An advanced question - sampling an oscillator's signal for analysis

[tautech]

FFS there's an easy way to quantify this.
Did you not follow my link in reply #163 or not understand its content ?

[dnessett]

FFS there's an easy way to quantify this.
Did you not follow my link in reply #163 or not understand its content ?

I had a lot of trouble understanding it, since the poster doesn't seem to be a native English speaker.

[dnessett]

Not that I think it important, but to get this red herring off the table, I found two 10' coax patch cables and attached them to each other to create a 20' length. I then ran the GPSDO delayed test. The results are not suprising. Here are the maximum and minimum phase differences.

max=57.6 degrees
min=35.4 degrees
variation = 22.2 degrees

Consequently, reducing the length of the delaying coax provides no insight into the problem.

So, I return to more interesting issues. I ran a 1st difference of the GPSDO 83' delayed data (picking every 20th data point, since the raw data has a data point each 5 ns) to see what phase differences occurred at each period of the signal. Figure 1 shows the results for the 1st 1000 periods. The Rigol has a lower bound of 1mV vertical resolution, so the data it captures has 1 mV resolution. This makes the phase difference signal appear as a step function, which is actually an artifact of the Rigol capture.

Figure 1 -

The plot shows a maximum of around 2.6 degrees phase difference on the top side and -2 degrees on the bottom side, so the bounds in the first 1000 periods is 0 -> 2.6 degrees. The max and min for the complete difference data vector are:

max = 3.6 degrees
min = -2.6 degrees

So, for the complete data set the bounds are 0 -> 3.6 degrees phase difference per period.

It doesn't appear there is much in the way of deterministic drift, although I am not a regression specialist. I continue to work on this question.

Added later: It may be of some interest to convert the phase difference in degrees to ns. Using the formula 100/360 (=.27777...) ns/degree, the 0 -> 3.6 degrees of per period deviation represents a maximum of 1ns jitter (= 1%) per period.

[tautech]

FFS there's an easy way to quantify this.
Did you not follow my link in reply #163 or not understand its content ?

I had a lot of trouble understanding it, since the poster doesn't seem to be a native English speaker.

All the info and clues you need to perform the same measurement are contained in the screenshot.
Please study it again in depth......every little snippet of info.

Clue, look at the Stats box and the # count = 1000s, so with infinite persistence the jitter (ch4) WRT 1pps (ch1) is under 4ns for a 1000s of a 10 MHz ref signal !
Very clever use of standard features in a DSO. 

Linked again below for simplicity:
https://www.eevblog.com/forum/testgear/is-bandwidthmemory-depth-a-waste-of-money-in-oscilloscopes/msg1648538/#msg1648538

[tautech]

So with what you're trying to achieve (big picture) is the most accurate 10 MHz ref you can so no doubt it'll be GPS disciplined, therefore using a similar method you can get some real #'s on how your system performs.

Hope that helps.

[dnessett]

All the info and clues you need to perform the same measurement are contained in the screenshot.
Please study it again in depth......every little snippet of info.

Clue, look at the Stats box and the # count = 1000s, so with infinite persistence the jitter (ch4) WRT 1pps (ch1) is under 4ns for a 1000s of a 10 MHz ref signal !
Very clever use of standard features in a DSO. 

I looked at the Rigol 1104Z manual, but could find no way to measure the skew between channels. So, I don't think I can perform the procedure suggested by the screen shot.

However, it is interesting that the test shown displayed 3.9 ns of jitter in the 1pps signal from a GPSDO. For a 10 MHz signal there are 100/360 (=.27777...) ns/degree. Applying this to the phase difference data I collected, the corresponding (maximum) jitter result is:

GPSDO delayed: 41.6 degrees = 41.6*.27777 = 11.555 ns
Rubidium delayed: 17.2 degrees = 17.2*.27777 = 4.7777 ns
GPSDO versus Rubidium = 18.4 degrees = 18.4*.27777 = 5.1111 ns

Of course, there are differences in the measurements. The 1 pps measurement was over 1001 seconds. The 10 MHz measurements are over 12 msec. Compared to the 1001 cycles for the pps measurement, the 10 MHz measurements are over 120,000 cycles.

