Call Search
     

New to Ham Radio?
My Profile

Community
Articles
Forums
News
Reviews
Friends Remembered
Strays
Survey Question

Operating
Contesting
DX Cluster Spots
Propagation

Resources
Calendar
Classifieds
Ham Exams
Ham Links
List Archives
News Articles
Product Reviews
QSL Managers

Site Info
eHam Help (FAQ)
Support the site
The eHam Team
Advertising Info
Vision Statement
About eHam.net

   Home   Help Search  
Pages: [1]   Go Down
  Print  
Author Topic: Is Down conversion better than up conversion receivers.  (Read 4822 times)
WB7CTI
Member

Posts: 1




Ignore
« on: November 11, 2012, 08:36:17 PM »

I'm use to thinking that a receiver with triple conversion was better than one that was double conversion.  Heck, the Kenwood TS-2000 has even a quad conversion receiver.  So how do I compare a down conversion receiver to these and what are the pros and cons of down conversion.
Logged
WX7G
Member

Posts: 6136




Ignore
« Reply #1 on: November 12, 2012, 03:09:53 AM »

An example of down conversion is the Elecraft K3. It uses down conversion so the 1st IF is at a low frequency and the roofing filter can be narrow (good). Many transceivers up convert to 70 MHz and the roofing filter is then wide (not as good).

I believe that with less conversions there will be less spurs (birdies).
Logged
G3RZP
Member

Posts: 4731




Ignore
« Reply #2 on: November 12, 2012, 04:06:52 AM »

Oh boy! What a question!  To cover it properly is a week's course.....

Double conversion came about because a low IF (needed to get selectivity) led to problems with image rejection. As the HF bands got more crowded, the lack of image rejection gave major problems. For exampl a 455kHz IF with high side injection meant that the image was 910 kHz higher than the tune frequency, and on 14 MHz, the image was slap in to the 19m HF broadcast band, where there were some very big signals. (Low side injection wasn't generally used, because it complicates the tracking of oscillator and signal frequency circuits, although some receivers used it on the top frequency range - something like 14 to 30 MHz.). The difficulty came with double conversion that the main selectivity was achieved further back in the receiver circuitry and so the levels of unwanted signals were bigger leading to problems of cross and intermodulation. Some receivers, in order to get good selectivity without the cost of crystal or mechanical filters for SSB used a low IF of 50 or 60 kHz. This needed multiple conversion - the Hammarlund HQ170 and 180 used a first IF of 3.035 on bands above 7MHz, then 455kHz, then 60 kHz. To get adequate rejection of the image of 60kHz from 3.035 would need quite a bit of filtering, so the intermediate conversion to 455 was needed.



Up conversion to an IF higher than the signal frequency became popular in the 1960's, although the Racal RA17 with its Wadley loop system used it in the 1950s. The advent of technology that could produce good, relatively narrow crystal filters was, with synthesisers, a driving force. The original idea, used in various receivers from Racal, Marconi, Redifon etc used an IF in the 30 to 40 MHz region, and it was hoped that by just using a low pass filter, you could get rid of those horrible tuned circuits that cost so much. Then it was found that the intermodulation performance was not good enough to handle the much larger signal power that the receiver mixer saw  because of it being hit with the whole of the HF spectrum, as well as problems with IMD in the crystal filter. Further, you could not at that time do narrow low VHF crystal filters, plus the problems of temperature drift in them. So double conversion was needed as a minimum to allow the use of lower frequency filters at 1.7 or 1.4 MHz or  455 or 100kHz, depending on what was historically used. The addition of pass band tuning and variable bandwidth meant another stage of conversion, but by that time, the selectivity in the early stages of the receiver provided enough protection to minimise intermod problems. But reducing the total power the early stages of the rx saw helped a lot, and the use of sub-octave filters menat that second order IMD was reduced.
 
The introduction of narrow band crystal filters with good IMD or the use of two identical filters coupled with 90 degree hybrids allowed an improvement in performance.

