LF: New Mixer Design by Dan Tayloe

Andre' Kesteloot akestelo@bellatlantic.net
Thu, 22 Oct 1998 15:17:10 -0400

> From: tayloe_d@juno.com (Daniel R Tayloe)
> To: "Low Power Amateur Radio Discussion" <qrp-l@lehigh.edu>
> Subject: ScQRPion Stinger Pacificon Design Entry Description
> [...] during a 6500 mile trip driving from Phoenix to New York to Florida
> and back again, I came up with an idea for a new kind of product detector
> that I *had* to try out.  Therefore
> this rig started out as a "proof of concept" prototype that happened to
> worked (and work well!). From there things kind of got out of hand, and it
> grew into a full blown transceiver.
> Some of the measured specifications of the receiver portion of the
> transceiver:
> MDS (Minimum Detectable Signal - measured by ear):      -136 dbm
> Two Tone Dynamic Range (2TDR):                          111 dB
> Third order intercept:                                  +30 dbm
> Receive Current Drain                                   60 mA.
> Compare these numbers with those of other rigs (info stolen from
> Jim Duffy, KK6MC & Az ScQRPion Kent Torell):
>      Unit       MDS     2TDR    Blocking
>      FT101E     142     60
>      TS520      132     63
>      R390       136     82
>      7553B      146     88         <--- Collins 'S' line radio
>      Corsair    128     90
>      GQ40       126     90      125
>      NN1G       132     90
>      Nor. 40    136     88      108
>      Sierra     132     88      100
>      FT1000     125     95      125
>      TS930      133     95
>      Breed      120     70      90
>      Hayward    128     95      125  <--- Hayward's criteria for contest
>      Mini R2    136     96         <---- from QST article
> The transceiver provides superior dynamic range, while satisfying
> requirements for both good sensitivity, and moderate current drain.
> The product detector has less that 1 dB of conversion loss from RF to
> baseband.  This is because it is not really a mixer, producing both sum
> and difference frequencies, but rather is a
> "switching integrator", converting most of the input signal to a baseband
> detected voltage.  Lowered conversion loss means that it may be possible
> to do without an RF preamp on the higher HF bands and still maintain a
> good noise figure.
> The product detector also has the property of being narrow banded.
> The detector acts like a bandpass filter centered on the frequency the
> detector is tuned to.  This provides an additional measure of high
> frequency roll off to the signals that are converted to baseband.
> This property is unlike any other detector I have worked with and
> represents "free" additional selectivity.
> The detector acts like an open circuit at the frequency it is tuned to.
> This means that it is no longer necessary to provide a current hungry post
> amplifier preamp in order to provide a broad band IF termination match
> across all frequencies to in turn have a high third order intercept.  This
> feature helps keep the drain current of the transceiver to reasonable
> value.
> The detector used in this radio is a new design, based around a 74CBT3253
> dual 1 of 4 analog mux.  This part is designed for switching
> microprocessor memory subsystems, and is rated for
> passing signals up to 300 MHz.  I found that it could pass signal of up to
> 4v pk-pk before distortion.
> I have dubbed this new product detector the "Tayloe" product detector.
> Hey, if Gilbert can have the Gilbert cell mixer, I can have the "Tayloe"
> product detector!
> Imagine a R/C low pass filter:
> input >------R--------------> output
>                        |
>                        C
>                        |
>                       Gnd
> At DC, the C looks like an open circuit, so the output looks like the
> input.  As the input signal frequency increases, C looks more like a
> short, causing the output to roll off to zero at high frequencies.
> Now, imagine using four Cs to ground instead of just one, and imagine a
> one pole, four position switch continuously sweeping from one C to the
> next in a continuously rotating pattern:
> input >------R--------------
>                            |
>                      1 of 4 switch------<----4xfo
>                     |    |    |    |
>                     C1   C2   C3   C4
>                     |    |    |    |
>                    Gnd  Gnd  Gnd  Gnd
> In the above diagram, the "One of four" switch is sweeping across the caps
> at a 4xfo rate.  Each of the four caps "sees" 1/4 of a cycle of the input
> signal.  If 4xfo is set to 28 MHz, each C will see 1/4 of a cycle of a 7
> MHz signal.  These little sample "bursts" of RF are averaged or "filtered"
> by the Cs.  If a signal of exactly 7 MHz is sent into the input, a DC
> signal will develop across each cap corresponding to the DC average of the
> RF sine wave seen during that 1/4 cycle.
> Anyway, the long and short of it all, is that the voltage across C1
> represents the 0 degree detected audio sampling of the input RF signal, C2
> represents 90 degree, C3, 180 degree, and C4 270
> degrees.  I use one op-amp to sum the 0&180 degree voltages to a single In
> phase (I) signal, and another to sum 90&270 degree voltages to a
> Quadrature signal (Q).
> Also note that just like the R/C low pass filter causes the signal to roll
> off as the frequency increases, this detector looks like an open circuit
> to signals that match the frequency it is tuned to.  As the signals move
> further to either side of this tuned detection frequency, the signals
> begin to roll off at the R/C rate.  My detector uses 50 ohms for R and
> 0.33 uF for the four detection caps. With a detection bandwidth of 2 KHz
> at 7 MHz, the detector has an equivilent "Q" of 7MHz/2KHz or 3500!
> Having an Inphase (I) and Quadrature (Q) audio sampling of the RF input
> signal, these two signal can be combined to provide a single image DC
> receiver.  The current receiver as greater than
> 45 dB of opposite sideband supression.
> This single circuit replaces an RF splitter, two mixers (both I&Q), and a
> 90 degree RF phase delay while being able to handle very large signal
> levels, in this case 4v pk-pk.
> The receiver is therefore a simplified "R2" single image, direct
> conversion receiver.  I shamelessly stole parts of his audio phasing
> design in order to get the audio 90 degree delay necessary for single
> image reception.
> On the other hand, the transmitter has been likewise greatly simplified
> and currently consists of two NAND gates and a MOSFET. The transmitter
> puts out 0-5w of adjustable power at 12v, up to
> 6.5w at 13.8v.
> This transceiver can be made to use only one tuned circuit, the low pass
> filter for the transmitter.  This it is reletively easy to construct very
> high performance, multiband designs.  Bands that differ by powers of two
> such as 80, 40, 20 and 10m are exceptionly easy to accomodate. For my
> first attempt, I will modify my current 40m transceiver to cover both 40
> and 80 meters.
> Oh, well.... Sorry for the long posting.  I guess I got carried away.  I
> will be writing up an article on this design for QRPp, so watch this
> spring for additional details.
> - Dan Tayloe, N7VE, Phoenix, AZ, Az ScQRPions, QRPL # 696
> -----------End Quote-----------------------