We assume that you have read the Overview by now, and hence you are familiar with the overall system block diagram.
As you can see, the QRP2001 is a direct conversion transceiver using the Tayloe N7VE product detector, feeding into a 7 pole Polyphase network to obtain single signal reception. When receiving, the desired sideband AF signal is filtered by a passive 2.4 kHz AF low pass filter and an active 300Hz high pass filter. It then goes to an AF amplifier with AGC, followed by another 8 pole active AF low pass filter with variable cut-off frequency and a conventional AF power amplifier.
The circuit diagrams for the QRP2001 main board are presented here in two parts. The first schematic covers the RF front end, Tayloe product detector and associated VFO drive logic, first AF gain stage, and the polyphase filter. The output from the polyphase filter feeds into the second schematic which covers the rest of the audio processing circuitry: the low-pass filter, high-pass filter, AGC circuit, SCAF, and audio power amp.
This product detector (Ref 1) uses both halves of the 74FST3253 fast bus switch in parallel, as a 1-4 analogue multiplexor. The RF input signal is continuously switched to each of the four Capacitors C13, C14, C15 and C18, and integrated over a quarter of the cycle. This results in four baseband audio signals, each phase shifted by 90 degrees from the previous one. Depending on which sideband is being received, each output may either lead or lag the previous one.
The 74FST3253 switch has a low 'on' resistance of around 5 Ohms, and by putting both halves in parallel we (theoretically) halve this value. The resulting demodulator offers a conversion loss of only -1 dB compared with -6 dB for a conventional DBM like the SBL1. The sensitivity is therefore excellent at 0.4 uV (-116dBm) for 10dB SINAD and unlike the QRP2000 an RF amplifier is not required, thus preserving the strong signal handling characteristics which Dan Tayloe measured as +30dBm 3rd order ICP. This is considerably better than the SBL1. Furthermore, this product detector does not require a RF power splitter to feed both RF mixers as in the QRP2000 design (Ref 2).
The product detector also acts as a Switched Capacitor Filter (SCF filter) or Commutating Filter at the RF input frequency with the pass band set by C13/14/15/18 and R7. Using the specified values this translates into a -3dB bandwidth of +/- 2.6 KHz away from the RF input frequency. With the switch cycling at 14.1MHz the bandwidth of this commutating filter is effectively 14097.4 KHz to 14102.6 KHz, equivalent to a Q of 2700.
Capacitors C13, C14, C15 and C18 should be good quality 5% polyester or equivalent. There is some evidence that it is worth hand-matching these capacitors to a closer tolerance (1 or 2%), but we are still investigating how significant this is.
AM detection, often a major problem with DC receivers in Europe on 40 Meters, is about the same level as the QRP2000 design, using SBL1 DBM's.
The Tayloe product detector requires a Local Oscillator signal which clocks the input signal to each of the four switch outputs once every cycle. Other designs have used a signal generator running at four times the required frequency, fed into a counter to provide the necessary drive signals. Our approach is a little more cunning, and uses two signals at the required frequency but in phase quadrature to drive the switch. As the relative levels of the two drive signals change through a cycle, all four combinations of input value (00, 01, 11, and 10) appear on the selection inputs to the switch. Note however that the sequence is 0,1,3,2, rather than 1,2,3,4 - this is easily compensated for by simply swapping two of the switch outputs.
To generate our two quarature VFO signals we still need to divide down a x2 VFO signal - however this is still significantly easier to generate than the the x4 signal which would have been needed otherwise. This is particularly relevant with the simple DDS VFO (detailed elsewhere in these pages) which can generate up to about 70MHz, but certainly could not cope with 120MHz!
U1, a quad XOR gate takes the Local Oscillator signal at twice the RF Input frequency. U1a together with R2, R3 and C5 help to encourage a unity mark-space ratio on the signals going into U1b and U1c. U1b is arranged as an inverter while U1c is just a buffer - hence the outputs from these devices are out of phase. These two signals are divided through flip-flops U2a and U2b, which are synchronised by U1d to force the required quadrature relationship. The flip-flop outputs directly drive the Tayloe bus switch select inputs.
The polyphase network is ideally suited to interface directly with the Tayloe product detector. Using standard 1% resistors and 5% tolerance capacitors values from John Heyes G3TDZ article (Ref 4) about 50 dB of sideband suppression is achievable. U4a and U5a are Low noise opamps (NE5532's or OP27's) with about 26 dB gain. U4b and U5b are unity gain inverters required to drive the polyphase network.
James Verduyn has produced a wonderful Excel spreadsheet (for Excel 97 or later) which makes it very easy to see the effect of component changes within the polyphase network. In fact, two different polyphase networks are modelled at the same time to make comparisons very easy and obvious. The component values are tabulated on the left of the spreadsheet and these values may be edited as you wish. Then just click on the button labelled 'Model the values' (towards the bottom left) and it will recalculate and display graphically the predicted sideband suppression for your networks. At the moment this spreadsheet does not allow for component tolerances, so practical results will generally be slightly worse than predicted.
The low level SSB/CW AF signal is taken from one output of the polyphase filter (in practice it should not matter which output) and feeds into U8a on the second schematic. This is another NE5532 Opamp with 14dB gain. U8a is followed by a 7th order elliptical low Pass filter, based on the KK7B R2 design (Ref 3) but scaled down from 3KHz to 2.4KHz. The subsequent 300Hz (active) low-pass filter around U8b is important to remove various oposite-sideband artifacts which creep through below the lower limit at which the polyphase filter is effective. The signal is then processed by the AGC stage around U9b, Q1 and Q2. This currently provides about 70dB of AGC range before distortion becomes significant, working right up to the point where earlier stages start to clip.
The AF output is then processed through an active 8 pole elliptic Low Pass filter implemented with a MAX7480 SCAF IC. The cut-off frequency of the SCAF is set at approximately 2.4KHz in SSB mode or 1KHz in CW mode. With a little extra complexity the cut-off can be made continuously variable, but we have tried to keep things simple in our design. The -6dB/-60dB shape factor of this filter is about 1.5, and has a sound which is well liked by amateurs who have used the QRP2001. It demonstrates that there is a cheaper alternative to obtain (almost!) DSP-like performance. The SCAF filter contributes some extra noise, however measurements indicate that this does not degrade the MDS of the receiver. If your interest is solely CW, you may want to use the 1 kHz CW low pass filter design from the KK7B article and make some further simplifications by dispensing with the active SCAF filter.
The final stage is a conventional AF amplifier using the TDA7052A. This device is less common than the usual LM386 or TDA2003 amplifiers, but it consumes a lot less current, is (in our experience) less prone to instability, and has the slight added convenience of voltage-controlled gain.
After a heroic struggle with various CAD packages, Jan has produced a PCB layout for the QRP2001 main board. It is a double-sided board, so there are separate images for the bottom layer and top layer. In an attempt to make these images relatively quick to load we have scaled them down, which has obviously lost some resolution. In any case, if you want to produce a PCB for yourself you will need just the track and pad patterns, without the component overlay layer. Please contact us directly if you want either better quality copies of these images, or raw track layouts for producing PCBs.
As noted elsewhere, work is still continuing to produce a transmit modulator which does justice to the receiver performance. However, the intention is that during SSB transmission the microphone signal will be amplified in a simple op-amp gain stage and then will use the same filtering and AGC stages as used during receive. However, the output from the AGC amplifier will be diverted to the input of the polyphase filter. Two of the polyphase inputs will be driven in-phase with the AGC output, and two will be driven through unity-gain inverting amplifiers. The outputs from the polyphase filter will be in accurate phase-quadrature ready to feed a phasing-style modulator.
