While writing and debugging a Frequency Reader firmware with support of a tuning knob, the test-team found out that the FT817 was not able to keep up with fast spins. When the external knob spins, a new frequency is sent over the serial CAT interface. If data is sent too often, the FT817 ignores it. Setting baud rate to 38400 instead of 9600 did not show any improvement.
The annoying part of this behavior is that the FT817 is "left behind" if the external knob is spun fast and abruptly stopped. This causes a mis-alignment of the external display and the actual tuned frequency. Any re-alignment software procedure would either re-tune the radio or change the external display, which does not look nice.
So I came up with a new software routine that does not send data too often, but it uses the fast rotation to increase the step (right now the increase is linear). But but but: how fast does FT817 accept a new frequency, without loosing a bit?
So I wrote a simple firmware that sends 200 frequency increments at a steady rate, each loop with a shorter delay between updates. Starting at 100ms (+ some uC time) down to 0ms, the "safe self-imposed delay" was determined to be 90ms (+ uC data processing, less than 1ms). Please note that this delay within the radio CPU applies to all those computer programs that interact with the FT-8x7 via the CAT interface (excellent HRD probably being the most popular).
Given the result of this simple experiment, the Frequency Reader Knob will send updates at a safe rate of 1 every 100ms. And vary step accordingly.
The "CAT reaction speed test" firmware is based on the Frequency Reader circuit, it requires at least an ATmega48 chip and it is available on request for those that want to repeat the test on their equipment (FT-8x7).