Radio receiver: RF Gain vs IF Gain vs IF Filter stages vs AF Gain?

[ELS122]

I assume you'd want enough audio gain so that the background noise of the RF+IF+detector stages could be turned up to "listenable levels" to ensure that even if the signal is barely over the noise floor, you wouldn't lack audio gain.
Then I assume that you'd want more IF stages so keep stations from blending together... But if that isn't a problem, for example with a directional antenna and not a lot of stations around, that wouldn't matter right? But how detrimental would be the random noise across the spectrum with few IF stages with not a lot of Q?

Now, what about RF gain vs IF gain? why not have all the RF gain, then mostly passive IF stages, since the sooner the gain source is, the less noise you're amplifying right?
Even considering since the RF stage wouldnt do much signal selecting/filtering, would it matter at all since the IF stages still do the filtering?
Or would high RF gain cause intermodulation because of non-linearities in the RF amplifier itself?

Also, why not do all the voltage amplification in the IF filter stages? - Have high transformer ratios to amplify the signal and then have high input impedance current amplifiers to drive the next IF stage. Wouldn't this have less distortion and noise than relying on active gain stages to get IF gain up?

[T3sl4co1l]

Think about dynamic range.

Putting all the gain in front (RF), now you get compression and mixing (IMD) even on moderately-distant stations.  It's utterly useless in the city.

Typically, just enough RF gain is used to compensate for mixer loss and such.  Enough to dominate the receiver's SNR, without reducing DR much.

DR may be a reason to keep the RF amp running pretty hot (i.e., high bias current) -- more bias current means larger signals are a smaller fraction of its operating point and thus less distortion (less compression, less IMD).  Assuming it's impedance-matched for maximum power output, of course.

Very high DR receivers might eschew the RF preamp entirely, going directly into the mixer -- a high-level one like a JFET double-balanced, or analog switch, or diode single-balanced with very strong LO, or even more archaic means (I don't think you could actually justify using vacuum tubes these days, for any level of NF or DR, but their large-signal behavior is at least one redeeming quality of them; if not their NF ).  Spectrum analyzers are often this way, sacrificing noise floor for SFDR; if you need noise figure instead, you can put your own LNA in front, so it's the best of both worlds (assuming you're so equipped, and don't mind switching things around as needed).

IF gain typically implies use of AGC.  This is convenient, as selecting out interfering stations means less chance for them to blow out your desired signal, as gain rises along the chain.

The RF amp can be (and often is) part of the AGC loop, blurring this distinction a bit.

You can indeed have quite a weak signal after the IF, and toss the remaining gain at AF, but this has some difficulties itself: there may be additional noise at AF (1/f flicker noise, hum/ripple, etc.), and it also assumes the detector works at all -- which, a square-law / diode detector needs at least some ~mV to do a reasonable job with, for example.  Still, you could have something like a synchronous (direct conversion) receiver, where the AF output is quite small, and just on account of raw gain, or ADC bits (per sample, or after much averaging/filtering), reasonable fidelity can be obtained.

IF gain is generally preferred, because simple VGAs (variable gain amps) have a lot of distortion.  Pushing that distortion to harmonics and filtering it out (as long as IMD doesn't get folded back into the passband) greatly simplifies things.  For example, using the exponential transfer function of BJT amplifier (which also gives V(AGC) ~ ln(input amplitude)).

So, you tend to arrive at something like the time-tested superhetrodyne receiver; or a couple quirks added for digital receivers, give or take when a lot of compute power is handy (SDR).

Tim

[ELS122]

Think about dynamic range.

Putting all the gain in front (RF), now you get compression and mixing (IMD) even on moderately-distant stations.  It's utterly useless in the city.

Typically, just enough RF gain is used to compensate for mixer loss and such.  Enough to dominate the receiver's SNR, without reducing DR much.

DR may be a reason to keep the RF amp running pretty hot (i.e., high bias current) -- more bias current means larger signals are a smaller fraction of its operating point and thus less distortion (less compression, less IMD).  Assuming it's impedance-matched for maximum power output, of course.

