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 > Other Discussions > Homebrew Equipment

Notices

Homebrew Equipment A place to show, design and discuss the weird and wonderful electronic creations from the hands of individual members.

Closed Thread
 
Thread Tools
Old 21st Apr 2022, 7:32 pm   #1
Alan_G3XAQ
Pentode
 
Join Date: Apr 2018
Location: Canterbury, Kent, UK.
Posts: 189
Default How much 455KHz gain is too much?

How much 455KHz gain is too much? Or for that matter, gain at any single frequency. Many classic receivers use multiple conversion, primarily for reasons of image rejection, signal selectivity and LO drift; but another factor is it limits the gain applied at any one frequency. This probably also helps reduce the chances of self oscillation.

For a 3.5MHz amateur band receiver I am inclined towards a low gain single conversion superhet front end with no RF stage and a "strong" mixer. I have been given a high quality 500Hz wide 455KHz filter which looks great for a CW receiver. But this leads me towards needing "lots" of IF gain. A valve approach will need perhaps 10mV-100mV IF delivered to the product detector and, even worse, maybe ten volts for the AGC detector. This is looking like needing 120dB or more IF gain. Gulp. 3xEF93 running flat out. Will this approach ever lie down? I know the G3PDM receiver has 3xEF183 in a 1.6MHz IF strip, but I have heard several reports of instability.

And will I ever keep the BFO out of the early IF stages and stop it swamping the AGC? I note that the Elecraft K2 with its 4.9HMz IF goes to the extreme of mixing the IF to another frequency for the AGC amplifier stages, presumably to avoid BFO/AGC interaction.

For just CW/SSB reception I think a 2xEF93 IF at 455KHz would be enough if I adopt a hexode/heptode product detector and lots of AF gain. For AGC I'm inclined to chicken out and use audio-derived AGC because I would only need a single (pentode?) AF AGC amplifier before the detector. I know audio AGC detector charging pulse are inferior to IF-generated AGC, but at least I feel a bit more confident that I wouldn't just be building a multivalve 455KHz oscillator. I would perhaps need to AGC the product detector and maybe add a PIN diode front end attenuator to get the desired 80+dB AGC control range without completely cutting off the IF amplfiers, but that doesn't look like a big deal.

Or is 455KHz low enough for my worries to be over the top?

Thanks,

Alan
Alan_G3XAQ is offline  
Old 22nd Apr 2022, 1:36 pm   #2
G0HZU_JMR
Dekatron
 
Join Date: Sep 2010
Location: Cheltenham, Gloucestershire, UK.
Posts: 3,077
Default Re: How much 455KHz gain is too much?

My background is with semiconductors and in my case it can be useful to measure the small signal s parameters of the device at the chosen operating point (in terms of HT and bias current).

This then allows a plot of GMAX on an RF simulator and this can be used to try and establish how much power gain can be achieved (without neutralisation) whilst still achieving unconditional stability at the operating frequency.

Down at 455kHz the GMAX numbers can become quite huge even for something like a 2N3904 biased at 3mA. The GMAX can be as high as 40dB down at around 455kHz in this case. However, if a 40dB gain design was attempted the amplifier design would typically be on a knife edge in terms of preserving unconditional stability where the K factor is >1. It wouldn't take much to spoil things if the stage was detuned slightly or if the bias point changed. Things could still be dodgy even if the design was reduced to 30dB gain.

Things get much worse if you apply AGC to the stage because the operating point will change and this will impact the stability margin. I'd expect the same issues to affect germanium transistors and also valves if they were measured on a VNA to extract s-parameters..
__________________
Regards, Jeremy G0HZU
G0HZU_JMR is offline  
Old 22nd Apr 2022, 6:38 pm   #3
Alan_G3XAQ
Pentode
 
Join Date: Apr 2018
Location: Canterbury, Kent, UK.
Posts: 189
Default Re: How much 455KHz gain is too much?

