FM converter (mixer) -- I don't understand

[michael1978]

It is difficult to simulate frequency mixers in spice. You can independently simulate the RF input impedance or the IF output network in the frequency domain, or oscillation of the LO in time, but not both time and frequency domains in the same run.

The way your example circuit was drawn effectively disguised the mixer in the LO where; Two signals are present on the non-linear emitter of Q3. One is the continuous oscillation of the LO, the other is the RF coupled into the emitter through C9. One of the cross-modulation products is selected by the tuned circuit in the collector, as the IF.

It is simple to understand once you look at the untangled circuit from the right direction.

Replace C5, 100 pF with C23, 300pF. I think I was thrown by having to guess the missing value for C14. What was the value of C14 before you put the solid black line of misunderstanding through it ?

Hi Baluncore my last question for this circuit, L4 and C16 are tuned to 92MegHz?

 

[Baluncore]

Hi Baluncore my last question for this circuit, L4 and C16 are tuned to 92MegHz?

That was just a wild initial guess for L4. My later version LTspice model oscillates and mixes OK, producing an IF output, but it has not been optimised or tuned to any particular IF or FM allocation. There is no point talking circuit values without frequency band and IF specifications.

I don't think we can arbitrarily assign a frequency of 92MHz to L4//C16 which would probably be in the middle of an FM RF band. The value of L4//C16 will also affect the LO resonant frequency. Why was a fixed value for L4 not specified? What is the purpose of L4//C16. What should the frequency and Q of L4//C16 be?

The RF frequency range is country dependent. It will be a VHF block, somewhere in the range between 66MHz and 108MHz.
The IF centre frequency is not specified, I would guess at 10.7MHz, but it could be as low as 5MHz or as high as 22MHz.
The LO range would be offset, probably above, but possibly below, the RF band by the IF centre frequency.

For 10.7MHz with C16=30pF, L4 would need to be 7.4uH.
For 30.MHz with C16=30pF, L4 would need to be about 1uH.
For 92.0MHz with C16=30pF, L4 would need to be 100nH.

 

[michael1978]

Thank you very much Baluncore, have a nice dayyyyyy

 

[michael1978]

That was just a wild initial guess for L4. My later version LTspice model oscillates and mixes OK, producing an IF output, but it has not been optimised or tuned to any particular IF or FM allocation. There is no point talking circuit values without frequency band and IF specifications.

I don't think we can arbitrarily assign a frequency of 92MHz to L4//C16 which would probably be in the middle of an FM RF band. The value of L4//C16 will also affect the LO resonant frequency. Why was a fixed value for L4 not specified? What is the purpose of L4//C16. What should the frequency and Q of L4//C16 be?

The RF frequency range is country dependent. It will be a VHF block, somewhere in the range between 66MHz and 108MHz.
The IF centre frequency is not specified, I would guess at 10.7MHz, but it could be as low as 5MHz or as high as 22MHz.
The LO range would be offset, probably above, but possibly below, the RF band by the IF centre frequency.

For 10.7MHz with C16=30pF, L4 would need to be 7.4uH.
For 30.MHz with C16=30pF, L4 would need to be about 1uH.
For 92.0MHz with C16=30pF, L4 would need to be 100nH.

THANKYOU A LOTTT...HAVE A NICE DAY...

 

[Zlatko1969]

Hi
I see that this topic is from last year.
But i am interested is this oscillator tested in practice and how work?
Also is there a way to force ocillator to work on 3V?
thanks

 

[Baluncore]

Welcome to PF.
What is your application for an oscillator? What frequency?

Why do you need this particular oscillator?

I see no reason why it should not work on a 3V supply, but component values will need to be changed. Do you have the free LTspice program?
https://www.analog.com/en/design-center/design-tools-and-calculators/ltspice-simulator.html

 

[Zlatko1969]

Hi
It is clear for what frequency is
It is for standard FM 88-108Mhz superhet receiver.
I am intersted how this oscillator perform in FM receiver because i build vary similar
which work on 3V battery.
In my circuit i used S9018 transistor in common base mode with (+) rail GND.

No i don't use LTspice than CircuitMaker free student version.

Also i am asking because in my version transistor 2N3904 not work.

 

[tech99]

It seems likely to me that the 2N3904 does not have sufficient gain at 3V with a small collector current. The transition frequency is quoted as typically 300 MHz, which is a bit low already.

 

[Baluncore]

Also i am asking because in my version transistor 2N3904 not work.

What circuit do you use now?
Please drag and drop your circuit diagram.jpg onto your next post.
We will look at your circuit to find the problem.