Very high DR receivers might eschew the RF preamp entirely, going directly into the mixer -- a high-level one like a JFET double-balanced, or analog switch, or diode single-balanced with very strong LO, or even more archaic means (I don't think you could actually justify using vacuum tubes these days, for any level of NF or DR, but their large-signal behavior is at least one redeeming quality of them; if not their NF ).  Spectrum analyzers are often this way, sacrificing noise floor for SFDR; if you need noise figure instead, you can put your own LNA in front, so it's the best of both worlds (assuming you're so equipped, and don't mind switching things around as needed).

IF gain typically implies use of AGC.  This is convenient, as selecting out interfering stations means less chance for them to blow out your desired signal, as gain rises along the chain.

The RF amp can be (and often is) part of the AGC loop, blurring this distinction a bit.

You can indeed have quite a weak signal after the IF, and toss the remaining gain at AF, but this has some difficulties itself: there may be additional noise at AF (1/f flicker noise, hum/ripple, etc.), and it also assumes the detector works at all -- which, a square-law / diode detector needs at least some ~mV to do a reasonable job with, for example.  Still, you could have something like a synchronous (direct conversion) receiver, where the AF output is quite small, and just on account of raw gain, or ADC bits (per sample, or after much averaging/filtering), reasonable fidelity can be obtained.

IF gain is generally preferred, because simple VGAs (variable gain amps) have a lot of distortion.  Pushing that distortion to harmonics and filtering it out (as long as IMD doesn't get folded back into the passband) greatly simplifies things.  For example, using the exponential transfer function of BJT amplifier (which also gives V(AGC) ~ ln(input amplitude)).

So, you tend to arrive at something like the time-tested superhetrodyne receiver; or a couple quirks added for digital receivers, give or take when a lot of compute power is handy (SDR).

Tim

What about using the IF transformer to amplify the IF signal. instead driving it with a current buffer. Same with RF front end. Wouldn't that have less IMD and noise than using active gain stages?
Although it would be hard to implement AGC.

[T3sl4co1l]

Transformers (or other matching networks) are a given, and they don't have (power) gain.

Tim

[ELS122]

Transformers (or other matching networks) are a given, and they don't have (power) gain.

Tim

Yeah but I mean, using current amplifiers... like a buffer. emitter follower or some other buffer to drive the transformer, then wind the secondary with far more turns to have higher output voltage.
This way you don't get the non-linearities of using something like a common emitter gain stage to get the voltage gain.
There's still power gain with a buffer. but unity gain buffers usually are more linear than common emitter gain stages. While afaik the IF transformer should be very linear in comparison.

[T3sl4co1l]

You should build more amplifiers (of more types).  I think you'll be (un?)pleasantly surprised to find out that amplifier configurations don't actually make much difference to the relevant parameters, and have many more downsides (like poor isolation and instability).  CE is most common for high power gain, and CB for isolation; and paired variations, like cascode or diff pair.  CC is uncommon because in the decreasing-hFE regime towards cutoff, the input impedance is a phase rotation versus the emitter (load) impedance; a capacitive load causes negative input resistance(!).  Cbe is relatively large, making isolation quite poor -- contrary to the DC case where it makes a fine buffer.

This is formalized with scattering parameters, and theorems like maximum stable gain, or gain-bandwidth (hrm, is that a theorem, or more an analysis? not sure).  Which also explains why tricks like neutralization works, which don't make any sense at LF.

And, all of them together, is roughly speaking why neutralization often takes the form of dumping some output power / efficiency / gain; a resistive "neutralization" can be expensive in terms of those factors, but it can be that bandwidth or stability (or linearity to some extent) is desired more than gain or gain-bandwidth.

Tim

[boB]

Is this receiver for FM or  AM and other linear type signals ?

FM needs LOTS of gain in the IF chain to clip the signal and reduce amplitude noise.

Otherwise, for AM or more linear modes, dynamic range is important.

Things to consider anyway.

boB

[radiolistener]

transformer doesn't have power gain. Its power on output is the same, as on input, with some little loss.
If you gain voltage with transformer, it leads to reduced current. And if you gain current, it leads to reduced voltage. So the power remains the same.

[David Hess]

In practical designs the RF amplifier in front of the first mixer is required to contain leakage of the local oscillator to meet FCC requirements, whether its gain is required or not.

