EMRFD Message Archive 3626
Message Date From Subject 3626 2009-10-22 10:43:25 Glen Leinweber IP3 specs , mixer specs One wonders if the way we specify and measure
amplifier or mixer parameters is proper - audiophiles
continually argue about spec validity.
One complaint involves IP3 measurements that are
described qualitatively as "well behaved" or "ill behaved"
where the intercept is found from measurements made
not far above the noise level, then extrapolated to the
far end (way beyond amplifier or mixer distortion levels).
Turns out that in-between, distortion products don't follow
a predictable relation to signal levels, or at least don't
follow the theory. So one amplifier or mixer having the
same IP3 as another may give more or less distortion
(at some signal levels). A graph of IP3 vs. signal level
tells a better story than just "IP3 intercept = +10dBm".
One always strives to quantify tersely, yet completely.
Guess I'm wondering if IP3 can be better quantified,
Same goes with mixers. We've discussed this briefly on
this list before, where one mixer generates a lot fewer
distortion products than another. A proper multiplying
mixer is better in this respect than a switching mixer.
But I can't find a spec that quantifies this aspect. When
one sees a really lousy mixer noise figure, one wonders
where all that extra noise comes from.
Of course the real goal of specs is to try to spotlight
those receivers that are a joy to listen to. Guess I'm
nit-picking incomplete specs that might be improved
to tell a better story. Can anyone suggest improvements?
3628 2009-10-22 13:24:14 w7zoi Re: IP3 specs , mixer specs Hi Glen and gang,
Yes, the specification of both receivers and of components (mixers, amplifiers, switches, and even filters) is complicated. A complete specification with graphs is the better way to do comparisons.
That said, it is extremely useful to have a single number, an "intercept" for a component. This is for the chore of prediction. If you know the intercepts, gains, and noise figures of all of the elements in a front end, where the front end is all of the stuff ahead of the bandwidth determining filter, you can then do calculations that will tell you what a system will do. I've generally had pretty good luck with this. For receiver work, one generally wants mixer intercepts that are measured at pretty low levels values determined at the higher levels where measurements are easier. In this vein, I've always liked actual dynamic range measurements for a receiver, for it is a worst case.
The situation is slightly different when you are trying to evaluate a receiver that might be used for measurements. You can often live with some low level sins in such a system, although it is not ideal.
This does not address the question of "what makes a receiver sound good?" You can have receivers with a stellar DR that sound like junk. Such is the case, for example, if a receiver has too much IF gain and a wimpy product detector. This can lead to intermodulation between the noise in the IF and a strong carrier. A 20 dB pad ahead of the product detector can sometimes clean things considerably. Often the measurement we want to do is the in-band intermodulation distortion. This requires a pair of tones that are only a hundred Hz apart or so, and a spectrum analyzer that has enough resolution to see the results. I'm told that the K3 from Elecraft is outstanding in this regard. I've never heard one and have certainly not measured one.
Mixer spurious responses are often well characterized. Take a look at the Mini Circuits data sheets for the mixers we often use. This is beyond the characterization for IMD and gain compression.
Rick and I have often batted the question around of just what it takes to make a receiver that "sounds good." It is really difficult to answer the question, but we both know that we recognize it when we do hear it. It is really hard to beat the sound of a good direct conversi
3629 2009-10-22 13:36:28 Leon Re: IP3 specs , mixer specs It's basically linearity isn't it? The more linear the whole system is, the better the performance.
3630 2009-10-22 15:34:15 Graham / KE9H Re: IP3 specs , mixer specs Mixers would not mix anything if they were linear.
3631 2009-10-22 15:42:17 leon Heller Re: IP3 specs , mixer specs ----- Original Message -----
3632 2009-10-22 15:51:58 Chris Trask Re: IP3 specs , mixer specs >BZZZT!! A linear analogue multiplier would make an ideal mixer as there
> Mixers would not mix anything if they were linear.
would be no spurious byproducts.
