|
Homebrew Equipment A place to show, design and discuss the weird and wonderful electronic creations from the hands of individual members. |
|
Thread Tools |
27th Mar 2014, 1:22 am | #21 | |
Nonode
Join Date: Jan 2009
Location: Papamoa Beach, Bay of Plenty, New Zealand
Posts: 2,943
|
Re: That being so, what does a Contemporary Hombrew Comms Rx Look like???
RW David, I suspect that your proposal would produce a receiver that handles like an AR88 but delivers performance that is in many respects well above it. But just as the AR88 was as state-of-the-art when it was new, your design would, I think, be state-of-the-art today, within the constraint of using the same topology.
I had written what follows before your post arrived, from the perspective that the goal is to match, but not significantly exceed AR88 capability, other than incidentally, and using technology that was easily available and not too exotic circa 1980. Quote:
Back to fets vs. valves, though, the Drake SPR-4 (fets) was allegedly as good as the R-4C (valves), although I have not seen hard data. And the Eddystone EC958 small-signal fet-based front-end might provide some comfort. Now by all accounts, the EC958 was very definitely not in the RA1772 class, and looked at from those heady heights, it might be easy to be dismissive about it. But by the standards of its time (1968-69, several years before the game-changing RA1772), it evidently had very good front end performance. It had a single RF stage, but a bandpass input circuit so was still 4-gang. An interesting aspect of the AR88 and like designs is that their very good front-end selectivity was probably driven primarily by the need for good image rejection at upper HF, but it also had the advantage of better protecting the mixer from out-of-band signals, and so improving the big signal handling capability. One wonders if some double conversion designs with higher 1st IFs and single RF stages, 3 gangs, had similar image rejection but poorer big-signal handling. It might be noted that Marconi typically used 2 RF stages and 4 gangs for its valve dual-conversion receivers, and that Eddystone used single RF stages but with bandpass inputs for its 830 and EA12 valve dual-conversion models. I suspect that not only image rejection, but also 1st mixer protection was a considered factor in both cases. Turning the telescope around, as it were, another approach to the primary question might be to take a “modern” upconversion receiver, such as the RA1792 (which incidentally had a 455 kHz 2nd IF rather than the by-then-customary 1.4 MHz) and look to see whether it could be simplified and de-featured to fit the desired performance level and handling characteristics. This could include the use of lesser devices where the prototypes delivered performance well beyond what is required and alternatives are significantly cheaper and/or easier to use and/or more readily available. Cheers, |
|
27th Mar 2014, 3:03 am | #22 |
Dekatron
Join Date: Apr 2009
Location: Oxfordshire, UK.
Posts: 4,923
|
Re: That being so, what does a Contemporary Hombrew Comms Rx Look like???
@David; just to clarify, the 3-pole "Cohn" filter mentioned at the start of the post is the same as you what you refer to as the pre-selector later on?
I Googled "Cohn" and there are many hits, so like you say, many strings on his bow. You cite mechanical capacitor tuning, with 3 gangs on the pre-selector, plus one on the mixer (?) being driven by stepper motor controlled by the DFC? Perhaps I should have kept the inclusion of the DFC a secret. I think I see before me, 4 gangs, driven not with a BC221 component, but a National HRO dial and PWD, of course . Curiously, I have those two items ready to go. B |
27th Mar 2014, 6:04 am | #23 | |
Nonode
Join Date: Jan 2009
Location: Papamoa Beach, Bay of Plenty, New Zealand
Posts: 2,943
|
Re: That being so, what does a Contemporary Hombrew Comms Rx Look like???
Quote:
Cheers, |
|
27th Mar 2014, 9:29 am | #24 |
Moderator
Join Date: Mar 2012
Location: Fife, Scotland, UK.
Posts: 22,798
|
Re: That being so, what does a Contemporary Hombrew Comms Rx Look like???
