UK Vintage Radio Repair and Restoration Powered By Google Custom Search Vintage Radio and TV Service Data

Go Back   UK Vintage Radio Repair and Restoration Discussion Forum > General Vintage Technology > General Vintage Technology Discussions

Notices

General Vintage Technology Discussions For general discussions about vintage radio and other vintage electronics etc.

Reply
 
Thread Tools
Old 22nd Feb 2021, 9:33 am   #1
SteveCG
Nonode
 
Join Date: Jun 2009
Location: Herefordshire, UK.
Posts: 2,287
Default Noise Figure data for VHF TV and FM front ends

I interested to find manufacturers Noise Figure / Noise Factor information on all the various valve and semiconductor tuners that have been created from the 1950s to the early 1980s. That is, from the early days of the Clydon TV 5 (Band I only, PCC84) via the VHF/UHF semiconductor tuners of the Dual-Standard TV era to the dual gate mostfet units, beloved of FM tuner buffs, with valve FM tuners positioned in between.

I've got some information from the editions of Wireless World which the American Radio History web-site kindly makes available. However information on units like the "Fireball" TV tuner, let alone HI-FI FM tuners, still eludes me.

Any suggestions for where to look - and any actual information on specific units for that matter - would be gratefully received.
SteveCG is offline   Reply With Quote
Old 22nd Feb 2021, 11:11 am   #2
Radio Wrangler
Moderator
 
Radio Wrangler's Avatar
 
Join Date: Mar 2012
Location: Fife, Scotland, UK.
Posts: 16,790
Default Re: Noise Figure data for VHF TV and FM front ends

Noise figure has long been a rather dubious specification. Unless you get data that was similarly measured and done with great care, the comparisons are likely to be misleading. A number of radio equipment manufacturers, lacking the machinery to measure it properly, were inclined to just quote the spec of their front end device.

Quite often you got expressions like .... so many microvolts input to produce XdB S/N ratio without mentioning what the input impedance, or the bandwidth was.

Someone with a collection of tuners could measure them nowadays, of course, but there are still difficulties.

Precision noise sources are normally only calibrated from 10MHz upwards. Noise figure meters like the HP 8970A only work down to 10MHz, and worst of all, the whole thing assumes it's measurement bandwidth (4MHz) is narrower than the thing under test.

The later Agilent N897x family of noise figure analysers share the 10MHz low frequency limitation as far as their internal preamps go, but the measurement bandwidth can be narrowed. I can't remember how narrow Andrew set the limitation in software, but I have used prototypes down to 100kHz where the DSP card is fitted.

I'd use a 15-ish dB ENR noise source with a fixed 5dB pad and then a minimum loss 50 to 75 Ohm pad. Measured carefully, the pad losses can be taken into account.

The 15 dB ENR sources give a small variation in output impedance between noise-on and noise-off states. This can modulate both the gain and the noise figure of the device under test in a way that puts a significant error into the results for low-noise devices. The 6dB ENR family of noise sources simply include an attenuator to dilute the Z variation. You can DIY this by adding a precision attenuator to a 15dB ENR family member, and this is an opportunity for 50 to 75 Ohm matching.

David
__________________
Can't afford the volcanic island yet, but the plans for my monorail and the goons' uniforms are done
Radio Wrangler is online now   Reply With Quote
Old 22nd Feb 2021, 8:24 pm   #3
Synchrodyne
Nonode
 
Join Date: Jan 2009
Location: Mt. Maunganui, New Zealand
Posts: 2,585
Default Re: Noise Figure data for VHF TV and FM front ends

As RW has suggested, obtaining a comprehensive and directly comparable set of the data you are seeking might be something of a quixotic quest. That said, much information is available about the design of, and device selection for TV and FM front end selection, often with references to noise factors, amongst the various other important parameters.

As a start, I suggest the following:

The book “VHF Television Tuners” by D.H. Fisher (of Pye), originally published by Heywood, c.1955.

The IRT paper “A Survey of Tuner Designs for Multi- Channel Television Reception", by D.J. Fewings and S.L Fife (English Electric), 1955.

The article “Use of New Low-Noise Twin-Triode in Television Tuners” by Robert M. Cohen, in RCA Review for 1951 March.

The article “A Nuvistor Low-Noise VHF Tuner” by G.C. Hermeling, in RCA Engineer for 1960 August-September.

The RCA paper “A Comparison of Solid-State and Electron-Tube Devices for TV-Receiver RF and IF Stages” by L.S. Baar and S. Reich, 1966 NEC.

The IEEE paper “A High-Performance VHF Solid-State TV Tuner” by Tae Hyung Moon, Zenith, 1969 July.

The IEEE paper “New VHF and UHF Varactor Tuners”, by G.W. Carter, RCA, 1976 August.


There are many pertinent or related articles in the American magazines available at the American Radio History website, and I’ll compile a list of those that I have found, but that may take a day or two. (I think the Fireball is in there somewhere.)

Some other books, such as those by Fink, are also useful, but I’ll first need to check which has what.

I’ll also follow with a list of FM front end references.

By the way, as far as I know, the Cyldon TV5, Band I only, used an EF80, ECC81 combination, whereas the TV12 (Bands I and III) used the PCC84, PCF80 combination.


Cheers,
Synchrodyne is offline   Reply With Quote
Old 22nd Feb 2021, 11:01 pm   #4
Synchrodyne
Nonode
 
Join Date: Jan 2009
Location: Mt. Maunganui, New Zealand
Posts: 2,585
Default Re: Noise Figure data for VHF TV and FM front ends

To divert for a moment from list compilation, I think that one can develop a broadly-based timeline for the evolution of TV tuners and the active devices used therein, this forming a backdrop to the progress in lowering noise levels.

In the immediate post-WWII era, American practice was to use the existing valve set, particularly the 6AG5 pentode and 6J6 double triode, both WWII era miniatures. In late 1947, GE released the 12AT7 double triode, specifically aimed at the FM and TV frequency changer applications, perhaps with more emphasis on FM than TV. Inter alia, it had lower microphony than the 6J6, highly desirable for FM, and also for split-sound TV receivers.