The percentage jitter for the 1 pps signal is 3.9*10E-7%. For the 10 MHz measurements the are respectively (GPSDO delayed) 11.5%, (Rubidium delayed) 4.77% and (Rubidium versus GPSDO) 5.11%, a much more serious deviation.

[tautech]

All the info and clues you need to perform the same measurement are contained in the screenshot.
Please study it again in depth......every little snippet of info.

Clue, look at the Stats box and the # count = 1000s, so with infinite persistence the jitter (ch4) WRT 1pps (ch1) is under 4ns for a 1000s of a 10 MHz ref signal !
Very clever use of standard features in a DSO. 

I looked at the Rigol 1104Z manual, but could find no way to measure the skew between channels. So, I don't think I can perform the procedure suggested by the screen shot.

You've got infinite persistence, cursors and a stopwatch haven't you ? 

[tomato]

Not that I think it important, but to get this red herring off the table, I found two 10' coax patch cables and attached them to each other to create a 20' length. I then ran the GPSDO delayed test. The results are not suprising. Here are the maximum and minimum phase differences.

max=57.6 degrees
min=35.4 degrees
variation = 22.2 degrees

Consequently, reducing the length of the delaying coax provides no insight into the problem.

Nothing could be further from the truth.  The measurement made with the 20' delay line is extremely revealing.  You need to think about what you're measuring, and why the above numbers are alarming. 

Hint 1: the above numbers have nothing to do with the stability of the oscillator.

Hint 2: 12 ms is the red herring.

[rhb]

If the phase difference between the signal from the same source traveling down two pieces of coax is not constant, then the experimental setup is flawed.

Most sources are 50 ohms. Many cheap scopes don't have a 50 ohm input, in which case you need a 50 ohm thru terminator at each scope input.  If you drive the two lengths of coax using a T rather than a splitter transformer you will get ringing from the mismatch at the T.  Very likely a 75 ohm cable TV splitter will do a better job than a T.  The T will have a reflection coefficient of 1/3.  The 75 ohm splitter an RC of 0.2, more than 50% lower.

Be wary of cheap Chinese thru terminators.  I bought some that seemed OK, but recently discovered that some of them are not.  So I need to test them all and discard the ones which are NG.

[dnessett]

If the phase difference between the signal from the same source traveling down two pieces of coax is not constant, then the experimental setup is flawed.

That is what I thought. But, when I studied the data, another possibility arose, which I think is the real answer. (And it has nothing to do with the length of coax used for the delayed signal)

I couldn't figure out how the phase difference between a signal and its constantly delayed image could vary as much as the data indicated. Then I noticed an artifact. (see Figure 1)

Figure 1 -

Figure 1 is an image produced by plotting the (1st 20000 data points in the) result of the following Octave code (where "p" holds the GPSDO delayed phase difference data).

Code: [Select]

Fc=10000000;
Fsam=500000000;
Fnyq=Fsam/2;
[b,a]=butter(6, Fc/Fnyq);
output=filter(b,a,p);
pf=output;
pfn=pf(:,2);
pfn=pfn.-mean(pfn);

This code normalizes the data by first applying a 5th order 10 MHz low pass Butterworth filter to it (to eliminate the 20 MHz superimposed signal) and then normalizing it by subtracting the mean from each element. I have marked with red lines prominent spikes in the data.

A free running oscillator would not have such spikes, but neither the Rubidium oscillator nor the GPSDO are free running oscillators. They are disciplined oscillators comprising a crystal oscillator that is periodically corrected by a reference signal. The periodicity of this correction (technically, its reciprocal) is commonly referred to as the servo loop bandwidth. My current hypothesis is the spikes represent periodic corrections to the frequency/phase of the crystal oscillator.