,----------------------. High Performance Mixers and
/ What's all this \ Amplifiers for RF Communications
/ extinct stuff, anyhow? /
\ _______,--------------' Chris Trask / N7ZWY
3633 2009-10-22 17:58:12 dixonglennb Re: IP3 specs , mixer specs Greetings all,
Well, we have two blanket statements:
1. > Mixers would not mix anything if they were linear.
2. >A linear analogue multiplier would make an ideal mixer as there would be no spurious byproducts.
Who is correct? Actually, both are:
A system is linear if and only if it passes the law of superposition, which basically states input signals do not interfere with each other. If you include the local oscillator signal in your input signal set and apply the law of superposition, you will find the local oscillator makes the mixer violate the law: all mixers (including a pure multiplier) are therefore nonlinear in this respect. Try the superposition test yourself: it will fail. We want this nonlinearity and couldn't mix without it.
If we do not include our local oscillator in the input signal set but instead apply superposition only to the multitude of signals at the antenna, we find that good mixers will, for the most part, pass the law of superposition. They are linear in this sense. An ideal multiplier and an ideal passive switching mixer can be shown to be perfectly linear in this way (square law mixers are not). When linearity measurements such as IP3 are measured and printed in data sheets, it is this superposition test they are applying (by a two-tone test applied to the single RF port).
Isn't it comforting to know we can talk about mixer (non)linearity either way and not be wrong?
Okay...back in my hole...still designing my Homebrew Challenge transceiver. Think I might be a little late...
3634 2009-10-22 18:06:04 dixonglennb Re: IP3 specs , mixer specs Well, I knew as soon as I posted I would find something that doesn't make sense...
When I wrote 'basically that input signals do not interfere with each other'.... by 'interfere' I meant cross modulation and not the kind of interference you get when someone calls CQ without listening to the frequency first...
3638 2009-10-23 11:19:53 ep_mand Re: IP3 specs , mixer specs 3639 2009-10-23 11:49:27 James Duffey Re: IP3 specs , mixer specs Glen and Wes - This is an interesting topic you started.
One thing that we will have to face in the future is that some of the
figures of merit we use to evaluate receivers in the analog regime are
inappropriate or have significantly different implications for
receivers based on analog to digital converters (A/D Converters), see
for example SM5BVZ's work here:
< http://www.sm5bsz.com/dynrange/qex/digital-imd.pdf >
and on his web page:
< http://www.sm5bsz.com/dynrange/intermod.htm >
where he discusses the problems of third order IMD and IP3 as figures
of merit with respect to receivers in general and includes A/D
converters as a sub set of this discussion. In modern A/d converters,
the IMD3 is essentially constant until the A/D saturates, then it
increases sharply, so that the IMD and IP3 figure of merit really
don't convey any significant additional information that the 1 dB
compression point or blocking dynamic range conveys. We clearly need
to think differently here.
I too have noticed that some receivers sound better than other
receivers with the same specifications or figures of merit. As Wes
notes, the direct conversion receiver is notable in this respect. I
think that this means that we should search for better figures of
merit. I am a bit reluctant to embrace strictly subjective listening
tests to denote differences in receiver performance, if one hears a
difference, it should be measurable and be able to expressed in
numbers. This is not a new idea with me, Lord Kelvin expressed it more
than 100 years ago.
The audio world collapsed into subjective evaluations of electronics
gear over 30 years ago in an irrational attempt to justify some
people's preferences for tube gear over solid state, and the result of
that is just plain silly in many cases today. I would hate to hate to
see the amateur radio world follow that example in a search for new
figures of merit. - Dr. Megacycle KK6MC/5
Cedar Crest NM
3641 2009-10-23 22:08:20 jr_dakota Re: IP3 specs , mixer specs This article may be 20 years old but it's still completely relevant
Science and Subjectivism in Audio by Douglas Self