Around 1948, RCA concluded that the existing TV RF amplifiers were unsatisfactory. The pentode (6AG5) was noisy, but easy to use. The triode (usually in the form of the 6J6) was quieter, but difficult to use. This spurred a development programme that resulted in the series-cascode circuit (originally called the driven grounded-grid circuit) and a corresponding double triode valve, the 6BQ7, released in 1951 . This was as easy to use, and had comparable gain to the pentode, but was as quiet as the triode.

The arrival of the cascode RF amplifier was timely in another respect. The contemporaneous upward movement of the American standard TV IF from 20-something to 40-something MHz indicated the use of pentode, rather than triode mixers, in order to avoid regeneration at the lower low band (Band I) channel frequencies, this due to the closeness of the new IF to those frequencies in combination with the relatively low Q of the IF bandpass circuit at the mixer output. But the pentode mixer was not such a good idea (and not really needed) at the high band (Band III) frequencies because thereat it was very noisy. That disadvantage could be mitigated through the use of a high-gain, low-noise RF amplifier, namely the cascode. The pentode mixer arrived in the form of the triode-pentode, in order to keep the TV front end valve count at two. The early examples were the 6X8 (RCA) and 6U8 (Tungsol).

The FM evolution was somewhat different. RCA had envisaged that its standard post-WWII miniature receiving valve range would cover FM as well as AM receivers, in the FM case largely in FM/AM receivers. Thus the 6BA6 pentode, 6BE6 heptode combination was slated for use at FM as well as at AM. (More complex FM-only receivers and tuners could use the TV valves.) Positioning the 6BE6 as an FM mixer was perhaps not one of RCA’s best moments. But on the other hand, the desirability of common front end valves for FM and AM was not to be denied. RCA’s initial response was to develop the 6SB7-Y as a heptode frequency changer with much improved (but still not brilliant) FM performance. It was an octal, because at that time any small signal valves that required more than 7 pinouts were of that form, the noval base not yet in view. (One has the impression that sudden arrival of the noval base was something of a surprise for RCA, who initially mentioned it en passant, as it were.) GE did differently, though, in that it offered the 12AT7 double triode for use as an FM-AM frequency changer in combined receivers. A double-triode AM frequency changer seems odd, but there was a precedent in that the Sylvania 7F8 (Loctal) had been so-used, including by Hallicrafters for communications-type receivers that included the FM band. One could say that with the 12AT7, precedence was given to FM performance over AM. With the 6BE6 and 6SB7-Y, it was the other way around. Zenith used the 12AT7 combined FM-AM frequency changer for a decade or more, both with and without AM RF amplifier. GE itself did differently, though, using the whole 12AT7 as an FM frequency changer, with one half as AM oscillator, the first IF strip pentode then doubling as AM mixer and FM 1st IF amplifier. Anyway, that shows that FM front end design was not totally divorced from that for AM.

RCA’s initial (and interim) response was twofold. It offered a noval version of the 6SB7-Y, namely the 6BA7, and issued an Application Note covering the use of the 6J6 double triode as an FM-AM frequency changer (probably not too brilliant from the FM perspective). Its definitive response was in the form of the previously mentioned 6X8 triode-pentode frequency changer of 1951, which was also configured for use as an FM-AM frequency changer (for which application a 150 mA heater version, the 19X8, was issued early on.) For AM frequency changing, the 6X8 was used as a triode-pentode frequency changer. For FM frequency changing, it could be used as a triode-pentode, or with the pentode triode-strapped, as a double-triode, depending on the desired noise contribution, which would depend upon what kind of RF amplifier was used. Because by then RCA was strongly of the viewpoint that RF pentodes should have a separate suppressor grid pinout, the 6X8 was so-configured, which meant that it had a combined cathode. On the other hand, the Tungsol 6U8 had separate cathodes, which meant that the pentode suppressor was internally connected to the pentode cathode. That made it quite versatile, and Tungsol offered it for various other TV applications. It was followed by a plethora of “utility” TV triode-pentodes.

Anyway, the cascode RF amplifier coupled with the triode-pentode frequency changer became the norm for American VHF TV tuners. The cascode also found its way into higher-performance FM receivers and tuners, although opinions seem to have been divided over the choice between double triode and triode-pentode frequency changers.

There was a third leg to the utility of the cascode RF amplifier. Early US UHF tuner practice, in the absence of suitably economic and long-enough-lived RF amplifier devices, was to avoid an RF amplifier stage, and instead use a crystal diode mixer (following a bandpass input), which was necessarily followed by a low-noise, high-gain IF stage. For the latter, the cascode was a good choice, and the cascode RF amplifier within the VHF tuner could be used, obviating the need for a separate valve. This also provided front-end agc at UHF as well as at VHF.

European VHF TV front end practice quickly adopted the American pattern of a cascode RF amplifier (e.g. PCC84) followed by a triode-pentode frequency changer (e.g. PCF80, PCF82). But there had been early Band I/Band III examples that used an EF80 pentode RF amplifier with an ECC81 double-triode frequency-changer. These must have been quite noisy on the Band III channels. Cyldon probably thought that the EF80, ECC81 combination was fine for the Band I-only case. It seems that for wideband (e.g. TV) applications, the cascode noise benefits are minimal at Band I frequencies. On the other hand, some receiver makers saw the cascode as better than the pentode above around 20 MHz in narrow-band (HF) applications.

During the later 1950s, alternative VHF TV RF amplifier valves were developed on both sides of the Atlantic, including guided-grid triodes, tetrodes, and the Nuvistor. Apart perhaps from the Nuvistor, I don’t think that these were any quieter than the cascode, just cheaper, and given the production scale of TV tuners, even fractional penny cost reductions were worthwhile. Frame-grid TV cascode and frequency changer valves appeared around 1960, these offering higher gains.

European FM front ends had a somewhat different development pathway. Early on, there was some limited use of AM-type triode-hexode and triode-heptode frequency changers. (For example, the 12AH8 was used in a BBC FM receiver design and in a Lowther FM tuner.) The ECH81 was claimed to be good up to 100 MHz, although Philips suggested a different way of using it, namely normally for AM but with the triode as a self-oscillating mixer on FM, the heptode section then being pentode-strapped as an FM IF amplifier. But the single-valve FM front-end, using a double-triode (ECC85) emerged around 1953-54. I am not sure who originated it, but I think it might have been Telefunken. This quickly became the European norm for FM front ends at what might be called the “cooking” level. It was use for some higher performance equipment, but there the pentode RF amplifier was popular, despite its noise disadvantage. The cascode FM RF amplifier seems to have been a minority in European practice. Perhaps best known amongst British equipment would be the Leak Troughline II, 3 and Stereo tuners. But before the Troughline II, Jason had also used cascode FM RF amplifiers.