Hello Jeremy. I have a NanoVNA but like most of these instruments it is designed for 50 ohm environments rather than the high impedances typical of valve circuits. IF strips coupled with the usual IF transformers don't seem amenable to low impedance probing as far as I can see. Or maybe I lack imagination?

I just got an old copy of Radio Receiver Design by Sturley arrive in the post today. Therein Table 5.1 on p180 suggests amplifications of at least 175 at 465KHz are feasible for a 3mA/V pentode so perhaps my initial pessimism is unfounded.

It then comes down to shielding between successive 465KHz IF stages and a continuing unease about the BFO signal escaping into the front of the amplifier chain. Does anyone have practical experience of these issues?

Alan
Alan_G3XAQ is offline  
Old 22nd Apr 2022, 6:39 pm   #4
G6Tanuki
Dekatron
 
G6Tanuki's Avatar
 
Join Date: Apr 2012
Location: Wiltshire, UK.
Posts: 13,951
Default Re: How much 455KHz gain is too much?

IF gain needs to be adequate.

How will this post-filter gain be realised? It's all very well having the world's=-best-filter but if you follow it by a chain of high-gain *wideband* amplifiers these will add noise as well as gain - meaning that [unless you're doing something-modern in the detector] the added noise-sidebands will beat together with the wanted signal in the detector and cause a raised background-noise level.

Some 1970s/80s/90s radios were particularly bad at this [Hello Icom - I'm looking at you!]

The answer is a 'tail end filter' between the last IF amp and the detector; this does not need to be as tight as the front-end filter...

In a 'classic' valve IF strip a series of tuned-circuit couplings did much the same. Though you really need to design the first-if-after-the-ftont-end-filter as if it was a front-end RF amp [low noise, resistance to intermod].

Also remember that your nice front-end filter needs to be terminated properly at both the signal and LO frequencies! You may have brilliant attenuation a few KHz either side of the filter's frequency but when you move out a few hundred KHz the attenuation can be a lot less... so signal-frequency and LO energy can be presented to the first IF amp in unhelpful ways.

[Personally I never 'got' the crude brute-force on-off-keying of digital data - which is basically what CW is! There's no continuous carrier so how can you recover a clock or control gain? Give me nice frequency- or phase-shift modulation any day]
__________________
I'm the Operator of my Pocket Calculator. -Kraftwerk.
G6Tanuki is offline  
Old 22nd Apr 2022, 7:09 pm   #5
Alan_G3XAQ
Pentode
 
Join Date: Apr 2018
Location: Canterbury, Kent, UK.
Posts: 189
Default Re: How much 455KHz gain is too much?

Thanks for your thoughtful comments, Tanuki

I'm thinking of conventional ex-broadcast-set IF transformers to couple the IF amplifier stages together. Yes, a proper tail-end filter would be nice but I don't have one so I must accept a 3dB degradation in S/N from noise in the 500Hz spectrum the other side of zero beat. At least total noise in an amplifier bandwidth of just a few KHz is unlikely to overload the product detector.

I don't see why the first IF stage needs to be particularly resistant to intermod in my paper design so long as it comes after the 500Hz filter and so long as the mixer has a bit of gain so as to swamp IF noise and filter loss.

From your Rolls Royce analogy, if you were a CW fan and so perhaps tempted to tread this path, would you go for three IF stages to be sure of enough gain and maybe add resistive loading to each stage to get "just enough" output?

Would you grit your teeth and go for full house IF-derived AGC needing maybe 10V of 465KHz into the detector? At least I suppose it would allow the signal to the product detector to be attenuated maybe 100:1 which would mitigate the BFO leakage problem to no worse than the 2-stage barely-enough IF approach. I confess that audio-derived AGC has proved good enough for me in the past because I tend to ride the RF/IF gain control and only rely on AGC as an ear defender.
Alan_G3XAQ is offline  
Old 22nd Apr 2022, 7:44 pm   #6
ms660
Dekatron
 
ms660's Avatar
 
Join Date: Apr 2011
Location: Cornwall, UK.
Posts: 13,454
Default Re: How much 455KHz gain is too much?

