High End Rcvr's, Some notes from G8MOB

[WA1HZK]

This was a response to an e-mail posted on the Premium-RX group. This guy had a lot to say about the rigs we stumble across so I thought it might be a good read for the group.
Keith
WA1HZK


 Dear Shane

 I read your email today chuckling and with a high element of sympathy.
 Put simply we both want the same things and we are both crying for the
 moon because no single receiver exists that I know of that encompasses all
 of your desiderata.

 I have never used a K3.  The review in the RSGB's RadCom last year by
 Peter Hart (a highly respected engineer and reviewer) sang its praises.
 Other users have done the same.  But, I query, what have been the
 comparable standards of comparison?  If it's other ham gear, then that is
 frankly not a big deal.  I have used some of industry's best and I reckon
 my ears and fingers know a good set or two.  Once you have used commercial
 and military radios made by the likes of Racal, Redifon, Plessey, Skanti,
 Marconi, Dansk Radio, Collins, Siemens and Rohde & Schwarz, all made for
 the high end of the professional market, the likes of Icom, Kenwood and
 Yaesu pale significantly, though in fairness the very latest offerings
 from Icom have been a great improvement.

 Great claims are made for "digital" but you should never forget the tale
 of the "Emperor's new clothes".

 Digital Audio Broadcasting ("DAB") is supposed to be the bees' knees for
 hifi broadcasting and in UK will almost certainly replace Band 2 FM in a
 few years time.  However, several commercial reviews have commented that
 the data stream is heavily compressed in order to cram umpteen dozen radio
 stations into the limited bandwith and that DAB is basically an MP3
 standard.  That's great for trannie radios and MP3s but it does not stand
 critical comparison with good quality analogue FM.  Everyone with sensible
 technical knowledge accepts that given current standards, the BBC would
 not have selected DAB today, but we are stuck with what is already a
 legacy digital system.

 Similar claims are made on HF for "digital".  Certainly the technical
 specs of selectivity achieved by digitising the IF and AGC look
 impressive.  But in my view the specs are merely the starting point.  The
 real test is the impression of the sound in your ears.  After all what
 earthy use has a radio unless to deliver quality audio to your ears?

 This said, most modern radios are not really designed for human operators.
 Most are under computer control and are listened to by computers owned by
 the security services for the usual unfortunate reasons or for data links
 for the commercial and military.  Human operators are far too expensive to
 keep on the payroll!

 I do the usual critical technical tests but my ultimate test is to listen
 carefully on good quality Sony or Senheisser headphones to a station that
 I know well, typically the BBC on 198 kHz and on 15,400 MHz.  Any trace of
 hum is a big minus issue.  Any trace of microprocessor squeaks and bumps
 is an even bigger minus.

 Any objectionable coloration is no good either.  Most of that will be
 caused by phase distortion in crystal filters or mechanical filters.  The
 true perfectionist will seek fully phase-compensated crystal filters.
 These can be very expensive.  They were used mainly for receiving fast
 data on HF where phase distortion across the passband will completely muck
 up the relative timing of the data pulses and may well render the signal
 unprintable.  I acquired a matched pair for my Marconi H2540 receiver for
 the princely sum of £10 at a flea market.  The factory list price from
 Marconi was £350 for the USB filter and £1,200 for the LSB filter plus
 postage plus VAT in the mid 1980s.  They improved the sound quality no
 end.

 Many amateur receivers and transceivers attempt to achieve better
 selectivity on SSB by using narrow SSB filters of 2.4 kHz bandwidth or
 even less.  Under conditions of strong adjacent interference they are
 useful but the resulting boxy sound of the human voice (to me) more
 resembles a cat being castrated.

 If you are going to listen to music or the human voice using SSB mode
 (useful when there is lots of QRM) you need to use better professional
 filters with a bandwidth of at least 2.7 kHz.  The SSB filters on my Racal
 RA1792 have a bandwidth of 2.9 kHz and sound very good.

 Great claims are made for using synchronous AM detectors.  A few work
 really well but they are complicated and expensive.  The simple designs
 are OK but often not much better than a traditional diode detector.  I am
 told that the Sherwood "add on" detector unit is excellent.  I use a
 fairly similar "one-off" professional design made by Surrey Electronics,
 but they had to stop selling under threat of patent infringement from
 another company called PhaseTrack.  The latter marketed a superb HF
 receiver specifically designed for commercial rebroadcasting called the
 "F1-2".  The IF selectivity was formed not by filters but by a complicated
 phasing system.  The result was total lack of any phase distortion, a flat
 audio bandwidth from 50 Hz to about 4.5 kHz and a total harmonic
 distortion under 0.5% across the whole of that bandwidth.  I have one and
 wow it is incredible.  I have yet to meet any DSP that will match that.
 It will come in a few years but as far as I am aware not at present.

