Loading ...
Sorry, an error occurred while loading the content.

Re: [softrock40] Re: How to measure IMD?

Expand Messages
  • FRANCIS CARCIA
    The ARRL spec is the 6dB gift method not the mil method. If you add 6dB to the two tone level why don t you add 6dB to the IP3 also since it is the result of
    Message 1 of 61 , Jun 15 6:59 PM
    • 0 Attachment
      The ARRL spec is the 6dB gift method not the mil method. If you add 6dB to the two tone level why don't you add 6dB to the IP3 also since it is the result of the two tones. Splain that Lucy.
      I'm attending the school of hard knocks trying to modify a 1200 watt MRI board to HF. Fairly easy to make power but a different issue to get good IMD on all bands.
      Frank WA1GFZ

      From: Jack Smith <Jack.Smith@...>
      To: softrock40@yahoogroups.com
      Sent: Friday, June 14, 2013 4:53 PM
      Subject: Re: [softrock40] Re: How to measure IMD?
       
      No, that's not an answer.

      Without the measurement protocol being identified, the dB down value is 6 dB ambiguous.

      If I say "I measured the 3rd order IMD from this transmitter at -30 dB" and provide no further information, you cannot know whether it's 30 dB down with respect to PEP or with respect to a single tone. You can guess, of course, but you can't be certain. If I'm trying to make the transmitter look good, maybe the reference is PEP, if I am conservative, perhaps it's with respect to a single tone. But the point being there is no a priori reason to assume one measurement standard was used over the other in the absence of evidence. If, in contrast, I say "measured 3rd order IMD is -30 dB with respect to PEP" or "measured 3rd order IMD is -30 dB with respect to each test tone" you can then make a meaningful comparison with other transmitters or amplifiers.

      Jack K8ZOA

      On 6/14/2013 4:37 PM, warrenallgyer wrote:
       
      Agreed.

      My understanding of IMD3 measurements is they are always referenced to single tone PEP or CW (they should be identical) levels if no other reference point is expressed.

      In other words "30 dB down" is understood to be 24 dB down from the two-tone level. "30 dB down from the two tone level" would be the same as "36 dB down"

