Up or down conversion: that is the question...

by Adam M. Farson VA7OJ/AB4OJ

As in all aspects of RF design, the choice of a high or an inband 1st IF is a compromise. (Down-conversion is a misnomer, because if a receiver with a 9 MHz 1st IF is tuned to 3.6 MHz it up-converts.) The move to a high 1st IF – with the Racal RA.17 as one of the pioneers – was underway whilst I was at Racal in the 1960's. The rationale for the high IF was straightforward; by “throwing” the image response to the other side of an IF well above the RF tuning range, the receiver designer was able to kill two birds with one stone. These were image/image-channel noise response and 1st IF leakage. At the same time, the design of a receiver or transceiver with unbroken HF coverage, free of birdies and spurs, became fairly easy and cost-effective. Amateur transceivers with a high 1st IF were amenable to “off-label” use in other radio services, thus creating additional markets for these equipments.

The high 1st IF architecture was more than adequate until contesting, using narrow-band modes without guard-bands, and often without any vestige of operating discipline, became the dominant form of amateur HF activity. All too frequently, transmitters were turned all the way up, yielding signals loaded with artefacts. Close-in operation raised the issue of IMD products and reciprocal mixing noise which could (and often did) mask that elusive “weak one” needed for an added “Q”. Thus, the contesting community – a relatively small but to all appearances a very well-heeled and vocal subset of the Ham fraternity at large – began pressuring OEM’s to mitigate these problems.

Reciprocal mixing is a function of LO phase noise. In this area, synthesisers have made light-years of progress since the days of the early, rather crude PLL architecture used in the HF rigs of the 1970’s and 1980’s. The military and the commercial wireless industry have been driving the DDS IC designs from which we now all benefit.

Third-order IMD products due to strong, close-spaced signals arise when these signals overdrive an active stage such as an RF or IF amplifier or mixer. In a typical superhet with multiple conversions, the second mixer is most vulnerable; thus, the 1st IF (“roofing”) filter must be sufficiently narrow to block unwanted strong signals from entering the second mixer and generating an IMD product in the IF channel. At a typical high 1st IF > 40 MHz, it is very difficult and costly to build a filter narrower than about 3 kHz which will have acceptable insertion loss, temperature stability and freedom from passive IMD (PIM). The very best professional receivers use a different approach; their first and second mixers have a very high linear operating region, and the roofing filter is designed for the lowest possible insertion loss and PIM consistent with sufficient bandwidth to pass the widest supported emission. For example, a typical R&S military receiver has an 8 kHz 1st IF filter at 48 MHz. This approach has a very significant impact on the price of the equipment, placing it beyond the reach of the amateur (except for rare surplus offerings).

Under pressure from the contesting marketplace, some – not all – amateur OEM’s have opted for a return to the pre-1960’s architecture using an inband 1st IF in the 8 - 11 MHz range with a narrow crystal filter following the 1st mixer as primary selectivity filter. This solves the 3rd-order IMD problem, with a caveat; many of the currently-available amateur transceivers with an inband 1st IF have a reciprocal-mixing dynamic range (RMDR) inferior to their 3rd-order IMD dynamic range (DR3). Thus, a radio with 100 dB DR3 (the “Holy Grail”) and 80 dB RMDR is an 80 dB radio, not a 100 dB radio. (To claim otherwise is false advertising.) In addition, all the old bugbears – poor image rejection, image-channel noise response and 1st-IF leakage – come roaring back with a vengeance, especially on bands close to the 1st IF! The narrow 1st IF crystal filter will also exhibit significant PIM in the presence of strong signals, which can negate the very benefit the filter was intended to confer.

Front-end bandpass filters with good shape factors, covering the amateur bands only, will mitigate the image and IF leakage problems, but at the expense of MDS due to their higher insertion loss. Such filters do not come cheap either.

Receiver design is like an inflated balloon: when one pokes one spot with a finger, another spot will bulge out. Poke too hard and “POP!”

I find the practice of “shopping on DR3” quite lamentable. All too often, other receiver parameters – to say nothing of the transmitter’s spectral purity - are utterly disregarded in the quest after this “magic number” – rather like buying a car on its 0-100 km/h times alone.

The advent of the direct-sampling SDR receiver and direct-sampling/digital up-conversion transceiver is forcing a complete re-think of receiver test methods. Traditional performance parameters such as DR3, 3rd-order intercept (IP3) and blocking dynamic range (BDR) are utterly meaningless in the context of a direct-sampling SDR in which the ADC is behind the antenna socket. Here are two presentations I have given on the subject:

HF Receiver Testing: Issues & Advances

Noise Power Ratio Testing: HF receiver performance evaluation using notched noise

In conclusion, the choice of receiver architecture in one’s rig purchase is a function of the intended application. For the casual multi-mode operator who also likes SWL activity, a radio set with a high 1st IF will be more than adequate. For the intensive CW or RTTY contester or DX-chaser, a radio with an inband 1st IF may commend itself, with the cautionary note that it may not perform as well on 40 or 30 metres as on other bands for the reasons alluded to above.

Adam Farson, VA7OJ/AB4OJ

ex: ZS1ZG (1962), ZS6XT (1964)

Copyright © 2014, A. Farson VA7OJ/AB4OJ. All rights reserved.
Last updated: 11/27/2015
Microsoft SharePoint Designer 2007.
Hit Counter