EMRFD Message Archive 11462

Message Date From Subject
11462 2015-08-09 07:11:26 peter_dl8ov The Black Art of DBM Mixer Termination
I have been thinking about how to terminate the IF Out port of a double balanced mixer over a working frequency of 1-30 MHz. It seems to be an easy way to lose sleep :(


1) I cannot use a diplexer because of the wide frequency range.


2) If I use one of the W7ZOI 2N5109 based amplifiers the mixer is properly terminated but I also amplify all of the unwanted mixer products.


3) If I use a low-pass or bandpass filter (even one configured for 50 ohms at the wanted frequency) then IMD will suffer because unwanted products outside of the passband will not see a fifty ohm resistive load.


4) If I use an attenuator (I have seen suggestions of 6 dB) then there will be a good match on the port but my wanted signal will be attenuated.


Please could somebody explain a way to solve my dilemma? 


Peter DL8OV

11464 2015-08-09 07:39:47 Ashhar Farhan Re: The Black Art of DBM Mixer Termination
how about an active diplexer :
a feedback amplifier with a crystal in the feedback loop, terminated in a 50 ohms dummy load. at all frequencies away from the crystal filter, it will exhibit 50 ohms impedance and the input impedance will shoot up at the operating frequency of the crystal, thus transfering all the RF energy to the crystal filter. 
i know that this means it will be tough to balance it out. so, just use the fischer hybrid coupler and expend a few on additional crystals.

- f

11465 2015-08-09 07:47:34 Eduardo Alonso Re: The Black Art of DBM Mixer Termination
hello peter,

a 3dB directional coupler (hybrid transformer) has the propierties you are looking for.

it stabilizates the impedance to 50ohm in a high bandwidth
it reduces the insertion losses in the postmixer filter

you can see an example here

http://martein.home.xs4all.nl/pa3ake/hmode/frontend-board.pdf

73


 
 

11466 2015-08-09 08:28:16 kb1gmx Re: The Black Art of DBM Mixer Termination

1) I cannot use a diplexer because of the wide frequency range.


Why not?  a high pass low pass combo will certainly terminate well outside the IF band pass.

Twin T also can do this as well as various hybrids.


2) If I use one of the W7ZOI 2N5109 based amplifiers the mixer is properly terminated but I also amplify all of the unwanted mixer products.


Always an issue, requires the amplifier not only have a good termination to all frequencies but also even gain over the range.  But it can be applied with the above filters to limit the range.  I have done this with excellent results (to the limits of the DBM imd).  The advantage of that is the prefilter amp has less demanding requirements that are easily satisfied.


The alternate to the 5109 is the CP666 power JFET from Crystalonics at 50ma. They are expensive 

but available.  The common gate amp using that part can provide both a 50 ohm input and wide 

bandwidth  with excellent input to output isolation.


3) If I use a low-pass or bandpass filter (even one configured for 50 ohms at the wanted frequency) then IMD will suffer because unwanted products outside of the passband will not see a fifty ohm resistive load.


Not true, conditionally.  If the filter are say a mix on high pass low pass and terminated the trick here 

is to route all unwanted signal to a 50 ohm termination.  The same hybrids (Fischer) used for IQ LO 

generation can be useful here as the 50 ohm resistor on one port is often referred to as the dump load, 

anything not absorbed by I or Q port ends up in the dump load and the source port sees 50 ohms.  This also works very well though the cost is about 3.5dB (split is 3db plus core loss of  .4-.5db). 


4) If I use an attenuator (I have seen suggestions of 6 dB) then there will be a good match on the port but my wanted signal will be attenuated.


This is also problematic for low noise receivers though useful deeper in the radio say at product detector

long after the  receiver noise figure has been established .  In the end you need a loss from the antenna 

to the IF input under 12Db or so less being better.   Do that entirely passive and not having a front end noise 

figure of 20db is a challenge but also doable.  It does require attention to detail.  Also EMRFD gives the 

template for many approaches.


5) use a mixer that tolerates the filter impedance out of band.  This excludes the DBM but does not

eliminate many active an FET switch based mixers.  Even if its is not 100% immune to load reactance

combined with less lossy termination approaches can still be effective.


6)  This applies most to out friend is EU countries though less so now that in years past. I believe some 

here in the USA and elsewhere also suffer this.  Their greatest source is not close in (in frequency) 

sources but close high power signals out of band (AMBC, SWBC, FMBC, TV).  This suggests that 

filtering out effectively signal that are not in the ham bands has great value.  This includes specific 

traps, narrow bandpass, narrow tuned bandpass specifically Preselectors), and other techniques 

to keep out of band signals from ever reaching The RF amp (if used) and mixer.


