Broadcast PLL - Veronica NRG 2025

[BriansBrain]

OK...
I have managed to reduce the output harmonics drastically by replacing the capacitors in the LPF circuit - as follows.

Veronica NRG 2025 - LPF - New Values.JPG

Now...
I have tried to balanced the Kallitron oscillator as much as I can.
Matched the transistors for hfe, and the relevant capacitors and resistors are accurate matched pairs.

Veronica NRG 2025 + Front end to LPF.jpg

However... I'm getting breakthrough of the ½f signal on the output.
Output on the Siglent Spectrum Analyzer.

Veronica NRG 2025 - Siglent .jpg

Any suggestions what to try

.

[King Croccy]

I don't know the answer but I am just curious as to why so many elements to the filtering? It looks like a 9 element filter, is the harmonics without it terrible?

[yellowbeard]

A high pass filter after C21 or maybe C31 might help - I sense you're going to stick a big amplifier after that or you wouldn't be worried about stuff thats 40dB down...

[jvok]

Would be interesting to know how much difference you got by matching the transistors

[Albert H]

I built similar ones with a little Soshin GFWB3 Band II filter between C21 and R25. The only thing that gets through is the Band II product. It has an insertion loss of just under 3dB, but there's plenty of signal there to drive the subsequent stages.

Make sure that your final FET isn't being overdriven with the preceding 2N4427 - I found that this was the principal cause of harmonic outputs.

Also, make sure that all the coils in your VCO are precisely the same.

[BriansBrain]

I don't know the answer but I am just curious as to why so many elements to the filtering? It looks like a 9 element filter, is the harmonics without it terrible?

Yes it's originally a 9 element filter in this NRG version.
Harmonics have now been reduced dramatically with the new values.

[BriansBrain]

A high pass filter after C21 or maybe C31 might help

OK
Consisting of what - can you suggest a circuit modification with component values

I sense you're going to stick a big amplifier after that or you wouldn't be worried about stuff thats 40dB down...

Yes it's going to be driving a 350W RF Linear

[BriansBrain]

Would be interesting to know how much difference you got by matching the transistors

When I built the board originally I had't matched anything in the Kallitron oscillator area, it was worse then around 34dB down.
When searching though posts regarding the NRG I found a couple of Albert H's that mentioned the matching of the component's.

Quote from Albert H

If the Kallitron oscillator is properly balanced, there is no breakthough of the ½f signal.

So I removed and replaced with matched ones R10, R11, R12, R13, C14, C17 and TR2 / TR3.

As you can see +/- 41.5dB down now

[BriansBrain]

Make sure that your final FET isn't being overdriven with the preceding 2N4427 - I found that this was the principal cause of harmonic outputs.

I have realized the Bias for the Final FET is critical regarding some nasty spurs

Also, make sure that all the coils in your VCO are precisely the same.

The coils are self wound and made and matched to the best of my ability

[reverend]

What is the control voltage feeding the VCO varicaps (measured at point P3). If this is too low (below around 3 or 4 Volts) the varicaps may be being driven into saturation and that will cause a large leap in all sorts of nasty products. Also the matching into the gate of the RD15 doesn't look optimal. You could try removing the 2u2 inductor and replacing it with L4, then making L4 a short circuit. You may also need to lower C31 to 33pF or less to provide a proper match. None of this will affect the balance of the oscillator but I have found that using half-frequency oscillators, mismatches and overdrives can cause all sorts of issues.

[BriansBrain]

What is the control voltage feeding the VCO varicaps (measured at point P3). If this is too low (below around 3 or 4 Volts) the varicaps may be being driven into saturation and that will cause a large leap in all sorts of nasty products.

Measured at point P3 = 2.7V

Also the matching into the gate of the RD15 doesn't look optimal.
You could try removing the 2u2 inductor and replacing it with L4, then making L4 a short circuit.
You may also need to lower C31 to 33pF or less to provide a proper match.

OK this will take a while to organize

[BriansBrain]

Also the matching into the gate of the RD15 doesn't look optimal.
You could try removing the 2u2 inductor and replacing it with L4, then making L4 a short circuit.
You may also need to lower C31 to 33pF or less to provide a proper match.

