schematic of, so I'll assume that's the benchmark. Following the EFC in, it looks like it goes to a 100K resistor and then tees to the 100pf varicap and a 15pf to the xtal. Other caps are attached as well, but it doesn't look like it's bypassed to ground anywhere along the EFC line. I see that there's another 100K to a 6.4V reference with a 6.8uf cap. So, that means that the EFC line ranges from +6.4 to -6.4? I haven't worked out the time constant, but that wouldn't seem to apply for a 10MHz signal riding on the EFC voltage. Like I said, I don't have a 10811 on my GPSDO. It's my faithful Trimble 34310-T. Still, I would imagine that they at least looked at HP's design. As to Hal's comment about probe pickup. I was careful to specify the X10 position of the probe. In the X1 position there was a signal that wasn't visible in X10, but should have been. So, I assumed that was some sort of induced signal. I'm using a generic cheap Chinese probe available on ebay. Bob From: Bob Camp To: Discussion of precise time and frequency measurement Sent: Thursday, August 7, 2014 6:11 PM Subject: Re: [time-nuts] Effects of noise on EFC line? - Resolved Hi Your EFC line is probably bypassed internally to the OCXO. A 3db modulation bandwidth beyond 1 KHz is unlikely. A modulation bandwidth below 100 Hz is quite possible. Next thing to consider is that the EFC does FM on the OCXO. Phase noise is PM modulation. FM is 1/Fmod relative to PM. If I go up a decade in frequency with constant FM, my PM sideband will go down by 20 db. Yes that’s for small modulation indexes. That’s very likely the case if we are dealing with noise. You can calculate exactly what PM sideband in dbc you will get from a 1 Hz tone at 1 mV p-p on your EFC. From that you can pretty quickly work out what this or that number of microvolts will do at this or that frequency. The answer is normally that the noise you have from a reasonable regulator or op-amp isn’t a big deal. Bob ___ 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] Effects of noise on EFC line? - Resolved
he is on the right truck, just look around in your testing environment, do you have shielded test set up, common ground for all the test gears ? On 8/7/2014 3:58 PM, Hal Murray wrote: b...@evoria.net said: So, I may throw another cap on it, but it seems to be clean down to what I can measure at the OCXO on my old Tek 455 with an X10 probe. Another thing to consider when chasing that sort of problem: How much are you picking up with your scope probe and/or its ground wire? The classic home brew better-probe is to use a chunk of 50 ohm coax feeding into scope with a 50 ohm terminator. (You get 2 of them if you cut a readily available connectorized cable in half.) Insert 950 ohms at the probe end to get higher input impedance at the cost of a 20:1 divider. http://www.signalintegrity.com/Pubs/straight/probes.htm It usually requires soldering at the DUT. For me, it also requires a mental shift. You have to think of your scope probe as a consumable rather than something that won't wear out if you take care of it. A chunk of coax will last a long time. Just cut off another 1/2 inch when the end gets too mangled. ___ 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] Effects of noise on EFC line? - Resolved
Hi Your EFC line is probably bypassed internally to the OCXO. A 3db modulation bandwidth beyond 1 KHz is unlikely. A modulation bandwidth below 100 Hz is quite possible. Next thing to consider is that the EFC does FM on the OCXO. Phase noise is PM modulation. FM is 1/Fmod relative to PM. If I go up a decade in frequency with constant FM, my PM sideband will go down by 20 db. Yes that’s for small modulation indexes. That’s very likely the case if we are dealing with noise. You can calculate exactly what PM sideband in dbc you will get from a 1 Hz tone at 1 mV p-p on your EFC. From that you can pretty quickly work out what this or that number of microvolts will do at this or that frequency. The answer is normally that the noise you have from a reasonable regulator or op-amp isn’t a big deal. Bob On Aug 7, 2014, at 6:58 PM, Hal Murray wrote: > > b...