Mike, thank you for commenting on much of the development and filling me in, on which code has what improvements, etc .? ?I wish I had the energy and time to learn the github stuff....but haven't yet.? ?Can only hope that the people who are working there READ the work that others are doing....such is life.
?
OK, I have tried to analyze the hardware absolute limitations that should be covered by the software delays in order to maintain the operation that Ashhar set up for going from receive to transmit without damage.? ?My full analysis is attached.
?
Basically
1.? ?In order to have less than 25% degradation of initial dot, @ about 25 wpm we need to get the turnon delay down to about 15 ms or less.??
2.? ?I don't know about the time requried to turn of channels in the codec, and the delays he put in are minimal, so leave alone.
3.? ?Driving the TX line high causes power to be applied to several lower power systems, and also to the PA bias circuitry, which might be the slowest part of the entire thing, and had a time constant of approx 0.5 millisecond.? ?I did some calculations to estimate the time to turn on Q23 fully, and it is down in the microseconds, so that doesn't seem to be a problem.? ?Providing 5 milliseconds here should be sufficient.
4.? ?Re- setting the Low Pass filters requires discharging as much as 6000 pf through the high voltage Mosfets, but they have an on resistance of about 5 ohms, so the time constant for that is in the tens of microseconds -- so again we don't need much time, maybe 5 milliseconds more.
?
With all that we should be able to ahve a turn on set of delays that add up to about 13 milliseconds, and that would be a very nice turn on.
?
I can't TEST it properly until I get a replacement IRFZ24N installed -- I don't have any with me at the vacation home because never anticipated this damage.? ?Some are coming from Amazon on Saturday just before we leave to drive back to Gainesville FL.? ?However, I can try the code just to see if anything happens with no real output power, later today.? ??
?
With these two fixes combined, I think it will be a HUGE improvement for CW ops up to about 30wpm which is the vast majority.? ?And so far implemented in the 32 bit code, but the same ideas should work in the 64 bit code that I examined.? ?I have all kinds of special stuff in my code so not sure I want to completely switch over just yet.? ?
?
Gordon KX4Z
?
?