 

[Zlatko1969]

Hi
2N3904 is a good transistor and work well in circuits like
regen or superregen receiver but you cannot found it in any
comercial oscillator/ mixer.
I have problem in current circuit even with S9018 not with transistor
than with transformer coupling capacitor which simply to much dump
oscillation( typical rush "hiss" is reduced too much.
Ok here is the schematic with grounded (+) rail.

 

[Zlatko1969]

So that is why I asking
Is it better to have grounded MINUS rail?

 

[Zlatko1969]

I forget to tell you L1 is a coil 19 Turns/dia 3mm - cca 280nH
what is purpose of this coil in combination with C 7 33pF ?
which resonantt freq is 52 Mhz ?
that is a really strange...

 

[tech99]

Hi
I have problem in current circuit even with S9018 not with transistor
than with transformer coupling capacitor which simply to much dump
oscillation( typical rush "hiss" is reduced too much.

Are you saying your circuit is super regenerating i.e. it is loudly hissing?

 

[Zlatko1969]

No ...it is typical "hiss " like in any comercial FM receiver when is not on station.

 

[tech99]

Are you saying the feedback capacitor is too small so that oscillation stops?

 

[Baluncore]

Ok here is the schematic with grounded (+) rail.

I cannot see any reason why your circuit should not work. The modeled local oscillator, LO, starts quickly and oscillates. The 2N3904 was used in the earlier simulation because it is a commodity model in LTspice, there will be better choices for low power RF circuits.

In your circuit, the emitter current is set by R3 =1k0. The base current is set by R1 = 150k. The ratio of those two resistors sets the share of the 3V supply, and so the operating point. But that ratio
is also sensitive to the current gain of the transistor.

RF = 88 to 108 Mhz superhet with IF = 10.7 MHz; ∴High side LO = 98.7 to 118.7 MHz.
Can you tell if your LO is oscillating? You might set your osc:mixer to the low frequency end of the band, then listen to your LO leakage mute a commercial FM receiver.

 

[Zlatko1969]

-Can you tell if your LO is oscillating?

Yes it oscillating because there is a hiss and i can tune few strongest station
but is unstable ,so that is why I ask is first version of this topic maybe better and more stable.

 

[Tom.G]

Yes it oscillating because there is a hiss and i can tune few strongest station
but is unstable

That describes a stage somewhere is doing un-controlled oscillation.

Ok here is the schematic with grounded (+) rail.

A few problems I see:
1) The circuit shows a GND symbol and a wire labeled GND (+) RAIL. What is happening?
2) The Base wiring in this circuit is a lot different than in post #1. For instance C6 here connects to GND (+) RAIL but the original circuit shows that capacitor connected to the Base of Q1.
3) L1 C7 resonate somewhere around 30-50MHz. Is that the frequency you need?

That miswire of C6 is probably a cause of the "unstable" you describe.
L1 C3 frequency could be the problem of "can tune few strongest station".

Cheers,
Tom

 

[tech99]

That describes a stage somewhere is doing un-controlled oscillation.

Zlatko: When you say "unstable", do you mean that the frequency varies too much, or do you notice unwanted oscillation of some type? Also, please notice the circuit questions from Tom.

 

[Baluncore]

I redrew the circuit to check the bias current circuit and the AC ground.

 

[Zlatko1969]

First of all
thank you guys for replys
tech99 yes i read Tom points and i think that have a right.
1) The circuit shows a GND symbol and a wire labeled GND (+) RAIL. What is happening?
- yes GND or grounding point is (+) ,simply cannot be on (-) because this colpitts oscillator require grounnding
on (+), for example if i connect 330pF to MINUS, or 5nF then oscillator stop to work
2) The Base wiring in this circuit is a lot different than in post #1. For instance C6 here connects to GND (+)
RAIL but the original circuit shows that capacitor connected to the Base of Q1.
- yes it is different and that is why i ask in first place and i want to know if circuit #1 is a maybe better
solution.Also 330pF is connected to base of transistor not on GND like in my case.
3) L1 C7 resonate somewhere around 30-50MHz. Is that the frequency you need?
- this resonant circuit is a mistery to me and i cannot explain to myself what is a purpose of
33pF + 280nH which gives cca 52Mhz resonant frequency
No , of course that i don't need it . I just need to build ordinary FM mixer/oscillator
which should mix local oscillator signal with incoming FM sinal, heterodyne effect
as usual Fif = fosc - fin should give cca 10.7mhz
That miswire of C6 is probably a cause of the "unstable" you describe.
L1 C3 frequency could be the problem of "can tune few strongest station"

 

[Baluncore]

When I simulate the redrawn circuit...
It takes time for the bias to settle with Vbase at about -2.0V. Once bias is settled, the LO amplitude varies in a sawtooth at the IF frequency. The IF and the LO are in series. It seems that high IF output voltage is robbing the LO of the voltage it needs to operate continuously.