The single-valve FM front did not cross the Atlantic in any material way until the late 1950s. Probably it was its economy, rather than its lower noise possibility than the pentode, which appealed. Unlike the case in Europe, there was further development of the concept. That included triple triodes, where the 3rd triode provided AFC (just before the adoption of crystal diodes for this job), dissimilar triodes, where the RF triode only was of the frame-grid type; double tetrodes; and pentode-triodes, for those (such as GE) who preferred pentode RF amplifiers.

By the time UHF TV reception was required in Europe, valves suitable for use in domestic equipment as RF amplifiers and frequency changers at these frequencies had become available. Initially the PC86 was slated for both applications, but the quitter PC88 was son developed for the RF stage. Given the combined gain of these UHF front-end stages, it was customary to feed the UHF IF output to the VHF tuner mixer, bypassing the VHF RF stage. Thus European and American practice differed in this regard. (Later, in the 1970s, American practice did change, but for what might be called external rather than intrinsic reasons.) Around 1960 or so, European valve makers (Mazda might have been the first) introduced TV VHF triode-pentode frequency changers that had remote (or perhaps semi-remote) cutoff pentode sections that allowed for the application of AGC bias. In part that might have been to extend the VHF AGC range. But I also suspect (but don’t know for sure) that it might also have been to obtain a measure of “front-end” AGC on UHF.

Relating the above to noise factors, it would seem that during the valve era, these could be made about as low as they needed to go for domestic VHF TV and FM reception applications with the arrival of the cascode circuit and associated valves, with any improvements thereafter of an incremental (or perhaps one should say decremental) nature.

Brief comments on the solid-state era to follow.


Cheers.
Synchrodyne is offline   Reply With Quote
Old 22nd Feb 2021, 11:30 pm   #5
Synchrodyne
Nonode
 
Join Date: Jan 2009
Location: Mt. Maunganui, New Zealand
Posts: 2,585
Default Re: Noise Figure data for VHF TV and FM front ends

Attached are some Mullard TV Tuner datasheets that include noise figures.
Attached Files
File Type: pdf Mullard TV Tuners.pdf (645.8 KB, 28 views)
File Type: pdf Mullard U321.pdf (262.2 KB, 19 views)
Synchrodyne is offline   Reply With Quote
Old 23rd Feb 2021, 1:26 am   #6
Synchrodyne
Nonode
 
Join Date: Jan 2009
Location: Mt. Maunganui, New Zealand
Posts: 2,585
Default Re: Noise Figure data for VHF TV and FM front ends

I think that the transistor era for TV and FM front ends is easier to categorize than was the valve era.

With VHF TV front ends, I don’t think that the typical three-transistor arrangement had much advantage over valves when it came to noise. But it was a lot worse in respect of signal handling and spurious responses, such as cross-modulation. That did not seem to much inhibit its adoption in Europe, but it was of major concern in the USA. Some makers held on to valve VHF tuners, at least for higher performance models, until mosfet-based types became available.

Given that domestic-application jfets were available before corresponding mosfets, some work was done, e.g. by TI, on jfet-based VHF TV tuners. Although good performance could be obtained, there were stumbling blocks. A cascode jfet pair was generally fine as RF amplifier, except that it did not have an adequate AGC range. A single jfet was an excellent mixer, but had little gain, necessitating an additional device for a supplementary gain stage. Thus relative economics were against the use of jfets. On the other hand, dual-gate mosfets did provide the answer. Initial use (by RCA c.1968) was for the RF stage, soon extending to the mixer stage. Some makers waited until protected-gate mosfets became available. With dual-gate mosfets, it was possible to match or better valve VHF tuner performance in respect of noise, gain, signal handling and spurious response. But I don’t think that there was a quantum leap in terms of noise performance. The Zenith paper mentioned above covers some of this ground. The subject solid-state tuner had a slightly lower noise factor than a good valve tuner at Band III, but slightly higher at Band I. Anyway, mosfet-based tuners provided American TV receiver makers with the opportunity to convert to solid-state without material performance penalty.

UHF was a different case, and in Europe two-transistor UHF tuners were adopted quickly once they became available, having much better noise performance than their valve counterparts, although signal-handling was more limited. In the USA, the transistor quickly replaced the valve as oscillator in their essentially passive UHF tuners.

The existing European bipolar VHF and UHF TV tuner layouts were more-or-less amenable to the use of varactor tuning, albeit with the possibility of some reduction in signal-handling capability. See WW 1976 January p.56 for a comparison of the Mullard ECL1042 (mechanical tuning) and ECL1043 (varicap tuning) in that regard.

But the advent of varicap tuning in the USA indicated that some changes be made. Firstly, it was no longer convenient to feed the UHF tuner output into the VHF tuner RF input. Probably this was because the VHF tuning range was not easily extended downwards to encompass the IF channel. With switched tuning of some form, the UHF IF input was essentially just another channel. This meant that additional gain had to be provided for the UHF channel. Typically this was a common base bipolar stage just after the mixer, and in the UHF tuner box, although sometimes in the VHF tuner box. I think the latter tended to apply where a conventional UHF tuner was used in conjunction with a varactor VHF tuner. An alternative was to use an active mixer, per European practice, and I think that Sylvania did this. The addition of varactor tuning to the UHF tuner itself necessitated the addition of an RF amplifier in order to maintain gain and Q of the input circuits. So UHF tuners would then have a transistor RF amplifier, a transistor oscillator, a diode mixer and a transistor IF pre-stage.

Dual-gate mosfets suitable for UHF RF amplifier use arrived c.1975. RCA was an early user, in its KRK226 tuner, described in the 1976 IEEE paper mentioned above. Somewhat oddly in retrospect, this retained the bipolar IF preamplifier of its all-bipolar predecessor, even though it fed the mixer dual-gate mosfet of the corresponding KRK228 VHF tuner. The KRK226 UHF tuner was said to have a better noise factor (7 dB at channel 83) than its bipolar predecessor, but comparative numbers were not given.