So why not place most of the gain at AF after all of the selectivity?  If there is too much gain at one frequency, then it becomes very difficult to isolate the output from the input of that stage, and oscillation becomes certain.  Gain is distributed at different frequencies to lower isolation requirements.

So the gain of each stage needs to be high enough to overcome the noise and loss of the next stage, and the total gain distributed between all stages needs to be enough for the detector or digitizer.

[ELS122]

Current buffer make more current, transformer ratio then make more voltage out of that amplified current.
More voltage = more pleasant sounds. get it now?

[jonpaul]

read the radio engineers handbook and ARRL re superhetrodyne RX design.

j

[ELS122]

read the radio engineers handbook and ARRL re superhetrodyne RX design.

j

That's like saying, read a dictionary to understand why a word like "ufoafjoa" doesn't exist.
That book talks about techniques that are used. I'm asking why the one in question isn't used.

[T3sl4co1l]

That's like saying, read a dictionary to understand why a word like "ufoafjoa" doesn't exist.
That book talks about techniques that are used. I'm asking why the one in question isn't used.

Take an upper-level radio systems course (or several of them, or the whole curriculum leading up to that if also necessary).  Or build the circuits and test them yourself.  Preferably both.

We can lead you to keywords and theory that explain what you seek.  But we cannot possibly write out the books that fully explore the field, nor can we pack it into your brain in a way that is meaningful.  Only you can do that.

Tim

[radiolistener]

More voltage = more pleasant sounds. get it now?

No, it doesn't works that way.

If you want higher sound volume, it needs higher power. The transformer cannot increase power. You're needs active circuit for that - power amplifier.

[Solder_Junkie]

You need to be careful with the distribution of gain, generally you want most of the gain in the IF and little, if any, ahead of the IF. One of the most successful receiver modules of recent times is the Hycas IF board by W7ZOI and WA7MLH. The board was covered in the Dec 2007 edition of QST magazine. The circuit and brief details are at this page: http://w7zoi.net/hycas-pcb.html

The gain of the Hycas IF board is 60 dB. A typical HF receiver using one of these boards is described in this web page:
https://www.qsl.net/g4aon/g4aon_rx/

SJ

[ELS122]

More voltage = more pleasant sounds. get it now?

No, it doesn't works that way.

If you want higher sound volume, it needs higher power. The transformer cannot increase power. You're needs active circuit for that - power amplifier.

For the 5th time maybe? a current amplifier increases power. You get the voltage gain from the transformer, get the power gain from the IF drive stages.
Instead of getting the voltage gain from the drive stages, you get the voltage gain in the IF transformer, with a higher ratio.

If you use a BJT with an HFE of 100 as an emitter follower, the current gain will be 100, the voltage gain will be 1 - 1/100.
The resulting power gain will be 99
Which is slightly more than 1

[CaptDon]

Voltage gain may be a factor if the following stage is very high impedance input such as the gate of a FET or the grid of a tube. The limit of how much the input to output ratio of the I.F. transformer has to do a lot with being able to maintain resonance at the I.F. frequency. Yes, you put a large number of turns on the secondary of the I.F. transformer giving an expected high voltage AND high impedance, but then the capacitive loading of the next stage begins to look more like a short or low pass filter and the entire I.F. transformer becomes a huge mis-match and can't be resonated. Voltage gain is easily obtained by the Q at resonance. Also, if the next stage is low impedance then you want current gain and probably a 1:1 I.F. transformer or even a 2:1 I.F. transformer which increases the current drive to the base of the next transistor while sacrificing voltage gain. I am sure my answer is not satisfactory to you as you have pre-conceived notions and only accept answers that agree with your summation of the design in question. The M.I.T. RadLab series of Radio and Radar circuit design explains all of the tradeoffs including mathematical examples and verbal sumations. May as well ask us why don't cars have square wheels and super suffisticated vibration damping? Of course it can be done, several reasons why it isn't.