I will need to think about how to fit those two signals into less than 2.5 volts.

 

[Baluncore]

I'm still thinking, meanwhile...
The following tuned circuits are connected by the non-linear transistor to make the LO/Mixer.
C3//L3 is resonant at the 10.7 MHz IF.
(C1+C4)//L2 is resonant at at the LO frequency.
L1//C7 should be resonant at the RF frequency, with low Q to accept all FM band.

It should not matter if C6 connects to the base or to AC ground. That is because this is a common base circuit, so the base is held at AC ground by C5. If the base is not firmly held to ground, then connection of C6 to the base would be an advantage as the RF tuned circuit would appear across the transistor's base emitter junction.

 

[Tom.G]

It seems that high IF output voltage is robbing the LO of the voltage it needs to operate continuously.

Try reducing the Q of the IF transformer; perhaps an R across L3, say 470R to 2k2. (Or just connect the secondary to the next stage. )

 

[Zlatko1969]

In some practical varsion of this oscillator i saw that is added ressistor
10k or 15k on primary side probably to dump unwanted oscillation .
so i will try that...

 

[Baluncore]

Well that is an interesting circuit.

The problem is with the LO and IF tuned circuits being in series. When first powered up, the LO starts quickly in a few microseconds. Then the amplitude of the IF increases gradually until it blocks the LO.

My model indicates that changing C2 from 5 pF to 10 or 15pF is sufficient to kill oscillation of the IF and stabilises the LO amplitude.

I changed L1 to 80 nH to be resonant in the RF range.

 

[Zlatko1969]

Ok
I will try to change bypass cap 5pF to 10 then i will see, also i will add
10k in paralele to primary coil of IFT.
Just to not forgot in some of versions i also see 100 ohm
collector resistor... probably also stabilises LO.

 

[Zlatko1969]

Guys
Is here someone who can give us better clear image of RFamp stage .
Image is too blur and i cannot see values of resistors and coils,capacitors

 

[tech99]

The oscillation of the circuit at IF, which blocks the LO each IF cycle, means that the circuit has become super regenerative. It will be very sensitive to signals but it will generate a lot of noise when no signal is present.

 

[Baluncore]

Is here someone who can give us better clear image of RFamp stage .

The circuit presented in post #1 was schematic and did not have values for all components.
First get the LO:Mixer operating, then redesign the RF front end. You need gain before the noisy mixer stage. Select an optimised RF transistor for the RF amplifier.

The aim with a receiver front-end is to use all available RF energy wisely. Where there is a Q problem you can adjust it simply by changing the reactance of the LC circuit components or by coupling more or less energy into the 1'st IF amp. The Q of the IF circuit will need to be about 10.7 MHz / 100 kHz = 107. Loading the IF input with a resistor wastes signal energy that could have been delivered to the IF strip.

As an example C3 = 47 pF brings the first IF transformer to resonance at 10.7 MHz. So the inductance of the transformer primary needs to be 4u7H. Then it has reactance XL = –Xc = 316 ohms. If you load that circuit with 1k0 you will bring the Q down to about 4, not the 100 needed.

For a Q of 100 the load should appear to be about 316 * 100 = 31.6 k. The 5:1 turns ratio transformer has an impedance ratio of 25:1. So the input impedance of the 1'st IF amplifier should appear to be 31k6 / 25 = 1k26 ohms. Otherwise juggle any of the values to get a happier solution.

 

[Zlatko1969]

Hi baluncore
i think that i understand your points.
I don't get it why i have problems because i use original
10.7Mhz IFT (orange) you may see in many FM receivers.
IF strip - amplifier work but it looks that I have big problems
between grounding.
In circuit shown here #1 is everything on MINUS rail so let say standard GND
when in my case is different GND of input RF amp and LO is on + RAIL of power supply

 

[Baluncore]

In circuit shown here #1 is everything on MINUS rail so let say standard GND when in my case is different GND of input RF amp and LO is on + RAIL of power supply

It is confusing, but I believe the AC coupling of the RF signal path with C8, and the IF with the transformer eliminates the supply problem.

I redrew your diagram using + and – power rails, with a separate AC ground symbol to clearly identify the DC bias circuit and the AC signal paths.

You are using one transistor for the LO and the mixer. Your one transistor is surrounded by RF, LO and IF, all of which must work together. There are always challenges with circuits that aim to minimise the number of active elements.

Reflex receivers were interesting because they used an amplifier for RF gain, then used the same amplifier again for audio output. https://en.wikipedia.org/wiki/Reflex_receiver

 

[Zlatko1969]

Yes i know that .
heh ..reflex receiver is easy to build and they work fine even with low selectivity
but this one is really confusing.
here is what i think that might work as RFpreamp