I think it was not until the 1980s, or even later, before the dual-gate mosfet found widespread use in European TV front-end practice, although I have a vague notion that there were exceptions, such as Tandberg in the 1970s on the VHF side. At least here in NZ, the difference in signal-handling at VHF was fairly obvious. With just two channels, one Band I and one Band III, a major make with a biopolar tuner showed serious cross-modulation with a signal level that a valve monochrome receiver handled easily, and in fact the bipolar unit required 18 dB of attenuation to clear the problem. That was in Auckland in the 1970s; I had a repeat performance about a decade later in Wellington with a later model from the same make. In the early solid-state days, only one maker in NZ (Pye) used a mosfet-based tuner. (I don’t know if it sourced from RCA or a Japanese maker.)

I suspect that in the VHF case, looking at a wide range of noise factor numbers might show that there is more scatter between makes and models, than amongst valve (cascode & later), bipolar and mosfet circuits. On the other hand, in the UHF case, bipolar generally should show a step improvement as compared with valves. Mosfets might have show a consistent but perhaps not as large further improvement, but that is uncertain. Quite where the American UHF tuners with passive signal paths and transistor oscillators would fit I don’t know. But that American practice did not change until it was forced to by the advent of varactor tuning suggests that they could be made quite enough, such that a change to the 1960s European bipolar style was not considered worthwhile.


The FM case was somewhat different, more to follow.


Cheers,
Synchrodyne is offline   Reply With Quote
Old 23rd Feb 2021, 4:15 am   #7
Synchrodyne
Nonode
 
Join Date: Jan 2009
Location: Mt. Maunganui, New Zealand
Posts: 2,585
Default Re: Noise Figure data for VHF TV and FM front ends

Quote:
Originally Posted by Synchrodyne View Post
There are many pertinent or related articles in the American magazines available at the American Radio History website, and I’ll compile a list of those that I have found, but that may take a day or two. (I think the Fireball is in there somewhere.)

Here is the list as it relates to TV tuners.

TV Tuner Articles.pdf


I haven’t checked thoroughly to see which articles contain useful information about noise factors. But I have pulled out a couple of pages that include noise factor graphs.

RE 195601 p.48.pdf RE 196104 p.44.pdf


As may be seen, variation of noise factor with gain reduction is another parameter of interest, and evidently one that can vary between different tuners with the same maximum gain noise factor.


Cheers,
Synchrodyne is offline   Reply With Quote
Old 23rd Feb 2021, 8:23 am   #8
SteveCG
Nonode
 
Join Date: Jun 2009
Location: Herefordshire, UK.
Posts: 2,287
Default Re: Noise Figure data for VHF TV and FM front ends

Synchrodyne,

Thanks for the data - plenty to study.

It might make you smile to know that I woke up this morning and realised that I had put the TV 12 front end valve (a PCC84) by mistake as being used by the TV 5 tuner which uses an EF80. I'm pleased though that somebody spotted and corrected my mistake! - that attention to detail is what makes this Forum really useful.

BTW: My interest in Noise Figures comes from wondering why the Cosmic Radio background ( Synchrotron radiation from within our Galaxy ) does not appear much in discussions of what puts a limit on VHF Radio and TV reception.
SteveCG is offline   Reply With Quote
Old 23rd Feb 2021, 10:24 pm   #9
Synchrodyne
Nonode
 
Join Date: Jan 2009
Location: Mt. Maunganui, New Zealand
Posts: 2,585
Default Re: Noise Figure data for VHF TV and FM front ends

Cosmic noise was mentioned in the English Electric paper, and I have attached the pertinent pages from it.

My take on that is that at Band III frequencies, cosmic noise was low enough to be swamped by receiver circuit noise, so for VHF TV front ends, the emphasis was on getting the latter as low as needed for satisfactory performance (as distinct from as low as possible). Apparently a relatively close approach to that target was possible with the valve cascode RF amplifier in 1951, and progress since then was mostly incremental. That may explain why cosmic noise was not much mentioned.

That said, I suspect that one might find non-trivial differences in device noise performance at the upper end of Band III, particularly in the 216 to 230 MHz band extension range. Even more so for the South African extension, to 254 MHz. The last-mentioned may have been in the region where mechanical tuning methods started to fall over, but fortunately by the time TV broadcasting started in RSA, in 1976, varactor tuning was well established.

Out of curiosity, I looked up the numbers for the Eddystone 990R and 990S communications receivers of the late 1960s, these being based upon bipolar circuitry. They formed a nominal VHF/UHF pair, the 990R tuning from 27 to 240 MHz in four switched bands, the 990S from 230 to 870 MHz in two bands, namely 230 to 510 and 470 to 870 MHz. The front end for the last-mentioned band looked suspiciously like a standard two-transistor UHF TV tuner. The separate front end for the 230 to 510 MHz looked as if it were a modified version of same. One could infer that 240 MHz was the point above which Eddystone saw the need to switch to UHF devices and tuning techniques to maintain performance. Noise factor for the 990R was given as less than 10 dB for all bands, suggesting that at 240 MHz it was getting close to that limit. For the 230 to 510 MHz band, noise factor was 8 to 10 dB. I’d expect the 8 dB to have applied at the lower end, so the 990S was just a bit better than the 990R in their frequency overlap range. (In this case I think that reasonably we may assume that the same measurement technique was used for both receivers.)


Cheers,
Synchrodyne is offline   Reply With Quote
Old 24th Feb 2021, 12:52 pm   #10
SteveCG
Nonode
 
Join Date: Jun 2009
Location: Herefordshire, UK.
Posts: 2,287
Default Re: Noise Figure data for VHF TV and FM front ends

Radio_Wrangler correctly points out the challenge that measuring Noise Figures represents, with many manufacturers rather dodging the issue.

So allow me to expand on my interest in this more:

I am not sure how much the contribution of radio noise from our Galaxy is appreciated when discussing 'Fringe Area' reception for Band I TV, Band II FM and even (just) Band III TV. I think recognition of its importance slowly grew over the years from the early days of TV and FM and is still relevant now to the (dwindling?) number of folk who still listen to FM radio.

