Elecraft › [MM]

XG3 device...

Classic

List

Threaded

27 messages Options

Alan Bloom

Re: XG3 device...

Hi Geoff,

The case we're considering is where the test signal(s) are near the
receiver tuned frequency so they are well within the passband of the
input filters.

If the receiver's in-band dynamic range is significantly affected by
non-linear filters in the front end, then that is a serious design
error. The distortion should be dominated by the mixer, not the filter.
If the filter is distorting then it doesn't matter if the generator has
harmonics - you're hosed anyway.

For in-band receiver testing, generator harmonics should not matter.
The only exceptions would be for receivers that do not have filters
protecting the front end or perhaps the exceedingly rare case where a
generator harmonic happens to fall on a receiver image or spur frequency
(easily fixed by changing the test frequency slightly). But in almost
all cases the harmonics should not matter.

Alan N1AL

On Fri, 2011-04-15 at 21:30 +0100, Geoffrey Mackenzie-Kennedy wrote:

> Hi Alan,
>
> Sorry that I was not here to reply to your previous e-mail.
>
> With respect, filter non-linearity and the level of harmonics coming from
> the signal sources are related in terms of the amplitude of the IMD products
> seen at the output of the filter.
>
> If the receiver being tested is perfect and linear, obviously it would not
> matter if the IMD test sources contained harmonics because, for example, the
> fundamental of one source (F1) has nowhere to mix with the second harmonic
> of the second source (2F2) to produce the third order product 2F2-F1 -
> assuming there is no IMD contribution from the test equipment.
>
> On the other hand if the receiver's signal path passes through a non-linear
> element such as an input filter, then of course there is 'mixer' available
> inside the filter to produce the natural third order product 2F2 - F1 as
> would be expected when using clean sources. If the sources produce harmonics
> however, then a twin third order product also at 2F2 - F1 would also be
> generated, where 2F2 is the second harmonic generated by the F2 source. The
> composite level of these two third order product as seen at the filter's
> output could be significantly different from that of the natural product
> depending on how they add.
>
> Although the second harmonic of a test source should be greatly attenuated
> by a typical front end filter, an unwanted 'mixer' can exist near the input
> of the filter, for example in the form of an inductor's core, and would be
> exposed to the full impact of the source's harmonics.
>
> Geoff
> GM4ESD
>
>
> On Friday, April 15, 2011, at 18:45 +0100, Alan Bloom wrote:
>
> > Sure, but if the filters are non-linear that's a whole different problem
> > that has nothing to do with harmonics in the signal source.
> >
> > The question was whether signal generator harmonics are normally a
> > problem when measuring a receiver. I claim they are not because they
> > are removed by the filters in the receiver front end.
> >
> > When I was working in the R&D lab of the Hewlett Packard Signal Sources
> > division, one of our competitors came out with a new signal generator
> > with digital modulation. The modulation was generated at an IF which
> > was heterodyned up to microwave frequencies without any filtering to
> > remove the image. So there was a zero-dB spur X MHz away from the
> > desired signal. They got away with it because it was far enough away to
> > be filtered out by the receiver under test.
> >
> > Alan N1AL
> >
> >
> > On Fri, 2011-04-15 at 13:11 -0400, Jack Smith wrote:
> >> Only to a first approximation are filters linear.
> >>
> >> Usually the major source of non-linearity is in the inductors. Even
> >> powdered iron is non-linear to some degree. Ferrite is worse, of
> >> course. Air is pretty close to perfect.
> >>
> >> Capacitors are also non-linear at some level, as C is not constant
> >> with applied voltage. NP0/C0G caps are very good, but again not
> >> perfect.
> >>
> >> Filter generated intermodulation problems are real and measurable.
> >> Been there and done that and have the Tee-shirt.
> >>
> >> Jack K8ZOA
> >>
> >>
> >> On 4/15/2011 1:01 PM, Alan Bloom wrote:
> >> > Hi Geoff,
> >> >
> >> > I guess I don't understand. Filters are linear devices. How can they
> >> > cause IMD? If the signal generator's harmonics are removed by the
> >> > receiver filters then it is the same as if the sig gen were "clean",
> >> > no?
> >> >
> >> > 73,
> >> >
> >> > Alan N1AL
>
>
>
>
>

______________________________________________________________
Elecraft mailing list
Home: http://mailman.qth.net/mailman/listinfo/elecraft
Help: http://mailman.qth.net/mmfaq.htm
Post: mailto:[hidden email]

This list hosted by: http://www.qsl.net
Please help support this email list: http://www.qsl.net/donate.html

wayne burdick

Re: XG3 and IMD testing

Administrator

The XG3's phase noise is a constant -105 dBc/Hz. This is pretty good
for a miniature, wide-range RF generator that runs from a 9-V battery,
and great for MDS tests (sensitivity), signal tracing, filter
alignment, and S-meter calibration. But it is not suited to high-level
IMD testing of receivers like K3's. That requires a signal generator
with a lower noise floor.

