Gaining a better understanding PN was the point of my question.
I am not focused at all on the 50 Hz sidebands. They are there, they are a problem. I am sure I can fix them. They are not an artifact of my receiver because the same receiver was used for both pix. They actually DO originate in the RXTX under test, if not in the 5170 itself then in the circuit that contains it.
I actually have a pretty clear picture of what PN is.... what I am asking for is a simple way to judge it by comparing the two pix. I pointedly inquired about the fatness of the center signal, not the presence or levels of the sidebands.
You clearly know more about this than I do. Otherwise I would not have asked.
Thank you for your response.
Warren Allgyer - 9V1TD
--- In email@example.com, Jack Smith <Jack.Smith@...> wrote:
> If you reviewed two automobiles and one had four flat tires and the
> other didn't, would it be reasonable to say the first one did not handle
> well compared with the second, without mentioning the flat tires?
> Yes, 50 Hz sidebands are a problem -- BUT THEY DO NOT ORIGINATE IN THE
> SI570. Or if they do originate in the Si-570 section they are caused by
> a problem within your receiver. There is no mechanism within the Si-570
> that causes 50 Hz sidebands as far as I know. I certainly saw no trace
> of those when I looked in some detail at the phase noise of an Si-570.
> Rather it indicates a deficiency elsewhere.
> So if you intend to compare the Si570 with the Si5338 you have failed if
> you focus on the 50 Hz sidebands.
> You can either fix the 50 Hz contamination or ignore the 50 Hz
> sidebands. What you cannot do and produce an analysis that has
> credibility is to say those are part of the Si570's characteristic
> performance. They are in fact an artifact of your receiver or test
> signal or something else that I can't determine without more work than I
> care to spend on it.
> Before you take this further, you really need to understand what PN is
> and how one measures it.
> Jack K8ZOA
> On 5/23/2013 8:20 PM, warrenallgyer wrote:
> > Thanks Jack. That is very clear and helpful.
> > So, if I understand you correctly, I should be more concerned about
> > the spreading of the central spike than about the sidebands? That
> > makes some sense to me. On the other hand, since the two signals under
> > consideration are used as the LO in a receiver, both the spreading and
> > the sidebands are introduced into and degrade the received signals.
> > The receiver in this case is a Softrocks RXII HF so it does have a
> > local Si5170 as well as op amps to consider.
> > My intent here was to compare the Si5170 to the Si5338, not
> > necessarily to quantify the PN so much as to quantify the difference
> > between the two. That is why I have signals of the same level and
> > frequency from the two devices.
> > The fact that the Si5338 has more spreading on the central spike than
> > does the Si5170, with identical receive conditions in both cases,
> > indicates to me the presence of more phase noise. Right?
> > If I could quantify the difference between the two, even if I could
> > not quantify the noise absolutely, that would be a major step forward.
> > Any thoughts?
> > I can access real test equipment to do this but what is the fun in
> > that? :-) . My goal is to be able to do meaningful measurements with
> > simple tools; something that a majority of people would have access to.
> > Warren Allgyer - 9V1TD
> > --- In firstname.lastname@example.org
> > <mailto:softrock40%40yahoogroups.com>, Jack Smith <Jack.Smith@> wrote:
> > >
> > > Warren:
> > >
> > > I did not mean to say that the central spike has 50 Hz contamination -
> > > rather that the spikes at +/- 50 Hz should be disregarded as they are
> > > not valid to indicate anything about the phase noise. Those must not be
> > > used as an indication of Si570 phase noise but rather indicate problems
> > > with the receiver, power supply filtering, connecting cables, etc.
> > >
> > > If you look at the central spike, you will see the typical spreading,
> > > but remember that the device the Si570 is driving (the receiver) has
> > its
> > > own noise, e.g., 1/f noise for op-amps which can look a lot like PN on
> > > the oscillator.
> > >
> > > So the most one can say is that the display indicates the sum of
> > various
> > > noise sources, one being Si570, but the others being internal and
> > > external. For example, what is the PN of the signal source you used to
> > > obtain the 700 Hz beat note? The display is, of course, the sum of the
> > > PN of the Si570 local oscillator plus the PN of the signal source it
> > > mixes with to produce the 700 Hz beat note.
> > >
> > > 50/60 Hz line noise related noise in a good signal generator can be 90
> > > dB down, or more, but that takes a great deal of attention to detail to
> > > achieve that. One of my HP 8657A signal generators has 60 Hz sidebands
> > > down 40 dB and a second HP 8657A has 60 Hz sidebands down 55 dB.
> > But, my
> > > HP 8568B spectrum analyzer has 60 Hz sidebands down 80 dB or more as
> > > evidenced by observing a high quality 10 MHz oven time base, where I
> > > don't see 60 Hz sidebands above the noise floor.
> > >
> > > Consequently before making any observations about PN of a particular
> > > device you MUST first establish the performance of the test equipment
> > > being used and verify its performance. To some degree this may seem
> > > circular, but it is possible to obtain useful performance
> > information on
> > > the test gear without reference to the device under test.
> > >
> > > Otherwise you are left in the position of not knowing what you see
> > comes
> > > from the device being tested or represent artifacts of the test
> > > equipment. And once you characterize the test setup, you will still, of
> > > course, measure the sum (in some fashion, I don't necessarily mean
> > > linear sum) of the device under test and the test equipment.
> > >
> > > The old rule of thumb used to be that if you intend to measure
> > something
> > > to a precision of X, the test equipment should be 10x better so that
> > the
> > > error contribution of the test equipment can be considered to be
> > > minimal. It may still need to be considered, but at least it does not
> > > dominate the measurement.
> > >
> > > Jack K8ZOA
> > >