The effect of this is the crystal oscillator free runs for a while and then experiences a movement in frequency/phase. Sometimes this movement is significant, which appears as a large change in the phase difference between the signal and its delayed image.

One question that presented itself is how could the free running oscillator drift so far in frequency as to require a significant correction? One possibility is a previous change overcorrected the error, which then requires a significant movement in the opposite direction. That is speculation, but it is at least plausible.

I eye-balled the distance between two spikes and it was about 1200 points apart. At 2 ns between data points, this represents about 2.4 usec separation. That would imply a servo loop bandwidth of ~417 KHz.

Since I do not have access to the circuit diagrams and design information for the GPSDO, this is still a working hypothesis. However, it is a plausible explanation for the significant differences in the phase difference data. I don't know any engineers who have designed either a GPSDO or a Rubidium oscillator, so I cannot ask them whether this hypothesis makes sense. If anyone reading this thread is such an engineer or knows someone with such experience, comments from them would be appreciated.

Added later: By the way, using the phase differences between a signal and its constantly delayed image is a good way to characterize the short-term stability of an oscillator. It eliminates the need for a reference oscillator, the use of which introduces errors in the measurements.

[tomato]

If the phase difference between the signal from the same source traveling down two pieces of coax is not constant, then the experimental setup is flawed.

That is what I thought. But, when I studied the data, another possibility arose, which I think is the real answer. (And it has nothing to do with the length of coax used for the delayed signal)

I couldn't figure out how the phase difference between a signal and its constantly delayed image could vary as much as the data indicated. Then I noticed an artifact. (see Figure 1)

Figure 1 -

Figure 1 is an image produced by plotting the (1st 20000 data points in the) result of the following Octave code (where "p" holds the GPSDO delayed phase difference data).

Code: [Select]

Fc=10000000;
Fsam=500000000;
Fnyq=Fsam/2;
[b,a]=butter(6, Fc/Fnyq);
output=filter(b,a,p);
pf=output;
pfn=pf(:,2);
pfn=pfn.-mean(pfn);

This code normalizes the data by first applying a 5th order 10 MHz low pass Butterworth filter to it (to eliminate the 20 MHz superimposed signal) and then normalizing it by subtracting the mean from each element. I have marked with red lines prominent spikes in the data.

A free running oscillator would not have such spikes, but neither the Rubidium oscillator nor the GPSDO are free running oscillators. They are disciplined oscillators comprising a crystal oscillator that is periodically corrected by a reference signal. The periodicity of this correction (technically, its reciprocal) is commonly referred to as the servo loop bandwidth. My current hypothesis is the spikes represent periodic corrections to the frequency/phase of the crystal oscillator.

No. Go back and re-read the above statement from rhb. There is no way a correction to the phase of the oscillator will produce those spikes.

The effect of this is the crystal oscillator free runs for a while and then experiences a movement in frequency/phase. Sometimes this movement is significant, which appears as a large change in the phase difference between the signal and its delayed image.

One question that presented itself is how could the free running oscillator drift so far in frequency as to require a significant correction? One possibility is a previous change overcorrected the error, which then requires a significant movement in the opposite direction. That is speculation, but it is at least plausible.

I eye-balled the distance between two spikes and it was about 1200 points apart. At 2 ns between data points, this represents about 2.4 usec separation. That would imply a servo loop bandwidth of ~417 KHz.

Loop bandwidths for GPSDO are orders of magnitude slower than this.  Regardless, those spikes aren't due to any servo correction.

Since I do not have access to the circuit diagrams and design information for the GPSDO, this is still a working hypothesis. However, it is a plausible explanation for the significant differences in the phase difference data. I don't know any engineers who have designed either a GPSDO or a Rubidium oscillator, so I cannot ask them whether this hypothesis makes sense. If anyone reading this thread is such an engineer or knows someone with such experience, comments from them would be appreciated.

See previous.

[rhb]

Feed the signals from the two different lengths of coax to your DSO.  Place cursors at the zero crossings.  Let it run for as long as you like.  The phase relationship should not change unless the coax is bad or you have a reflection problem.  Until you can get a reliable signal to the instrument there is no point in speculating about possible causes of artifacts in the phase measurements.