Quote:
Originally Posted by Alan_G3XAQ View Post
It then comes down to shielding between successive 465KHz IF stages and a continuing unease about the BFO signal escaping into the front of the amplifier chain.
One of the most famous WW2 receivers (the R1155) ran the BFO at half the IF frequency, the 2nd harmonic was picked off and used for the BFO injection to the detector, this also prevented the IF signal from locking the BFO.

Might be worth fiddling with.

Lawrence.
ms660 is offline  
Old 22nd Apr 2022, 8:35 pm   #7
commie1
Triode
 
Join Date: Apr 2018
Location: Buckley, Clwyd, Wales, UK.
Posts: 32
Default Re: How much 455KHz gain is too much?

Quote:
Originally Posted by Alan_G3XAQ View Post
For a 3.5MHz amateur band receiver I am inclined towards a low gain single conversion superhet front end with no RF stage and a "strong" mixer.
I'd be interested in how you intend to limit the image frequency? This is precisely why the original HRO uses 3 tuned tanks and 2 rf stages in its front end, also the mixer(multiplier) will be protected from mixer overload because the front end limits the input signal under a.g.c/rf control.

If I were you, I'd start looking at the HRO schematic. Some would argue that 'its complicated' but I'm afraid, that is what works.

Darren 2w0epi
commie1 is offline  
Old 22nd Apr 2022, 9:07 pm   #8
G0HZU_JMR
Dekatron
 
Join Date: Sep 2010
Location: Cheltenham, Gloucestershire, UK.
Posts: 3,077
Default Re: How much 455KHz gain is too much?

Quote:
Hello Jeremy. I have a NanoVNA but like most of these instruments it is designed for 50 ohm environments rather than the high impedances typical of valve circuits. IF strips coupled with the usual IF transformers don't seem amenable to low impedance probing as far as I can see. Or maybe I lack imagination?
It's definitely going to be much more challenging to use a VNA with valves. The nearest I have got to this is to measure two port s-parameters for dual gate MOSFETs such as the BF981 and BF988 with a VNA. Because of the really high impedances below about 20MHz it is only realistic to measure down to about 20MHz.

However, where the K factor is less than 1 (usually at VHF and below in this case) the GMAX gain (the green curves) will increase 10dB per decade as you go down in frequency. See below for a BF988 measured on my VNA and there is also the manufacturer's data overlaid. The agreement for the green GMAX curves is quite close up at VHF so I think I made a good measurement here. The manufacturer didn't bother to try to measure below 50MHz probably for the reasons I gave above. You can see that I tried to measure below 50MHz with fairly good results down to about 20MHz.

You can see that GMAX can be extrapolated to be about 54dB at around 1MHz. My GMAX curve flattens off below 10MHz and this is caused by the limits of my VNA. Usually the GMAX of a transistor falls at 10dB/decade if K<1 and it falls at 20dB/decade if K>1.

I don't think this could be done with a nanovna but some of the modern nanovna versions are probably not far off. This really is pushing beyond the limits of pretty much any 50R VNA but useful things can be learned at lower frequencies through extrapolation.
Attached Thumbnails
Click image for larger version

Name:	BF998 GMAX.jpg
Views:	44
Size:	92.0 KB
ID:	255873  
__________________
Regards, Jeremy G0HZU
G0HZU_JMR is offline  
Old 22nd Apr 2022, 10:13 pm   #9
Alan_G3XAQ
Pentode
 
Join Date: Apr 2018
Location: Canterbury, Kent, UK.
Posts: 189
Default Re: How much 455KHz gain is too much?

Quote:
Originally Posted by commie1 View Post
I'd be interested in how you intend to limit the image frequency? This is precisely why the original HRO uses 3 tuned tanks and 2 rf stages in its front end, also the mixer(multiplier) will be protected from mixer overload because the front end limits the input signal under a.g.c/rf control.
The key is my limiting the coverage to 3.5 to 3.55 MHz or so. A 3 pole fixed tuned bandpass preselector designed with Zverev's book (David) or a software implementation of the k's and q's (the rest of us) can give much the same image performance as the HRO. I'll not be using a noisy 6C6 mixer so I don't need the RF amplifiers that were embedded in the HRO preselector to achieve an acceptable overall noise figure.