[ELS122]

Voltage gain may be a factor if the following stage is very high impedance input such as the gate of a FET or the grid of a tube. The limit of how much the input to output ratio of the I.F. transformer has to do a lot with being able to maintain resonance at the I.F. frequency. Yes, you put a large number of turns on the secondary of the I.F. transformer giving an expected high voltage AND high impedance, but then the capacitive loading of the next stage begins to look more like a short or low pass filter and the entire I.F. transformer becomes a huge mis-match and can't be resonated. Voltage gain is easily obtained by the Q at resonance. Also, if the next stage is low impedance then you want current gain and probably a 1:1 I.F. transformer or even a 2:1 I.F. transformer which increases the current drive to the base of the next transistor while sacrificing voltage gain. I am sure my answer is not satisfactory to you as you have pre-conceived notions and only accept answers that agree with your summation of the design in question. The M.I.T. RadLab series of Radio and Radar circuit design explains all of the tradeoffs including mathematical examples and verbal sumations. May as well ask us why don't cars have square wheels and super suffisticated vibration damping? Of course it can be done, several reasons why it isn't.

Thank you for mentioning a book that apparently actually discusses things of the sort.

I repeated my question because the responses were ignorant of the fact that current amplifiers amplify power. and only kept repeating that there is no power gain, which there obviously is.
Besides, not every detector circuit requires power.

[T3sl4co1l]

You seem to be stuck on, or have missed the underlying point, of this:

amplifier configurations don't actually make much difference to the relevant parameters, and have many more downsides (like poor isolation and instability).

Perhaps you found it too brief to be worth consideration, an introductory or throw-away statement; perhaps you lack the understanding to appreciate the fact.  I don't know.  (And you aren't telling us what you lack to make the connection, leaving us to guess at best.)

A current amplifier + transformer has more downsides than a voltage amplifier -- which may still have current gain anyway, and which may not require a transformer at all.

CE amplifiers have more freedom of compensation/neutralization, feedback / distortion reduction / gain reduction, and impedance matching, so we tend to use them most often.

Sure, let's talk h_FE.  A BJT might have h_FE of 100 at DC.  It might also have h_fe of 100 at say 10kHz, but it's dropping by 1MHz (-3dB or so, maybe?) and runs out at f_T (h_fe = 1), say 100MHz.  It's a dominant pole cutoff (more or less due to C_cb * R_bb' predominantly I think, but also physical limits), and independent of circuit limitations which may cut lower (like C_cb * R_L setting output GBW).  An emitter follower is a terrible (general purpose) choice at 10kHz already because who even makes (let alone budgets for) transformers at that frequency, but it's even worse at 10MHz where gain is dropping sharply.  (Adjust numbers appropriately for any given part in question.)

Riddle: how do you get power gain from a transistor above fT?  By your premise, this is impossible!  In fact it's not (but it isn't terribly useful, the efficiency or bandwidth is poor).

You imagine that the difference might be distortion, or noise; it is not, the device has the same fundamental input-referred noise limits regardless of topology.  You get say 1nV/rtHz of noise in series with the base, and it doesn't make the tiniest bit of difference whether the base-emitter is wired to the left or right side of the input, it's always and necessarily in series regardless!

Put still another way: the noise figure of a device is fundamental to the device, regardless of arrangement or topology, depending only on the bias and impedance matching (and frequency range, since noise generally doesn't have flat frequency statistics).  You put in a low-noise transistor, you bias it up as needed, match it to the source and load, and that's what you get.  Sure you might get lower nV/rtHz if you run it hotter (higher Iq), but you get higher pA/rtHz (current noise) too: it just means a lower source impedance for noise match.  Meanwhile, the noise figure hasn't changed much.

RF design is all about eliminating confounding specifics, like the voltage or current in a circuit, and considering the flow of power back and forth through a network.  In general, we don't care about impedance, because impedance can be matched with simple RLC networks, transformers at the worst -- and if impedance isn't matched to its ideal value (give or take acceptable tolerance), it ain't our components to blame!  This is why we have dimensionless figures like gain and noise factor, s-parameters and etc.

Tim

[ELS122]

You seem to be stuck on, or have missed the underlying point, of this:

amplifier configurations don't actually make much difference to the relevant parameters, and have many more downsides (like poor isolation and instability).

Perhaps you found it too brief to be worth consideration, an introductory or throw-away statement; perhaps you lack the understanding to appreciate the fact.  I don't know.  (And you aren't telling us what you lack to make the connection, leaving us to guess at best.)