You will get apparent jitter in a stable signal because the scope is interpolating the trigger point and the measurement points.  The 40 pS pulser is far less stable than the 33622A or GPSDO, but it *appears* to have less jitter because it has a very fast edge.  I don't know the jitter spec for Leo's GPSDO, but the 33622A is specified at less than 1 pS but the RTM3K indicated ~24 pS standard deviation for the time period.  That's not real.  It's a DSO artifact.

7042 shows the 33622A hooked up with what should be a good piece of coax,  but there is an obvious mismatch.   There is not 30 pS of jitter in the 33622A.   The GPSDO has a faster rise time so the step is more pronounced as seen in 7037, but when I connected the GPSDO directly the step went away.  So my "To Do" list got testing and culling BNC cables added. 

Tomato is absolutely correct, though not very clear.  The first requirement is to verify that you can get accurate signals to the test device.  There is no reason to assume that the inputs to the AD board are actually 50 ohms.  Or that anything else is 50 ohms.  10 MHz is not all that high, but it is still RF and can be confusing because of the speed.  I makde the mistake of buying 10 Chinese BNC cables.  They make great 50 MHz notch filters, but are useless for anything else.

I suggest you start by sweeping your cables on the spectrum analyzer.  If at all possible do the cal with a known high quality N cable.  My "To Do" list already has testing a bunch of Chinese adaptors of which I know at least one is bad and I suspect there are others.

[dnessett]

Feed the signals from the two different lengths of coax to your DSO.  Place cursors at the zero crossings.  Let it run for as long as you like.  The phase relationship should not change unless the coax is bad or you have a reflection problem.  Until you can get a reliable signal to the instrument there is no point in speculating about possible causes of artifacts in the phase measurements.

You will get apparent jitter in a stable signal because the scope is interpolating the trigger point and the measurement points.  The 40 pS pulser is far less stable than the 33622A or GPSDO, but it *appears* to have less jitter because it has a very fast edge.  I don't know the jitter spec for Leo's GPSDO, but the 33622A is specified at less than 1 pS but the RTM3K indicated ~24 pS standard deviation for the time period.  That's not real.  It's a DSO artifact.

7042 shows the 33622A hooked up with what should be a good piece of coax,  but there is an obvious mismatch.   There is not 30 pS of jitter in the 33622A.   The GPSDO has a faster rise time so the step is more pronounced as seen in 7037, but when I connected the GPSDO directly the step went away.  So my "To Do" list got testing and culling BNC cables added. 

Tomato is absolutely correct, though not very clear.  The first requirement is to verify that you can get accurate signals to the test device.  There is no reason to assume that the inputs to the AD board are actually 50 ohms.  Or that anything else is 50 ohms.  10 MHz is not all that high, but it is still RF and can be confusing because of the speed.  I makde the mistake of buying 10 Chinese BNC cables.  They make great 50 MHz notch filters, but are useless for anything else.

I suggest you start by sweeping your cables on the spectrum analyzer.  If at all possible do the cal with a known high quality N cable.  My "To Do" list already has testing a bunch of Chinese adaptors of which I know at least one is bad and I suspect there are others.

Thanks for the useful suggestions.

I checked all of the coax cables using the TG on my SA and they were fine. I also checked the BNC Ts and 30 dB pads (see below) and they were fine as well. However, in the process of disassembling and reassembling the set up, I noticed some of the BNC to SMA adapters were loose. I tightened them up and that cleaned up the signal. I performed the input signal cursor test you suggested and it remained stable for an hour.

I think it is best at this point to document the test setup and seek constructive criticism of it. Figure 1 shows it in wide angle. The delaying coax and the GPSDO are off-camera to the left.