OK this will take a while to organize

Done......... OK........First test it's lowered the ½f signal by about 7dB

I will give further more in depth updates tomorrow

[Albert H]

Tuning_Indicator_tidied-V2.png

We added the above tuning indicator to the next version of the PLL Pro series (the version that was never completed because of Stephen's illness and demise). It allowed you to easily optimise the tuning of the VCO trimmer capacitor - just tweak until both LEDs are out.

You really should consider getting a Soshin GFWB3 filter (available through plenty of Ebay sellers). They have virtually no response below 70 MHz, so the ½f will be attenuated by at least 24dB if you install it at the output of the oscillator, making the output carrier purity mostly dependent on the alignment of the output FET and its associated filter.

It's also worth considering inter-stage screening and additional supply rail filtering.

[BriansBrain]

We added the above tuning indicator to the next version ........

You really should consider getting a Soshin GFWB3 filter (available through plenty of Ebay sellers)......

It's also worth considering inter-stage screening and additional supply rail filtering.

Thanks a lot mate........ much appreciated

[Shedbuilt]

You really should consider getting a Soshin GFWB3 filter (available through plenty of Ebay sellers). They have virtually no response below 70 MHz, so the ½f will be attenuated by at least 24dB if you install it at the output of the oscillator, making the output carrier purity mostly dependent on the alignment of the output FET and its associated filter.

Was this by inserting the GFWB3 directly between stages Albert; without any additional matching for the 75 Ohm in/out of the GFWB3 itself ?
I was also wondering about an "old school" series resonant trap filter (tuned to 1/2f of course); either across the output, or possibly between stages. It'd obviously have to be tuned to (half of) the specific frequency. Between stages would stop the 1/2f early, so would help reduce all the odd harmonics of the 1/2f too (as would the Soshin filter obniously).

[Shedbuilt]

You really should consider getting a Soshin GFWB3 filter (available through plenty of Ebay sellers). They have virtually no response below 70 MHz, so the ½f will be attenuated by at least 24dB if you install it at the output of the oscillator, making the output carrier purity mostly dependent on the alignment of the output FET and its associated filter.

Was this by inserting the GFWB3 directly between stages Albert; without any additional matching for the 75 Ohm in/out of the GFWB3 itself ?
I was also wondering about an "old school" series resonant trap filter (tuned to 1/2f of course); either across the output, or possibly between stages. It'd obviously have to be tuned to (half of) the specific frequency. Between stages would stop the 1/2f early, so would help reduce all the odd harmonics of the 1/2f too (as would the Soshin filter obniously).

Thinking reactances: especially around the trap filter resonance, may well cause issues for inter-stage use.

[Albert H]

One experiment I tried (some years ago) was a parallel ½f trap at the output of the oscillator - I tried (and failed) to come up with a "roofing" filter that would cover 43 to 54 MHz with a flat(ish) passband. I almost succeeded, but only with a design that would end up as more complex than the rest of the exciter!

The output impedance of the VCO isn't a bad match into the Soshin filter, and I increased the resistor into the base of the subsequent transistor to 68Ω (as I remember), and the slightly attenuated output from the filter was nicely clean into the next stage.

Other modifications I made to the board included putting a metal flange right across the board, which provided heatsinking for the regulators and the output FET (in the HVF06 version), and also formed a screen to separate the output filter from the rest of the board.

[Shedbuilt]

I just noticed they changed the “under decoupled emitter”. The changes made will reduce gain slightly (quite possibly the intention), but also reduce the frequency at which gain starts to increase (it’s designed so that the gain is still rising with increasing frequency above band II, and falls off with falling frequency; partly to compensate for falling transistor gain with rising frequency). It might be worth removing C28, and check the results (including 1/2f level and output flatness across band II). I don’t think it’ll change anything drastically, but might give a slight reduction in 1/2f.

[Albert H]

The sloped gain profile was to compensate for the falling gain of the BLW60C towards the top of the band. It ensured that the output power was 45 Watts right across the band!

[Shedbuilt]

The sloped gain profile was to compensate for the falling gain of the BLW60C towards the top of the band. It ensured that the output power was 45 Watts right across the band!

Thanks Albert. I remembered that it was to compensate for falling output device gain, but forgot the BLW60C specifics. The decreasing gain with decreasing frequency will obviously continue beyond the bottom of Band II, but probably only about 4 or 5dB down by 1/2f vs f; I'd imagine.