Moderm RF transistors in SMD packages.

[1608cc]

One commercial transmitter had VCO semirigid coax between two 5mm thick pieces of Armaflex like foam.

Will try with this - 3mm ID and RG316.

obraz_2025-02-03_004515164.png

[pukje62]

Albert:

Is the push-push setup for 2 x BFG135 something like this:
Can we also use the BFG591 in the same setup ?

[Albert H]

That's pretty much the same!

[Krakatoa]

BFQ19 is the same silicon as BFR96 but in SOT89 package, if it is of any help.
In fact, there are many transistors of the old through hole times that went SMD but was never stated in the datasheets. For example, the BF199 is now the BFS20, the BF324 is the BF824, all same silicon, so, in some way, they never went obsolete, only the package!

[EFR]

https://www.eevblog.com/forum/rf-microw ... mplifiers/

[Shedbuilt]

I was going to mention MRF4427, but I see that's also listed as obsolete now.

[jvok]

Some of the no-name chinese fabs make SMD equivalents of the ss9018 labelled as s9018 (only one s). No idea if they're any good

https://www.lcsc.com/mobile/products/Bi ... _z=n_S9018

[EFR]

Some of the no-name chinese fabs make SMD equivalents of the ss9018 labelled as s9018 (only one s). No idea if they're any good

https://www.lcsc.com/mobile/products/Bi ... _z=n_S9018

I have those, almost same as BC547

[reverend]

A couple of 2SC3357 in parallel should get you a Watt or so.

[nrgkits.nz]

Use the LMX2541 VCO, it's what I used in my last SMD design, very low phase noise and not microphonic. I had one running on the bench connected to a 300W PA with no shielding or anything and there was no hum, it was rock solid. This chip operates in the GHz range, and the output is divided down to your intended output frequency, this eliminates RF feedback and hum. I use a 470uF capacitor in the PLL line and modulate the MPX directly onto this, no overmod with a 30Hz square wave.

[zulu53]

Hi reveverend,
Would you please share a 12V application circuit using a couple of 2SC3357 as suggested?

[reverend]

As long as you use a bit of emitter degeneration (i.e. place a resistor in the emitter of each transistor to ground) to stop one transistor from becoming thermally overloaded, you can pretty much replace any output transistor with more than one. Doesn't work for higher power circuits as you typically need to ground the emitter directly. For lower power, it's no biggy.

The circuit below is for a 1 Watt HF amplifier but the princples are pretty much the same. See the separate emitter resistors (and bypass capacitors) on each output transistor?

hf_amp.gif

Note that the input matching will be different to a single transistor as you have two base/emitter junctions in parallel so the impedance will be lower.

[zulu53]

Thank you for the circuit and explanations.

[reverend]

Where? I don't see any post about that here (or you mention another topic?).

Anyway - varicap voltage high as 30V is practically useless. Basing on BB809 characteristic, 1-10V range give us 30pF deviance of capacitance, where 10-30V only 8pF. Also the highest capacitance (equals to lower freqency) is at lower bias voltage.

It can be difficult to use varicaps at low voltages (anything under about 3 to 4 Volts) as, if the RF across them is more than a volt or so, they will begin to conduct (they are diodes after all) and produce all kinds of nastiness. Maintaining the bias at 3 or more Volts ensures that this does not happen. This is one of the reasons that many designers use two varicaps back-to-back, as this reduces the RF voltage across them by a factor of 2, but also reduces the capacitance by a factor of 2.

The other thing to keep in mind is the stray capacitances in the circuit. In a typical Colpitts oscillator, you might have 15 pF or more which forms the capacitive divider plus any additional capacitance from the transistor used. The J310 used in the circuit above loads the inductor with around 6pF. If we add stray capacitance in the circuit, let's assume that makes about 10pF in total. Using your curve, and assuming we don't go below 3V bias and let's also assume we are using a single varicaps (or 2 pairs of 2) and a 100nH inductor:

• With 3V bias we get a total capacitance of 36 pF (10pF stray plus 26 pF from the varicap) so the resonant frequency is ~85 MHz
• With 10V bias the total capacitance is 23 pF and the resonant frequency ~105 MHz
• With 30V bias, the total capacitance is 15pF and the resonant frequency ~130 MHz

Note that the frequency range from 3 - 10 V bias is 20 MHz (85->105) whereas the frequency range from 10 - 30 V bias is 25 MHz (105->130).

Whilst a 3-10 V swing might just work, having a wider voltage swing improves the frequency range. The more significant the stray capacitances in the circuit, the more important it becomes. It's not impossible to cover the whole 88 - 108 MHz range with a lower voltage swing (several PLL do this) but a wider voltage swing is useful.

[Albert H]

The HUGE problem with high sensitivity varicaps is HUMMMMMM. If your varicap tunes 87.5 - 108 MHz with a 3 - 10V swing, you're only going to need a couple of millivolts of modulation for 75 kHz deviation. If you're modulating the same varicaps as you're using to VCO tuning, you'll have terrible problems keeping it quiet and stable. Roger (Broadcast Warehouse) found this out the hard way when he moved to an "at-frequency" exciter. When he was using frequency multiplication, the sensitive bits of the VCO were working at a different frequency to the nearby output power amplifier, so there was little or no proble,. When he moved to "at-frequency", the exciter output had fewer other products (from the multiplication) but he could never entirely eliminate the hum.

My approach, on the PLL Pro IV that we were working on just before Stephen died, was to switch fixed capacitors into the VCO, using logic lines from the PLL programming to switch in capacitors as you went down the band. This allowed us to eliminate the VCO trimmer capacitor, but greatly increased the complexity of the circuit.....

Most competently designed commercial exciters use high bias voltages on their varicaps. The Bext box I've just mended had a 36V supply for the varicap bias, and used frequency multiplication too. It had a remarkably quiet carrier, and the deviation was highly linear, leading to extremely low distortion. It's one of the best commercial exciters I've seen for a while!