A current amplifier + transformer has more downsides than a voltage amplifier -- which may still have current gain anyway, and which may not require a transformer at all.

CE amplifiers have more freedom of compensation/neutralization, feedback / distortion reduction / gain reduction, and impedance matching, so we tend to use them most often.

Sure, let's talk h_FE.  A BJT might have h_FE of 100 at DC.  It might also have h_fe of 100 at say 10kHz, but it's dropping by 1MHz (-3dB or so, maybe?) and runs out at f_T (h_fe = 1), say 100MHz.  It's a dominant pole cutoff (more or less due to C_cb * R_bb' predominantly I think, but also physical limits), and independent of circuit limitations which may cut lower (like C_cb * R_L setting output GBW).  An emitter follower is a terrible (general purpose) choice at 10kHz already because who even makes (let alone budgets for) transformers at that frequency, but it's even worse at 10MHz where gain is dropping sharply.  (Adjust numbers appropriately for any given part in question.)

Riddle: how do you get power gain from a transistor above fT?  By your premise, this is impossible!  In fact it's not (but it isn't terribly useful, the efficiency or bandwidth is poor).

You imagine that the difference might be distortion, or noise; it is not, the device has the same fundamental input-referred noise limits regardless of topology.  You get say 1nV/rtHz of noise in series with the base, and it doesn't make the tiniest bit of difference whether the base-emitter is wired to the left or right side of the input, it's always and necessarily in series regardless!

Put still another way: the noise figure of a device is fundamental to the device, regardless of arrangement or topology, depending only on the bias and impedance matching (and frequency range, since noise generally doesn't have flat frequency statistics).  You put in a low-noise transistor, you bias it up as needed, match it to the source and load, and that's what you get.  Sure you might get lower nV/rtHz if you run it hotter (higher Iq), but you get higher pA/rtHz (current noise) too: it just means a lower source impedance for noise match.  Meanwhile, the noise figure hasn't changed much.

RF design is all about eliminating confounding specifics, like the voltage or current in a circuit, and considering the flow of power back and forth through a network.  In general, we don't care about impedance, because impedance can be matched with simple RLC networks, transformers at the worst -- and if impedance isn't matched to its ideal value (give or take acceptable tolerance), it ain't our components to blame!  This is why we have dimensionless figures like gain and noise factor, s-parameters and etc.

Tim

You're right, my master

[BrianHG]

Is this receiver for FM or  AM and other linear type signals ?

FM needs LOTS of gain in the IF chain to clip the signal and reduce amplitude noise.

Otherwise, for AM or more linear modes, dynamic range is important.

Things to consider anyway.

boB

Really?  I was under the impression that the RF/IF AGC for FM was fast enough / outside the bandwidth of the receiver, just trying to maintain a proper amplitude so that a proper sine to sine phase comparison can be done for precision fine demodulation can be done.


Wont just over cranking up the gain beyond a set clipping range will cause interference like a lightning bolt, of a stronger adjacent broadcast just flood and saturate your IF section causing inter-modulated distortions?

[T3sl4co1l]

You can do it that way, but it's a lot easier just letting it saturate -- and putting in enough stages to guarantee this even at zero input (effectively an oscillator with terrible phase noise, but feedback is strictly suppressed, it's not oscillating as such).  IMD I think doesn't matter as it has to be 3rd order or higher, and 2nd order is dominant from the single-ended stages typically used; not to say it's wholly absent, but it's an ever-lower level, which combined with the selectivity per stage and natural noise immunity of FM, certainly makes for an acceptable system.