Moving to DSP has again meant a low IF, and avoiding image and simplifying synthesiser design means up-converting and then coming down to a suitable frequency for DSP, which requires multiple conversion. Again, the image of the second IF can make it desirable to have an intermediate conversion, depending on exactly how the architecture is arranged. At the same time, especially if covering a very wide frequency range, the first IF may well too low to give good image rejection at the top end of the range - a good rule of thumb is that the IF should be about 10% of the signal frequency, and never less than 5% in a good receiver. Additionally, the lower the IF, the nearer the local oscillator frequency is to the signal, and leakage of phase noise from the oscillator can be a problem - this is specifically so with direct conversion, where the local oscillator leakage means the  noise mixes with itself down into the  baseband. This can be the major sensitivity limitation in that architecture.

Bear in mind that the selectivity needs to be as near the antenna as possible, so that unwanted signals are rejected before they can cause intermodulation. So the conflict occurs between needing the low IF for selectivity (or these days for the DSP) and the high IF for image rejection.

Somw 30 years ago, a marine receiver was designed which used a wideband front end, a one bit sampler at 500 MHz followed by successively higher bit sampling at lower frequencies - a sort of digital multiple conversion rx. It met the bench tests beautifully, but when placed on a ship in Rotterdam, they had to make a special (and expensive!) stop in Southampton to have it removed and a conventional rx fitted -  the capability of handling the number of big signals that the whole of the HF range can provide led to the thing being unuseable.

Another problem with multiple conversion is internal spurious responses - 'birdies', as opposed to external spurious such as image. Where there are two conversions, internal spurious can be of the form

M times first LO plus/minus N times second LO = first IF or second IF or signal frequncy or image of first or second IFs.

Obviously, bringing in more oscillators gives more possible spurious. For a general purpose HF receiver, fixing  these can be a nightmare - one rx design my team did required the front end PCB to be re-laid out 7 times until we killed the earth loops that were giving us trouble.




OK, hope this helps.......
Logged
N3QE
Member

Posts: 2289




Ignore
« Reply #3 on: November 12, 2012, 04:16:30 AM »

The main advantage of down conversion in a ham rig, is that with a first IF in the 9 MHz ballpark, the first IF filter can be very "tight", in fact the bulk of selectivity can be done right there at the start of the first IF. By getting the selectivity done right at the start of all the IF chains, the downconversion rig gives substantially better performance in the presence of other strong signals on the band.

When up conversion is used, the first IF might be in the 45 MHz or even higher ballpark, and getting good tight filtering there is not so easy. These first filters in an upconversion receiver are often called "roofing filters" because they have traditionally been rather broad and are followed up on, in the 2nd or 3rd or 4th IF with other filters (crystal or mechanical) that actually provide the bulk of selectivity.

To confuse the matter... folks in the past few years have taken to calling the first filter in a downconversion rig the "roofing filter" too. When in fact it may be the only filter needed to provide the  IF selectivity. Somehow to me "roofing filter" always meant that it was the initial filter and cleanup and final filtering was done at later stages.

Logged
TANAKASAN
Member

Posts: 933




Ignore
« Reply #4 on: November 12, 2012, 04:22:46 AM »

I always thought that a roofing filter got it's name because it's at a high frequency and everything useful is below it, for example a 45 MHz roofing filter in a 1 MHz to 30 MHz receiver. Obviously Elecraft build houses differently with a roof between the ground and first floor Smiley

Tanakasan
Logged
G3RZP
Member

Posts: 4731




Ignore
« Reply #5 on: November 12, 2012, 04:30:17 AM »

A problem with HF crystal filters is IMD. Because of necessity the crystal is thinner, for any given voltage, the mechanical stress is greater and  it starts to deviate from Hooke's Law at a lower input level. Typically, 9 MHz filters have an input IP3 of around +15 or +18 dBm for signals outside the pass band but still fairly close.

The final finishing of the crystal also has an effect on the IMD, and you can turn filters round (i.e. swap input and output) and get an improvement. Ferrite transformers in the filter are another definite No-No.

1.4MHz filters are noticeably better. Early mechanical filters weren't very good on IMD, but over the last 30 years, have improved remarkably.
Logged
WX7G
Member

Posts: 6136




Ignore
« Reply #6 on: November 12, 2012, 07:24:16 AM »

G3RZP, that is great information and fascinating. before, I did not know that the IMD performance of crystal and mechanical filters was such a limitation. Is this the dominant limitation of the K3 transceiver?

