There seems to be some difference of opinion on how to properly
terminate Tayoe detectors (and other similar circuits) at audio
frequencies. See the following published examples:

(1)G0BBL et al (QRP2001): load resistors are 220 ohms
(2)Hans Summers: load resistors are 1k
(3)9A2HL (on Hans Summers' website): Approx. 0 ohms, detector goes
directly to op-amp pins
(4)AC5OG (recent QEX series, Part 1): Approx. 0 ohms, detector goes
directly to op-amp pins
(5)AC5OG (recent QEX series, Part 4): 60 megohms (doubly balanced
circuit).

In addition references 1-3 use a series 47 ohm resistor at the input
of the detector, which seems unnecssary to me - the 50 ohm antenna
should perform the function of the filter resistor, shouldn't it ?

I did some Spice modelling and found that my detector was well matched
to the 50 ohm RF source for audio load resistances of a few hundred
ohms (470 ohms worked well), and the conversion loss was a bit more
than 3 dB when matched. With somewhat lower resistances the match was
not as good but the conversion loss was reduced - 220 ohms may be
close to optimum for both reasonable input match to 50 ohms and
conversion loss. With higher (and much lower) resistances both
conversion loss and input match degraded. The integrator capacitors
were 0.1uF in this model.

Note that I am defining conversion loss as the relationship between
the total power delivered to the 4 load resistances and the forward
power at the input. With high load resistances the voltage at each
load is indeed closer to the input voltage than for lower load
resistances, but most of the power incident at the mixer is reflected
back to the antenna. Therefore the conversion loss (defined in terms
of power, which is what matters for noise figure) goes up.

I would be interested in the comments of those of you with some
experience with these detectors. Also, can I assume (for the purpose
of computing noise figure of the audio preamps) that the source
impedance seen by the audio amps is simply equal to the load
resistance that gives the best match at the mixer input ?

Steve, VE3SMA

In article ,
(Steve Kavanagh) writes:

There seems to be some difference of opinion on how to properly
terminate Tayoe detectors (and other similar circuits) at audio
frequencies. See the following published examples:

(1)G0BBL et al (QRP2001): load resistors are 220 ohms
(2)Hans Summers: load resistors are 1k
(3)9A2HL (on Hans Summers' website): Approx. 0 ohms, detector goes
directly to op-amp pins
(4)AC5OG (recent QEX series, Part 1): Approx. 0 ohms, detector goes
directly to op-amp pins
(5)AC5OG (recent QEX series, Part 4): 60 megohms (doubly balanced
circuit).

In addition references 1-3 use a series 47 ohm resistor at the input
of the detector, which seems unnecssary to me - the 50 ohm antenna
should perform the function of the filter resistor, shouldn't it ?

I did some Spice modelling and found that my detector was well matched
to the 50 ohm RF source for audio load resistances of a few hundred
ohms (470 ohms worked well), and the conversion loss was a bit more
than 3 dB when matched. With somewhat lower resistances the match was
not as good but the conversion loss was reduced - 220 ohms may be
close to optimum for both reasonable input match to 50 ohms and
conversion loss. With higher (and much lower) resistances both
conversion loss and input match degraded. The integrator capacitors
were 0.1uF in this model.

Note that I am defining conversion loss as the relationship between
the total power delivered to the 4 load resistances and the forward
power at the input. With high load resistances the voltage at each
load is indeed closer to the input voltage than for lower load
resistances, but most of the power incident at the mixer is reflected
back to the antenna. Therefore the conversion loss (defined in terms
of power, which is what matters for noise figure) goes up.

I would be interested in the comments of those of you with some
experience with these detectors. Also, can I assume (for the purpose
of computing noise figure of the audio preamps) that the source
impedance seen by the audio amps is simply equal to the load
resistance that gives the best match at the mixer input ?

