30 November 2009

A classic VXO with a CMOS gate

I have been captured by the fact that a TTL/CMOS gate can serve as a XTAL oscillator, as well as frequency multiplier. That was few years back. Now my crazy idea of building a 6m FM transmitter took me into the world of homebrew voice communication, whereas I have always limited myself to CW.

The CMOS oscillator is available on all datasheets and seems to be pretty reliable. I have also found on my high school book formulas for computing both loading capacitors and resistor! (in a soon-to-be-published article in this blog)

Once I solved the SixBox microphone amplifier clipping distortion (too much amplification), I had to get a proper frequency deviation for 6m NBFM: about 9 kHz on 50 MHz, that's 3 kHz at the XTAL fundamental frequency.

I tried different varicaps, from a BB105 to a IR LED through ordinary 5mm RED LED. In order to measure VXO frequency change I drove the varicap through a resistive potentiometer. The XTAL had a 4.7uH molded choke in series, to reduce the overall oscillator Q. The following diagram plots two configurations:

(The vertical axis is in kHz, while horizontal is the reverse voltage applied to the diode.)

"Config A" had 115pF load plus the BB105 plus the 4.7uH.
"Config B" had 33pF load plus any diode plus the 4.7uH.

In both cases the varicap is in series with a 100pF DC blocking capacitor.

Curiously the same values of "Config B" were obtained with a BB105, a 1N4003 and several LEDs. The usable deviation at the fundamental frequency is about 1.6 kHz, which results in low and distorted modulation at 50 MHz.

In another attempt I repeated these tests without the molder choke and the swing was even less. This effect light up a possible solution: use a larger inductor in series with XTAL!

When I replaced the 4.7uH with a 10uH choke the swing extended to 3kHz at the fundamental frequency, with the same curve as shown in the picture above.

Where do I stand now? I have the ability to obtain a proper deviation at 50.7 MHz but I have to restrict the varicap modulating voltage between 7-8V and 10V.

18 November 2009

Electret microphone output

For my reproduction of the FredBox/SixBox I wanted to know how much voltage output to expect from an electret microphone. Actually my mind is already on a 6m QRP FM transmitter, but the output swing is good to be known.

The oscilloscope comes handy for this kind of measurements. A resistor, capacitor and a surplus electret microphone do the job. Then, speak into it. Oh, don't forget a power supply!

My daughter really enjoyed screaming into the microphone and observing changes on the o'scope screen. I couldn't have tested it without her.

At her loudest scream the microphone outputted 200mV peak. On the DSO she also learned which button would freeze the trace. It was a very instructive experiment for all of us!

16 November 2009

When conditions are good ... P2!

I need to share my luck last Friday when I worked P29CW on 30m CW using the FT-817 at 5W and my homebrew balcony antenna at 8th floor (picture1 and picture2 of the antenna).

I headed to the 30m QRP calling frequency, but some rare station was having a pile-up. Another pile up was found on the 10.12x MHz. So I tuned the band for some lone CQ'er and heard this P29CW at 15-16 WPM, 559 to 599 ... I copied the callsign 10 times since I never heard of such a prefix. I replied and he came back with IZ1Z? I sent my call twice and we exchanged reports.

Since I thought he was some European special prefix I didn't mention I was QRP.

Then I opened a paper on my desk that lists countries without a QSL bureau and there it was: Papua New Guinea. What a night!

That was 13800 km with 5W out. The antenna might have an efficiency of 50%, making it a 0dBi radiator according to some simulation software.

So, real QRP DX is possible when conditions are good: it's just that both propagation and crowd need to cooperate. :-)

Now back to my version of the SixBox.

12 November 2009

FT817 keypad - first quirk solved

I produced a new firmware version that fixes the quirk reported in the previous post (mode not changed when recalling an onboard memory on a "far" band). It now uses a larger delay in the command sequence (200ms instead of 100ms) and the FT-817 seems to be happy with it and behave as expected.

All firmwares released since 2009-November-12 will carry the new delay. Freely available firmware for download will be updated shortly.

Thank to Cristian in YO-land for reporting the non-blocking issue.

10 November 2009

FT817 keypad - first quirk reported

A builder of my FT-817 keypad has reported me a problem using onboard dynamic memories and "far" QSYs. It was reported while testing a 38400 baud keypad version (default is 4800 baud), but the glitch is caused by RTX internals and not the keypad firmware itself.

I could reproduce the same behavior but not deterministically! Try it yourself:
  • write 3700 kHz LSB in onboard memory #1
  • write 145500 kHz FM in onboard memory #2
Set your VFO to, say, 20m band. Recall memory #1, then recall memory #2: it might happen that the mode stays LSB and doesn't change to FM.

Quick fix: recall memory #2 and the mode is set properly.

Cause. I think the internal CPU of the FT-817 is busy doing something else when asking for a large QSY and ignores the mode change command. The 100ms delay between the two is not enough.

Fix: new firmwares include a 200ms delay that seems to make both the operator and FT-817 happy.

Freely available firmware will be updates shortly.

Since this glitch is non-blocking for keypad functions and doesn't hang the radio I will not issue a recall statement for already shipped chips.

04 November 2009

Doubling to 6m and filtering

Previous measurements on my base oscillator for a simple 6m TX were encouraging and I started thinking of making a DSB transmitter instead of AM. I added a parallel LC circuit at the output of my EX-OR mixer/doubler and started seeing an improvement on the oscilloscope.

Then I checked at the spectrum analyzer and results were encouraging.

Frequency After LC [dBm] Delta [dBc] Before LC [dBm] Delta bef [dBc]
50,35 7,5
75,53 -29,5 -37

100,7 -14 -21,5 -6 -11
125,88 -17 -24,5

151,05 -33 -40,5

The 50 MHz output on 50 ohm has increased to 7.5 dBm (through the 7dB pad means 14.5 dBm = 28 mW) and I can tune the delay line to almost kill the 75 MHz product while keeping all other harmonics more than 20 dB below the carrier. Before adding the LC the second order harmonic was down 11 dBc, another positive improvement.

Now I have too many options:
  • try it on the air as a CW transmitter
  • build a PA and AM-modulate it
  • feed this signal to an NE602 and generate DSB
In any case I need to start working on the low frequency part.