Now firstly what do I mean by Galactic radio noise? It is a fact that our galaxy is a source of radio emission - Karl Jansky discovered this in the 1930s when investigating sources of interference to long-distance, short-wave radio links. The embryonic discipline of Radio Astronomy explored it more in detail in the later 1940s and 50s. What was found was that electrons between the stars travel in circular orbits, as a result of the galactic magnetic field, producing radio radiation called synchrotron radiation. We are bathed in this radiation from all directions - with the most coming from the direction of the centre of the Galaxy and the least coming from the direction of the Galactic poles. On an AM sound receiver the radiation just sounds like random hiss; on an analogue TV screen it looks like random noise .

When we erect ANY aerial (not in a tunnel !) to receive TV or FM we receive this radiation - we cannot avoid it. However its importance depends on several factors and these have changed over the years with changes in the performance of receivers.

The early valve TVs were not very sensitive, as forum members will have read especially in the descriptions of pre-war TVs, so this galactic noise was hardly apparent, let alone receiver noise. The challenge was to get a strong enough signal to drive the sets so a fully contrasted picture could be seen. However as time progressed with the introduction of 'Fringe' models and then low noise transistor front ends coupled with higher gain IF stages, so receiver noise could be seen on blank channels, especially Band I channels, with the aerial disconnected. Now when the aerial was connected so there was a perceptible rise in both the visual noise level and also (for TV standards that used AM sound ) the audible hiss. This meant that the galactic noise was setting the effective limit for the viable operation of TV when the received signal from the transmitter was weak and not the receiver's internal noise performance.

A similar tale can be told for early valved FM radios; again the challenge was to get sufficient signal to give full limiting. However later (transistorised) sets, especially from the later 1960s onwards had higher gain IF stages and so could fully limit on just the receiver noise. For these sets it is not possible to tell if the aerial is plugged in or not and so it not so evident that galactic noise is still the limiting factor in fringe area reception - but it is. The introduction of FET front ends reduced the noise figures of the tuners further resulting in impressive sensitivities when connected to a signal generator's output. Nevertheless, connected to an aerial, the galactic noise exceeded the receiver generated noise and so set the limit.

Let us put some ideas and measurements in place to explore these points.

Firstly a receiver's Noise Factor, f, is the ratio of the input signal to noise to the output signal to noise. For a perfect receiver which introduced no noise the noise factor is 1. The commonly used term Noise Figure, NF, is simply 10 * log( f ), where log is log to the base 10. So a receiver with an f value of 2 has a noise figure of 3 dB. The noise factor is related to an alternative measure of noise performance, the receiver's apparent input noise temperature Tr. Tr is defined as Tr = ( f-1 ) * To, where To is a reference temperature of 290 K on the Absolute temperature scale ( 0 C = 273 K ).

Consider a receiving set-up where the aerial and the feeder cable connecting the aerial to the receiver are loss free and where there is no galactic noise to be received. Then if the aerial is delivering a signal of equivalent thermal temperature Ts, the signal to noise ratio is Ts / Tr. If now we acknowledge the presence of galactic noise of temperature Tg then the signal to noise ratio is now Ts / ( Tr + Tg ). Using the Noise factor definition we find that the effect of the galactic noise is like an extra noise factor of 1 + Tg / Tr.

A valve tuner (using a PCC89 say) has a noise figure of about 6 dB, which is a Tr of 870 K. (The BBC used NFs between 8 and 10 dB in its VHF examples in some of their training manuals).

In comparison the galactic noise Tg has the following approx values:

Min (Galactic pole) Max (Galactic centre) Average

40 Mc/s 7000 K 65,000 K 21,000 K

100 Mc/s 700 K 9,000 K 2100 K

200 Mc/s 130 K 1,700 K 390 K

Data from "Antenna Applications Reference Guide", Johnson & Jasik (eds), McGraw-Hill, 1987, Page 6-10, Fig 6-3.

A related data source is "Radio Astronomy", J.D. Kraus, McGraw-Hill, 1966, Page 237, Fig 7-1.

Since an aerial's polar diagram is unlikely just to encompass the Galactic centre or just the Galactic poles then let us consider using an average between these extremes, say a factor of 3 greater than the minimum, as representing what might be experienced in practice.

We see that at 40 Mc/s (the bottom of Band I TV ) the galactic noise completely dominates the receiver noise. It follows that having a lower noise receiver will not improve the quality of the final TV picture.

For 100 Mc/s we are in the domain of FM sound broadcasting. Even so we see that the galactic noise is a dominant factor and so using an ultra low noise front end ( a fashionable late 70s selling point) will still not improve the weak signal performance on the fringe.

At 200 Mc/s we are back to TV ( Band III TV ). Here we note that a low noise front end will be of use.

I could go on more - but there is one final point worth making: that it always helps to use a high gain aerial as the Ts increases with increasing aerial gain, whereas the Tg is effectively a constant for the aerial gains used for domestic reception.
SteveCG is offline   Reply With Quote
Old 24th Feb 2021, 5:19 pm   #11
Radio Wrangler
Moderator
 
Radio Wrangler's Avatar
 
Join Date: Mar 2012
Location: Fife, Scotland, UK.
Posts: 16,790
Default Re: Noise Figure data for VHF TV and FM front ends

One large Achtung!

Noise factor/noise figure are defined as the degradation in signal to noise ratio when a signal passes through a device, say an amplifier.

Noise figure works in log units, noise factor in linear units.

So if a signal to noise ratio of (pick a number, any number...) 34dB goes into a tuner and then we measure 32 dB on the IF output of that tuner, then the tuner has a noise factor of 2dB. Simples!

Well, not really. There is an elephant in the room. Quite a generously sized one.

You measure the signal and the noise coming out of the tuner and that's unambiguous.
You measure the signal going into the tuner, but what's this about putting noise IN? who said you had to put noise in along with the signal? And the resulting noise factor value depends on how much you chuck in.

Nasty!

There is an undocumented assumption floating around here. The definition of noise figure assumes that the signal carries with it thermal noise associated with an ideal resistor at room temperature (and they pinned the tail on the donkey or something to pick 290K for room temp)

So If you have a tuned with NF of 3dB, and your signal source is at 290K in a comfortable lab, the tuner will be contributing an equal amount of noise to the signal source. If we improved the tuner and really made it completely noiseless, the S/N ratio at teh IF could only improve 3dB

It begins to look like going for very low noise figures, below 1dB, looks pointless. Who would notice the difference?