I'd suggest you take a look at Dan Tayloe's own article in
RF Design for 1 March 2003. You can get a printable copy at

http://www.rfdesign.com

If you search for "Tayloe." Unfortunately, the figures aren't
available for downloading (I have my own paper copy) but the
details on the bits and pieces are all there. Good article.

Len Anderson
retired (from regular hours) electronic engineer person

In article ,
(Steve Kavanagh) writes:

There seems to be some difference of opinion on how to properly
terminate Tayoe detectors (and other similar circuits) at audio
frequencies. See the following published examples:

(1)G0BBL et al (QRP2001): load resistors are 220 ohms
(2)Hans Summers: load resistors are 1k
(3)9A2HL (on Hans Summers' website): Approx. 0 ohms, detector goes
directly to op-amp pins
(4)AC5OG (recent QEX series, Part 1): Approx. 0 ohms, detector goes
directly to op-amp pins
(5)AC5OG (recent QEX series, Part 4): 60 megohms (doubly balanced
circuit).

In addition references 1-3 use a series 47 ohm resistor at the input
of the detector, which seems unnecssary to me - the 50 ohm antenna
should perform the function of the filter resistor, shouldn't it ?

I did some Spice modelling and found that my detector was well matched
to the 50 ohm RF source for audio load resistances of a few hundred
ohms (470 ohms worked well), and the conversion loss was a bit more
than 3 dB when matched. With somewhat lower resistances the match was
not as good but the conversion loss was reduced - 220 ohms may be
close to optimum for both reasonable input match to 50 ohms and
conversion loss. With higher (and much lower) resistances both
conversion loss and input match degraded. The integrator capacitors
were 0.1uF in this model.

Note that I am defining conversion loss as the relationship between
the total power delivered to the 4 load resistances and the forward
power at the input. With high load resistances the voltage at each
load is indeed closer to the input voltage than for lower load
resistances, but most of the power incident at the mixer is reflected
back to the antenna. Therefore the conversion loss (defined in terms
of power, which is what matters for noise figure) goes up.

I would be interested in the comments of those of you with some
experience with these detectors. Also, can I assume (for the purpose
of computing noise figure of the audio preamps) that the source
impedance seen by the audio amps is simply equal to the load
resistance that gives the best match at the mixer input ?

I'd suggest you take a look at Dan Tayloe's own article in
RF Design for 1 March 2003. You can get a printable copy at

http://www.rfdesign.com

If you search for "Tayloe." Unfortunately, the figures aren't
available for downloading (I have my own paper copy) but the
details on the bits and pieces are all there. Good article.

Len Anderson
retired (from regular hours) electronic engineer person

(Avery Fineman) wrote in message ...

I'd suggest you take a look at Dan Tayloe's own article in
RF Design for 1 March 2003. You can get a printable copy at

http://www.rfdesign.com

If you search for "Tayloe." Unfortunately, the figures aren't
available for downloading (I have my own paper copy) but the
details on the bits and pieces are all there.

Unfortunately, not only are the figures missing but so are most of the
equations, so I am unable to follow the analysis.

Good article.

It does look interesting...I'll have to try to track down a paper
copy. One of the problems with being laid off...I'm not getting my
subscription to RF Design !

Steve VE3SMA

(Avery Fineman) wrote in message ...

I'd suggest you take a look at Dan Tayloe's own article in
RF Design for 1 March 2003. You can get a printable copy at

http://www.rfdesign.com

If you search for "Tayloe." Unfortunately, the figures aren't
available for downloading (I have my own paper copy) but the
details on the bits and pieces are all there.

Unfortunately, not only are the figures missing but so are most of the
equations, so I am unable to follow the analysis.

Good article.

It does look interesting...I'll have to try to track down a paper
copy. One of the problems with being laid off...I'm not getting my
subscription to RF Design !

Steve VE3SMA

Some quick notes on the application of the Tayloe mixer to answer some
of your questions:

Recovered power is not relevant for this detector depending on the kind
of post detector pre-amp is used. Detected power might be more a
concern if you were driving something like a discrete bipolar transistor
post detector amplifier.

