This is good work. What would make is even more useful is to show your hardware, as the drive strength is just one variable, the parameters of the wiring carrying the quadrature clock signals is another. On one design, I found just the opposite - reducing the drive strength improved the IQ suppression, which I attribute to reducing reflections and mutual coupling between the traces. The QSD/QSE is deceptively simple - the key to performance lies in these details!
@mariovano
Жыл бұрын
Hi Paul! I totally agree about the importance of the drive signal rise time! In the current PCB of my "radiohat" design - which uses a nearly identical QSE, I find that 4ma is enough - but I had to work at the design of the tracks between the 5351 and modulator to get that result. In earlier prototypes with sloppier output wiring, I had a very hard time getting a decent rise time or exceeding 30db suppression. The current board is 4 layer with good ground planes and I manually had to balance the track lengths and routing to get clean operation at 4ma. Keep up your great work... M
@paulh0029
Жыл бұрын
Hi Mario, indeed I also use 4 layers boards with 2 ground planes in the middle. I did not balance (intentional) the tracks but they are more or less parallel. But a good tip. I had a different approach before using a divide by 4 method. But did not get good results with that. Thanks
Пікірлер: 3