      Warren Allgyer - 9V1TD

      --- In mailto:softrock40%40yahoogroups.com, Jack Smith mailto:Jack.Smith@... wrote:
      >
      > Let me try it again.
      >
      > Of course the end result is the same. BUT when you read a specification
      > that the 3rd order IMD products are 28 dB down and no other information
      > is provided, what do you do?
      >
      > If the measurement is from a ham, maybe it's ARRL definition. Maybe it's
      > not.
      >
      > The point being you can't say X dB down is adequate unless you define
      > the measurement protocol, right?
      >
      > Jack K8ZOA
      >
      > On 6/14/2013 4:06 PM, warrenallgyer wrote:
      > >
      > > The ARRL spec and the commercial/military spec are exactly the same.
      > >
      > > The ARRL spec of -30 dB is with respect to the PEP output. The spec of
      > > -24 dB is with respect to the PEP of the SSB signal resulting from a
      > > two tone modulating signal.
      > >
      > > IMD3 products that are 24 dB down from the PEP resulting from a two
      > > tone modulating signal are, by definition, 30 dB down from the PEP
      > > resulting from single tone modulation.
      > >
      > > Warren Allgyer - 9V1TD
      > >
      > > --- In mailto:softrock40%40yahoogroups.com
      > > <mailto:softrock40%40yahoogroups.com>, Jack Smith <Jack.Smith@> wrote:
      > > >
      > > > Well, this is certainly interesting but it seems to me to be wrong.
      > > >
      > > > First, the spectrum analyzer plot looks to me exactly like a DSB - SC
      > > > signal. Note the symmetry and that you can see the two individual test
      > > > tones symmetrically displaced from where the suppressed carrier would
      > > > normally be. I suspect the transceiver needs to be driven with an I & Q
      > > > audio signal to cancel the unwanted sideband and that Chris, from what
      > > > he has said, is driving the transceiver with only one audio signal. All
      > > > my work with the Softrock and other SDRs is on the receive side, so I
      > > > could well be wrong about the need for I & Q drive to the transceiver,
      > > > but this surmise certainly matches the spectrum analyzer plot.
      > > >
      > > > As far as how one specifies the permitted IMD level, a couple of
      > > > observations ...
      > > >
      > > > 1) As far as I know only the ARRL believes the proper measure of IMD
      > > > suppression is with respect to PEP or the fictitious carrier. The
      > > > standard commercial / military measurement protocol defines IMD
      > > > suppression with respect to one of the two test tone levels. This is 6
      > > > dB less suppression than calculated under the ARRL's methodology.
      > > >
      > > > 2) Keep in mind that the IMD will only get worse as subsequent
      > > > amplifier stages are added. At the milliwatt level in the exciter the
      > > > output should be exceptionally clean, and I've seen and measured better
      > > > than -60 dBc distortion in the case of an Elecraft K3 at the
      > > transverter
      > > > output. The K3 is much worse when looked at either the 10 watt or 100
      > > > watt stage outputs, so I question whether -30 dB ARRL method or -24 dB
      > > > commercial/military spec at the output of a low level stage is
      > > > acceptable at the power level involved here. It may be marginally
      > > > acceptable if measured at the transmitter output , including all
      > > > amplifier stages but it's certainly not close to state of the art. (The
      > > > K3 isn't close either, as Elecraft's choice of 12V DC power is not the
      > > > way to improve 3rd order distortion. You really need a 24 or 48V FET
      > > > output stage for that purpose.)
      > > >
      > > > Indeed, there is a normal 3:1 (in dB) relationship between drive and
      > > IMD
      > > > for 3rd order analog stages, and a bit of drive reduction will be
      > > > beneficial.
      > > >
      > > > Jack K8ZOA
      > > >
      > > >
      > > >
      > > > On 6/14/2013 10:19 AM, warrenallgyer wrote:
      > > > >
      > > > >
      > > > > Hi Jack
      > > > >
      > > > > I am going to argue with you a little bit. These are not sidebands
      > > but
      > > > > they are the result of two tones modulating the carrier, resulting in
      > > > > two discrete RF frequencies at 550 and 1330 on one side of the
      > > carrier
      > > > > frequency. If Chris did not have sideband suppression there would be
      > > > > another pair of these signals at 550 and 1330 Hz on the opposite side
      > > > > of the carrier frequency.
      > > > >
      > > > > The maximum acceptable IMD3 products are 30 dB below CARRIER
      > > level. By
      > > > > definition, a two tone signal with equal input levels on both tones
      > > > > produces signals at the two frequencies each with a level 6 dB below
      > > > > the single tone carrier. Because Chris's IMD3 products are at 23 dB
      > > > > below the two tone level, they are actually 29 dB below the carrier
      > > > > and therefore within smelling distance of the goal.
      > > > >
      > > > > My experience has shown the IMD3 level is very much dependent on how
      > > > > hard you drive Q6 and this is determined by the audio input level. I
      > > > > am guessing a 1 dB reduction on audio drive would result in at
      > > least a
      > > > > 3 dB reduction in IMD3.
      > > > >
      > > > > I am not an expert... I am self taught on this subject,and my teacher
      > > > > has a number of issues. So I am very willing to be corrected by
      > > > > someone better educated on the subject.
      > > > >
      > > > > Best regards.
      > > > >
      > > > > Warren Allgyer - 9V1TD
      > > > >
      > > > > --- In mailto:softrock40%40yahoogroups.com
      > > <mailto:softrock40%40yahoogroups.com>
