VCO for FM transmitter

[wakoko79]

We are to design a VCO for an fm band.. I chose 91.7MHz (200kHz bandwidth)..

I have a working design, a Clapp oscillator. The oscillator is modulated by a varactor (MV104) centered at 2V (1V to 3V), so the capacitance range is ~59pF to 39pF.

I think no matter what topology (split capacitance) I use, the main equation for resonant freq won't change: f=1/(2pi*L*C), where C is the series equivalent resistance of the 2 feedback capacitors and the modulating capacitor (Cd).

I made an excel file so I can see the values. I have a 220nH inductor (air core, I coiled it) so I would like to work with it.

With analysis, I also found that:
high L/C ratio is better = higher L = automatic lower Ctotal (for a given freq) = lower capacitance change *needed* for a given BANDWIDTH = lower Cd(varactor*fixed* + parallel cap)


The problem is that the Cd needed for 91.6MHz and 91.8Mhz have a difference in the order of tens of femtofarad for L=~1000nH... is this OK? Will the varactor be ok if it is paralleled with another capacitor to achieve the lower capacitance swing? I feel like it will be swamped or something...

Please help!!

 

[Nigel Goodwin]

We are to design a VCO for an fm band.. I chose 91.7MHz (200kHz bandwidth)..

I have a working design, a Clapp oscillator. The oscillator is modulated by a varactor (MV104) centered at 2V (1V to 3V), so the capacitance range is ~59pF to 39pF.

I think no matter what topology (split capacitance) I use, the main equation for resonant freq won't change: f=1/(2pi*L*C), where C is the series equivalent resistance of the 2 feedback capacitors and the modulating capacitor (Cd).

I made an excel file so I can see the values. I have a 220nH inductor (air core, I coiled it) so I would like to work with it.

With analysis, I also found that:
high L/C ratio is better = higher L = automatic lower Ctotal (for a given freq) = lower capacitance change *needed* for a given BANDWIDTH = lower Cd(varactor*fixed* + parallel cap)


The problem is that the Cd needed for 91.6MHz and 91.8Mhz have a difference in the order of tens of femtofarad for L=~1000nH... is this OK? Will the varactor be ok if it is paralleled with another capacitor to achieve the lower capacitance swing? I feel like it will be swamped or something...

Please help!!

Stop thinking and do it!! - but use commonsense in the choice of coil, you're NOT going to reach anywhere near 90MHz with a 1mH coil.

Try a few turns of thick wire around a pencil - between 4 and 8 turns.

 

[wakoko79]

Stop thinking and do it!! - but use commonsense in the choice of coil, you're NOT going to reach anywhere near 90MHz with a 1mH coil.

Try a few turns of thick wire around a pencil - between 4 and 8 turns.

That made me laugh haha. Well, I do have an actual circuit on a PCB (hexagon pads for RF). Here's the schematic:
View attachment 65740

This was the old circuit I did. I think this is still in the fm band (89MHz maybe?). But the BW is much much bigger back then(2MHz vs 200kHz now). So it's much more difficult since the circuit needs a higher Q. But the main problem is really the capacitor swing needed to achieve the BANDWIDTH limits, the capacitance is just damn too small. I also don't know how to EXACTLY compute for the needed loop gain (I think I solved this previously, I have to find my notes first).

I don't intend to have an inductor as big as 1mH. But I read that higher L (also higher impedance) means better operation, since there is less circulating current in the LC tank. So I think 220nH is good.

So going back to my point, the capacitance swing needed is just too small that it may be swamped by other caps (maybe the feedback caps?). Will this just be fine? (As of now, I don't have the necessary capacitors, so I'm going theoretical now.)

Thanks

 

[Nigel Goodwin]

I don't intend to have an inductor as big as 1mH. But I read that higher L (also higher impedance) means better operation, since there is less circulating current in the LC tank. So I think 220nH is good.

So going back to my point, the capacitance swing needed is just too small that it may be swamped by other caps (maybe the feedback caps?). Will this just be fine? (As of now, I don't have the necessary capacitors, so I'm going theoretical now.)

