EMRFD Message Archive 5478
Message Date From Subject 5478 2010-11-29 00:23:50 Nick Receiver planning-filling in the blanks
some months ago I decided to improve my understanding on some issues that most of the times I take for granted. Having in hand a good set of books (EMFRD,IRFD,SSD) and the internet, I try to understand "the why's and how's" of RF design.
One of the domains on which I feel I am overloaded and ready to crash at the moment is receiver planning!
I thing I have some gray areas that need some clarification and I would be glad if someone could push me to the right direction.I will try to keep the story as short as possible (and still it is too long,sorry for this).
From what I am reading, as a receiver front end we consider all the stages before the main selectivity stage, for example, in a superhet I assume it is everything before the first IF crystal
filter. The front end defines the receiver noise figure and dynamic range. By using the cascaded equations and the cascade.exe program I followed all the examples in EMRFD and SSD and I have understood how each stage affects the total performance.My question is, why during this calculations we do not take into account the IF and AF stages? they also contribute in noise, and I can (only) guess that they may also limit the dynamic range.
Up to which stage the cascading calculation should be performed and how we put into the calculations the performance of the stages following the main selectivity (if it is required) ?I mean, when I know the NF, IIP3 and gain of the IF amp should I plug them into cascade.exe or I will get wrong results?
The only thought that seems reasonable to me is that after the main selectivity,in contrast to the fronted which is exposed to multiple signals, only the desired signal is present and as a
single singal it cannot generate IMD on the IF amplifier.Still, the IIP3 of the IF amp should be large enough to prevent gain compression.So, in total, the IF and AF stages need to provide in a well distributed manner, the rest of the required gain (that is, the total required gain minus the total front end gain)...but I still do not understand completely how they affect the receiver performance,and thus cannot understand how to define their performance requirements (mainly the max NF,IIP3 of IF amp) so that the required receiver performance can be achieved.
Am I thinking correct or I am miissing something?If there is something wrong in the way I understand it, could you give me a small kick to the right direction?
thanks in advance for your time,
Nick,SV1DJG
[Non-text portions of this message have been removed]5479 2010-11-29 07:50:42 Roelof Bakker Re: Receiver planning-filling in the blanks
RADCOM for August and September 2009 has a two part article "Noise
considerations in Receiver Design", written by GJ3RAX.
It shows that it is good practice to use an extra filter at the end of the
IF chain.
An other idea is to use distributed selectivity in the form of a four pole
ladder filter between the mixer and the first IF amplifier, between each
IF amplifier and between the last IF amplifier and the detector.
73,
Roelof, pa0rdt5480 2010-11-29 08:34:38 Leon Heller Re: Receiver planning-filling in the blanks 5481 2010-11-29 09:06:13 Weddig, Henning-C... Re: Receiver planning-filling in the blanks
concering the AGC action:
a narrowband filter in the loop will cause a big time delay and then
possibly render the AGC loop unstable!!!
See my findings in the German magazine "UKW Berichte" and the English
translation in VHF communication!
Regards
Henning Weddig
DK5LV
Am 29.11.2010 17:34, schrieb Leon Heller:
>5482 2010-11-29 09:31:32 Tim Re: Receiver planning-filling in the blanks
e.g. for a simple diode mixer + crystal filter + IC IF amp configuration, it's pretty easy to hear the noise from say a MC1590 as a limiting factor to system performance.
Or use an op-amp as the first gain stage in a direct conversion receiver and be limited by its noise performance (Fig 6.29).
Or compare a good audio output amp with a not so good one and hear the harmonic distorti5483 2010-11-29 09:47:21 William Carver Re: Receiver planning-filling in the blanks
IF amplifier noise from the opposite sideband. Since time delay of that
filter decreases stability of the AGC feedback loop, I put AGC detector
BEFORE the filter. Not a big problem since AGC threshold is much higher
than MDS.
But today an image-reject mixer (product detector) is so simple, and 20+
dB of opposite sideband is so easy to achieve, the post-IF filter might
be replaced with (say) a Tayloe mixer or a more conventional image
reject mixer..