A "strong" mixer such as push-pull triodes or a beam deflection valve should accept everything in the 50KHz preselector passband without significant distortion. As I mentioned, an AGC controlled PIN diode attenuator ahead of the mixer might be a suitable bandaid if signal overload proves to be an issue although it won't help if signals outside the IF filter are causing IMD. Maybe a front panel switched attenuator would be necessary to cover that eventuality.

Well, that's the theory anyway.
Alan_G3XAQ is offline  
Old 22nd Apr 2022, 10:20 pm   #10
Radio Wrangler
Moderator
 
Radio Wrangler's Avatar
 
Join Date: Mar 2012
Location: Fife, Scotland, UK.
Posts: 22,799
Default Re: How much 455KHz gain is too much?

You can do these sorts of measurements at spot frequencies by using matching networks, but at lower and lower frequencies the impedance transformation ratios get higher and higher and this forces the matchers to be higher and higher Q.... so lossier. Essentially you need to produce a pair and try them back-to-back to deduce the effect of their strays and losses.

Theoretical gain goes through the roof for this class of device at lower frequencies, but to access that gain you need high ratio step-up transformers (usually resonant networks) and as the Q has to get very high, so does the criticality of tuning. So the theoretical high gain is provably possible, but you make more trouble if you try to access it.

One trouble I used to have with a GHz part that had Zin of essentially 2pF with a very small real part (AH312) was that at low (for it) frequencies, if you tried to create a matching network, the real part presented to the device was insufficient to keep it stable. So yes it was excellent at the target frequency, but in doing so, you failed to keep it stable at much lower frequencies. You just can't win.

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  
Old 23rd Apr 2022, 9:42 pm   #11
commie1
Triode
 
Join Date: Apr 2018
Location: Buckley, Clwyd, Wales, UK.
Posts: 32
Default Re: How much 455KHz gain is too much?

Quote:
Originally Posted by Alan_G3XAQ View Post
I'll not be using a noisy 6C6 mixer so I don't need the RF amplifiers that were embedded in the HRO preselector to achieve an acceptable overall noise figure.
Let me just check, you think the HRO front end is present to reduce noise figure generated from the 6c6 valve mixer? I have already said, the HRO front end is a high Q 3 tank circuit used for both amplification/attenuation and provides excellent image rejection up to about 20m, it provides at least -26dB image rejection on 10m. Further, the HRO receiver is one of the quietest receivers ever designed.
commie1 is offline  
Old 24th Apr 2022, 1:43 am   #12
G0HZU_JMR
Dekatron
 
Join Date: Sep 2010
Location: Cheltenham, Gloucestershire, UK.
Posts: 3,077
Default Re: How much 455KHz gain is too much?

I had a quick look at the EF93/6BA6 datasheet and this looks to be a pentode valve. My valve experience pretty much tops out at the triode but I assume one advantage of the pentode is the low anode to grid capacitance and this makes it easier to get good gain with good stability.

See below for a very crude VCCS model of a U310 JFET based on the typical gm and inter electrode capacitances plus some leg inductance. This is compared the the GMAX of a real U310 measured on a VNA. The agreement is quite close and it suggests just over 30dB might be possible at 455kHz.

With this good agreement in mind, I wonder if it's possible to model the pentode in a similar way across the SW bands? I tried loading up a VCCS model of the EF93 based on the inter electrode capacitances in the EF93 datasheet and a gm of -4.5mA/V.

This valve has a very low anode to grid capacitance and this boosts the GMAX to over 50dB at 455kHz with this model. To model it I've simplified the device to a triode with the same tiny feedback capacitance of a few fF. I'd imagine this valve would benefit from a special base that aims to preserve the low feedback capacitance.

