Feedthrough capacitors in XXI century

Since I am planning to go QRO with my 4m CW TX (5W RF out), I thought it would be a good precautional idea to shield the main oscillator. So, if it is going to be boxed, I need a way to let allowed electrons in and out:

• power supply
• RF out (more a square than a sine wave)
• frequency control (capacitor or varicap)

Remembering projects seen in the past two decades, I imagined feedthrough capacitors would be ideal interfaces for my signals. But, which values amongst those that I could easily locate? 1000pF, 150pF, 22pF.

The DC line does not mind being decoupled to ground with 1nF, so that one would do. As an alternative, I have been suggested to simply drill a hole through the shield and decouple +V to ground with 100nF. Plain simple and equally effective.

The RF out would be totally shorted to ground by a 22pF. And 1000pF imposes an unacceptably high capacitive load to my oscillator. Alternatives are:

• use a passthrough RF connector, like SMA
• use a small piece of RG174 coax with the braid soldered to the shield/ground
• make the 22 or 150pF part of an output filter

Given that my circuit needs a square wave signal rather than a sine wave, I will probably use the short piece of coax.

Last but not least, the frequency control. Not sure whether this will be a variable capacitor or a varicap. In the former case I could install the capacitor inside the shield and let out only the component shaft. The varicap control voltage instead is DC, so the same rule as DC line applies (standard decoupling).


(My) Conclusion. Feedthrough capacitors are a useful tool that is becoming harder and harder to source. Practical alternatives are available to the hobbyst with a little imagination (or guidance from The Wise Ones), so don't give up!

Calibrated signal generator at harmonics

Even the XG2 manual mentions that the circuit produces useful harmonics at predefined levels. Using the 14.000 MHz XTAL and the IC706MKiiG as a test instrument I recorded the following:

1. 14 MHz: S9
2. 28 MHz: S8
3. 42 MHz: S6
4. 56 MHz: S3
5. 70 MHz: no signal, barely audible

I cross-checked 2nd and 3rd harmonic readings with FT-817 too.

Single frequency calibrated signal generator

This piece of lab instrument was built after the curiosity to measure (... estimate ...) the sensitivity of an homebrew receiver. As described in a previous post about measuing RX sensitivity with simple tools, I tried to reproduce the Elecraft XG2.


The local shop did not have the 1.22V zener diode and a couple of 1% resistors were not available, but in the end I could get it working with a LM317T regulator at minimum voltage, giving about 1.268V. Having an higher supply voltage means that the generated signal is slightly more powerful than advertised (according to my calculations the difference lies within 1dB).

I used a single 14.000 MHz XTAL which resonates steadily at about 14.0045 MHz.

The completed circuit on a piece of recycled copper-clad board was powered with two exhausted CR2032 button cells (theoretically 3+3 = 6V). Current drain from the battery measured 4.98mA load independent, which is 20x higher than XG2's 250microA but still acceptable for portable use (just in case).

When feeding my commercial receivers with the generator output I could read the S9 signal, a bit on the high side, FT817 indicating S+ few times. Is that caused by the 48mV higher supply voltage, some "hand" effect increasing the output coupling or a calibration error on the receiver's meter?

I took only one test at S1 level and my FT817 went down to S0: need to cross check the result with other receivers.

Another measurement done was to check the signal level on 14 MHz harmonics, which I will publish in a future post. With the 14 MHz XTAL the S9 signal is readable up to the 4th harmonic at 56 MHz

NB: FT-817 conditions were IPO ON (no antenna preamplifier), ATTenuator OFF and NARrow filter OFF.

New frequency formatting for I.F.R. - dial-in 100 Hz

An update to yesterday's post. I fixed the frequency visualization when the operator is supposed to type the 100 Hz value too (it is a user configuration option).

"Small" 0's are the "locked" digits, all other can be typed in.

I feel a new I.F.R. demo video is needed.

New frequency formatting for I.F.R. - dial in

After fixing the frequency display with dots and right-alignment on the Interactive Frequency Reader, it was time to improve frequency dial too.

The picture below shows how it looks like when typing 145'550 kHz, with the trailing zero still awaiting for key pressure. Last three zero's are shown slightly differently since they cannot be modified.

I am also working on a meaningful way to show the frequency being typed when the 100 Hz digit is also required (a configuration parameter allows to enable/disable 100 Hz dial in).

Measuring RX sensitivity with simple tools

Yesterday I had the 'scope ON and it occurred to me that I could visualize the output of a simple 20m CW RTX I had designed and built back in 2005.

While it measured 100mA at 12V input key down, I could see a relatively well shaped sinewave of 7,4Vpp, which should equate to ~140mW on 50 ohm (ouch! That's 11% overall efficiency!).

Later on, I wondered if and how I could measure the receiver sensitivity, which I remember sounding good in my ears. I did some internet searches and I would need a 14 MHz source in the ballpark of  -100dBm. My lowest power generator is FT-817's 500mW, +27dBm: 130dB is too much attenuation to be implemented! Even commercial fixed attenuators don't go beyond 30dB (source JFW catalogue). Should I  build a Colpitts oscillator and compute its output power with the oscilloscope? Are there other techniques to generate really low power signals? I have the advantage that I need a single frequency...

Fellows over at GQRP reflector suggested three already tested circuits/products:

• Elecraft's XG2 (-107 and -73dBm)
• Norcal's S9 (-107 and -73dBm)
• a test generator published on SSDftRA book

Both commercial kits documentation include schematic and parts list. They are also quite similar, so I will go with either circuit, depending on parts I can locate. I will tend to use 1% precision parts whenever available.

To complete the test, on the other side of the receiver I would probably use a computer software doing audio spectral analysis and look for a 20dB S+N/N signal. Or a DVM.


More ideas came up around the little RTX too...