You need to go back and rethink it - I don't know where you've read that higher L is good? - a balanced ratio of L to C is good, and 0.22mH is still massively too big. Your coil needs to be in the uH range.

EDIT:

Please accept my apologies, I was in a rush and misread your post, 220nH or so should be fine, but really as you need to handwind the coil (thick wire in free air) then you don't need to know it's accurate value

 

[JimB]

Two possible ways to reduce the "sensitivity" of the MV104:

1 Reduce the value of the 10pF capacitor.

2 Move the connection of the MV104/10pF from the junction of L1/C3 to the junction of L1/C1.

JimB

 

[WTP Pepper]

I assume you are pulling the frequency by DC applied to the MV104. Bad idea as you are altering the frequency by load.
Move the Varactor to the tuning elements prior to the transistor for some isolation.

 

[JimB]

I assume you are pulling the frequency by DC applied to the MV104.

I also hope that he is.

Bad idea as you are altering the frequency by load.

No, it is a good idea and generally accepted practice.
The diode is reverse biassed, nothing is being "loaded".

Move the Varactor to the tuning elements prior to the transistor for some isolation.

Uh???
What "tuning elements prior to the transistor" ?
And what will be "isolated" from what ?

JimB

 

[RCinFLA]

You want to go with a physical coil that will yield the highest Q. For an air wound coil this will usually be in the XL in the 50 to 100 ohms range. The width to length aspect ratio of about 1 with a turn spacing gap about equal to wire diameter. Use wire gauge in the 16 to 22 range. There are some rough coil calculators on-line that can get you in ball park.

Don't take off your output load from the tank circuit. Unless you ensure a very light load you will always have frequency pulling with load variation, even if it is only buffer input capacitance due to RF drive level variation.

I suggest to try a Colpitt configuration. Collector is normally RF grounded but you can put a small value resistor (10 to 50 ohms) in series and pick off your output across it. It will minimize output loading changing oscillator freq. The collector series resistance does lower loop gain of oscillator a bit but if you keep it small it will not be siginficant. If you need more RF level then put an amplifying buffer from the collector load resistor output.

As to tank circuit. Using a 100 to 130 nH coil should yield good results. If you are only interested in a fixed channel then use mostly fixed cap and the feed back caps to provided the tuning. Roughly the contribution from the feedback caps are their series equivalent capacitance. You can throw in 5 to 8 pF for the Cbe of a reasonable RF transistor. De-embed the varactor with a series cap to get FM modulation you need.

All of this is for not if you are required to use a 2N431 transistor. It is a dinosaur with an Ft of about 25 MHz. It is almost impossible to find a transistor that bad today. It will be very hard to get enough gain to make a reasonable oscillator at 92 MHz. Suggest something like a 2N918 or other with an Ft of at least 500 MHz. Even the common 2N3904 general purpose transistor commonly used for audio has an Ft of 300 MHz.

 

[WTP Pepper]

The diode is reversed bias at DC. In such a circuit it is an AC component whose value is changed by a DC bias. That's what varactors do! The load will depend upon what is connected and this will affect the effect the varactor diode has.

As for what is being loaded, maybe an aerial that comes into contact with anything thus changing it's impedance without an isolator and thus pulling both the deviation and C/F.

Prior to the transistor I mean on the base side thus helping isolation via the transistor. It will still pull but the pull will be less.

The posted circuit is a bad example. There are better ways to make a load vairiable stable VCO with a single transistor without having the major tuning component on the output stage.

 

[Nigel Goodwin]

The posted circuit is a bad example. There are better ways to make a load vairiable stable VCO with a single transistor without having the major tuning component on the output stage.

As it's only one small section of a complete design you can't really say it's a bad example - presumably there will be a suitable buffer following the VCO before further stages, which you would be with pretty well any oscillator style - I suspect it's probably the way it's drawn that you don't like?, I can't say I find it very 'likeable'.

 

[WTP Pepper]

Yes if it's buffered then things will be different. Just going off the information posted.