Logged
AC5UP
Member

Posts: 3903




Ignore
« Reply #7 on: November 12, 2012, 08:08:02 AM »

Some 30 years ago, a marine receiver was designed which used a wideband front end, a one bit sampler at 500 MHz followed by successively higher bit sampling at lower frequencies - a sort of digital multiple conversion rx. It met the bench tests beautifully, but when placed on a ship in Rotterdam, they had to make a special (and expensive!) stop in Southampton to have it removed and a conventional rx fitted -  the capability of handling the number of big signals that the whole of the HF range can provide led to the thing being unuseable.

Glad you mentioned this as I see it as an example of a good idea that was premature for the available technology. It's easy to assume a good idea poorly implemented is a poor idea based on the outcome. Sometimes you have think through the obvious.

I'm not as current on SDR design as I'd like to be, but I am aware of how sophisticated the tools of home digital audio & video have become over the past decade. There was a time when 8 bit arcade game music with blocky animation could attract teenagers and their pocket change. Today their kids take HD video for granted and assume media was always clean, clear, instantly accessible and played on disposable hardware.

The cost and speed of processors will continue to improve and with it the capability of SDR designs. As with digital audio there will be landmark products that present a compelling marketing advantage. Remember when CD players cost more than a turntable? And the discs were almost twice the cost of a vinyl album? The sound was cleaner despite the earliest A/D converters, the noise floor was incredibly low and the discs didn't wear out. Today the CD player is all but obsolete but from a marketing standpoint that was the foot-in-the-door SONY and Philips were looking for. Then came the iPod. And smart phones that did more than make phone calls... They texted, e-mailed, surfed the web, took pictures, could read a bar code or a credit card. VCR's became DVR's. Today you can buy a Hi-Def digital camcorder for about the same money you paid for your first AM / FM Walkman. Etc. Etc. Etc.

Ten years ago, could you have imagined?  Imagine what amateur radio could be ten years from now if the same level of innovation was applied.

What I see holding back hobby SDR radio is twofold: It's a niche market. For every Ham that might be tempted to buy the first really, really good SDR rig there are how many potential smart phone buyers? A hundred? A thousand? Are you still wondering why the money chases smart phones instead of smart radios? Then there's the concept of market inertia. Digital broadcasting arrived in the US about two years ago. Try finding a digital receiver off the shelf at an electronics store. You can order them through the web, but as far as a normal retail item? Not enough demand to make it worthwhile. Amateur radio would be even worse because... How to phrase this... Some of us are really set in our ways. Anything new that has a learning curve will be doomed as a niche product within a niche market and considered too risky by the manufacturers.

OTOH, like computers, you will see independents working out of a virtual garage developing specialty boards.

You'll also hear the argument that digital radio won't take off until the FCC adopts a modulation standard. I can see that in part, but I also know that no microphone, speaker or camera is 100% digital and that hasn't stopped anyone from developing digital mastering. Kinda' interesting to realize that damn near everything created by Industrial Light & Magic started out as an analogue signal... Or, as an old-school Panavision camera might say,  " Luke... I am your father "

I think Pogo understood the roadblocks to digital amateur radio better than we do today... And that was 60 years ago!
Logged

Never change a password on a Friday                
KQ6EA
Member

Posts: 609


WWW

Ignore
« Reply #8 on: November 12, 2012, 09:50:49 PM »

@WX7G....
I was talking to the guys from Elecraft a year or so ago at one of the "HAMCONS", and he told me they had to work directly with INRAD to design the crystal filters used in the K3.

The first go-round of the filters had bad internal IMD, which made the performance K3 worse than they knew it was capable of.

It was one of those "DUH" moments for me, as I had never considered the filters could cause IMD worse than the circuits they were used in.

73, Jim
Logged
G3RZP
Member

Posts: 4731




Ignore
« Reply #9 on: November 12, 2012, 11:21:29 PM »

Back in 1973, I spent some time working with a French company on getting a 38MHz, 9kHz wide filter with a shape factor of 2.5:1 (6 - 60dB), insertion loss of less than 4 dB and an intercept point of +28dBm. They needed special techniques of finishing the crystals to get the IMD.

