Wednesday, July 2, 2014

Transverter rework

I have spent this weekend to rework the 10 GHz transverter. After one year since I finished it (but only three months of operation) it was time to fix some minor problems it showed.

Annoying sidebands

When I mounted the transverter one year ago, I used a Avenger PLL-based LNB with a RTL2832 dongle to monitor my output signal. The received signal was tremendous, so strong that when I noticed some sidebands around my main signal I didn't get bothered. I thought there where produced by the big signal into the DVB-T dongle...

But one day, in one of the operations, my signal where received into the local WebSDR, and Iban (EB3FRN) send me one screenshot of my signal into the WebSDR waterfall.

My 10GHz signal in Madrid's WebSDR, showing sidebands

If the sidebands were visible on a WebSDR 50km away, obviously I had a problem. I have no 10GHz spectrum analyzer, so I tried to use the DVB-T dongle as a cheap spectrum analyzer. I tested the 10 MHz reference with my HF transceiver, the FT-817 and Agile PLL signal with the dongle. The all were clean. The other multiplier signal could not be checked, because were out of dongle's range, but the sidebands where visible on 10 GHz. Where they came from? Maybe from the 9936 MHz multiplier?

Coincidentally I was testing the output power using a detector diode and (fortunately) an oscilloscope instead a multimeter when I noticed something immediately: The DC output signal from the diode was strongly modulated by a several kHz signal. My 10GHz output was AM modulated by an unknown signal !

Later, I noticed a strong ripple over the DC power line. It was 12 volts, but with an ugly square-like microsecond signal over it almost 2-3 volts peak to peak. I checked power draw. The expected DC current with some 6-7-8 amps microsecond peaks with a few kHz cadence. After a few tests, I found the culprit: The DC-DC inverter used to get 28 volts for the coaxial relay from the 12 volts supply.

The problematic (bad designed?) step-up converter

I tried to filter out this signal but I was unsuccessful. The solution come from Ramiro, EA4NZ, who gave me a similar DC-DC converter, but this one based on the LM2577 integrated circuit. What a difference! This new DC-DC draws less current and do not produce ugly big amplitude peaks on the DC line, just a few mV 400kHz sine wave signal, very easy to filter out to negligible levels.

Now the transverter runs clean, with no sidebands at all.

Local oscillator leakage.

I had local oscillator leakage. Then I installed a pair of pipe cap filters and it was reduced a lot. Then Luis, EA4BGH; gave me a sheet of Eccosorb SF-10.0, a microwave absorber. It's like a sheet of very heavy grey rubber. After gluing it to the internal side of transverter's cover, LO leakage was completely eliminated. Now it is really well under my detection limit, even on Eyal Gal's logarithmic output.

Output power

Every ham I know using one of these Eyal Gal modules affirm they output more than a watt. My unit seemed to be somewhat lazy, because the most I had out of it where 900mW cold, and 600-700mW when warmed.

After playing a bit with the module's gain, and attenuation before mixer, I can confirm the Eyal Gal module can reach almost 2W output (at saturation) and a comfortable 1.2 - 1.4 watts level on linear operation even running hot. In fact, it can do even AM if output is adjusted to best linearity ( 1.2W for my module ).

I have updated all these info on my Eyal Gal entry.

No comments:

Post a Comment