Infinite input impedance (such as the + side of an op-amp) allows for
best detected voltage and best adjacent frequency roll-off.
Instrumentation amplifiers do this best, but are more noisy when run at
reasonable gains. Remember, the detector acts as a bandpass filter.
Best frequency roll off gives the best high signal performance.

You can also present feed the "-" side of an op-amp, and use the
detector more like a current source. The detector bandwidth is more
flat when this is done.

The best low noise compromise seems to be to feed one side into "-" and
one side into "+" on the post detector amplifier. The balance is
compromised a bit, and the detector roll off is not a sharp, but it does
give best sensitivity without having to resort to an instrumentation amp
with 1000x gain, which would kill the dynamic range of the front end.
"Low noise" instrumentation amps are kind of noisy unless they are run
high gain. A gain of 15x is the right kind of ball park for the voltage
gain of the post detector preamp.

When a 50 ohm source is used, the 4:1 multiplexing action makes each
output look like a 200 ohm impedance (neglecting switching times and
switch resistance), so 220 ohms is close to the best power match.

Any resistance in series with the following preamplifier introduces
noise, which is not a good thing when best sensitivity is the
objective. Some series resistors in the inputs to an op-amp, can give
better frequency roll off (high signal level performance) at a cost of
some sensitivity (a trade off). Modeling can show you the roll off
trade, calculations can give you the noise voltage (sensitivity) impact.
220 ohms is a high here (over double the noise voltage of a 50 ohm
system). This resistance matters less if you are simply interested in
ok sensitivity performance, which is a "don't care" on something like
40m and lower.

The 50 ohm on the input to the detector is not needed. The system
impedance provides an equivalent input "R" for the detector caps to work
against.

A low noise op-amp is desirable for the post detector amplifier. 50 ohm
system noise is around 0.85 nV/squarerootHz. Some opamps have noise
performance to match this such as the LT1115 (0.85 nV/squarerootHz), but
are about $4 each and you will need two. I have heard that some folks
have built their own low noise preamps using 2N3904s, but I have not
really tried that.

More common decent op-amps are in the 4 nV/squarerootHz range, such as
the LM833 or the NE5532, which are under $0.50 in a dual opamp package.
Recognize that 0.85 nV/SqrtHz noise rating compared to a noise rating of
4 nV/SqrtHz is a 13.5 db (sensitivity!) difference.

Perfect receiver sensitivity (3 db S+N/N) is about -148 dbm with a 500
Hz bandwidth. In practice, I have gotten -142 dbm sensitivity
performance using a pair of LT1115s and no RF preamp. With this rig on
20m, QRP cw signals in a contest could be heard weakly but clearly that
were not there on a K2 (A/B comparison using the same antenna). With a
NE5532, expected performance is in the -134 to -136 dbm range.

Beware when an op-amp claims low noise! The specification sheet will
give the noise vs. frequency curve. Many "low noise" op-amps, are only
low noise at 100 KHz. Be sure to check the noise curve at 300 to 1000
Hz!

- Dan, N7VE

Steve Kavanagh wrote:

There seems to be some difference of opinion on how to properly
terminate Tayoe detectors (and other similar circuits) at audio
frequencies. See the following published examples:

(1)G0BBL et al (QRP2001): load resistors are 220 ohms
(2)Hans Summers: load resistors are 1k
(3)9A2HL (on Hans Summers' website): Approx. 0 ohms, detector goes
directly to op-amp pins
(4)AC5OG (recent QEX series, Part 1): Approx. 0 ohms, detector goes
directly to op-amp pins
(5)AC5OG (recent QEX series, Part 4): 60 megohms (doubly balanced
circuit).

In addition references 1-3 use a series 47 ohm resistor at the input
of the detector, which seems unnecssary to me - the 50 ohm antenna
should perform the function of the filter resistor, shouldn't it ?

