Message	Date	From	Subject
7802	2012-09-21 19:51:39	jwharding3	FET Balanced Amp Mixer Termination
Very glad to see a rekindling of mixer interest. Might I have some consideration of the Harris RF590 first mixer termination using a balanced amp - rather than two frontend xtal filters are shown in current designs. This would seem a further adaptation of Dr Rohde's CP643 FET termination from Ham Radio article. Am I correct that the hybrid splitter ahead of the Harris balanced amps assures the mixer sees the 50 ohm resistor at all frequencies? Does the use of the CP643 in Rohde's design also insure that all legs of the splitter are 50 ohms except the mixer input? I note that neither the Harris balanced amp nor the Rohde FET termination has a diplexer. Is one advisable? Could the guru's please provide some thoughts on the RF590 - design, with the use of CP643's, I am considering use with a 5 to 9 MHZ IF. I just feel the second filter could be better used later in the signal chain rather than just to provide an impedance matching tool. I have added a folder "Mixer Term Balanced Amp" that shows both circuits - Best wishes- John
7808	2012-09-22 14:31:58	William Carver	Re: FET Balanced Amp Mixer Termination
Your question, John, prompts this reply. However I am an amateur, not a guru, so while I think it's essentially correct, someone who really IS a guru might say "not quite". Learning is a never-ending process so I'll be watching for info that improves my own understanding. A bridged-T diplexer, with series tuned circuit to the next stage, typically has a Q of 1. If once is careful to use good VHF layout, the leg with a 50 ohm resistor to the parallel cpacitor of the LC can be a pretty good termination to the sum, and all the VHF signals that can be circulating in a diode mixer and need termination. While I don't think a switching mixer (H-mode, etc) is as sensitive to termination, the RLC diplexer is inexpensive and simple insurance. The CP643 is an elegant way to provide a broadband termination to the mixer. Its output (amplified) signal is also broadband, so any need for selectivity has been passed to the next stage, as in "passing the buck". The lossless feedback amplifiers don't have the VHF bandwidth of the CP643. The CP643 has a higher noise figure than a lossless feedback amplifier. A Q=1 LC diplexer nicely isolates the lossless feedback amplifier from the mixer at VHF where its input impedance is probably far from 50 ohms, while the combination has a noise figure of perhaps 1.2 dB for an HF i.f. frequency, several dB better than a CP643. But either amplifier is still seeing a wide spread of signals because there's no narrowband selectivity ahead of it. I suspect the Harris folks, faced with the issue of producing matched VHF filters, decided to build a very strong first IF amplifier and defer selectivity to a point where they could use a single filter. IT IS MY GUESS that the splitter, dual amplifiers and combiner was chosen to get 3 dB more output power capability, rather than for an impedance matching concern. That arrangement also allows one amplifier to fail, and the receiver still works. Quadrature splitters that drive IDENTICAL circuits do have a 50 ohm input impedance regardless of the impedance of the embedded circuits, whether amplifiers or filters. So those circuits can be narrowband filters with wild variation in input impedance but the mixer sees 50 ohms. So everything operates at 50 ohms. Wonderful except for needing two filters. "Something for nothing" just isn't in the cards. Double the number of filters allows you to move your selectivity ahead of any gain stages. You pay with two filters. Or build a stronger first i.f. amplifier (single ended or the one Harris used) and you only need one filter. Be aware, however, there is a price: filter crystals have third order intercepts just like active stages, often LOWER. So hitting them with signals amplified by "X" db increases their IMD by 2X dB. Whereas embedding them ahead of amplifiers, between two splitters, REDUCES the signal level by 3dB because each filter gets half of the i.f. power. So IMD is REDUCED by 6 dB if the filter is the first thing to produce IMD in the radio. If you don't contest with 5 element monoband yagis on 150 foot towers, etc you might never see a situation where the two-filter radio was crucial. The two filter setup is definitely harder to pull off. On the other hand, modest reduction of the amplitude of off-frequency signals seen by the first amplifier stage is all that's needed to protect it, and subsequent stages. Simple 2 or 3-pole ladder filters can do that protection job without costing an arm and leg. Interesting stuff. W7AAZ

Your question, John, prompts this reply. However I am an amateur, not a
guru, so while I think it's essentially correct, someone who really IS a
guru might say "not quite". Learning is a never-ending process so I'll
be watching for info that improves my own understanding.

A bridged-T diplexer, with series tuned circuit to the next stage,
typically has a Q of 1. If once is careful to use good VHF layout, the
leg with a 50 ohm resistor to the parallel cpacitor of the LC can be a
pretty good termination to the sum, and all the VHF signals that can be
circulating in a diode mixer and need termination.

While I don't think a switching mixer (H-mode, etc) is as sensitive to
termination, the RLC diplexer is inexpensive and simple insurance.

The CP643 is an elegant way to provide a broadband termination to the
mixer. Its output (amplified) signal is also broadband, so any need for
selectivity has been passed to the next stage, as in "passing the
buck".

The lossless feedback amplifiers don't have the VHF bandwidth of the
CP643. The CP643 has a higher noise figure than a lossless feedback
amplifier. A Q=1 LC diplexer nicely isolates the lossless feedback
amplifier from the mixer at VHF where its input impedance is probably
far from 50 ohms, while the combination has a noise figure of perhaps
1.2 dB for an HF i.f. frequency, several dB better than a CP643.

But either amplifier is still seeing a wide spread of signals because
there's no narrowband selectivity ahead of it.

I suspect the Harris folks, faced with the issue of producing matched
VHF filters, decided to build a very strong first IF amplifier and defer
selectivity to a point where they could use a single filter. IT IS MY
GUESS that the splitter, dual amplifiers and combiner was chosen to get
3 dB more output power capability, rather than for an impedance matching
concern. That arrangement also allows one amplifier to fail, and the
receiver still works.

Quadrature splitters that drive IDENTICAL circuits do have a 50 ohm
input impedance regardless of the impedance of the embedded circuits,
whether amplifiers or filters. So those circuits can be narrowband
filters with wild variation in input impedance but the mixer sees 50
ohms. So everything operates at 50 ohms. Wonderful except for needing
two filters.

"Something for nothing" just isn't in the cards. Double the number of
filters allows you to move your selectivity ahead of any gain stages.
You pay with two filters.

Or build a stronger first i.f. amplifier (single ended or the one Harris
used) and you only need one filter. Be aware, however, there is a price:
filter crystals have third order intercepts just like active stages,
often LOWER. So hitting them with signals amplified by "X" db increases
their IMD by 2X dB. Whereas embedding them ahead of amplifiers, between
two splitters, REDUCES the signal level by 3dB because each filter gets
half of the i.f. power. So IMD is REDUCED by 6 dB if the filter is the
first thing to produce IMD in the radio.

If you don't contest with 5 element monoband yagis on 150 foot towers,
etc you might never see a situation where the two-filter radio was
crucial. The two filter setup is definitely harder to pull off.

On the other hand, modest reduction of the amplitude of off-frequency
signals seen by the first amplifier stage is all that's needed to
protect it, and subsequent stages. Simple 2 or 3-pole ladder filters can
do that protection job without costing an arm and leg.

Interesting stuff.

W7AAZ