Bill W7AAZ
On Mon, 2010-11-29 at 16:50 +0100, Roelof Bakker wrote:
> Hello Nick,
>
> RADCOM for August and September 2009 has a two part article "Noise
> considerations in Receiver Design", written by GJ3RAX.
> It shows that it is good practice to use an extra filter at the end of the
> IF chain.
>
> An other idea is to use distributed selectivity in the form of a four pole
> ladder filter between the mixer and the first IF amplifier, between each
> IF amplifier and between the last IF amplifier and the detector.
>
> 73,
> Roelof, pa0rdt5484 2010-11-29 11:19:04 Leon Heller Re: Receiver planning-filling in the blanks 5487 2010-11-29 12:02:37 William Carver Re: Receiver planning-filling in the blanks
delay could be under 1 millisecond and perhaps not unsettle the AGC loop
too much. And being passive is an endearing quality. But is DOES need to
reject noise from the opposite sideband or that noise will be folded
into the detector output and become inseparable.
"Zeros" don't affect time delay, so perhaps a half-lattice configuration
adjusted to have a notch in the opposite sideband would accomplish the
desired noise reduction even if the filter center frequency was at 9+
MHz.
But an image-reject mixer, even with a very simple I-Q combiner, could
provide 20 dB of image noise reduction. Of course it adds at least a
dual op amp and would not be passive. But you need to have a product
detector anyway.
The wonderful thing about homebrewing is one can choose a method that
suits one's inventory of components, or satisfies one's instrumentation
or design philosophy. Either approach can be made to work. I've done
noise filters, I'm looking forward to making an image reject product
detector.
Bill W7AAZ
On Mon, 2010-11-29 at 19:18 +0000, Leon Heller wrote:
>5489 2010-11-29 15:31:19 Wes Re: Receiver planning-filling in the blanks
Your thinking seems to be the same as that used to assemble the program CASCADE08. I'm pleased to see that the program is being used.
One of the critical assumptions made when writing the program is that filtering is absolute. Hence, when a signal is out of the passband of a filter, absolutely nothing gets through to the stages that follow. Receivers are not that simple, but it's a good start. This is why two tone dynamic range can be such a strong functi5493 2010-11-30 02:44:56 Nick Re: Receiver planning-filling in the blanks 5494 2010-11-30 02:49:43 Nick Re: Receiver planning-filling in the blanks
5497 2010-11-30 13:03:48 Wes Re: Receiver planning-filling in the blanks
You said, in part:
> crystal filter, so at the input of the IF amp the power level can vary between (-140.50 + 10) to (-140.50+10+98.17), that is -130.50dbm to -32.33dbm, before any detectable IMD products are generated into the front-end. If the IF amp has an IIP3 of +5dbm then its 3rd order.....
You are adding 10 dB to the levels. Generally, the definitions that I've been using are based upon 3 dB additions. That is, the MDS is a mere 3 dB above the noise floor of a receiver. We increase the strength of two input signals or tones to produce an IMD response that is also 3 dB above the noise floor. The two tone DR is then defined, in the program and in EMRFD, as the difference in dB between one of the two equal "loud" tones and the MDS. Adding 10 dB to the levels is complicated, for the IMD products increase at a cubic rate while a response from a signal near the MDS will increase linearly. See the discussion on page 6.29.
You can build your own sources for IMD testing. Some rather strong crystal controlled sources are shown in Chapter 7. I have also had good luck in doing IMD measurements with the general purpose generator of Fig 7.27 on page 7.15. The key to getting good performance is the common base buffer amplifier that follows the oscillators. The reverse isolation is generally very good with a common base stage.
MDS can be measured with a separate source like that in Fig 7.31. I still have the same source that was presented in a 1975 QST paper on DR definition. I've checked the calibration of the source from time to time and it continues to hang in there.
Anyway, don't hesitate to build your own sources. It's not all that difficult and it opens up a world of possibilities.
73, Wes
w7zoi