Also there is no harm or foul in having and RF amp that can be switched out for those times where

adjacent frequency interference is not an issue but noise figure (sensitivity) is.


All of the above suggests with some reality there are few solutions that work for all cases without 

some compromise or balancing of priorities in the receiver.  What works for 160 though 40 is useless 

at 10M and may be less than optimum for 600M.



Allison






11467 2015-08-09 08:46:53 in3otd Re: The Black Art of DBM Mixer Termination
Hello,
as Allison said a low-pass/high-pass combination will work as a wideband diplexer.;
here is a link to an example schematic http://oi59.tinypic.com/98gcq0.jpg

and here is its response http://oi58.tinypic.com/rk7hxj.jpg

 

73 de Claudio, IN3OTD







11468 2015-08-09 09:04:37 peter_dl8ov Re: The Black Art of DBM Mixer Termination
Thank you for the replies but I need to point out that this is a mixer in a transmitter. The LO Port is fed with 45-75 MHz and the RF port is fed with a fixed 45 MHz signal with the output from the IF Port.

Crystal based termination solutions will therefore not work but I like the idea of LPF/HPF based circuits. I seem to remember that there was a power amplifier designed by Bill Sabin that used a similar system.

Peter DL8OV
11469 2015-08-09 09:12:26 Chris Trask Re: The Black Art of DBM Mixer Termination
>
>Thank you for the replies but I need to point out that this is a mixer in a transmitter.
>The LO Port is fed with 45-75 MHz and the RF port is fed with a fixed 45 MHz signal with
>the output from the IF Port.
>

I suspected that this was the application. With that LO and RF scheme you have eliminated the problem of signal leakage to the IF port as you can easily filter that out. I have done the same thing with wideband receiver front ends.

That being the case, could you describe the details of the spurious outputs? Are they just a matter of the usual close-in IMD products, or something more serious?


Chris Trask
N7ZWY / WDX3HLB
Senior Member IEEE
http://www.home.earthlink.net/~christrask/
11470 2015-08-09 09:34:19 John Marshall Re: The Black Art of DBM Mixer Termination
The challenge is to properly terminate the mixer so as to obtain its best IMD performance without compromising the signal-to-noise ratio. In a receiver you might want to go to extremes to keep the noise figure low but in a transmitter the signal-to-noise ratio should be high enough coming out of the mixer that you can tolerate some loss. Why not go with a 6 dB attenuator followed by some combination of bandpass filters and broadband amplifiers?

John, KU4AF
Pittsboro, NC

11471 2015-08-09 10:00:17 Ashhar Farhan Re: The Black Art of DBM Mixer Termination
it is a myth that transmit mixers can take losses. actually, they can't. any attenuation in the output of the transmit mixer, will drop the IMDR by that much in the succeeding stages. consider this :
a 7dbm mixer, with 15dbm IIP3, should have all the IMDs below -50dbc to be safely make it through the PA at -40dbc. That would mean, that the input levels should be at -10dbm, and the output at -17dbm. 
An amplification from -17dbm to 50dbm (100 watts output) is a gain of 77db. any noise generated in the early transmitter stages could easily be substantial in the output of the transmitter.  consider a 30db noise figure. -174 + 35 + 33 is appox -100dbm, with 77db amplification is will become -27dbm, or 3 microwatt. 
that is substantial noise being transmitted out of band. enough to drown any other ham down the block.

- f

11472 2015-08-09 10:15:09 Ashhar Farhan Re: The Black Art of DBM Mixer Termination
oops. i meant 2 uW

- f

11473 2015-08-09 12:21:36 John Marshall Re: The Black Art of DBM Mixer Termination
OK, I make it -174 dBm +35 +33 is -106 dBM. And -17 to 50 dBm is 67 dB gain rather than 77 dBm. So -106 dBm plus 67 dB is -39 dBm or 0.126 uW. Perhaps they'll only hear it next door! Nevertheless, point taken - noise is important in the transmit strip. What noise power would be acceptable in a 100 Watt transmitter?

John, KU4AF
Pittsboro, NC

11474 2015-08-10 05:30:44 k1rf_digital_stev... Re: The Black Art of DBM Mixer Termination
Don't quite understand your comment "1) I cannot use a diplexer because of the wide frequency range."  If you rally need the wideband output, you can design an excellent wideband diplexer with Tonne Software's Diplexer program. See: Diplexer - from Tonne Software

  You can design very high order diplexers with this software and it even displays return loss.  Have fun winding lots of toroids!!


"Digital Steve", K1RF