Very High Performance Image Rejecting Direct Conversion Receivers

Or how can an 11 ma receiver out perform the worlds best ham transceivers?

Dan Tayloe, N7VE

NC2030 20m Prototype

SCAF Tune CW Speed Volume Keyer Program

Main Tuning

Frequency Read-Out

RIT

Spot Switch

Keyer Input

Headphone Jack

NC2030 Receiver Specs

MDS (3db S+N/N): -135 dbm (0.1 uV) Receiver Bandwidth (-6db): 350-800 Hz IP3 DR: 93db (2KHz), 105 db (5 KHz), 109 db (10 KHz) BDR: 119db (2 KHz), 128.5db (5 KHz), 139db (10 KHz), 142db (20 KHz) Image rejection: ~ >45 db over the band Receiver current drain: ~11 ma at 12v

BDR: A Comparison 5 KHz

NC2030 at full sensitivity, ranks among the best rigs which were measured with their RF pre-amps off. Note: With the RF pre-amp on, the K2 suffers a 7 db degradation in blocking Even at only 2 KHz, the NC2030 performs at least as well as all but two rigs measured at 5 KHz.

Blocking DR: A comparison (vs. K2)

150 db
KHz Low Side 119 122 125.5 128.5 133 139 142 118 122 125 130 137 142 High Side

20 KHz 10 KHz 140 db 7.5 KHz

2 3 4 5 6 7 10 20

130 db

NC2030 Blocking DR does not plateau

- Rejection keeps improving 120 db

K2 plateau shows IF amp saturation

- Signals on the other side of the band (300+ KHz away) can still cause blocking

110 db
*Extracted from QST K2 expanded report

=> NC2030 blocking is a bit worse close in, much better further out

IP3DR: A comparison 5 KHz

NC2030 at 5 KHz is 13 db better than the best. NC2030 at 2 KHz is still better than all the rest at 5 KHz. Not a true apple-to-apples comparison since NC2030 is at full sensitivity while other rigs have pre-amps off

13 db

IP3 DR: A comparison (vs. K2)

110 db
KHz 2 Low Side 93 98.5 102 105 107.5 109 109 High Side 93.5 98.5 102 106 109 109 109

10 KHz 100 db 5 KHz 90 db

20 KHz

3 4 5 6 7 10 20

80 db

2 KHz

IP3DR is noticeably better than the best radios available (K2/Orion) 70 db NC2030 appears 17 db better 2 KHz away (93 db vs. 76 db) NC2030 appears 18 db better 5 KHz away (105 db vs. 87 db)

60 db
*Extracted from QST K2 expanded report

Typical Superhet Front End

Step Step Atten Atten Band Band Pass Pass filter filter RF RF Preamp Preamp First First Mixer Mixer First First IF IFAmp Amp Crystal Crystal filter filter

VFO VFO

This is a simplified view, but represents many superhet receiver front ends The large signal performance is set in the sections before the radio brick wall filtering (Xtal filter)

Superhet Front End Bandwidth

Not BW Limited Step Step Atten Atten Not BW Limited Band Band Pass Pass filter filter RF RF Preamp Preamp First First Mixer Mixer First First IF IFAmp Amp No significant BW Limits 500 Hz BW (CW) Crystal Crystal filter filter

300 500 KHz Band width

VFO VFO

RF preamp, first mixer, and first IF amp sees all signals in the entire band all at the same time. Wide front end bandwidth is the main reason preamps are turned off and attenuators are kicked in during a contest.

Phasing DC Front End

1.5 KHz BW Band Band Pass Pass filter filter 300 - 500 KHz BW Tayloe Tayloe Quadrature Quadrature Detector Detector VFO VFO I IAudio Audio Preamp Preamp Q QAudio Audio Preamp Preamp 1.5 KHz BW 90 90Degree Degree Phase PhaseShift Shift 90 90Degree Degree Phase PhaseShift Shift Wide BW RC RCActive Active Audio AudioFilter Filter 500 Hz BW (CW)

The narrow bandwidth direct conversion detector allows few signals to get to the audio preamps. The audio preamps also has a narrow bandwidth, thus off frequency signals are attenuated even further prior to the receiver brick wall audio filter

Band View: Superhet Vs. DC RCVR

Signal Strength 300 to 500 KHz Bandwidth Superhet Front End BW

Frequency DC Rcvr Front End BW Big Gun Contest Signals (Field Day?)