I really am out of my depth here but if it is possible to model it like this then the GMAX for the EF93 does appear to be quite high at 455kHz. I also tried modelling a 12AU7 triode and this was much worse. The feedback capacitance limits the GMAX quite badly.
Attached Thumbnails
Click image for larger version

Name:	U310_GMAX.gif
Views:	37
Size:	47.0 KB
ID:	255960  
__________________
Regards, Jeremy G0HZU
G0HZU_JMR is offline  
Old 24th Apr 2022, 8:39 am   #13
Radio Wrangler
Moderator
 
Radio Wrangler's Avatar
 
Join Date: Mar 2012
Location: Fife, Scotland, UK.
Posts: 22,799
Default Re: How much 455KHz gain is too much?

In a tetrode or a pentode, some of the electrons from the cathode wind up on the screen grid, the rest make it through to the anode. The separation into anode and screen grid currents is quantised into whole electrons and has a degree of randomness. In this way tetrodes and pentodes suffer from 'partition noise' and are significantly noisier than triodes.

Triodes, on the other hand have significant feedback capacitance and to be able to use them for much gain at RF and still have them stable requires 'neutralisation' where a bridge arrangement is used to cancel Cag. These are normally not very wide band arrangements and so a bit of a problem for an HF receiver. The other escape is to use the triode in a grounded grid circuit.

The HRO and AR88 have multiple RF stage valves to serve as buffers between RF selectivity sections and to provide high impedances to allow those sections to be run at high Q. The gain of those valves is mostly cancelled by the losses of running resonators set for narrow bandwidths.

If you go for 455kHz IF, then the image is only 910kHz away from your wanted frequency, and you want lots of image rejection...

HRO and AR88 have plenty low enough noise floors across most of their coverage, but do show limitations above the 20m band. The AR88 sported ultra low loss coil formers (polystyrene!) for its upper most band.

Gain is significantly cheaper than selectivity.

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  
Old 24th Apr 2022, 11:00 am   #14
Alan_G3XAQ
Pentode
 
Join Date: Apr 2018
Location: Canterbury, Kent, UK.
Posts: 189
Default Re: How much 455KHz gain is too much?

Thanks Jeremy. So it looks as if 40dB gain at 455KHz from an EF93 is not too unreasonable.

Overall I'm inclined towards a low loss fixed tuned "minloss" preselector and a strong mixer with just a few dB gain feeding my Inrad 455KHz CW filter backed by three EF93 IF stages, each with AGC control.

From discussions with G3RZP it's not clear whether delivering 10v or more to the AGC detector is just a hang over from AM receivers or whether the large attenuation into a hexode/heptode product detector is needed to avoid BFO leakage swamping the IF but that approach has been widely adopted in classic receivers so I'm thinking of avoiding the low IF gain, high AF gain, audio-derived AGC approach. It might be less likely to be unstable but would need additional measures to get adequate AGC range and the audio AGC would never be as good as IF-derived AGC.

Thanks everyone for your input.
Alan_G3XAQ is offline  
Old 24th Apr 2022, 7:10 pm   #15
G0HZU_JMR
Dekatron
 
Join Date: Sep 2010
Location: Cheltenham, Gloucestershire, UK.
Posts: 3,077
Default Re: How much 455KHz gain is too much?

Hi Alan I hope the info is useful but I'm not sure how valid my first attempt at the model of the pentode will be though. It's very much a 'page 1 of the logbook' model and not to be trusted really.

Note that the EF93 model was in common cathode and the U310 GMAX plots were in common source. Usually the U310 or J310 is used in common gate when used as an RF amplifier and the GMAX plot will be quite different in grounded gate.

I'm not sure how well my VNA copes when generating a GMAX curve for a JFET in common gate but usually it produces a flat GMAX figure of about 20dB for the U310 or J310. This then rolls off like a low pass filter from just under 100MHz and the GMAX dips to about 14dB by 433MHz. In reality, the achievable gain will be a bit less than this due to passive losses in the matching networks.