Figure 1 -

Figure 2 shows the connections to my Rigol DS1104Z. The input from the DUT comes into channel 2, which is T'ed to the cable carrying the signal to the AD8302 PC board (see Figure 3). The input signal arrives at a T connecting it to one of the AD8302 board inputs through a 30 dB pad and moves past it to one side of the delaying coax. The delaying coax returns to the AD8302 board to another T connected to the board's other input through another 30 dB pad. It then moves past it to channel 1 on the Rigol, which is T'd between the channel and a 50 ohm terminator. Channel 3 of the Rigol is connected to the probe that is shown in Figure 3 connected to pin 5 of the header and grounded at pin 6.

Figure 2 -

Figure 3 -

Figure 4 shows a close up of the AD8302 on its PC board. Pins 1-7 are on the left side. Pins 1 and 7 are ground and pin 4 is +5V. Pins 2 and 6 are the AD8302 inputs, which are fronted by capacitors. It is not apparent what value these are, but one of the test setups described in the spec suggests using 1nF. The impedance for both inputs is spec'd at 3K Ohms/2 pf. Pin 8 (lower right corner) has the 68 pf capacitor I soldered to it. It is a messy solder job because I tried several other capacitor values before settling on the one shown. I stopped trying new values because I was worried the pads on the board would lift.

Figure 4 -

Figure 5 shows the layout of the chip taken from the spec.

Figure 5 -

I am in the process of analyzing the new data and will provide the results in a new post.

[tomato]

I think it is best at this point to document the test setup and seek constructive criticism of it.

1) You've got some termination issues.  You can't just connect your signal to the AD chip with BNC tees, because the AD inputs are terminated with 51Ω resistors. You need to connect via splitters or directional couplers.

2) Why in the world do you have 30dB attenuators on the inputs of the AD chip?

[rhb]

I think it is best at this point to document the test setup and seek constructive criticism of it.

1) You've got some termination issues.  You can't just connect your signal to the AD chip with BNC tees, because the AD inputs are terminated with 51Ω resistors. You need to connect via splitters or directional couplers.

A mild understatement. 

1) Tee + terminator != thru terminator.  That little stub rings like mad, but because it's short you can't see it on the Rigol. Can't see it on my 200 MHz Instek either.  But it's there.

#6 40 pS pulser to 50 ohm thru

#7 same but Tee + terminator

#8 Tee+terminator but with a short BNC cable between the Tee and the terminator

The white reference trace in #7 & #8 is the trace in #6

#9 pulser feeding a Tee and BNCs w/ thru terminators.  One cable is a couple of inches longer.  Again, the reference trace is #6.  Note the apparent increase in gain as we now have approximately 25 ohms terminating the pulser rather than the 50 it needs.

To summarize:  You cannot make meaningful measurements with things connected the way you have them.  I suggest a quick review of transmission lines.

I also suggest being *very* wary of cheap Chinese RF connectors.  I bought a bunch and am finding that there are lots of flaky units waiting to confuse things.

[dnessett]

Thank you for your comments and question. I will address them in reverse order.

1) You've got some termination issues.  You can't just connect your signal to the AD chip with BNC tees, because the AD inputs are terminated with 51Ω resistors. You need to connect via splitters or directional couplers.

2) Why in the world do you have 30dB attenuators on the inputs of the AD chip?

Since you are asking about attenuators on the AD8302 inputs, I presume you have read the device data sheet. If that presumption is correct, then you know the input power range is 0 dBm to -60 dBm (with respect to a 50 ohm load).

I have several oscillators I want to characterize using the test set-up. There are (among a larger set) the GPSDO, which outputs 1.25V P-P sine wave, the Rubidium, which outputs a 1 V P-P sine wave, and an OCXO, which outputs a 50% duty cycle square wave from 0 to 3.5V. The RMS voltage of a 50% duty cycle non-negative square wave is V_P-P/sqrt(2). So, the RMS voltage of the OCXO output is ~2.47V. Looking into a 50 ohm load its power is V_rms²/R ~= 6.1/50 = .122 watt =~20.6 dBm. So, a 20 dBm attentuator just misses the mark, which implies the next common attenuator value of 30 dB. That is why I put them in front of the AD8302 inputs.