Superhet RF preamp/Mixer/IF Amp sees all signals at full strength

Must remain linear with the sum of all the signals on the band This is hard! RF pre-amp on/off, Attenuators, Variable IF amp gain Requires a lot of power to stay linear; IF amp often uses 50 to 100 ma

DC receiver sees only a fraction of the band

Must remain linear over just a few of the many signals on the band Only the close in signals are problems; -16 db, 5 KHz away, -40 db at 20 KHz A much easier problem!

Superhet RF/IF Preamps

RF pre-amp, No heat sink IF amp w/ heat sink

IF amps are normally built to handle more power than RF Amps

RF and IF amps are typically 50 ohm in, and low Z out

These are both power amplifiers

Wide band, high signal linearity amps require lots of power RF pre-amps are not normally designed to survive large in band signals
Which is why they are useless and get turned off in a contest First mixer can only handle so much power out of RF preamp anyway Superhet performance measured with RF Preamp off for a reason

DC Receiver Detector/AF Preamps

I-Q audio preamps

High performance quadrature detector (3v pk-pk max)

Detector has ~0.9 db of conversion loss rather than the typical high performance superhet 6 to 8 db mixer loss
Thus, RF preamp not needed to overcome first mixer loss Allows receiver to have both high sensitivity & large signal performance

AF Pre-amp is low Z in, high Z out, voltage amplifier

Voltage amplification takes less power than power amplification

Detector/AF preamp rolls off relatively quickly

16 db down at 5 KHz, 27 db at 10 KHz, 39 db at 20 KHz

Superhet Brick Wall Filters

RF preamps and IF amps must have power limits because of crystal filter limitations
Typical crystal filter, 5 crystals

Crystals used in xtal filters typically 10 mW

~1.4v RF limit, blocking limit of ~140 db BDR

Crystal power limitations may contribute to close in IP3 problems FT243 crystals might make superior filters
Old FT243 crystals handle much higher power levels

DC Receiver Brick Wall Filters

NC2030 8 pole low pass filter High voltage, very high dynamic range brick wall RC filters are easily constructed
Caps typically 50v 1/2w resistors common Op amps typically +/- 18v (36v)

8 pole low pass filter

R/C filters: Lots of Rs and Cs! With a 3v receiver chain, NC2030 has ~13 db better IP3DR and similar BDR to the best available rigs
And this is at full sensitivity, not RF Pre-amp off!

Superhets can be simple

RF Input Audio Output

RF Preamp & First Mixer (NE602)

Brick Wall crystal filter (no IF amp)

However, this is not a high performance superhet

DC Receivers can be simple also

RF Input

49er Receiver Schematics

Audio Output

RF Preamp & Detector (NE602)

Brick Wall L/C filter (no AF preamp)

However, this is not a high performance DC Rcvr

A High Performance Phasing DC Receiver (NC2030) has a Price

Lots of parts, with many Rs and Cs!
- ~280 out of 360 total parts are Rs and Cs

175 Capacitors 108 Resistors 25 Inductors 19 ICs

5 op amps, 5 LDO voltage regulators, 5 digital ICs, 2 uPs, 1 SCAF, and 1 switching regulator

17 Transistors 17 diodes 2 crystals High performance DC Receiver (NC2030) is more complex than a typical superhet
- But higher performance and less power!

Quadrature Detector
RF Input 50 ohm 0-3v max 270 Audio out 90 Audio out 180 Audio out 0 Audio out One 50 ohm input, Four 200 ohm outputs No power gain Output ~0.9x Input due to integration on detector caps ~0.9 db loss Diode mixer : 6 db of conversion loss typical

- Clocks route RF input to 1 of four Detector Caps at a 4x rate - Each det. cap. averages cycle of RF Audio! - Four blade ceiling fan w/ strobe light analogy

Quadrature Detector Outputs

- Note that 0 & 180 and 90 & 270 outputs are mirror images of each other. - These pairs (such as 0 & 180) are summed differentially via + & - inputs of op-amps

NC2030 5 KHz Blocking Calculations

- The simple RC roll off of the Detector and AF preamp is somewhat gradual, but 16 db of attenuation greatly helps BDR (and IP3DR also) - AF Preamp has 66x of voltage gain (36 db) - 16 db roll off at 5 KHz leaves 20 db of gain (10x) - With 3v pk-pk max audio output, RF input blocks at 0.3v (-6 dbm) 5 KHz away - Using -135 dbm sensitivity, BDR = 135 6 = 129 db - Actual measured result 128.5 db BDR at 5 KHz