Actually, people would. Your antenna is outside and is directional. It is looking at a distant transmitter, just above the horizon.

In most of the world, the ground is at less than 290K, and half of the beam area of your antenna pattern is peering at outer space. Yes, there is some warmth in the atmosphere, yes there are hot spots wandering around the sky, but the background is famously just a whisker cooler than 3K. So the overall noise temperature seen by your antenna is well down on 290K

If you're into satellite TV then the land surface drops out of view of your antenna, and the noise level gets a lot colder. You want a receiving set up that is quieter than the space it's viewing. Quieter than a cosy 290K is not good enough.

So, beware. Noise figure and factor have an implicit assumption of cosy room temp built into them, and it's not always a helpful thing.

So now you know why people bust guts to get noise figures significantly less than 1dB for special applications.

Serious people tend to talk not of noise figures, but of equivalent noise temperatures... they don't have the hidden assumption built-in and they fit more easily in your imagination.

David
__________________
Can't afford the volcanic island yet, but the plans for my monorail and the goons' uniforms are done
Radio Wrangler is online now   Reply With Quote
Old 25th Feb 2021, 10:54 am   #12
SteveCG
Nonode
 
Join Date: Jun 2009
Location: Herefordshire, UK.
Posts: 2,287
Default Re: Noise Figure data for VHF TV and FM front ends

Radio_Wrangler,

I agree with you about using equivalent noise temperatures. By the way, the Galactic Temperatures I quoted are the equivalent thermal temperatures, not the physical temperatures, of the electrons in the Galaxy. By this I mean the thermal temperature is that which produces an identical noise power to the synchrotron radiation at the aerial at the receiving frequency of interest.

This Galactic synchrotron radiation intensity falls off rapidly with increasing frequency, so by the time we have got upto UHF TV frequencies it is about 30 K equivalent temperature. At microwave frequencies the galactic contribution is negligible. So I agree with your analysis of what the aerial is seeing for UHF TV and higher frequencies, but for VHF the Galactic radiation contribution cannot be neglected.
SteveCG is offline   Reply With Quote
Old 26th Feb 2021, 11:47 am   #13
SteveCG
Nonode
 
Join Date: Jun 2009
Location: Herefordshire, UK.
Posts: 2,287
Default Re: Noise Figure data for VHF TV and FM front ends

This ITU publication ( ITU-R P.372-7 ) may be of interest concerning noise sources:

https://www.itu.int/dms_pubrec/itu-r...2-S!!PDF-E.pdf

See especially Page 6, Figure 3; also Page 17, Section 6 including Figure 12.

Finally Page 19 for detail on what the Sky looks like at a frequency of 408 Mc/s.

Last edited by SteveCG; 26th Feb 2021 at 11:49 am. Reason: added explicit ITU title
SteveCG is offline   Reply With Quote
Old 26th Feb 2021, 5:56 pm   #14
G6Tanuki
Dekatron
 
G6Tanuki's Avatar
 
Join Date: Apr 2012
Location: Wiltshire, UK.
Posts: 9,660
Default Re: Noise Figure data for VHF TV and FM front ends

This has been a fascinating thread: though not really that much into TV/FM [my VHF world has always been amateur- and mobile-radio].

Pentode RF amps were kinda common in first-generation mobile radios [and military VHF stuff] - the likes of the EF91 or EF95/6AK5. Low-noise-figure always seemed at-odds with 'variable-mu' AGC-controllable gain - you could have one or the other, not both.
The 'nearness' of the FM Band-II broadcasts could cause problems with intermodulation/harmonic-generation; there was one particular mobile radio that had a big problem with this - the spacing between the Home/Light/Third-program transmitters meant they could force their way through a mobile-radio whose front-end was optimised for 96-102MHz and the intermodulation caused an untunable 'sprog' precisely at the 10.7MHz IF!

Poor hams used UHF-TV valves [PC88 etc] or sometimes a balanced-neutralised-twin-triode [6J6 or 12AT7] for the RF amp and mixer on 144MHz. If you were rich you bought a 6CW4 Nuvistor and then spent ages tweaking it to get the best balance of gain and noise-figure. If you were lucky you could borrow a pulsed noise-generator...

Trransistors: The later Pye Cambridge/Westminster series were a revelation: Texas Instruments had produced the GM0290 Germanium P-N-P transistor! A couple of these, in a narrow-band front-end, followed by a GEX66 diode mixer [pumped by a simple 3rd-overtone xtal oscillator using an OC170] - and high-band VHF receiver-generated noise was no longer a thing to concern us.

The first UHF Pye Pocketfones [PF1] used a passive front-end feeding a diode mixer on 450MHz and worked just fine.

Then the Pye Westminster appeared, with 2N3819 FET RF amplifiers in the high-band VHF version. Hey, Wow! Welcome to 1968! [Low-band versions had to make-do with a bipolar transistor RF amp].

[sorry for my ramble back through the memory-registers....]
G6Tanuki is offline   Reply With Quote
Old 27th Feb 2021, 2:26 am   #15
Synchrodyne
Nonode
 
Join Date: Jan 2009
Location: Mt. Maunganui, New Zealand
Posts: 2,585
Default Re: Noise Figure data for VHF TV and FM front ends

Something to note is that in general, the articles and other sources that discussed the introduction of jfets and mosfets to VHF TV and FM front ends seldom placed lower noise as the primary reason for the move from bipolar to unipolar devices. Rather it was the reduction of spurious responses and improved ability to handle large signals that were the key issues. Not that noise was never mentioned, but implicit was that good bipolar devices in general were seen as being quiet enough, but mostly much inferior to valves when it came to spurious responses and signal handling. The use of fets took this aspect of performance back to valve level, with no noise penalty as compared with bipolar devices, and sometimes with even lower noise. The latter initially seemed to be treated as a bonus, although later probably exploited for the highest performance FM front ends.

When it comes to noise figures, Ambit quoted this parameter for most of the FM front ends included in its catalogues. The Ambit FM front ends covered a range of performance levels and used a variety of devices types. The noise figures are mostly in the range 5.5 to 8 dB, with one top-end model down at 4 dB. I have attached a composite excerpt from Catalogues 1 through 3.