73,
Wayne
N6KR

______________________________________________________________
Elecraft mailing list
Home: http://mailman.qth.net/mailman/listinfo/elecraft
Help: http://mailman.qth.net/mmfaq.htm
Post: mailto:[hidden email]

This list hosted by: http://www.qsl.net
Please help support this email list: http://www.qsl.net/donate.html

wayne burdick

Re: XG3 and IMD testing

Administrator

And by the way, the XG1 and XG2 also have more or less square-wave
output. This is where the harmonics come from that allow them to put
out a useful signal on higher bands.

For the intended applications, this waveform is not an issue, because
the receiver under test will virtually always have a low-pass filter
at its antenna input.

73,
Wayne
N6KR

On Apr 15, 2011, at 3:19 PM, Wayne Burdick wrote:

> The XG3's phase noise is a constant -105 dBc/Hz. This is pretty good
> for a miniature, wide-range RF generator that runs from a 9-V battery,
> and great for MDS tests (sensitivity), signal tracing, filter
> alignment, and S-meter calibration. But it is not suited to high-level
> IMD testing of receivers like K3's. That requires a signal generator
> with a lower noise floor.
>
> 73,
> Wayne
> N6KR
>
> ______________________________________________________________
> Elecraft mailing list
> Home: http://mailman.qth.net/mailman/listinfo/elecraft
> Help: http://mailman.qth.net/mmfaq.htm
> Post: mailto:[hidden email]
>
> This list hosted by: http://www.qsl.net
> Please help support this email list: http://www.qsl.net/donate.html

n7ws

Re: XG3 device...

In reply to this post by Alan Bloom

Getting a little far afield here and maybe a little pedantic (can you be just a little pedantic?) mixer technology has progressed to the point that the mixer no longer has to be the limiting factor.

Input filters and post mixer crystal filters are now the focus.

"After having concentrated on the H-Mode mixer circuit for quite some time and looking at the results obtained with the combination of FSA3157 switches and T1-6T transformers, it was clear that at this point the mixer was not the weakest chain in the receiver's frontend anymore." from: http://www.xs4all.nl/~martein/pa3ake/hmode/frontend_intro.html

--- On Fri, 4/15/11, Alan Bloom <[hidden email]> wrote:

> From: Alan Bloom <[hidden email]>

> Hi Geoff,
>
> The case we're considering is where the test signal(s) are
> near the
> receiver tuned frequency so they are well within the
> passband of the
> input filters.
>
> If the receiver's in-band dynamic range is significantly
> affected by
> non-linear filters in the front end, then that is a serious
> design
> error. The distortion should be dominated by the
> mixer, not the filter.
> If the filter is distorting then it doesn't matter if the
> generator has
> harmonics - you're hosed anyway.
>
> For in-band receiver testing, generator harmonics should
> not matter.
> The only exceptions would be for receivers that do not have
> filters
> protecting the front end or perhaps the exceedingly rare
> case where a
> generator harmonic happens to fall on a receiver image or
> spur frequency
> (easily fixed by changing the test frequency
> slightly). But in almost
> all cases the harmonics should not matter.
>
> Alan N1AL
>

David Cutter

Re: XG3 and IMD testing

In reply to this post by wayne burdick

For my education: what is the mark-space ratio to develop optimum even and
odd harmonics? Does this have any merit?

David
G3UNA

> And by the way, the XG1 and XG2 also have more or less square-wave
> output. This is where the harmonics come from that allow them to put
> out a useful signal on higher bands.
>
> For the intended applications, this waveform is not an issue, because
> the receiver under test will virtually always have a low-pass filter
> at its antenna input.
>
> 73,
> Wayne
> N6KR
>
>

Geoffrey Mackenzie-Kennedy

Re: XG3 device...

In reply to this post by Alan Bloom

Hi Alan,

Due to visitors arriving tomorrow, I must make this my last post on the
subject we have been discussing.

I believe that this discussion started after Jack had warned that a square
wave source can get you into trouble if used to make 2nd and 3rd
intermodulation measurements, and noted that MiniCircuits AN-00-008 offered
an explanation.

As a summary,

In the last paragraph of my last post I suggested that an "unwanted" mixer
mixer can exist near the input of the filter, and that it would be exposed
to the full impact of the source's harmonics. I should stress that this is
not speculation, but fact derived from many diagnostic tests on filters
during design.

In the case of two-tone IMD tests using sources whose harmonic content is
high, any "unwanted" mixer near to the filter's input will generate a host
of products, and some of these will pass through the filter and contaminate
the test results.

