Finally, some real solid progress on my transverter for 137kHz. It wasn’t by my efforts alone, because at the end of the day this is still my first build of any kind of transmitting device – so there’s a lot to be learnt. Thanks goes out to Owen (VK1OD), Dale (VK1DSH) and Dimitris (VK1SV) for their help and tolerance of my spam.
After completing the build a few weeks back, things just weren’t right. First and foremost, the output from the transverter was not a sine wave.
From here I ended up focusing where it all started way back at the LO and the attenuator for the input from the FT-817. Basically, the LO and RF going into the SBL-1 mixer were way high. General guidelines seem to be that for a balanced mixer you want an LO of 7dBm and an RF of 1dBm. In it’s original config RF was way up at 13.75dBm and LO was up around 9.65dBm.
From here I reduced the voltage to the LO module by a bit more and got the LO closer to 7dBm (slightly below) and built a new attenuator to offer 25.5dB as on low power the FT-817ND was actually feeding in 28.3dBm rather than the 27dBm (500mW) that is commonly believed. Even then though, the design only used a ~15dB attenuator.
So using the online pi network calculator and some nice big 1W resistors from Jaycar I had a nice new attenuator:
From here, although things were better at the start fo the chain, the output was only slighly improved. So I started to focus on the PA module and eventually on the load inductor (L4) for the IRF510. Basically, seeing this was acting ultimately as an RFC I was feeling that it was letting too much of the fundamental frequency through and as a result the secondary harmonic (254kHz) was becoming too strong and destorting the output.
The design used 40t on a iron powder core (possibly a T-106-2, but there’s a bit of mis-information on the schematic), but this seemed to yield a rather low impedance at 137kHz. Seeing this was not being used for tuning etc, it was early on pointed out to me that using an iron powedered core could be a waste and that a ferrite would do. Well, now wanting to also increase the inducatance a ferrite core seemed the way to go.
After some trial and error, I settled on using an 18mm Jaycar Ferrite Toroid with 20t. This definitely showed improvement (as did the previous attempt with just 10t) and the result of the output waveform was:
After this, my thinking led to was there enough load for the IRF to feed into (which in this design, comes in the form of the low pass filter – LPF) and indeed was the LPF doing a good enough job at filtering. After using a few calculators etc. it seemed that the simple 3 pole LPF didn’t really match the frequency and thereby was probably not providing sufficient filtering nor was providing a suitable load.
The plan then came about to replace the LPF with a 5-pole Chebyshev filter in the hope of getting the filter to bring the harmonics to a suitable level and provide a suitable load (hoping to increase the output power). For this I used this online calculator and came up with a design for a 5-pole cheby with cutoff frequency of 150kHz and hopefully a bit over 30dB of attenuation at the second harmonic.
The benefits of this LPF were immediate. The resulting waveform was finally the long sort after smooth sine wave and power output even increased slightly. As a result, I’ve finally got a transverter for 2200m!
As you can see, the power is a bit low at only about 1.6W (not the 8W advertised) but it’s a start. From looking at the PA module it looks like it could definitely do with some work (bit of biasing and coupling/buffering at both the input to the PA and output) but I think now I might move onto trying to setup a usable antenna system for 2200m.