Noise figures were seldom quoted for complete FM tuners though. Typically there was a single-point signal-to-noise ratio sensitivity, and sometimes quieting curves. I guess that these numbers were obtained using a signal generator, thus essentially devoid of any galactic noise input. For the early solid-state era, if comparably-based numbers could be found, an interesting comparison might be had amongst the Revox A76 (dual-gate mosfet), Sony ST-5000 (bipolar), Sony ST-5000F (jfet), B&O Beomaster 5000 early (bipolar pnp) and Beomaster 5000 late (jfet).

The noise figure was quoted for the Radford FMT-2, at 5.5 dB. This had the D&W 341AFC front end, bipolar, four-gang, with the bandpass at the input, presumably to offer maximum protection to the RF amplifier. The Radford FMT-3 used the D&W 341AFC/FET/2 front end, essentially the dual-gate mosfet version of the 341AFC, and retaining the bandpass at the input. (More common mosfet practice, as with the Revox A76, seems to have been to place the bandpass at the interstage.) Unfortunately, I can’t find the noise figure for the FMT3. The late FMT2 was optionally available with a fet-based front end, presumably the 341AFC/FET/2. The rationale for this option was given as: “The F.E.T. model is being made available primarily for foreign markets where conditions demand the highest performance in respect of cross-modulation, image rejection and adjacent channel selectivity.” Nothing was said about noise, which inclines me to think that it was not much different to that of the standard model with bipolar front end. Presumably the mosfets had less adverse effect (at higher signal levels) on the performance of their associated tuned circuits, as well as lower production of spurs.

In the communications arena, Pye did not quote noise figures for its R401 (AM) and R402 (FM) base station receivers, which had dual-gate mosfet RF amplifiers and mixers. What it said was: “Helical resonators and field effect transistors in the RF section give the receiver high selectivity with good signal-to-noise performance and outstanding protection against blocking and cross-modulation.” That suggests that low noise level was not the primary reason for the choice of mosfet devices.

Noise factors quoted by Eddystone for its 2nd and 3rd generation solid-state VHF communications receivers were:

Eddystone 1990R (VHF to 500 MHz, dual-gate mosfet RF amplifier) Typically 4 dB, not worse than 10 dB at any frequency.
Eddystone 1990S (UHF 440 to 1000 MHz, bipolar 1st RF amplifier, dual-gate mosfet 2nd RF amplifier) Typically 10 dB.
Eddystone 1995 (VHF & UHF, 20 to 1100 MHz, front-end devices unknown) Typically 10 dB.


Cheers,
Attached Files
File Type: pdf Ambit #1-3 FM Front Ends.pdf (917.1 KB, 13 views)
Synchrodyne is offline   Reply With Quote
Old 27th Feb 2021, 8:05 am   #16
Radio Wrangler
Moderator
 
Radio Wrangler's Avatar
 
Join Date: Mar 2012
Location: Fife, Scotland, UK.
Posts: 16,790
Default Re: Noise Figure data for VHF TV and FM front ends

Once you get to this level of involvement in front end design, you'll find that designers all have their own pet spreadsheets for calculating overall performance figures from figures for each individual block.

Calculating progressive gain is pretty simple.

Calculating progressive noise figure is also simple. You can use the Friis Equation recursively, or you can just keep tabs on noise power density and total noise power within the narrowest bandwidth so far in the block diagram. This latter has a problem. The style of having all your main selectivity in a block ahead of IF amplification means front end stuff is well filtered, but broadband noise from the IF will hit the demodulator. It's all too easy to have IF noise dominating the audio S/N and to not be aware of this from your calculations. You have to consider how demodulators convolve broadband noise.

So, it's starting to get a bit sticky.

But looking into gain and noise floor isn't the whole job, you also have to calculate the effects of non-linearity and this can be modelled in many ways.

I did my pet spreadsheet when I was designing the Agilent Noise Figure Analyser. This was an instrument for measuring the noise figure (or factor) of a device under test or an entire system under test. It used the classic 'Y-factor' method. This method is particularly susceptible due to errors cauased by compression. Howard Moss, the designer of the classic HP 8970 Noise Figure Meter, had some good empirically demonstrated rules of thumb for relating compression to signal level and 1dB compression point, and I included these. Because the signal I was handling was noise, and therefore had to be seen as BOTH a power density AND a total power, the issue of signal to noise ratio was hilarious.

Arriving at dynamic range figures is interesting. You can claim almost any figure you feel like, if you choose appropriate definitions for what you mean by 'dynamic range'. In reading specs, you have to pay very great attention to just what is NOT said!

So, my pet spreadsheet has a lot of stuff in it that I added to handle the needs of the NFA.
Which makes it quite different to the block diagram design functions of commercial RF design CAD packages. I find it slightly ironic that what my spreadsheet helped design measures the noise parameters of devices, and those results get used by the fearsomely expensive CAD software that *I* can't afford!

I later used it to do the Trig TT21, TT22 transponder receiver section, and that of the TY91, TY92 COMM transceivers, but those didn't need all the finesse in compression and noise versus noise ratios. The receiver designs carried through into later models.

Back to FM tuners etc., The stations in the FM band are regularly spaced, so this means that odd-order intermod products are not only in-band, but turn up precisely centred on other channels, perfect for making swirly birdie sound effects.

There is a website, the Tuner information center (note american spelling) that contains an immense amount of info on most tuner models. It's interesting, but it is a two-edges sword.

They have 'shootouts' where tuners are ranked against each other in terms of perceived audible characteristics in the opinions of various listeners. There is no management to handle the differences in listeners, and different people grade different small comparison sessions which are then assembled into an overall pecking order. The result may seem good to some people, but it's really bad science and quite likely to be misleading.

It's good info in terms of telling you what's inside each box... how many gangs, how many RF and IF stages, what sort of demodulator, what sort of IF filtering and any added cleverness.

The second failing of the site is that they haven't really handled the difference between the environments the tuners are needed to handle. In some places people need very low noise, very sensitive tuners to work on distant signals. In other places tuners are needed that can handle massive numbers of strong signals in a city stuffed with fiercely competing FM stations. You can compromise, but you can't be the best of both of these worlds at once.