Tim

[CaptDon]

I can think of one example where the turns ratio becomes part of a voltage gain and impedance matching circuit. I have a home built tube preamplifier for the 20 meter ham band (14MHz) and the very front end is one of those old 'AirDux' or 'B&W' ham 'coils'. It is about 3/4" diameter and maybe 1.25" long. Probably about 16 turns total. The 'bottom' of the coil is grounded, the 50 ohm antenna input connection is about three turns up from the bottom and the top of the coil feeds the grid of a 6AK5 (maybe 9003?) vacuum tube with a self biased cathode resistor and its bypass capacitor. The entire coil is resonated by a variable trimmer capacitor. This gives me selectivity because of resonance but it also helps to 'magnify' the voltage going to the grid of the 6AK5 due to resonant Q factor. The voltage at the grid is indeed higher than the voltage at the antenna input. This is however mostly due to impedance transformation and not exactly turns ratio. Due to the Q of the circuit the grid signal voltage will be much higher than what the simple turns ratio would predict. B.T.W., the plate of the 6AK5 is also resonated and coupled to a 6C4 cathode follower circuit to provide a 50 ohm output to the receiver. I don't have any neutralization on the 6AK5 and if the grid resonator and plate resonator are tuned at the very highest throughput gain the unit breaks into self oscillation!! (as may be expected) Overall, the preamp works great for pulling in those weak signals!! Would be amazing to try the same circuit concept with some of the modern low noise FET's in place of the 6AK5.

[T3sl4co1l]

Very similar setup on this project I did a while ago,

https://seventransistorlabs.com/Radio_20m/index.html



Though the tapped input coil is tuned (C + varicap), curiously (or perhaps not) the input impedance and/or gain is actually fairly flat above the band of interest.

This is using 5702, which is basically submini 6AK5.  20m band.

The effect is, I think, due to the coil's relatively high coupling, the Q factor is low, and there's (a) zero(es) in the transfer function, basically to say that the bottom half of the coil acts to load the source inductively while the top half acts in series with the grid, and the tuning cap kind of doesn't matter much in the process.  So, instead of having a sharp peak, it kind of drops off slowly.

Note ferrite bead; with the crufty layout, it otherwise oscillates at 400MHz or so -- stub length to the tuning cap, more or less.


Also have a VHF radio built with a 6688 RF amp,

https://www.seventransistorlabs.com/Images/FMRadio3.jpg
https://www.seventransistorlabs.com/Images/FMRadio2.jpg

It didn't do as well as I had expected/hoped; I'm not sure what the difference is exactly, if gm is lower than expected, or there's plate losses somehow, or what.  Grid losses are fairly high but that's kind of just specsmanship, they never really tell you quite what to expect -- note that C_in is measured cold, so doesn't include space charge or beam capacitance, which are significant (in this case, about 1.5 times stronger than C_cold alone!).  So I didn't get all that much voltage gain from the input matching network, nor the plate (to mixer grid) network.

Hm, I don't have any schematics of this handy, so, it's not exactly easy to follow.  I don't have the FM BCB coils handy it seems, but I do have the input coils for 144MHz modeled:



C4 || R2 being the grid equivalent.

All the tuning coils are pluggable so I can reconfigure the radio, it's supposed to be flexible; antenna input is on the right, and the input coil on the phenolic tube.  6688 socket is under the shield/divider, with screen supply through the 680R, grid and cathode on the right side, plate on the left.  Mixer (middle) is 6J6, single balanced; LO is 6C4 (left); output (IF1) coils middle bottom, here tuned for 30MHz (rather high, but selected to minimize images at a few typical bands of interest).

Don't actually have a lot of solid state radio projects, partly because, I kinda just don't care, one is enough; but also I can't seem to avoid tremendous amounts of harmonics / IMD / images / crosstalk / etc. when using them; tubes are just... easier, I'll give 'em that!

Tim

[CaptDon]

Many years back a fellow gave me a gallon sized plastic bag filled with those miniature tubes. There was all kinds, dual triodes, some pentodes, many different numbers. They were loose in the bag, none in boxes. I believe they were U.S. Military surplus left behind from one of the 'rear echelon' service depots. If I recall they were all standard 6.3v filament power. All appeared new with long uncut leads. I think I sold them at a ham radio flea market here in the states? Was about 30 years ago probably. I can remember a time when the Gaithersburg Maryland hamfest nearly filled the entire Montgomery County Fairgrounds!!! I sold an Intellec 8-MOD-80 fully populated computer system there and the guy was wild over it!! Had several programs on paper tape to go with it. I always suspected it may have been Christopher Titus who bought it for the museum?? I think I found it in a dumpster dive? It was fully working including the 1702 (?) front panel eprom programming socket.