I wanted to address the pad issue first, since its existence complicates the analysis of the termination problem. I designed the input circuit to the AD8302 board with the idea of comparing two signals, a reference oscillator signal and a device (oscillator) under test signal. I didn't design it for comparing a signal with its delayed image; a situation I was forced into when the Rubidium/GPSDO comparison test yielded counter-intuitive results. I will admit, I didn't give the delayed coax configuration enough thought. In my defense I only will say that I was hunting down a bug and in the heat of the chase was more interested in getting some hints about what was going on than engineering interface circuits. And it worked, so it had that advantage.

In any case, now is a good time to investigate how the input circuit might affect the results I seek. Figure 1 illustrates the termination topology for both cases.

Figure 1 -

In the original setup, both inputs had the same configuration, which is shown in the Figure. I measured the through resistance and to-ground resistance of the pad with the following results: 1) through resistance = 94 ohms; 2) to-ground resistance = 50 ohms. The AD8302 has a 51 ohm resistor to-ground in front of each signal input. So, the effective resistance from the coax to ground is 25 ohms in series with 94 ohms = 119 ohms.

The length of each coax cable is 3'. The wavelength of a 10 MHz signal is Corrected 7-13-18:98.4 feet 64.9 feet (velocity factor of RG-58=.66 and wavelength in vacuum of 10 MHz~=98.4 feet => wavelength in RG-58 is 98.4*.66=64.9 feet). So, each coax is much less than even a quarter wavelength and in practice can be ignored as a source of standing waves need not be treated as a transmission line. This greatly simplifies the computation of termination resistances, since the resistor network at the AD8302 input is effectively directly connected to the terminating resistance at the scope. If you do the math, I should have an 86 ohm terminating resistor at the scope to achieve an effective to-ground resistance of 50 ohms to the coax. I don't have a BNC terminator with an 86 ohm resistor and I really didn't want to build one; but I may reconsider if turns out to be important. In any case I used a standard 50 ohm terminator, which gives an effective coax termination of 35 ohms. Given the short distances of the coax, I think this is probably OK.

The delay coax setup is significantly different. The total coax length between the oscillator and scope is about 89', which is very near the full wavelength of the 10 MHz signal. Consequently, it is likely that a standing wave will occur due to the mismatched terminations. Since analyzing complex termination toplogies on coax is a skill gained from experience and since I don't have that experience, I will leave the math to others. Instead, I decided to simply measure some factors and see the results, rather than compute them.

One thing to keep in mind: the AD8302 separates the amplitude and phase data. So, a standing wave (which generates a DC bias as a function of position along the coax) should not affect the measurement I am making. There may perhaps be other effects that do, but if so I don't know of them.

Figure 2 shows the spectrum of the signal presented to input 1 of the AD8302. This is after it comes from the oscillator and moves through the BNC T and 30 dB pad. As is evident, there are no significant spectral anomolies present. Figure 3 shows the spectrum of the signal presented to input 2 (after it travels through the 83' of coax and then the BNC T and pad). Again, there are no significant spectral features that suggest a problem.

Figure 2 -


Figure 3 -

This leads me to believe the phase difference data should be uneffected by the termination issues you raise. I am, of course, open to clear arguments that suggest otherwise.

[dnessett]

I think it is best at this point to document the test setup and seek constructive criticism of it.

1) You've got some termination issues.  You can't just connect your signal to the AD chip with BNC tees, because the AD inputs are terminated with 51Ω resistors. You need to connect via splitters or directional couplers.

A mild understatement. 

1) Tee + terminator != thru terminator.  That little stub rings like mad, but because it's short you can't see it on the Rigol. Can't see it on my 200 MHz Instek either.  But it's there.