 Diode AM detectors are regarded as terribly old fashioned but that is
 mainly because some listeners don't know any better.  The very best diode
 detector is on the ancient GEC BRT400E of the 1950s.  The recovered audio
 quality is staggering.  This is not just me.  I quote a well known hifi
 reviewer and ham, the late Angus McKenzie G3OSS with whom I had many an
 exchange of views.  These sets were used extensively for monitoring by the
 BBC, and BBC engineers demanded the highest professional standards.  Alas
 the remaining ones are pretty bashed up and will need a great deal of
 renovation.  They are also very big, hot and heavy.

 The venerated RCA AR88D still has an excellent reputation for AM quality
 with several selectable bandwidths and will hold its own up to 10 MHz but
 the selectivity is poor by modern standards, it weighs 100 lbs and will
 almost certainly need realignment.  No one would really want to use one
 for SSB.  The build quality however is superb and these radios originally
 cost a fortune.

 The valved Collins gear such as the R390 and 51J-4 are mechanically
 gorgeous and were even more expensive.  You won't find better construction
 anywhere. They are heavy to tune (dozens of gears and moving slug racks
 inside) and the recovered audio is not so good.  But they are fun to use
 if you like big panels, big meters and decent sized meters (as I do).  An
 R390 plus the Sherwood synchro detector unit is said to be superb.

 In similar vein, the Racal RA17L is a nice old valved receiver but the
 audio is only OK, not outstanding, and they all need restoration after 40
 years +. However, their usefulness (and audio quality) increases mightily
 when married up with an external SSB adaptor such as the RA121 or RA298, a
 VLF converter such as the RA137 (listening down to 12kHz), a digital
 display adaptor (displays to 10Hz) and a bargraph tuning adaptor (displays
 to 10Hz). But you are then looking at a pile of gear 19 inches wide and
 about 22 inches high.  Mind you it will impress all your visitors !!!

 Most (if not all) the valves are easily available, in contrast with
 numerous ICs.  Repairs do not demand watchmaker skills except if you have
 to sort out the complex mechanics of the tuning drive of the R390.

 WJ sets are of superb construction and look handsome but many are getting
 on in years.  Some parts may be difficult to source.

 My ultimate advice is to get either the newer Racal RA1772 (of the
 mid-late 1970s (but only if in good condition) or the much newer RA3701.
 You are assured of excellent built quality, a bomb-proof front end, decent
 IF filters, 8kHz AM filter, superb velvety tuning feel with light finger
 tip control and are generally nice to use.  They have no unobtainable ICs
 or other parts.  The 1772 uses a low distortion AM detector based on a
 constant current circuit.  The 3701 uses a synchro AM detector.  Both have
 IF outputs for external adaptors.  The frequency display in both is to
 10Hz and stability can be measured in fractions of a Hz per week.  Neither
 has IF shift or a notch filter.  Few professional receivers until the
 latest bunch of DSP rigs have these since such functions can be
 implemented at little or no extra cost within the mathematics of the DSP.

 Two other sought-after radios with excellent audio are the Plessey PR2250
 and 2280.  However, there are few around and many of the Plessey-made ICs
 inside them have not been manufactured for many years.

 I was not impressed with the 340.  The tuning is awful compared with the
 Racals and in the UK their cost is huge.  Others swear by them.  You need
 to try one out and see if you can live with the tuning.  The knobs and
 displays are good.

 The current "must have" is the WJ 8711A.  I tried out the original version
 at an exhibition when they first came out and was horrified at how
 mechanically flimsy it was, the cheapness of the knobs and meter and the
 number of spurious noises I could hear.  This view was echoed in a
 detailed technical review in SWM where many unkind (but accurate) words
 were voiced. The new A version is said to be a lot better but is just as
 flimsy.  I prefer more mechanically robust rigs. I commend the words of
 Julius Caesar (if you remember your Shakespeare): "Let me have men about
 me that are fat............Yond Cassius has a lean and hungry look".

 Your needs may also be met by the Rohde & Schwarz EK085 but big, big
 money, rare even on eBay, and repair bills ........... more big money!

 I hope this helps.

 Regards
 Michael O'Beirne
 G8MOB

[WA1HZK]

Here's a follow up.
Keith


Dear Bill,


I tend to agree with everything that you and Michael said.


Although I am a DSP user both at my job and at home (see my article in July/August 2005 QEX) I prefer to listen to receivers that use analog elements in the signal processing chain.


It's really hard to put into scientific terms, but they seem to sound better.  I've compared DSP based receivers like a K3, HF-1000, and IC-7800 with analog receivers like an R-9000, R8B, and WJ-8718A.  If you use high quality earphones (I like the Sennheiser ones too; I use HD490s) the analog receivers just sound better.