" I suspect that your proposal would produce a receiver that handles like an AR88 but delivers performance that is in many respects well above it. But just as the AR88 was as state-of-the-art when it was new, your design would, I think, be state-of-the-art today, within the constraint of using the same topology. "
Yup. That's what I was aiming for as an example. In its day, the AR88 had a single downconverter architecture limited to the only modes they used at the time and the realisation was state of the art. Bazz, the 3-pole tuneable filter that seems to have become named Cohn needs three gangs and would do the job pretty well. What I was suggesting as a further, somewhat OTT variant would be a second 3-pole filter between RF amp and mixer giving 6 poles of total preselection. These filter designs were certainly in print in the 1970s, so if not 'vintage', are certainly 'classic'. Funnily enough, I have a big diecast box with a stepper motor on it and inside is one of those ceramic-shafted 4-gang capacitors that I think may have been RA17 spares. I want all the rotors grounded, so I've machined-up and fitted a new brass shaft, so all the grounding contacts are now in parallel. I'm using this for occasional work on developing a serious preselector for a grossly OTT radio project. G3ROO and myself refer to as the Doomsday Box. I think that's a reference to the probable completion date. Yes, those AD variable gain amps are Barrie Gilbert's work. There are transistor trees in there with current-voltage conversion diodes, the true Gilbert Cell. Analog Devices wanted him so much they opened a whole new division out in the US North West close to where he was living while working for Tektronix. Perhaps the Gilbert ought to be the unit of recognition for an engineer? Though I think the name's already taken. Anyway such a unit, like a Farad, would be a lot too large for general use. Plan B for variable gain with very low distortion is a mix of ancient and modern. ORP12 photoresistors lit by super bright LEDs. I've found you need to run them at the bright end of their range because their response gets slower and slower at low light levels. The AR88 architecture done afresh was only one possibility. The same job could be done to the RA1772 for example. Racal missed a number of tricks. One big one could have been done with the same components they'd used... if only they'd known. David
__________________
Can't afford the volcanic island yet, but the plans for my monorail and the goons' uniforms are done |
27th Mar 2014, 9:39 am | #25 |
Moderator
Join Date: Mar 2012
Location: Fife, Scotland, UK.
Posts: 22,798
|
Re: That being so, what does a Contemporary Hombrew Comms Rx Look like???
Sorry, I forgot to mention the RA1792.
It is already a heavily de-featured radio compared to the RA1772. I have a 1792, and I'd like a 1772. The 1792 is a very poor relation. Cheapened off wherever they could to get it down to a price. Even the bearing bush for the tuning knob is plastic and wears out. It lacks a preselector and it's wideband into the mixer. Surplus dealers charged a lot of money for 1792s because it said Racal on the panel and they were newer. THey were designed to only be just good enough for a specified, not very demanding purpose, and done by a different branch of Racal. David
__________________
Can't afford the volcanic island yet, but the plans for my monorail and the goons' uniforms are done |
27th Mar 2014, 10:27 am | #26 |
Dekatron
Join Date: Apr 2012
Location: Wiltshire, UK.
Posts: 13,951
|
Re: That being so, what does a Contemporary Hombrew Comms Rx Look like???
Definitely some interesting thoughts emerging: I'm not so sure whether an IF in the 450KHz range is really the way to go though - it does complicate the need for serious pre-first-mixer selectivity which in turn means more signal-frequency tuned-circuits with all the associated bandswitching/tracking issues.