Anyway, it makes gripping reading provided you have a good supply of pinches of salt and are aware of what to watch out for.

David
__________________
Can't afford the volcanic island yet, but the plans for my monorail and the goons' uniforms are done
Radio Wrangler is online now   Reply With Quote
Old 27th Feb 2021, 11:45 am   #17
SteveCG
Nonode
 
Join Date: Jun 2009
Location: Herefordshire, UK.
Posts: 2,287
Default Re: Noise Figure data for VHF TV and FM front ends

Synchrodyne - thanks for the Ambit info. It seems that the EF5804 at typically 3.5dB NF (could be as low as 2 dB) was the best in the late 70s.

I agree with both Radio Wrangler and Synchrodyne that for many applications the ultimate in S/N was not an overriding priority. It is only when you get to be 'Beyond the Fringe' for any signals that overall system Noise Figure ( which includes, Sky background, receiver Noise Figure, cable losses etc, using the Friss formulation) really comes into play.

If I can digress slightly about how the Sky looks at Radio Wavelengths compare to the Visual.

In the daytime the light from the Sun utterly dominates, yet move to radio wavelengths and the Sun really fades out of view (unless there is a solar storm/flare). What the radio astronomers found was that what was seen in the visible - stars, planets, the Moon and many galaxies - hardly could be detected. Rather, at shorter radio wavelengths, the sky was dominated by 3 compact radio sources which were named Cassiopia A, Cygnus A and Sagittarius A after the optical constellations in which they were located. Finding their exact optical counterparts was a challenge as nothing in the general direction of them looked 'obvious' - even to big optical telescopes. I mention all this as an example where extrapolating from the optical to the radio does not work.

A parting point: a lot of radio astronomy is done during daylight hours, which makes for a more normal life!
SteveCG is offline   Reply With Quote
Old 1st Mar 2021, 4:25 am   #18
Synchrodyne
Nonode
 
Join Date: Jan 2009
Location: Mt. Maunganui, New Zealand
Posts: 2,585
Default Re: Noise Figure data for VHF TV and FM front ends

A couple of papers quote 5 dB as the target noise figure for VHF TV front ends. Presumably at this level, the minimum usable picture signal point is determined by the noise (galactic, etc.,) that arrives with it, rather than the noise created by the front end.

These papers are:

“A comparison of Solid-State and Electron-Tube Devices for TV-receiver RF and IF stages” L.S. Boar and S. Reich, RCA; IEEE, 1966.

“Solutions to the Cross-Modulation Problems in Solid-State VHF-TV Tuners” Sam Weaver and Donald Wilcox, TI, IEEE 1967 July.

The latter proposed the use of jfets in VHF TV tuners. The reasons why this idea did not catch on are included in the previously mentioned Zenith paper.

Another TI reference quoted 6 dB as the target noise factor for this application. This was the book :Circuit Design for Audio, AM/FM and TV”, of 1967, and covering bipolar devices.

RCA fielded quite an array of TV front ends under a multiplicity of KRK-numbers. Just before 1968, its top-of-the-line VHF model, the KRK-140, had a nuvistor RF amplifier followed by a cascode bipolar mixer. In 1968, the KRK-142 was generally similar, except that a dual-gate mosfet replaced the nuvistor RF amplifier. So for this stage, RCA went direct from the best consumer-type valve technology to the mosfet. The following KRK-155 was similar, except that it had varactor tuning and a different UHF input arrangement. Then came the KRK-228, which had a dual-gate mosfet mixer in place of the cascode bipolar type. This conferred a 20 dB improvement in the channel 6 colour beat rejection, which was the bellwether spurious response parameter for American VHF TV tuners. Nothing was said about an improvement in noise factor, which was quoted as 4.0 dB low band and 3.5 dB high band.

The UHF tuner counterpart to the KRK-155 was the varactor-tuned KRK-160, using bipolar devices. Hitherto RCA had used the standard American form, with a single transistor oscillator and diode mixer. The KRK-160 was followed by the KRK-226, with a dual-gate mosfet RF amplifier replacing the previously used bipolar device. This was said to be about 12 dB better on cross-modulation (channel 25 on channel 19), which seemed to be the main reason for using a mosfet. It also had a lower noise factor (7 dB at channel 83) , although the difference there was not quantified.

I suspect that a key driver for the improved spurious signal rejection performances of the KRK-228 and KRK-226 was the increasing use of CATV, with relatively high signal levels and the presence of equal level signals on most/all channels. But one could say that the use of dual-gate mosfets also meant that adequately low noise and low spurious response could easily cohabit.


Cheers,
Synchrodyne is offline   Reply With Quote
Old 1st Mar 2021, 8:07 am   #19
Radio Wrangler
Moderator
 
Radio Wrangler's Avatar
 
Join Date: Mar 2012
Location: Fife, Scotland, UK.
Posts: 16,790
Default Re: Noise Figure data for VHF TV and FM front ends

The Mullard "Transistor audio and radio circuits" book also pushed bipolar circuitry for VHF/FM front ends, mostly on current consumption grounds and tried to blind people with science in terms of blocking, crossmod etc, but didn't really offer comparative data with the best examples. I think I detected marketing footprints

DAvid
__________________
Can't afford the volcanic island yet, but the plans for my monorail and the goons' uniforms are done
Radio Wrangler is online now   Reply With Quote
Old 1st Mar 2021, 8:56 am   #20
Nuvistor
Dekatron
 
Nuvistor's Avatar
 
Join Date: Aug 2013
Location: Wigan, Greater Manchester, UK.
Posts: 8,363
Default Re: Noise Figure data for VHF TV and FM front ends

I can’t contribute to this thread but thank you to the members who are posting, it’s a really interesting thread.
__________________
Frank
Nuvistor is online now   Reply With Quote
Reply

Thread Tools



All times are GMT. The time now is 9:19 am.


All information and advice on this forum is subject to the WARNING AND DISCLAIMER located at https://www.vintage-radio.net/rules.html.
Failure to heed this warning may result in death or serious injury to yourself and/or others.


Powered by vBulletin®
Copyright ©2000 - 2021, vBulletin Solutions, Inc.
Copyright ©2002 - 2021, Paul Stenning.