In common gate the power gain will be a function of the drain load resistance including the internal resistance of the JFET. I think this is why it produces a fairly flat GMAX response through LF and up into VHF as the JFET will often be unconditionally stable if the gate connection is very short (SMD package) and so it will be the internal drain resistance that limits the GMAX figure. I'm not sure how well I can measure this, especially for a valve in grounded grid. I'd expect the internal anode resistance to be very high.
__________________
Regards, Jeremy G0HZU
G0HZU_JMR is offline  
Old 24th Apr 2022, 9:45 pm   #16
G0HZU_JMR
Dekatron
 
Join Date: Sep 2010
Location: Cheltenham, Gloucestershire, UK.
Posts: 3,077
Default Re: How much 455KHz gain is too much?

Interesting stuff from David about the partition noise in tetrodes and pentodes. Choosing the most suitable valve for each stage of a receiver must be quite a challenge!
__________________
Regards, Jeremy G0HZU
G0HZU_JMR is offline  
Old 24th Apr 2022, 11:36 pm   #17
G0HZU_JMR
Dekatron
 
Join Date: Sep 2010
Location: Cheltenham, Gloucestershire, UK.
Posts: 3,077
Default Re: How much 455KHz gain is too much?

There's some good info about vintage SSB receivers in the Collins Book "Fundamentals of Single Side Band".

http://www.collinsradio.org/cca-coll...gle-side-band/

I guess many people will have already seen this book but it's the first time I've seen it. The full book is a 49Mb download in the link above.
__________________
Regards, Jeremy G0HZU
G0HZU_JMR is offline  
Old 25th Apr 2022, 1:19 am   #18
Radio Wrangler
Moderator
 
Radio Wrangler's Avatar
 
Join Date: Mar 2012
Location: Fife, Scotland, UK.
Posts: 22,799
Default Re: How much 455KHz gain is too much?

Look for books by Pappenfus Sabine and Schoenike These were Collins books often with a specialist per chapter.

Bill Sabine was a nice chap and contributed to the old rec,radio.amateur.misc and .tech newsgroups from a public library's internet facility. He knew Hugh Walker (of 'sources of intermodulation distortion in diode ring mixers' fame) fairly well from correspondence between them. Hugh used to be my boss.

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  
Old 25th Apr 2022, 10:42 am   #19
Alan_G3XAQ
Pentode
 
Join Date: Apr 2018
Location: Canterbury, Kent, UK.
Posts: 189
Default Re: How much 455KHz gain is too much?

Quote:
Originally Posted by G0HZU_JMR View Post
There's some good info about vintage SSB receivers in the Collins Book "Fundamentals of Single Side Band".

http://www.collinsradio.org/cca-coll...gle-side-band/
I was unaware of this digital copy, which is available with permission from the copyright holders (go up a level in the URL for details). This is great. The other very desirable Collins book is Single Sideband Principles and Circuits but this one seems to still be under full copyright protection and even scruffy second hand copies are pricey.
Alan_G3XAQ is offline  
Old 25th Apr 2022, 2:14 pm   #20
Radio Wrangler
Moderator
 
Radio Wrangler's Avatar
 
Join Date: Mar 2012
Location: Fife, Scotland, UK.
Posts: 22,799
Default Re: How much 455KHz gain is too much?

Quote:
Originally Posted by G0HZU_JMR View Post
Interesting stuff from David about the partition noise in tetrodes and pentodes. Choosing the most suitable valve for each stage of a receiver must be quite a challenge!
Electrons are literally boiled off from the surface of the cathode, so there is a thermal noise component, then there is the resistivity of the cathode coating, its temperature and Boltzman's constant. Then there is partition noise at G2, finally the electrons arrive like hail on a tin roof at the anode, making shot noise.

Sometimes, transistors ARE the answer...

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  
Closed Thread

Thread Tools



All times are GMT +1. The time now is 5:06 pm.


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 - 2024, vBulletin Solutions, Inc.
Copyright ©2002 - 2023, Paul Stenning.