Detector Clock Drive

Clock A

Clock B

90 180 270

90 180 270

- Need to switch to each of four outputs every RF cycle, cycle dwell time on each detector output - Two phase clock used to get four output states

Detector Clock Drive Circuit A

Four states 00 01 11 10 Not a straight binary counter
1x TX Output

4x RF LO Square-wave Input

- 4x frequency source used with digital dividers - Advantage: Accurate clocks, excellent opposite sideband rejection over a very wide range - Disadvantage: Dividers are a bit power hungry

Clock Drive Circuit B (NC2030)

1x RF LO Sine-wave Input

- 1x frequency source used with L/C delay section - Advantage: Uses much less power than dividers - Disadvantage: Bandwidth limited, USB rejection good over a limited range (i.e., CW portion of band)

I Q USB and LSB Outputs

USB I,Q

LSB I,Q - I (0, 180) and Q (90, 270) are 90 degrees apart - USB/LSB depends on which leads the other

90 Shift Phasing Network

- Two stage R/C phase shift network
Q in

- Both sides cause phase shift - One side starts first - 2nd trails 1st by 90
Sum Out

- Limited sideband rejection range - Rejection range optimized for CW bandwidth (500 Hz)

I in

Phasing How to Get 90 Shift

90

- One side starts falling in phase after the other - The late side is adjusted to be exactly 90 late - The 90 difference is good for a limited range

USB After 90 Phase Shift

USB I,Q before phasing

USB I,Q after phasing - After phase shifting, I & Q opposites of each other - Phasing outputs sum to zero USB suppressed

USB Rejection Plot

Unfiltered USB Rejection

- USB rejection varies across audio bandpass - Smallest USB rejection at 150 &650 Hz, ~ 55 db down - Filtering improves high & low frequency rejection

Filtered USB Rejection

- Rejection shown is best case

- LO clock uses L/C phasing - Causes USB rejection to vary across band > 45 db across the band typical

LSB After 90 Phase Shift

LSB I,Q before phasing

LSB I,Q after phasing - After phase shifting, I & Q are in phase - Phasing outputs sum to 2x LSB enhanced

LSB Audio Response Plot

SCAF LPF not included

- 6db at 350 & 800 Hz; 60 db at 50 Hz & 1.6 KHz - Does not include the additional 40 db of variable SCAF LPF attenuation - Main RC filter designed for low audio ringing

LSB Audio Response Plot

Actual Band Noise 30m
SCAF SCAF wide open

- High side audio roll off is very step - SCAF cleans up high frequency roll off even when wide open

SCAF SCAF at ~700 Hz >40 db SCAF Limit

- SCAF very good at removing a high side interferer when needed - Noise below 100 Hz is a
sound card issue

DC Receiver Pwr Consumption

Quadrature detector voltage driven not power driven as required by diode mixers.
74CBTLV3253 is a dual 4:1 analog bus switch

First low noise audio preamplifier outputs are voltage outputs, not power, as needed by superhets 3v receiver powered by a 3v & 5v switching supply, giving a 3x power savings over simple linear regulation from 12v

DC Receiver Pwr Consumption

VFO and VXO; 3 ma LO mixer; 1.6 ma LO filter amp; 9.5 ma LO squaring & detector driver (74AHC00); 0.8 ma Quadrature detector (Tayloe Mixer); 4.4 ma First audio LNA & phase shift network; 7.8 ma High and low pass RC filters and headphone drivers; 2 ma SCAF variable audio low pass filter; 1 ma

=> Roughly 30 ma total receiver drain at 3v supply

- 14 ma for the LO subsystem, 16 ma for the receiver line up

=> 11 ma at 12v into the 3v & 5v switching supply

Conclusions
DC receivers have a performance advantage over superhets because: 1. DC quadrature det has lower loss (1 vs. 6 db)
DC does not need an RF amp for high sensitivity The superhet mixer can be 100s of MHz wide The superhet has a wide bandwidth IF amp (>1 MHz?) Superhet good to ~2v pk-pk because of its crystal filter DC filter is good to 36v pk-pk signal DC can have superior large signal capabilities (20+ db higher than current 3v NC2030)

2. DC detector has a limited ~1.5 KHz bandwidth 3. DC AF amp also has ~1.5 KHz bandwidth 4. DC receiver uses R/C active filters, not crystals