#6 40 pS pulser to 50 ohm thru

#7 same but Tee + terminator

#8 Tee+terminator but with a short BNC cable between the Tee and the terminator

The white reference trace in #7 & #8 is the trace in #6

#9 pulser feeding a Tee and BNCs w/ thru terminators.  One cable is a couple of inches longer.  Again, the reference trace is #6.  Note the apparent increase in gain as we now have approximately 25 ohms terminating the pulser rather than the 50 it needs.

To summarize:  You cannot make meaningful measurements with things connected the way you have them.  I suggest a quick review of transmission lines.

The figures you posted show the effects of various input hardware on square waves. Right now I am working with fairly pure sine waves. When I get around to measuring a square wave, I may need to revisit some of the issues you raise, but I don't think they are germane at the moment.

Also, it is beneficial to remember that, presently, I am measuring phase difference data, not baseband data. Changes in amplitude are filtered out by the AD8302, so at least the amplitude component of the ringing you illustrate will not have an affect on the measurement. In addition, the ringing displayed is of significantly higher frequency than the fundamental frequency. I apply a 10 MHz low pass filter to the data gathered, which means the effects of such ringing will fail to survive the post processing phase.

[tomato]

Thank you for your comments and question. I will address them in reverse order.

Since you are asking about attenuators on the AD8302 inputs, I presume you have read the device data sheet. If that presumption is correct, then you know the input power range is 0 dBm to -60 dBm (with respect to a 50 ohm load).

I have several oscillators I want to characterize using the test set-up. There are (among a larger set) the GPSDO, which outputs 1.25V P-P sine wave, the Rubidium, which outputs a 1 V P-P sine wave, and an OCXO, which outputs a 50% duty cycle square wave from 0 to 3.5V. The RMS voltage of a 50% duty cycle non-negative square wave is V_P-P/sqrt(2). So, the RMS voltage of the OCXO output is ~2.47V. Looking into a 50 ohm load its power is V_rms²/R ~= 6.1/50 = .122 watt =~20.6 dBm. So, a 20 dBm attentuator just misses the mark, which implies the next common attenuator value of 30 dB. That is why I put them in front of the AD8302 inputs.

Yes, I looked at the data sheet. Although the input range is -60 to 0 dBm, the noise in the phase output increases significantly as the amplitude of the input signal decreases.  Your GPSDO is +6 dBm and your Rb is +4 dBm.  Aren't those the oscillators you've been using in all your tests?  You're throwing away signal and compromising your S/N for no reason.  Save the 30 dB attenuator for when you are using the OXCO.

I wanted to address the pad issue first ... now is a good time to investigate how the input circuit might affect the results I seek

You're making things too complicated again.  A properly designed attenuator terminated by 50Ω will appear as 50Ω at it's input. 

The problem is that your signal sees 25Ω at the BNC tee, because it is split into two paths that are both 50Ω.  You need a splitter or directional coupler instead of the BNC tee.

This leads me to believe the phase difference data should be uneffected by the termination issues you raise. I am, of course, open to clear arguments that suggest otherwise.

Your goal is to measure phase changes that correspond to time delays measured in ps.  Proper termination is electronics 101; good luck if you choose to ignore it.

[dnessett]

You're making things too complicated again.  A properly designed attenuator terminated by 50Ω will appear as 50Ω at it's input. 

The problem is that your signal sees 25Ω at the BNC tee, because it is split into two paths that are both 50Ω.  You need a splitter or directional coupler instead of the BNC tee.

OK. I need some help finding a splitter that satisfies the requirements you think important. Will this one work?

[rhb]

It doesn't matter what the signal is.  A transmission line is a transmission line.  The advantage of the square wave it that it makes the effect of reflections more obvious.  With a pure sine wave, all you get are phase shifts.

I made up that example because it highlighted the salient issue.  You cannot collect valid data with your test setup.  To make it work you'll have to get an exemption from God on the laws of physics.  Not a likely event.  Get some 50 ohm splitters.

You do not have a pure sine wave. Sit down and do the algebra for a sine wave and a 2nd and 3rd harmonic reflected at the end of a cable.  Then consider the effect of a fundamental frequency shift on the harmonics and what that phase effect will be.