I did some Spice modelling and found that my detector was well matched
to the 50 ohm RF source for audio load resistances of a few hundred
ohms (470 ohms worked well), and the conversion loss was a bit more
than 3 dB when matched. With somewhat lower resistances the match was
not as good but the conversion loss was reduced - 220 ohms may be
close to optimum for both reasonable input match to 50 ohms and
conversion loss. With higher (and much lower) resistances both
conversion loss and input match degraded. The integrator capacitors
were 0.1uF in this model.

Note that I am defining conversion loss as the relationship between
the total power delivered to the 4 load resistances and the forward
power at the input. With high load resistances the voltage at each
load is indeed closer to the input voltage than for lower load
resistances, but most of the power incident at the mixer is reflected
back to the antenna. Therefore the conversion loss (defined in terms
of power, which is what matters for noise figure) goes up.

I would be interested in the comments of those of you with some
experience with these detectors. Also, can I assume (for the purpose
of computing noise figure of the audio preamps) that the source
impedance seen by the audio amps is simply equal to the load
resistance that gives the best match at the mixer input ?

Steve, VE3SMA

Some quick notes on the application of the Tayloe mixer to answer some
of your questions:

Recovered power is not relevant for this detector depending on the kind
of post detector pre-amp is used. Detected power might be more a
concern if you were driving something like a discrete bipolar transistor
post detector amplifier.

Infinite input impedance (such as the + side of an op-amp) allows for
best detected voltage and best adjacent frequency roll-off.
Instrumentation amplifiers do this best, but are more noisy when run at
reasonable gains. Remember, the detector acts as a bandpass filter.
Best frequency roll off gives the best high signal performance.

You can also present feed the "-" side of an op-amp, and use the
detector more like a current source. The detector bandwidth is more
flat when this is done.

The best low noise compromise seems to be to feed one side into "-" and
one side into "+" on the post detector amplifier. The balance is
compromised a bit, and the detector roll off is not a sharp, but it does
give best sensitivity without having to resort to an instrumentation amp
with 1000x gain, which would kill the dynamic range of the front end.
"Low noise" instrumentation amps are kind of noisy unless they are run
high gain. A gain of 15x is the right kind of ball park for the voltage
gain of the post detector preamp.

When a 50 ohm source is used, the 4:1 multiplexing action makes each
output look like a 200 ohm impedance (neglecting switching times and
switch resistance), so 220 ohms is close to the best power match.

Any resistance in series with the following preamplifier introduces
noise, which is not a good thing when best sensitivity is the
objective. Some series resistors in the inputs to an op-amp, can give
better frequency roll off (high signal level performance) at a cost of
some sensitivity (a trade off). Modeling can show you the roll off
trade, calculations can give you the noise voltage (sensitivity) impact.
220 ohms is a high here (over double the noise voltage of a 50 ohm
system). This resistance matters less if you are simply interested in
ok sensitivity performance, which is a "don't care" on something like
40m and lower.

The 50 ohm on the input to the detector is not needed. The system
impedance provides an equivalent input "R" for the detector caps to work
against.

A low noise op-amp is desirable for the post detector amplifier. 50 ohm
system noise is around 0.85 nV/squarerootHz. Some opamps have noise
performance to match this such as the LT1115 (0.85 nV/squarerootHz), but
are about $4 each and you will need two. I have heard that some folks
have built their own low noise preamps using 2N3904s, but I have not
really tried that.

More common decent op-amps are in the 4 nV/squarerootHz range, such as
the LM833 or the NE5532, which are under $0.50 in a dual opamp package.
Recognize that 0.85 nV/SqrtHz noise rating compared to a noise rating of
4 nV/SqrtHz is a 13.5 db (sensitivity!) difference.