      > > > > <mailto:softrock40%40yahoogroups.com>, Jack Smith <Jack.Smith@> wrote:
      > > > > >
      > > > > > Chris:
      > > > > >
      > > > > > I looked at your images this morning - you seem to have no unwanted
      > > > > > sideband suppression - the upper and lower sidebands are
      > > identical in
      > > > > > level - like you are operating in DSB-SC (double sideband,
      > > suppressed
      > > > > > carrier) mode. I don't know if this is your intention or not, but
      > > > > that's
      > > > > > what it looks like.
      > > > > >
      > > > > > Your oscilloscope envelope image also does not look like a
      > > standard SSB
      > > > > > signal, but some of that is due to the sweep speed - the envelope
      > > > > > display should be with a sweep speed much slower than the value
      > > needed
      > > > > > to resolve the individual RF cycles. You are looking for the
      > > modulated
      > > > > > envelope display, so something on the order of 500us/div will be
      > > in the
      > > > > > neighborhood of the desired range, not 20ns.
      > > > > >
      > > > > > It looks as if you have about 23 to 25 dB 3rd order intermodulation
      > > > > > suppression, measured from the test tone. That's not all that great.
      > > > > >
      > > > > > Also, the higher order IMD products are much higher than one would
      > > > > > normally like, particularly if you will be operating with an
      > > amplifier.
      > > > > >
      > > > > > Before going further, I would try to find the reason the signal
      > > is in
      > > > > > DSB mode, not SSB.
      > > > > >
      > > > > > One normally measures with a span around 10 KHz, and RBW of 100 or
      > > > > 30 Hz
      > > > > > - looks like you have an HP8568 or 8566 spectrum analyzer (I use an
      > > > > > 8568B here) - in order to resolve the individual products better
      > > than
      > > > > > with the 300 Hz RBW used in most of your images. But that's
      > > something
      > > > > > to be looked at after you get it running in SSB mode.
      > > > > >
      > > > > > Jack K8ZOA
      > > > > >
      > > > > >
      > > > > >
      > > > > > On 6/14/2013 7:36 AM, Chris Wilson wrote:
      > > > > > >
      > > > > > >
      > > > > > >
      > > > > > > >
      > > > > > > >
      > > > > > >
      > > > >
      > > http://www.funkamateure-dresden-ov-s06.de/index.php?article_id=259&clang=0
      > > > > > > >
      > > > > > >
      > > > >
      > > http://www.home.agilent.com/upload/cmc_upload/All/ENA_IMD_Measurement_Summary.pdf?&cc=GB&lc=eng
      > > > > > > > http://www.maurymw.com/pdf/applib/5C-043.pdf
      > > > > > >
      > > > > > > > Not comprehensive guides but should give some insights.
      > > > > > > > Two-tone test and look at the Spectrum Analyser for additional
      > > > > products.
      > > > > > > > 5C-043.pdf shows what you should see and the difference
      > > between the
      > > > > > > > carrier peak and the 3rd order product peak as seen on the
      > > Spectrum
      > > > > > > > Analyser gives you the figure.
      > > > > > > > 73 ... Sid.
      > > > > > >
      > > > > > > 14/06/2013 12:33
      > > > > > >
      > > > > > > Thanks Sid and Alan.
      > > > > > >
      > > > > > > I made an Audacity dual tone file with combined frequencies of
      > > 1330
      > > > > > > Hz and 550 Hz, and checked the resulting file looked clean
      > > from the
      > > > > > > audio card, on a `scope. I then ran it into the Softrock
      > > Ensemble II
      > > > > > > RXTX audio input. I measured the TX output at the aerial
      > > socket via a
      > > > > > > 30dB attenuator into the SA. I played with RBW and reference level
      > > > > > > settings as I wasn't sure of the best way to show the results. The
      > > > > > > image *55.jpg is with the Softrock at full 1 Watt power with
      > > sensible
      > > > > > > modulation levels. The earlier SA scans are with about 750 mW
      > > output.
      > > > > > > The `scope scan is with the Softrock at full power. I can't
      > > see any
      > > > > > > clipping. I sniffed the output into a dummy load. I am sure I am
      > > > > > > doing some things wrong, criticism welcome. Thanks! Once I am more
      > > > > > > confident of technique I want to look at the output after power
      > > > > > > amplification.
      > > > > > >
      > > > > > > http://www.gatesgarth.com/IMD-SCANS/imd.zip
      > > > > > >
      > > > > > > --
      > > > > > > Best Regards,
      > > > > > > Chris Wilson. 2E0ILY
      > > > > > >
      > > > > > >
      > > > > >
      > > > >
      > > > >
      > > >
      > >
      > >
      >

    • warrenallgyer
      cause IMD, as measured using the ARRL protocol, is specified with reference to PEP, which is accepted to be 6 dB higher than the two tone level. I would be
      Message 61 of 61 , Jun 18 4:31 AM
      • 0 Attachment
        'cause IMD, as measured using the ARRL protocol, is specified with reference to PEP, which is accepted to be 6 dB higher than the two tone level.

        I would be interested to see the mil protocol and specification if you have it.

        Lucy
        (AKA Warren Allgyer - 9V1TD)


        --- In softrock40@yahoogroups.com, FRANCIS CARCIA <carcia@...> wrote:
        >
        > The ARRL spec is the 6dB gift method not the mil method. If you add 6dB to the two tone level why don't you add 6dB to the IP3 also since it is the result of the two tones. Splain that Lucy.
        > I'm attending the school of hard knocks trying to modify a 1200 watt MRI board to HF. Fairly easy to make power but a different issue to get good IMD on all bands.
        > Frank WA1GFZ
        >
        >
        >
      Your message has been successfully submitted and would be delivered to recipients shortly.