I'll let tomato answer this: 

Is there any reason to expect that the amplitude of the harmonics of a physical oscillator will be constant over time?  What is the consequence of the harmonics not being  constant amplitude relative to the fundamental.  Both of you should be able to quote chapter and verse.  This is basic day to day math in a laboratory setting.

[tomato]

OK. I need some help finding a splitter that satisfies the requirements you think important. Will this one work?

You're woking at 10 MHz, so I'd look for a splitter with lower minimum frequency than that one.

Pick up a few attenuators (10dB, 6dB, 3dB).

You're trying to monitor the signals with both the oscilloscope and AD chip at the same time? Use the splitter to split the output of your source.  Connect the two signals to the scope inputs (hi-Z) with BNC tees, then continue on and terminate at the AD chip. Use as little attenuation as possible at the AD chip. (Ultimately, you'll get rid of either the scope or the AD chip when you want to make serious measurements.)

[dnessett]

While waiting for power splitters and directional couplers to arrive (hopefully sometime next week), I did some reading about test setups for measuring oscillator stability. One of the calibration steps described requires the measurment of phase noise bandwidth. Since I had time on my hands, I did a quick check of this for the GPSDO, Rubidium and Rigol DG1022. One thing surprised me, so I thought I would forward the results to this thread for comment.

I am sure to get blasted unless I state the following. My intention was to explore the phase noise bandwidth issue, not to obtain definitive measurement values. That is a later goal So, the following information is best viewed as exploratory.

Figures 1 shows the noise floor of the spectrum analyzer for a band from 9 MHz to 11 MHz using a RBW of 30 Hz.

Figure 1 -

Using noise markers, I measured the phase noise bandwidth of the oscillators (Figures 2 (GSPDO), 3 (Rubidium), and 4 (Rigol)) from 9 MHz to 11 MHz.

Figure 2 -

Figure 3 -

Figure 4 -

Pay no attention to the dBm values, I did not setup the SA to obtain valid power values (for example, the internal attenuator is set to -20 dB). My interest was in the bandwidth of the phase noise, not its power. Here is a tabulation of the pertinent data.

Oscillator	Lower	Upper	Bandwidth
GSPDO	9.589	10.402	813 KHz
Rubidium	9.669	10.328	659 KHz
Rigol	9.688	10.309	621 KHz

I found it surprising that the GPSDO had the largest phase noise bandwidth. While the difference in bandwidth between the Rubidium and Rigol is probably insignificant, it is still a bit surprising that the Rigol had the lowest value.

[borghese]

I think you're measuring the phase noise of the spectrum analyzer; a good oscillator has> 150 dBc at 1 kHz offset.

[dnessett]

I think you're measuring the phase noise of the spectrum analyzer; a good oscillator has> 150 dBc at 1 kHz offset.

I don't have phase noise specs for the GPSDO, since it is an ebay special. But, for the others:

FEI FE5650

-100 dBc @ 10Hz
-125 dBc @ 100 Hz
-145 dBc @ 1 KHz

The spec doesn't indicate, but the dBc figures are almost certainly dBc/Hz.

Rigol DG1022

-108 dBc/Hz @ 10 KHz.

The FEI FE5650 spec comes closest to your suggestion that a good oscillator displays a phase noise of ~150 dBc/Hz at 1 KHz. On the other hand, I bought it on ebay and it is 20 years old, so it may not meet its original specifications.

Nevertheless, the kernel of your comment is correct. I have a Siglent SSA3021X, which has a phase noise figure of < -95 dBc/Hz @ 10 KHz, <96 dBc/Hz @100 KHz, and , < -115 dBc/Hz @ 1 MHz. From what I have read, the phase noise figure of a spectrum analyzer measuring phase noise of a device should be at least 10 dB less than that of the device. So, using it to determine the phase noise bandwidth of the oscillators is not going to work out.

Any suggestions how to achieve this measurement in some other way?

[borghese]

You can read "Choosing a Phase Noise Measurement Technique" from HP or Agilent.