CW Filter Stopping Point ?

Steve N4LQ

In Alpha Version for Ssdr 1.4.15 # 953 it states that 0 Hz is now the stopping point for the low side of the CW filter. It does seem to be closer to zero than before however with the bandwidth set for 3.0K I can still hear about 300 Hz on the opposite side of zero. 
I can manually slide the upper side of the bandwidth over to the center-line and eliminate but from the specs it looks like this should not be needed. 
Has this been implemented or am I misunderstanding the intent? 
Note: WIth slightly more narrow filtering the undesired beat disappears. Could this be something to do with the filter skirt overlapping the center?

6500 Ssdr etc.

Tim - W4TME

"Could this be something to do with the filter skirt overlapping the center?"

Yes. Wider filters, like 3kHz, do not have sharp skirts as the narrower filters because (a) they are not really needed and (b) to minimize the signal propagation latency in a wide filter.  Since CW is a timing specific mode, we have to make these considerations.

Paul Christensen, W9AC

This was covered in an earlier thread on CW filter optimization.  K5UA and I had suggested this:

"Keep the filter symmetrical on the CW offset until the filter width approaches twice the CW offset, then any additional filter width gets added to the high side of the filter, just like in SSB mode. Vice-versa for LSB CW mode."

The latest SSDR now does this automatically.  For example, selecting the 2.8K filter with a 600 Hz CW side-tone offset in CW-L should produce a filter that is -2400 Hz and + 600 Hz.  Just the opposite for CW-U. 

On wide CW filter settings, the small amount that's heard (and to N4LQ's point) on the opposite side of zero is likely the result of the current DSP filter tap setting.  On my SunSDR2-Pro, it's possible to increase the tap count to the point where there's absolutely nothing heard on the other size of zero.  As neat as that sounds, the Sun is not a full QSK transceiver.  Increasing the tap count will introduce more latency.  And to Tim's point, I'm not sure if Flex could give us the ability to change DSP taps without significantly upsetting QSK system timing that they're worked so hard to perfect.

Paul, W9AC 

Steve N4LQ

Thanks guys. I'm quiet happy with the filter now. Having the center positioned as before was annoying. Great work.

Sergey R5AU

Hi Paul, the next is not directly in relation to the filters tap
but you mentioned abt SunSDR where interesting, sorry a bit Off-Topic:

IC7800/7851:
https://www.youtube.com/watch?feature=player_detailpage&v=-sVpoPJKgo8
SunSDR2(not PRO version):
http://www.youtube.com/watch?v=BacBq36CNg4

Concerning latency in SunSDR2, Yuri UT4LW  measured the latency in CW mode with 500Hz filter with different DSP taps:

Minimal DSP tap:  35ms
Default: 40ms
Maximum DSP tap: 70ms

Tim - W4TME

But we are talking about 3kHz filters, not 500 Hz in this context.  It takes more time to process a wider filter.

Sergey R5AU

Tim , sure, i wrote a  note "a bit off-topic" = just fyi 
sorry

Tim - W4TME

No problem and no apologies are necessary, Surgery.  Just adding a clarifying comment for those who are reading it. Хорошего дня

Paul Christensen, W9AC

One option to explore at some point down the long SSDR road map is to allow users the ability o change DSP taps by mode, However, fast modes like CW QSK would stay with fixed tap parameters.  To keep users from getting into trouble, a "DSP Tap Default" button could be added in the menu.

I would also like to see the ability to quickly change between CW-L and CW-U by simply "right-clicking" on the CW button in the slice menu.  Each right-click would toggle between CW-L and CW-U.

Paul, W9AC