VCO oscillates correct frequency but 30dbm less pwr than sim

Hello,

I am having difficulty with the VCO I have designed.
The VCO oscillates at the same frequency predicted by simulations but there is a whopping 30dbm less power available.

I have tried different impedance matching techniques to the output. I have tried increasing the bias current with still little output power.

The board consists of 1.6mm FR4 with 35um copper traces. This is what I simulated in momentum. I did assume a perfect ground plane on the bottom of the board when in reality in some places it is broken by other traces.


The picture below shows the actual board covered with BNC connectors for testing.



Does anyone have any ideas as to what the cause is or what I should try to get this working?

I think it is worth mentioning that I altered the circuit to bypass the BNC connector I was measuring from yet I was still picking up the signal. ie I guess not all power was being delivered to the desired load.

I am using BFR182 transistors. The linear and non linear models were available in ADS

Thankyou

You don't show a load or impedance matching network. You also don't tell at which node you measured
an output level and what's the actual power.

Re: VCO oscillates correct frequency but 30dbm less pwr than

I assume you really meant 30dB less output not 30dBm

the BFR182, assuming 50% efficiency, could output ~ 125mW, ~ 20.5dBm

as FvM asked ... what level are you getting out of it ?


Dave

Re: VCO oscillates correct frequency but 30dbm less pwr than

Hey,

During simulation with a 50ohm load connected to net A1, I achieved -10dbm at 432.6MHz.

When I measured the output power with my spectrum analyzer I measured -37dbm at 432.6MHz. The input on the spectrum analyzer has a 50ohm impedance.

According to simulations with a 50ohm load connected to net A1 without any other impedance matching components, I should be getting -10dbm at 432.6MHz.

Cheers

Re: VCO oscillates correct frequency but 30dbm less pwr than

My friend, it is only natural that you get such small power out of your VCO, as with 3V supply you will not get proper bias conditions. The transistor will require something of the order of 10mA collector current, which with 680 ohms at the emitter would drop some 6.8Volts.
There is nothing wrong with the emitter resistor value as such, smaller values would load your circuit appreciably. You need to increase the supply voltage to leave the necessary Vce and adjust the 100K resistor for the required Ie.
Another thing is you need to isolate the varicap from the driving circuitry (pll or whatever) either by a resistor or preferably by an inductor, otherwise the Q of your tank circuit will be diminished as soon as you connect anything to it.

-10 dBm sounds basically realistic with the shown bias conditions, even with severe impedance mismatch. But I also expect strongly
non-linear operation and thus very bad VCO performance.

I don't think, that the problem has to do with the varicap drive circuit. Decoupling is actually present but also additional components
at oscillator output, so I wonder where the analyzer has been connected.

I assume, that you performed a wideband scan, so you won't be fooled by a PLL sideband? Or did you operate the VCO in open loop?

I also assume, that you know how to use a spectrum analyzer. Some combinations of
settings will result in an uncalibrated level display...

Re: VCO oscillates correct frequency but 30dbm less pwr than

FvM wrote:

I don't think, that the problem has to do with the varicap drive circuit. Decoupling is actually present

Click to expand...

Do you mean the 33nH inductor? Because as I see it, this is not decoupling, it is the tank circuit inductor. The 15pF parallel cap also does not qualify as decoupling, since it would then shunt the varicap.

To decouple the varicap, and the rest of the tank circuit for that matter, would require the addition of a series inductor or resistor between the diode and the driving circuitry.

But the dc operating point needs to be taken care of.

Impedance matching at the collector, while potentially desirable, cannot make or break a VCO.

Re: VCO oscillates correct frequency but 30dbm less pwr than

Just add a buffer at the output of the VCO and problems with matching and output power will be resolved.
Generally in Coplitts VCO's is better to pick-up the output from the emitter/source instead collector/drain.

Re: VCO oscillates correct frequency but 30dbm less pwr than

darcyrandall2004 said:

The board consists of 1.6mm FR4 with 35um copper traces. This is what I simulated in momentum. I did assume a perfect ground plane on the bottom of the board when in reality in some places it is broken by other traces.

Click to expand...

Breaking the ground plane is bad. This can introduce significant inductance, because now the return path is "somewhere". If have seen several designs which failed because of this mistake. You must design the ground below the trace with the same care as the trace above, they are equally important.

What you can do for testing: add capacitors across the slot in the ground so that your RF signal can pass.

Re: VCO oscillates correct frequency but 30dbm less pwr than

What you can do for testing: add capacitors across the slot in the ground so that your RF signal can pass.

Click to expand...

Brilliant idea.

As for the the varicap, it isnt shown but it is actually driven by a 470ohm resistor in series with the PLL Loop filter. I can post the image if you wish.

My friend, it is only natural that you get such small power out of your VCO, as with 3V supply you will not get proper bias conditions.

Click to expand...

Il try vary the supply voltage and see how the output power changes. I have in the past pulled apart handheld pagers and they have tended to use similar bias levels.

Just add a buffer at the output of the VCO and problems with matching and output power will be resolved.
I actually do have a buffer at the output of the VCO but unfortunately it doesnt produce enough power to overcome the -30dbm loss.

Generally in Coplitts VCO's is better to pick-up the output from the emitter/source instead collector/drain.

Click to expand...

Good idea ill try this aswell.

Thanks for the ideas, I didnt expect to find many alternatives to what I have already tried.

Cheers

Click to expand...

Re: VCO oscillates correct frequency but 30dbm less pwr than

I see, that the ground plane layout is really bad. With a small change in component placement, you could add traces to short the huge gaps.

Even with the present layout, I don't believe the - 37 dBm output level. But didn't tell any details about the measurement conditions.

Did you check the harmonic distortion in your simulation? In my opinion, it's terribly high.

Re: VCO oscillates correct frequency but 30dbm less pwr than

I agree harmonics will be bad, 2nd harmonic will only be down 3-4dB. Design should make -10dBm easy so I suspect something else going on causing the low level output.