High Frequency VCO -- Capacitor bank

Papabravo

Joined Feb 24, 2006
16,772
It does not sound like that would be a useful method. I'm not entirely sure what you have in mind. A schematic diagram would be most helpful in understanding the question.
 

ronsimpson

Joined Oct 7, 2019
1,450
Thanks for the picture that helps!
There is a switch in the "capacitor bank" that sets the amount of capacitance. C or 2C or 4C.

Here is an example of a oscillator/ signal generator. There is a large know that sets the frequency and a small knob, bottom left, that also sets the frequency in big jumps. In the picture the big know goes from slightly below 1 to slightly above 10 while the little knob is 1, 10, 100, 1000, 10000.
1626965753248.png
In your example the "bank switch" is 1,2, 4 and the knob is 1 to 2. So when bank=1 knob set the frequency any where from 1 to 2. But when the switch is set to 2 the knob sets from 2 to 4. Then when the switch is set to 4 the knob set the frequency in the 4 to 8 range.
 

MrAl

Joined Jun 17, 2014
8,470
Please check this
You get a much, much wider range of frequencies with multiple capacitors in any oscillator that uses capacitors as part of the frequency tuning circuit.

I think the confusion was that the "capacitor bank" isnt really a capacitor bank or at least it shouldnt really be called that. A better more descriptive name would have been "capacitor selection switch matrix" or something like that. As Papa said, it is used to select the value of the capacitor used for oscillation so it sets the range of the oscillation fine adjustment. So you can look at it as a coarse frequency adjustment.

Now the different capacitor values you can select are binary weighted, so if you close the top switch you select "C" so get a value of 1 x C, but if you instead close the next switch you get 2 x C, and if you close both switches you get 3 x C because capacitors in parallel add, and if you were to also close the 4C switch you would get a capacitance of 7 x C, and that is because all three capacitances would be in parallel so they would all add.
Because they are binary weighted you can get capacitance values of 1C, 2C, 3C, 4C, 5C, 6C, 7C or even zero C if there is a built in default value of C elsewhere in the circuit. This gives you 7 or 8 coarse ranges of frequency selection.

The reason for multiple capacitors ie because oscillators only adjust so much by other means so by using multiple cap values you can get a much, much wider range of frequencies.

Here is a slightly better drawing...

HighFrequencyOscillator-1.gif
 
Last edited:

Papabravo

Joined Feb 24, 2006
16,772
It is even more subtle than that because identical capacitors in series ha a value of C/2. So C || C = C*C/(C+C) = C/2
Capacitors in parallel add so:
C/2 + C = 3C/2 and so forth. So there a 7 switch combinations that represent valid tuning ranges. The fine tuning is done with an (old school) varactor, nowadays we use varactor diodes in a very small package.

The use of the term "capacitor bank" had me confused, thanks for the schematic.
 

MrAl

Joined Jun 17, 2014
8,470
It is even more subtle than that because identical capacitors in series ha a value of C/2. So C || C = C*C/(C+C) = C/2
Capacitors in parallel add so:
C/2 + C = 3C/2 and so forth. So there a 7 switch combinations that represent valid tuning ranges. The fine tuning is done with an (old school) varactor, nowadays we use varactor diodes in a very small package.

The use of the term "capacitor bank" had me confused, thanks for the schematic.
Yes thanks it is less than the actual multiples of C it has to also be divided by 2.
So we can get C/2, 2C/2, 3C/2, 4C/2, 5C/2, 6C/2, 7C/2 and perhaps 0C/2 with an unseen default capacitor somewhere else in the circuit.
 

Tesla23

Joined May 10, 2009
477
This is very common in modern PLL/VCO chips, this is from the ADF4351 datasheet that sums it up nicely:

1626998549474.png

3 VCOs, with 16 sub-bands each - achieved with switching in capacitors, covering 2200MHz - 4400MHz, along with respectable phase noise performance and all in a 5mm x 5mm package.

The ADF4351 is quite old, newer devices have capacitor banks giving hundred(s) of sub-bands. The ADF5610 datasheet shows a bit more more detail, and there are many others from ADI and other manufacturers.
 

MrAl

Joined Jun 17, 2014
8,470
This is very common in modern PLL/VCO chips, this is from the ADF4351 datasheet that sums it up nicely:

View attachment 244173

3 VCOs, with 16 sub-bands each - achieved with switching in capacitors, covering 2200MHz - 4400MHz, along with respectable phase noise performance and all in a 5mm x 5mm package.

The ADF4351 is quite old, newer devices have capacitor banks giving hundred(s) of sub-bands. The ADF5610 datasheet shows a bit more more detail, and there are many others from ADI and other manufacturers.
That looks interesting.
There is another one i got to use. It only went up to 40MHz but it was digital 32 bit word controllable. It came on a complete PC board with filters and everything and it was under 20 dollars USD. Dont remember the part number offhand. It produces a fairly clean sine wave and because of the 32 bit word the frequency increments were very small so you could effectively dial in just about anything you needed.
Havent used it a while now though.
I can find the part number if anyone is interested, but again it only goes up to 40MHz but it is crystal controlled so very stable.
 

Tesla23

Joined May 10, 2009
477
That looks interesting.
There is another one i got to use. It only went up to 40MHz but it was digital 32 bit word controllable. It came on a complete PC board with filters and everything and it was under 20 dollars USD. Dont remember the part number offhand. It produces a fairly clean sine wave and because of the 32 bit word the frequency increments were very small so you could effectively dial in just about anything you needed.
Havent used it a while now though.
I can find the part number if anyone is interested, but again it only goes up to 40MHz but it is crystal controlled so very stable.
Sounds like a DDS, nothing to do with VCOs.
 

MrAl

Joined Jun 17, 2014
8,470
I guess that's why they have different names: VCO and NCO
Yes we know they (may) have different names, but i figure that someone interested in a VCO may be interested in the digital version. As i am sure you know, it's just a matter of how they are tuned (digital word or voltage).
With that in mind, a good Analog to Digital converter chip would immediately convert the digital version to the analog version, and with very precise frequency selection.

If you wanted to complain about something you should have mentioned the frequency difference which i think would be 100 to 1 or even 1000 to 1 so someone that needs a 4GHz oscillator may not be interested in a 40MHz oscillator, but then with a little PLL magic the lower could control the higher, with fairly precise frequency selection.

So really my post was about related areas of interest and could in fact end up being part of a solution.
So i think drawing a critical distinction from one to the other may be a little too critical, but hey if you want to mention it i guess there's nothing wrong with that either and i like to see members freely voicing their opinions :)
 
Top