@evoria.net said: >> So, I may throw another cap on it, but it seems to be clean down to what I >> can measure at the OCXO on my old Tek 455 with an X10 probe. > > Another thing to consider when chasing that sort of problem: How much are you > picking up with your scope probe and/or its ground wire? > > The classic home brew better-probe is to use a chunk of 50 ohm coax feeding > into scope with a 50 ohm terminator. (You get 2 of them if you cut a readily > available connectorized cable in half.) Insert 950 ohms at the probe end to > get higher input impedance at the cost of a 20:1 divider. > http://www.signalintegrity.com/Pubs/straight/probes.htm > > It usually requires soldering at the DUT. For me, it also requires a mental > shift. You have to think of your scope probe as a consumable rather than > something that won't wear out if you take care of it. A chunk of coax will > last a long time. Just cut off another 1/2 inch when the end gets too > mangled. > > > -- > These are my opinions. I hate spam. > > > > ___ > 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] Effects of noise on EFC line? - Resolved
b...@evoria.net said: > So, I may throw another cap on it, but it seems to be clean down to what I > can measure at the OCXO on my old Tek 455 with an X10 probe. Another thing to consider when chasing that sort of problem: How much are you picking up with your scope probe and/or its ground wire? The classic home brew better-probe is to use a chunk of 50 ohm coax feeding into scope with a 50 ohm terminator. (You get 2 of them if you cut a readily available connectorized cable in half.) Insert 950 ohms at the probe end to get higher input impedance at the cost of a 20:1 divider. http://www.signalintegrity.com/Pubs/straight/probes.htm It usually requires soldering at the DUT. For me, it also requires a mental shift. You have to think of your scope probe as a consumable rather than something that won't wear out if you take care of it. A chunk of coax will last a long time. Just cut off another 1/2 inch when the end gets too mangled. -- These are my opinions. I hate spam. ___ 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] Effects of noise on EFC line? - Resolved
Thanks to everyone for their comments, and especially to Bill, WB6BNQ for cast a jaundiced eye on my schematic. The problem turned out to be in the design of my 5V rail on the PCB. There is a 2.5 inch run from the 5V regulator over to the hex driver (74HCT365) that outputs two signals of either 1MHz or 5MHz, as well as four 10MHz outputs. About .5" before the hex driver, I tapped off 5V to power both a thermistor and the output op-amp. I hadn't counted on the impedance of about 2.5" of trace at 10MHz, and the noise the hex driver would put on it. What I did was cut the tee at both ends, and put a jumper from the thermistor & op-amp directly to the 5V regulator. The 10MHz noise is now gone. There is some noise from the DAC output that looks to be at the PIC's clock frequency, but that was filtered out by the voltage divider and .1uf cap at the EFC pin of the OCXO. I did learn something interesting, though. The Vref output of the OCXO (Trimble 34310-T) has about 4mv of 10MHz on it. That was also cleaned up by the voltage divider/corrector. So, I may throw another cap on it, but it seems to be clean down to what I can measure at the OCXO on my old Tek 455 with an X10 probe. Bob From: Bob Stewart To: Time Nuts Sent: Tuesday, August 5, 2014 1:36 PM Subject: [time-nuts] Effects of noise on EFC line? I've run into a noise problem on the EFC line of my GPSDO engine at the frequency of the oscillator. I've traced the source down to the 74HCT365 I'm using to output the 1(or 5)MHz and 10MHz signals. When I pull it, the EFC quietens down a lot. I'm seeing about 50mv of 10MHz noise at the output of the op-amp that feeds the EFC voltage divider at the OCXO. The voltage divider is corrected by the VRef from the OCXO with a simple circuit using temp-co'ed resistors. On the 0.1uf cap at the OCXO's EFC pin, I'm seeing about 5mv of 10MHz signal. I've considered switching the HCT out for a 74LS365, assuming my drive levels are compatible. Unfortunately, I don't have one "in stock", and "I'm way out of my pay grade", as they say. I've also thought about putting a 100uh inductor in series with the EFC line. I wonder if I'll have to isolate the 74xx365 chip's VCC through an inductor? Any thoughts? I'm also wondering what the impact of this level of on-frequency noise will be? Is the impact somewhat mitigated, since it's at oscillator frequency? I don't have anything better than an HP 8558B to look at the output of the board. I'm not quite ready to generally share my schematic with the list, but I can make individual exceptions. Bob - AE6RV ___ 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] Effects of noise on EFC line?
