Want to learn more about Ham Radio?
SUBSCRIBE to the Ham Radio DX channel
Further to previous posts about the KL Reference 12MHz to 10MHz and Simulcast Experiments, I had finally got around to purchasing a BG7TBL GPS receiver module due to some motivation in a video by Joe KC2IRV who is doing the same thing.
First order was to modify the KL.... I had a spare 6M conversion laying around which I decided to modify further for this purpose.
The original 12MHz TCXO was removed.
7DB low power mods were then performed. I had decided to do this as potentially the repeater may be located at a solar site. It also removed all the congestion of the main board. I removed the receiver and decided to make this a TX only unit.
There is a bit going on here. The voter board is on the left. Next on the right, is the 9.6MHz OCXO needed to keep the clock into the RTCM/Voter stable. The other OCXO is the 10MHz which I relocated into the box.
Wired in the Allstar RTCM/Voter board. Fairly straightforward as all that was required was power, discriminator audio, transmit audio and PTT. Squelch action is done in software. I may need to run CTCSS decoding in the future, so there is an option for that. The RTCM/Voter board can generate CTCSS encode in software too. I had issues with getting enough TX audio level out of the board on my first trial. The voter board can generate enough TX audio level to drive the radio to 3KHz deviation, but is maxed out at around 4KHz using the 5-Tone modulation input.
On advice from another who uses the RTCM I ran the TX audio into the Encode input of the TX module. I also shorted R340B (27K) in series with this line to remove attenuation and probably some high frequency roll off and so that the audio is direct to the modulator. The Allstar software takes care of pre-emphasis and deviation limiting.
I then programmed the EPROMS for 10MHz reference use as documented earlier. This proved to be straightforward after writing down what had been done previously.
I placed this transmitter at one of the sites and ran my test unit on my bench at home on low power to create an overlap area close to my house. 6 metres propagates well, so I had to turn the power down on the exciter to around 40mW. I zero beat the transmitter on the bench to match the one on the mountain. When generating a CTCSS tone, I could hear what seemed to be the two tones beating out of phase with one another. Audio was distorted and or chopped up in some parts of the overlap, especially when stationary. I then had a look at the time delay between the transmitters and my receiver. Radio travels at the speed of light, or 5.4 microseconds per statute mile (1.6km). It worked out that I was really close to my TX at home, but far from the TX on the mountain (20km). This is a delay of 67.5uS. Above a delay of 83uS with a 1KHz tone, things start to degrade. At 2KHz the max is 42uS.
You can read more about the theory here - http://www.simulcastsolutions.com/userfiles/file/simulcastforums/technology_Simulcast_Theory_F20.pdf
The VOTER boards have a Simulcast Launch delay function which enables a delay of one or more of the transmit audio times to suit users in an overlap area to make sure the audio arrives in phase and in time. I adjusted the delay of the TX unit at the mountain site to around 330 (67.5uS). This made some improvement, but the audio was still tearing and poor in some areas.
Next I decided to have a look at the audio response of the KL unit on my bench from 300Hz to 3000Hz. All audio needs to be in time, but also in phase, deviating up and down the same on each transmitter. I setup a reference tone of 1KHz to equal 3KHz of deviation and plotted the response. These are the results.
I made a mistake in my initial measurements. The low end of the audio frequency range deviates higher. This is because my test sets HPF was set to 50Hz, so an accurate measurement was not made. But it is corrected later - further on that soon. The CTCSS Input was my audio input, used on both transmitters at the time. As you can see it rolls off rather sharply after about 1800Hz. There is a LPF after the clipper where the audio was being injected, so I decided to adjust this to see if I could get a better response. I could. You can still see some rolloff after 2300Hz, due to the LPF, so it wasn't perfectly flat. Any variations here between the transmitters would cause additional distortion.
I then modulated directly on top of the varactor diode on the VCO board. You can see pretty much a flat curve. I then redid these measurements with the proper HPF selected in the test set. In addition, I also lifted R403 on the VCO board and injected audio there. I couldn't modulate from the VOTER board directly onto the diode as it caused the VCO to fall out of lock. The level was also way too high anyway, therefore injecting at R403 gives me a bit more range to set levels. Schematic below of this.
As you can see, low frequencies are more flat now - although there is some variance due to the HPF being directly at 300Hz in the HP8924c. The level should stay constant (within 10%) of 3KHz deviation across the whole audio frequency range. As you can see it rolls off a little bit more that 2700Hz at a 3K tone. For now, I'm modulating at R403 as it's much flatter, and will be easier to get the other transmitter matched to this. That is the next step and is where I am up to on this project.