Perfect receiver sensitivity (3 db S+N/N) is about -148 dbm with a 500
Hz bandwidth. In practice, I have gotten -142 dbm sensitivity
performance using a pair of LT1115s and no RF preamp. With this rig on
20m, QRP cw signals in a contest could be heard weakly but clearly that
were not there on a K2 (A/B comparison using the same antenna). With a
NE5532, expected performance is in the -134 to -136 dbm range.

Beware when an op-amp claims low noise! The specification sheet will
give the noise vs. frequency curve. Many "low noise" op-amps, are only
low noise at 100 KHz. Be sure to check the noise curve at 300 to 1000
Hz!

- Dan, N7VE

Steve Kavanagh wrote:

There seems to be some difference of opinion on how to properly
terminate Tayoe detectors (and other similar circuits) at audio
frequencies. See the following published examples:

(1)G0BBL et al (QRP2001): load resistors are 220 ohms
(2)Hans Summers: load resistors are 1k
(3)9A2HL (on Hans Summers' website): Approx. 0 ohms, detector goes
directly to op-amp pins
(4)AC5OG (recent QEX series, Part 1): Approx. 0 ohms, detector goes
directly to op-amp pins
(5)AC5OG (recent QEX series, Part 4): 60 megohms (doubly balanced
circuit).

In addition references 1-3 use a series 47 ohm resistor at the input
of the detector, which seems unnecssary to me - the 50 ohm antenna
should perform the function of the filter resistor, shouldn't it ?

I did some Spice modelling and found that my detector was well matched
to the 50 ohm RF source for audio load resistances of a few hundred
ohms (470 ohms worked well), and the conversion loss was a bit more
than 3 dB when matched. With somewhat lower resistances the match was
not as good but the conversion loss was reduced - 220 ohms may be
close to optimum for both reasonable input match to 50 ohms and
conversion loss. With higher (and much lower) resistances both
conversion loss and input match degraded. The integrator capacitors
were 0.1uF in this model.

Note that I am defining conversion loss as the relationship between
the total power delivered to the 4 load resistances and the forward
power at the input. With high load resistances the voltage at each
load is indeed closer to the input voltage than for lower load
resistances, but most of the power incident at the mixer is reflected
back to the antenna. Therefore the conversion loss (defined in terms
of power, which is what matters for noise figure) goes up.

I would be interested in the comments of those of you with some
experience with these detectors. Also, can I assume (for the purpose
of computing noise figure of the audio preamps) that the source
impedance seen by the audio amps is simply equal to the load
resistance that gives the best match at the mixer input ?

Steve, VE3SMA

One additional item to model is using highly reactive antennas.

The Lowpass rolloff point caused by the integrating capacitors is
controlled by the effective antenna impedance.

While one can assume 50 ohms, more or less, for a matched antenna, using
the Tayloe detector in a general coverage system should be calculated to
assume the antenna has massive mismatches at most frequencies.

I did some simplistic modelling and the results were interesting. To
control the detector rolloff, I'd be tempted to use a low gain RF preamp and
perhaps a small attenuator pad feeding the mixer....

Besides, as I recall, the Tayloe detector responds to harmonics of the LO
and so some frequency selective flters in the RF preamp might be useful.

All in all, The Tayloe detector has some benefits, but as with most
detectors, adding some components can enhance the system performance.


Jim Pennell
N6BIU

One additional item to model is using highly reactive antennas.

The Lowpass rolloff point caused by the integrating capacitors is
controlled by the effective antenna impedance.

While one can assume 50 ohms, more or less, for a matched antenna, using
the Tayloe detector in a general coverage system should be calculated to
assume the antenna has massive mismatches at most frequencies.

I did some simplistic modelling and the results were interesting. To
control the detector rolloff, I'd be tempted to use a low gain RF preamp and
perhaps a small attenuator pad feeding the mixer....

Besides, as I recall, the Tayloe detector responds to harmonics of the LO
and so some frequency selective flters in the RF preamp might be useful.

All in all, The Tayloe detector has some benefits, but as with most
detectors, adding some components can enhance the system performance.


Jim Pennell
N6BIU