For EFC drive, I prefer to use a buffer fed from an "analog" power, so that the power line noise is not as efficiently inserted into the EFC line. Cheers, Magnus On 08/05/2014 10:00 PM, Chris Albertson wrote: On Tue, Aug 5, 2014 at 11:36 AM, Bob Stewart wrote: I've considered switching the HCT out for a 74LS365, assuming my drive levels are compatible. Unfortunately, I don't have one "in stock", and "I'm way out of my pay grade", as they say. I've also thought about putting a 100uh inductor in series with the EFC line. I wonder if I'll have to isolate the 74xx365 chip's VCC through an inductor? Any thoughts? Not just an inductor but a "low pass filter" (maybe CLC) designed reject anything outside the bandwidth of your control signal. How often do you update the EFC? Once per second? then your bandwidth is 0.5 Hz. That EFC is very close to DC if your uP is only updating EFC once per second. Your filter should be physically close to the OCXO and have a very large time constant. In addition to this I have a 0.1uF cap soldered directly to the EFC and Ground pins of the OCXO. You say you have detected 10MHz noise, but you can bet there is some 60Hz too. About Vcc, I assume each chip has a bypass that is physically right on the Vcc pin. That is enough. Well almost. You need a digital and an analog power supply, sort of. Don't power both off the same voltage regulator. They can share a transformer But make sure the digital stuff has it's own power rail and it's own voltage reg. Then you do the same with the grounds, digital and analog and tie them together at one place near the power supply. Tuns out these GPSDOs are way-simple now. Just a few parts needed. You can use a serial interface DAC to drive EFC. Chris Albertson Redondo Beach, California ___ 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] Effects of noise on EFC line?
and use not to large capacity, because above the serial resonance frequency [trace inductance, own inductance] it will be inductive! 73 Alex KJ6UHN On 8/5/2014 12:16 PM, Jim Harman wrote: I was able to quiet things down a lot by putting a 100 ohm resistor in series with the 74HC output. If these guys drive more than a few inches of wire, they ring like crazy. Also make sure you pay close attention to the bypassing of the 5V supply. Make your bypass cap lead lengths as short as possible. On Tue, Aug 5, 2014 at 2:36 PM, Bob Stewart wrote: I've run into a noise problem on the EFC line of my GPSDO engine at the frequency of the oscillator. I've traced the source down to the 74HCT365 I'm using to output the 1(or 5)MHz and 10MHz signals. When I pull it, the EFC quietens down a lot. I'm seeing about 50mv of 10MHz noise at the output of the op-amp that feeds the EFC voltage divider at the OCXO. The voltage divider is corrected by the VRef from the OCXO with a simple circuit using temp-co'ed resistors. On the 0.1uf cap at the OCXO's EFC pin, I'm seeing about 5mv of 10MHz signal. I've considered switching the HCT out for a 74LS365, assuming my drive levels are compatible. Unfortunately, I don't have one "in stock", and "I'm way out of my pay grade", as they say. I've also thought about putting a 100uh inductor in series with the EFC line. I wonder if I'll have to isolate the 74xx365 chip's VCC through an inductor? Any thoughts? I'm also wondering what the impact of this level of on-frequency noise will be? Is the impact somewhat mitigated, since it's at oscillator frequency? I don't have anything better than an HP 8558B to look at the output of the board. I'm not quite ready to generally share my schematic with the list, but I can make individual exceptions. Bob - AE6RV ___ 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] Effects of noise on EFC line?