If you go for a higher IF and a RF-robust mixer it opens up the possibility for fixed-tuned bandpass filters for each band (maybe you need four of them to cover 28-30MHz)? at which point you no longer have front-end-tracking issues. I'm surprised nobody's really mentioned short-term local-oscillator stability: whether you're aiming to receive SSB or CW this is a big issue. if I had to design a seriously-stable LO but wasn't allowed to use anything digital I'd probably come back to the old trick of a low-frequency oscillator (tuning let's say 500KHz-1MHz) then put this into a balanced-mixer with the output of a crystal osc, one crystal per band. The advantage here is that you can focus all your attention on designing that one low-frequency oscillator for best mechanical/electrical/thermal stability, linearity-of-tuning-scale and a nice silky-smooth drive without needing to worry about bandswitching it [bandswitches are to my mind a no-no for _any_ variable-frequency oscillator] but then by mixing its output with a suitable crystal you shift the osc to the required LO frequency without compromising the stability. It also gives the benefit of a constant tuning-rate in terms of KHz-per-degree-of-rotation, which would mean you could have one really nicely engineered and easy-to-read tuning dial (filmstrip-style?) whose calibration remained consistent. Detectors are another interesting area to ponder: a balanced product-detector would attract most of my attention but for receiving the occasional full-carrier-AM stations I would either still put the AM through one of the SSB filters (thereby lopping off the other sideband and the carrier) then resolving it as SSB. Or I can possibly see the signal-to-noise-ratio advantages of using a PLL CIO and demodulating both sidebands coherently - though that adds complication and cost because it means also having a 6KHz-wide IF filter available to let both sidebands through. |
27th Mar 2014, 2:27 pm | #27 |
Moderator
Join Date: Mar 2012
Location: Fife, Scotland, UK.
Posts: 22,798
|
Re: That being so, what does a Contemporary Hombrew Comms Rx Look like???
The narrow preselector will out-perform a bandpass fixed filter even with a strong mixer and it's for fun, so I'd go for the fancy way.
Downconverting to 455kHz, was from the AR88 architecture. I wouldn't do it nowadays. Downconverting to 1.4MHz would be better and I have a *huge* collection of 1.4MHz crystal filters for all modes, so this route is interesting to me. The issue is whether the gap in coverage is OK or not. Upconverting to 40-60MHz (McKay Marine did 90MHz!) and then back down to 1.4MHz is the next possibility. I have some Marconi filters at 68.6MHz designed for just this purpose (nice round 70MHz 2nd LO) But maybe a set of narrow mode-specific first IF filters would be good for reducing off channel signals later on. This is IC7800 style. Can I build SSB and CW filters at 60 MHz? Yes. Is it easy? well it's not impossible but probably would involve expensive custom quartz. Going down to 1.4MHz IF for tight mode specific filters is good for me, but I stocked up when the market flooded with 1.4MHz filters. Mentioning the IC7800. It has narrow first IF filters at 60-ish MHz and then comes straight down to maybe 15kHz to use high dynamic range ADCs ahead of DSPs. Actually, I like this (I don't like their price, though, and I once carried one for a couple of hundred yards... Japan is approaching another AR88 characteristic) On another front, Yaesu have done tuneable preselectors with permeability tuning by stepper motors, but only a single pole. How about a multipole bandswitched version? It sounds like fun and we are in it for fun. No-one is paying salaries and calling the shots. We can pick the best old techniques and the best recent ones and not have to take short cuts which spoil the ship for a ha'porth of tar. David
__________________
Can't afford the volcanic island yet, but the plans for my monorail and the goons' uniforms are done |
27th Mar 2014, 4:17 pm | #28 |
Dekatron
Join Date: Nov 2011
Location: Surrey, UK.
Posts: 4,385
|
Re: That being so, what does a Contemporary Hombrew Comms Rx Look like???
Picking what we like from old and new (and not forgetting that our AR88 baseline was actually quite a straightforward radio, there's little in it (oh, alright, the crystal filter and VR150 maybe, neither of these too difficult!) that Mr. Average Joe who mended late 'thirties table radios wouldn't have been pretty familiar with), how about having what Mr. Racal and Mr. Collins would have liked to have had 65 years ago- a straightforward (TTL even?) DDS providing switchable 1MHz steps into a high-level switching 1st mixer, followed by a stable, relatively low frequency continuous tuning VFO for second conversion. Keeps the technology straightforward and understandable for the intrigued hobbyists like myself but still good enough for stable and repeatable SSB reception- definitely not something on the AR88's original brief. The VFO could even have a highly resolving kHz readout but it would be nice to have a VFD readout to 100Hz as well- like the 4 3/4 digit, 40MHz one that Ambit did in the late 'seventies that the schoolboy me couldn't afford :-(
I have a feeling that trying to replace the AR88 is going to be like trying to replace the DC-3 or the VW Combi- quite a few have tried, and many of these have found themselves deciding that the original was very difficult to better! |