I've matched bandwidths, put both receivers on parallel taps of a multicoupler, disabled as much DSP processing as possible and the analog receivers still sound better.  Although DSP based AGC can really screw up a receiver (my WJ-8712 is impossible to listen to) that's not all there is to it.


Since I'm not a communications engineer (I leave that level of analysis to folks like Rob Sherwood) I haven't dug into why this is true.  But I have some prime suspects:


1.  Quantization effects due to the number of effective bits in the A/D (not just bits, but effective ones).
2.  DSP algorithms are only an approximation of linear systems due to effects like a zero-order hold.
3.  For cost reasons receivers simplify the algorithms/filters to fit into lower performance DSP chips.
4.  Then there is the AGC.


Clearly the newer DSP based radios sound much better than the older ones.  This makes sense as the processors are more powerful and the A/Ds have more resolution and sample at faster rates.  Maybe without a direct comparison they sound good indeed.  But I'd still rather listen to BBC an R-9000 or WJ-8718A than an IC-7800 or K3.


A while ago I decided to put the "tubes vs transistors" argument to the test.  I designed and built a stereo amp out of both technologies.  Well, the transistor amp measures better in every regard.  But the tube amp sounds better.  One argument for this is that tubes tend to generate harmonic distortion due to their quadratic term in the transfer function.  Transistors, on the other hand, have an exponential term in theirs.  That makes the distortion more annoying.  Another argument is that the harmonic distortion "fills in" the harmonic sequence of music, so it sounds richer. 


I suspect the difference between DSP based and analog based receivers is a little more complicated.


BTW:  I try to turn off every digital device that I can in the house while I listen to SW.  Every one of them, including the little switching power converters, generates a little noise somewhere in HF.  I wish my neighbors would occasionally turn off all their noise-makers as well.


73's,


Gary WA0SPM

[Fred k2dx]

Another Racal that is impressive is the RA 6778c. Not a well known radio I guess due to it's scarcity. Theres are a *few* of them around (in Canada mostly), as they were made in the U.S. for Canadian intelligence service to monitor Ruski sigs over the pole. Outstanding construction. Osterman states one of the most heavily shielded receivers produced as well as Tempest qualified. The filters are very good, 8 KHz down to 150 Hz. (8KHz seems to be the widest due to early roofing) The 'feel' is what you would expect in such a premium receiver. The weak point is the internal audio amplifier / speaker, I use good phones on either front panel jack. Line audio out is available of course. Analog receiver, no DSP but full flexibilty of microprocessor control - 99 memories, sweep, scan, computer control ready, etc.

I have to say I prefer it slightly to my Harris RF590 or WJ 8718. IMHO if you try one it won't disappoint.

[WA1HZK]

Some more input from the Brits...

Good evening all.

I agree wholeheartedly with Gary's piece this evening.

There is the added factor that human perception is incredible difficult to
analyse and measure.  Resistors are unthinking and will follow Ohms Law
regardless and receivers will
follow the IP3 rules but we humans are umpteen times more sophisticated in
the way we function and seldom function entirely as predicted.  How can any
machine sensibly compare the tonal qualities of a Yahama piano with a
Broadwood with a Steinway?  Of course one can take spectrum measurements but
what do they prove?  As I type this I am listening to a relay on the BBC of
"Rigoletto" direct from the New York Met.  How can any machine possibly
compare the qualities of the voices of the baritones singing Rigoletto and
Monterone?


Harmonic audio distortion sounds not unpleasant.  As a professional musician
friend of mine told me not long back, this distortion mostly brightens up
the top end of the spectrum.  Sometimes he deliberately adds some to his
"creations" to alter the effect.  This distortion is given mostly by valved
amplifiers.


Intermodulation audio distortion, in contrast, sounds absolutely awful.  I
joke with him that this is "Stockhausen music".  I have read that transistor
amps are far more prone to such distortion.  The ear is particularly
sensitive to this distortion and may react disproportionately to what lab
measurements of IM would suggest.


There are of course the super keen audiophiles with super deep pockets who
are prepared to pay outrageous sums for valved amps.  I saw a review of one
the other day retailing for over £14,000.  Presumably even the rich
bankers/footballers/Russian billionaires and the like would hesitate before
parting with such money unless the quality was unexcelled.


So where is this ramble heading?  Only that technical measurements are but
part of the story.


73s
Michael
G8MOB

[WA1GFZ]

Gary the Racal man told me the 6778 does not use a processor so controlled by discrete logic. This makes it a lot more complicated but quieter. He has fixed a number of them. Toronto Surplus has had one for sale for years. The 1772 is also a nice RX but older. My hot Rod 6830 has dual 8 KHz roofing filters. It does strip off a lot of broad band crud. Most Racals have 600 ohm balanced audio if you want it.