I left that company before the receiver went into production, and it didn't because their parent company went bust.......Once it was bought by someone else, that rx would have been competition for existing product, so the programme got stopped.

The company I went  to had test fixtures where they tested incoming filters for IMD, noted which way round was best and then installed the best filters in receivers and the not so good ones in exciters.

There was, at that time, a PhD student at Bath University researching the subject: he had such a good offer to go and work for a US company that he never did finish the PhD!

I haven't yet heard a true DSP based rx that didn't, to my mind, have a somewhat 'harsh' sound to the audio. Maybe it's just me.
Logged
WA3SKN
Member

Posts: 5497




Ignore
« Reply #10 on: November 13, 2012, 07:12:56 AM »

The term "roofing filter" used to indicate an up conversion process, but is now being used as a generic term for any bandwidth determining filter, whether an up conversion or downconversion process is used.
A new process that made 70 MHz filters affordable was developed, making up conversion practical.  The original reason that superhetrodyne was developed was to use cheaper low frequency filters for selectivity.  By up-converting, less filter (and loss) is required (read cheaper!).
Most up-converters use a 3 or 4 pole filter, while 8 pole filters have become standard for the downconversion process.  And note that the Hilberling has 16 pole filters... part of the reason for the cost!
Both have plus and minuses.
73s.

-Mike.
Logged
G3RZP
Member

Posts: 4731




Ignore
« Reply #11 on: November 13, 2012, 07:53:03 AM »

Using a high IF makes the synthesiser design easier for an HF rx, as the percentage tuning range is reduced. Was it the KWM380 that used something up around 100MHz?
Logged
AC5UP
Member

Posts: 3903




Ignore
« Reply #12 on: November 13, 2012, 09:41:22 AM »

The original reason that superhetrodyne was developed was to use cheaper low frequency filters for selectivity.

If you're thinking TRF and Neutrodyne to Superheterodyne you're on the right track but there is a little more to it... Adapting to a changing radio audience and market advantage.

A classic high end Radiola had three or more tuning stages operating at RF. Each stage added some gain and selectivity but also meant every station change required peaking three or more L/C circuits. The advantage was that a clever user could peak the stages wide for best fidelity and you can find old adverts that talk about superhet sideband clipping as a disadvantage. In the days when only a handful of stations were on the air and radio was a hobbyists pursuit TRF's were adequate, but as the dial filled up selectivity and convenience became more important. Especially at night when peaking three or more stages was even more of a PITA. The superhets required only two knobs (tuning and volume), used fewer tubes, worked well with an internal loop antenna, and were smaller and cheaper.

Price, performance and convenience are a salesman's best friend and both RCA and The Campbell's Soup Company wanted as many radios in US homes as they could manage. The old brass variable condensers were a fine example of machine work and the honeycomb coils were a marvel to look at, but it all came at a price....

http://www3.telus.net/radiomuseum/projects/FadaC75A/finChas800.jpg  (Note the thumbscrews for the A, B and C batteries)

http://www.radiomuseum.org/forumdata/users/4942/coils/245uH_180-44Litz_44.jpg
Logged

Never change a password on a Friday                
G3RZP
Member

Posts: 4731




Ignore
« Reply #13 on: November 13, 2012, 09:54:08 AM »

You can't get the selectivity with a high IF without crystal filters - there's a limit to the Q that can be got in a coil. So to get the necessary selectivity, you had to go to a superhet, with highest IF at which you could get a reasonable Q for the narrower bandwidth. For communications receivers, it was the National HRO and FB7 for the high end of the market, Hammarlund for the middle, and Hallicrafters for the cheapest stuff.

Professional comms receivers started going to double superhets in the late 1940s, although the Collins tuneable IF approach was unique to them for some time. It was proposed in the RSGB Bulletin in about 1943, though, for amateur band receivers.

Although Armstrong is credited with inventing the superhet, the idea was first proposed by a French guy in 1909: the triode was not then available so it couldn't be built. Much like radar, with Telemobiloscope patent of 1904 by Christian Hulsmeyer.
Logged
Pages: [1]   Go Up
  Print  
 
Jump to:  

Powered by MySQL Powered by PHP Powered by SMF 1.1.11 | SMF © 2006-2009, Simple Machines LLC Valid XHTML 1.0! Valid CSS!