In the case of single-tone tests using a similar source, this "unwanted"
mixer in the filter can still affect the accuracy of test results, by how
much will depend, of course, on the characteristics of this "unwanted" mixer
(which are related to the filter's design), and the level of the incoming
test signal's fundamental and each harmonic.Please bear in mind that this
"unwanted" mixer is NOT protected by the selectivity of the filter.

If in this second case the fundamental test frequency is F1, then the
harmonics of F1 should not get through the filter. However because the
"unwanted" mixer is exposed to the incoming signal's fundamental and its
harmonics, among the products generated will be 2F1- F1, 3F1 - 2F1, 4F1-
3F1 etc, all of which will get through the filter and contaminate the test
results.

However if the single-tone test signal is a true and symmetrical square
wave, the amplitude of the incoming even order harmonics should be small
which would help to reduce this problem, leaving the "unwanted" mixer's
"self generated" products at F1.

Lastly, with regard to front end filters and dynamic range, I can assure you
that the design of the front end filters as used in a high performance HF
receiver is not casual. What is causing some concern is that the gap between
robust front end filters and top range mixers (H-Mode class), in terms of
their IMD performance, is closing. Useful front end filters tend to be large
in size!

73,
Geoff
GM4ESD

On Friday, April 15, 2011, at 22:43 +0100, Alan Bloom wrote:

> Hi Geoff,
>
> The case we're considering is where the test signal(s) are near the
> receiver tuned frequency so they are well within the passband of the
> input filters.
>
> If the receiver's in-band dynamic range is significantly affected by
> non-linear filters in the front end, then that is a serious design
> error. The distortion should be dominated by the mixer, not the filter.
> If the filter is distorting then it doesn't matter if the generator has
> harmonics - you're hosed anyway.
>
> For in-band receiver testing, generator harmonics should not matter.
> The only exceptions would be for receivers that do not have filters
> protecting the front end or perhaps the exceedingly rare case where a
> generator harmonic happens to fall on a receiver image or spur frequency
> (easily fixed by changing the test frequency slightly). But in almost
> all cases the harmonics should not matter.
>
> Alan N1AL
>
>
>
> On Fri, 2011-04-15 at 21:30 +0100, Geoffrey Mackenzie-Kennedy wrote:
>> Hi Alan,
>>
>> Sorry that I was not here to reply to your previous e-mail.
>>
>> With respect, filter non-linearity and the level of harmonics coming from
>> the signal sources are related in terms of the amplitude of the IMD
>> products
>> seen at the output of the filter.
>>
>> If the receiver being tested is perfect and linear, obviously it would
>> not
>> matter if the IMD test sources contained harmonics because, for example,
>> the
>> fundamental of one source (F1) has nowhere to mix with the second
>> harmonic
>> of the second source (2F2) to produce the third order product 2F2-F1 -
>> assuming there is no IMD contribution from the test equipment.
>>
>> On the other hand if the receiver's signal path passes through a
>> non-linear
>> element such as an input filter, then of course there is 'mixer'
>> available
>> inside the filter to produce the natural third order product 2F2 - F1 as
>> would be expected when using clean sources. If the sources produce
>> harmonics
>> however, then a twin third order product also at 2F2 - F1 would also be
>> generated, where 2F2 is the second harmonic generated by the F2 source.
>> The
>> composite level of these two third order product as seen at the filter's
>> output could be significantly different from that of the natural product
>> depending on how they add.
>>
>> Although the second harmonic of a test source should be greatly
>> attenuated
>> by a typical front end filter, an unwanted 'mixer' can exist near the
>> input
>> of the filter, for example in the form of an inductor's core, and would
>> be
>> exposed to the full impact of the source's harmonics.
>>
>> Geoff
>> GM4ESD

Jack Smith-6

Re: XG3 and IMD testing

In reply to this post by David Cutter

A classic comb generator products equal level harmonics.

In the time domain, it is an impulse waveform--in the limit infinitely
narrow. It can be approximated by differentiating a fast rise/fall
square wave with an RC circuit.

If you search "comb generator" you can find many implementations on the
Internet.

Jack K8ZOA

On 4/16/2011 8:15 AM, David Cutter wrote:

> For my education: what is the mark-space ratio to develop optimum even and
> odd harmonics? Does this have any merit?
>
> David
> G3UNA
>
>
>> And by the way, the XG1 and XG2 also have more or less square-wave
>> output. This is where the harmonics come from that allow them to put
>> out a useful signal on higher bands.
>>
>> For the intended applications, this waveform is not an issue, because
>> the receiver under test will virtually always have a low-pass filter
>> at its antenna input.
>>
>> 73,
>> Wayne
>> N6KR
>>
>>
> ______________________________________________________________
> Elecraft mailing list
> Home: http://mailman.qth.net/mailman/listinfo/elecraft
> Help: http://mailman.qth.net/mmfaq.htm
> Post: mailto:[hidden email]
>
> This list hosted by: http://www.qsl.net
> Please help support this email list: http://www.qsl.net/donate.html
>