Performance of the Maxon CP-530 at 9600 baud
Transmitter Keyup Performance
Much of the gain in moving to 9600 baud packet operation can be lost if the radios are slow stabilize after PTT is asserted. For example, if the radio needs a 300ms TXDelay (which is the case with many synthesized rigs), the TXD is longer than the time it takes to send a 256 byte packet. That means that half the channel time is wasted! Ideally, we'd like a system that allows a TXD of 50ms or less. Most of the changes I've made to the CP-530 radio are designed to achieve this.
After initially modifying the CP-530, I discovered that attaching the BP-9600 modem to the radio would cause a transmitter frequency shift of several kilohertz. This is because the output circuit of the modem ends with a 10k pot (used to set output level) that goes to ground. Since the circuit was DC-coupled, the pot in the modem upset the bias voltage on the varactor diode that serves as the modulator, and that caused the frequency to shift.
The cure was to put a coupling cap in series with the TX data input. That solved the intial problem, but resulted in an equally troublesome transient frequency shift every time the radio keyed up. It took over 150ms for the transmit frequency to stabilize:
Horizontal: 50ms/div -- Vertical: 2kHz/div
The problem is that the bias voltage on the varactor is switched by the PTT line, and is a fairly high impedance (it's fed by a 47k resistor, with 33k to ground). When the radio keys up and the bias goes from zero to a couple of volts, the coupling cap (which goes to ground through that pot in the modem) starts charging and pulls the bias voltage down. Once the cap's charge nears the bias voltage, things equalize and the bias voltage returns to normal. The problem is that the time constant is too darned long for a reasonable TXDelay.
The answer is to modify the circuit so the bias voltage is not switched but rather is present all the time. This avoids the charging cycle, and the results are dramatic (note the different time scale of this shot; 1ms/div instead of 50ms/div!):
Horizontal: 1ms/div -- Vertical: 1kHz/div
Triggered by PTT
Now, the radio is on frequency within a few milliseconds. Time to full power is also very quick:
Horizontal: 1ms/div -- Vertical: uncalibrated
Triggered by PTT
(RF Detector generates negative voltage; lower deflection means more power.)
The radio is thus on frequency and up to full power within 3ms after keying. I don't have a picture yet, but when keyed by the BP-96 modem the data is stable within about 7ms after power is present. The total delay between PTT and clean data is thus just about 10ms.
After modification, the radios will talk to each other with a TXD of 3 (using the Tigertronics BP-96 modem and Baycom software). With that setting, the Baycom software shows 1 flag being received on each packet. With a TXD setting of 4, Baycom shows about 1 or 2 flags being recieved (I wonder if that's correct, or if the software is calculating flags based on 1200 baud timings and is thus off by a factor of 8). Regardless, in the real world, a TXD of 4 or 5 will be appropriate.
This part of the page is under construction.