06-02-2019, 01:38 PM
For the sort of DDS applications that Frank and I are using the absolute performance isn't terribly critical. If we really need better performance we've probably got the resources to do it by the brute force method of wider and longer trigonometric tables.
The great thing about using a well designed core such as the Xilinx one is that you get various tricks to improve performance at the click of a mouse. If you need low spurii then the taylor series corrections and trading off spurs for noise are things that I'd rather not have to design unless it was really necessary.
Yet another approach is to buy a commerically available software defined radio. Trevor has done some work here, mainly on the receiving side I think. The module he used isn't cheap (over £100) but it's very powerful and a great prototyping platform.
The great thing about using a well designed core such as the Xilinx one is that you get various tricks to improve performance at the click of a mouse. If you need low spurii then the taylor series corrections and trading off spurs for noise are things that I'd rather not have to design unless it was really necessary.
Yet another approach is to buy a commerically available software defined radio. Trevor has done some work here, mainly on the receiving side I think. The module he used isn't cheap (over £100) but it's very powerful and a great prototyping platform.
www.borinsky.co.uk Jeffrey Borinsky www.becg.tv