On Tue, Aug 5, 2014 at 11:36 AM, Bob Stewart wrote: > I've considered switching the HCT out for a 74LS365, assuming my drive levels > are compatible. Unfortunately, I don't have one "in stock", and "I'm way out > of my pay grade", as they say. I've also thought about putting a 100uh > inductor in series with the EFC line. I wonder if I'll have to isolate the > 74xx365 chip's VCC through an inductor? Any thoughts? Not just an inductor but a "low pass filter" (maybe CLC) designed reject anything outside the bandwidth of your control signal. How often do you update the EFC? Once per second? then your bandwidth is 0.5 Hz. That EFC is very close to DC if your uP is only updating EFC once per second. Your filter should be physically close to the OCXO and have a very large time constant. In addition to this I have a 0.1uF cap soldered directly to the EFC and Ground pins of the OCXO. You say you have detected 10MHz noise, but you can bet there is some 60Hz too. About Vcc, I assume each chip has a bypass that is physically right on the Vcc pin. That is enough. Well almost. You need a digital and an analog power supply, sort of. Don't power both off the same voltage regulator. They can share a transformer But make sure the digital stuff has it's own power rail and it's own voltage reg. Then you do the same with the grounds, digital and analog and tie them together at one place near the power supply. Tuns out these GPSDOs are way-simple now. Just a few parts needed. You can use a serial interface DAC to drive EFC. Chris Albertson Redondo Beach, California ___ 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] Effects of noise on EFC line?
I was able to quiet things down a lot by putting a 100 ohm resistor in series with the 74HC output. If these guys drive more than a few inches of wire, they ring like crazy. Also make sure you pay close attention to the bypassing of the 5V supply. Make your bypass cap lead lengths as short as possible. On Tue, Aug 5, 2014 at 2:36 PM, Bob Stewart wrote: > I've run into a noise problem on the EFC line of my GPSDO engine at the > frequency of the oscillator. I've traced the source down to the 74HCT365 > I'm using to output the 1(or 5)MHz and 10MHz signals. When I pull it, the > EFC quietens down a lot. I'm seeing about 50mv of 10MHz noise at the > output of the op-amp that feeds the EFC voltage divider at the OCXO. The > voltage divider is corrected by the VRef from the OCXO with a simple > circuit using temp-co'ed resistors. On the 0.1uf cap at the OCXO's EFC > pin, I'm seeing about 5mv of 10MHz signal. > > > I've considered switching the HCT out for a 74LS365, assuming my drive > levels are compatible. Unfortunately, I don't have one "in stock", and > "I'm way out of my pay grade", as they say. I've also thought about > putting a 100uh inductor in series with the EFC line. I wonder if I'll > have to isolate the 74xx365 chip's VCC through an inductor? Any thoughts? > > > I'm also wondering what the impact of this level of on-frequency noise > will be? Is the impact somewhat mitigated, since it's at oscillator > frequency? I don't have anything better than an HP 8558B to look at the > output of the board. > > I'm not quite ready to generally share my schematic with the list, but I > can make individual exceptions. > > > Bob - AE6RV > ___ > 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. > -- --Jim Harman ___ 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] Effects of noise on EFC line?
OK my 2 cents and others will have better comments. You can use a far larger inductor in the efc line to try to reduce the noise. EFCs tend to be integrated by some cap. So they are slow moving compared to the noise. Better design is separate analog and digital supplies. Regards Paul WB8TSL On Tue, Aug 5, 2014 at 2:36 PM, Bob Stewart wrote: > I've run into a noise problem on the EFC line of my GPSDO engine at the > frequency of the oscillator. I've traced the source down to the 74HCT365 > I'm using to output the 1(or 5)MHz and 10MHz signals. When I pull it, the > EFC quietens down a lot. I'm seeing about 50mv of 10MHz noise at the > output of the op-amp that feeds the EFC voltage divider at the OCXO. The > voltage divider is corrected by the VRef from the OCXO with a simple > circuit using temp-co'ed resistors. On the 0.1uf cap at the OCXO's EFC > pin, I'm seeing about 5mv of 10MHz signal. > > > I've considered switching the HCT out for a 74LS365, assuming my drive > levels are compatible. Unfortunately, I don't have one "in stock", and > "I'm way out of my pay grade", as they say. I've also thought about > putting a 100uh inductor in series with the EFC line. I wonder if I'll > have to isolate the 74xx365 chip's VCC through an inductor? Any thoughts? > > > I'm also wondering what the impact of this level of on-frequency noise > will be? Is the impact somewhat mitigated, since it's at oscillator > frequency? I don't have anything better than an HP 8558B to look at the > output of the board. > > I'm not quite ready to generally share my schematic with the list, but I > can make individual exceptions. > > > Bob - AE6RV > ___ > 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.
