Pan adapter waterfall slightly off?

Alan - KA4B

I have noticed when using WWV to check calibration that the waterfall display appears to be about 5 Hertz higher then the smoothed spectral display.  You really have to zoom in to see it, but it appears consistent.  Has anyone else observed this?  I am using the latest version of SSDR.

Walt - KZ1F

Did you opt for the gpsdo option? If you Google Flex wwv calibration you'll find lots of threads on this subject. It does come up periodically.

Alan - KA4B

No Walt.  The problem is not the calibration, but that the waterfall display appears to be shifted about 5 Hertz to the right of the spectral display.  It is easiest to see if you average the spectral display and zoom all the way in.  the gpsdo should not have any effect on this.

Mike NN9DD

I have seen that but only when using cw skimmer. I quick shake of the pan adapter trued them up

Mike
N9DFD

Walt - KZ1F

Sorry, misunderstood. However, the news gets better. The waterfall, even though the control width is the same as spectrum display the geometry is different as far as the tile presented. The tile is the collection of data points that span, or otherwise need to be mapped to pixels. What happens is you get a pixel that straddles two tiles entries. To marry the two displays is an approximation. Google Flex, 'Stu Phillips' waterfall. It will point you to a thread Stu did on processing waterfall tiles. You could try adding tile to the search. I, and some others, wrote software to process data from the radio. Stu was likely the first non Flex software person to do that. It gets a tad convoluted but he did a reasonable job explaining it. I do recall that was something Flex worked on to improve. I suspect 5Hz is about as spot on as it'll get.

Walt - KZ1F

This is it: https://community.flexradio.com/flexradio/topics/generating-a-pan-adaptor-and-waterfall-display

Alan - KA4B

Mike, can you clarify what '1 quick shake of the panadapter' is?  Are you joking?

Alan - KA4B

Walt, it appears if a person wished to use WWV for checking calibration they would need to use the peak in the spectral display and ignore the waterfall, because the waterfall is a secondary source and may be off.  Would that be your interpretation?

Mike NN9DD

No it actually worked or,was perceived to work with cw,a few weeks ago. What I mean is. "Click in the pamadapeter and hold down the button and shake the mouse back and forth a few times". It really did work I was running cw skimmer and bridge 1.6 and the current version of SSDR. It was 3 or more weeks go Mike N9DFD

Walt - KZ1F

I think I have to agree with that Alan. The waterfall intent is to reflect relative strength of a signal compared to noise or adjacent signals. Through interpolation it pretty much matches up to the spectrum display above it but pretty much isn't the same as perfectly.

DrTeeth

@ Alan KA4B

I did notice something similar a while back. Is this https://community.flexradio.com/flexradio/topics/minor-graphical-inconsistency similar to your issue?

Walt - KZ1F

Yes, that is the less technical answer Guy. One is what. the other is why.

Alan - KA4B

Guy, it is the same in principle, but what I was looking at was the center frequency on WWV on the panadapter and the center of the red line below on the waterfall.  They were different by about 5 Hertz when fully zoomed in.  As near as I can tell the center of the signal on the waterfall seems to be 2 pixels to the right of the center of the signal on the panadapter. At normal width of about 300 kHz that represents an error of about 300 Hertz.  So tuning using the waterfall is not very precise in this case.

Alan - KA4B

Kevin, notice that your yellow line is two pixels to the left of the center of the red line in the waterfall below.  This is my point.  It seems to always be that way. Often the waterfall makes it easier to fine tune a signal.  But you will always be 5 to 300 Hertz high (depending on zoom level) if you do that.

Tim - W4TME

This behavior is due to calculating the display width and having to round with an even display width.  We are aware of the issue and it is in our bug tracker.

James Del Principe

Five Hertz? Two pixels? You guys are killing me. Radios used to have 5 KHZ calibrations. A Collins had 1 KHZ which was a miracle. How about if you just say "This Flex is a wonder" and go work an ATNO. Anyway, thanks for making this old guy laugh. Very Best 73, Jim

James Del Principe

Not sure of my point?  OK, I was speaking tongue in cheek. You see the Flex, in my opinion, is just a wonder. It amazes me that it can perform so very well. My 6500 is head n shoulders better than any rig I have ever owned and that is a lot of radios. To say a signal is off by 5 Hertz or two pixels is so different than my personal experiences over the 53 years of Ham radio. I just closed out a 46 year career in electronics - 4 years in commercial military radio and 42 years in cardiovascular x-ray. I have seen electronics go from relay logic (we called them Potter processors) and vacuum tubes to multiple industrial computers that directly convert x-ray to a giga bit stream to process in ways that would make your head spin. Those systems cost over one megabuck each and up. My Flex, while not quite pocket change, was really cheap compared to some of the radios out there.......and in closing, at my age, I cannot see two pixels.   So pardon me if I chuckle when someone says his display is off by 5 HZ or two pixels.   Now, do you see my point?   Do we still have a sense of wonder for what is going on with this amazing device?   73, Jim

James Del Principe

Kevin, never lose your sense of wonder.  It will lead you through a lifetime of adventure.   Jim

Alan - KA4B

James there is no question that this is a fantastic radio.  I wouldn't have anything else as my primary radio.  However, as I said above, if you are looking at 300 kHz of bandwidth those two pixels represent 300 Hertz.  If you have it set for 150 kHz across the screen, then those two pixels represent 150 Hertz.  I have noticed the discrepancy many times when trying to see if I am on the same frequency with others on the net.  You can easily hear if you are only 30 Hertz off. The waterfall would be the easiest way to visually see what frequency people are on if it were synchronized with the panadapter. As it is I have to zoom all the way in to get close, and then zoom back out.  Its not a big complaint, just a nit that I have observed many times.  I appreciate all the comments.

Walt - KZ1F

Alan, this is a math problem. A panadapter can see 14MHz of spectrum, if I recall the specs. If your monitor has 1280 pixels along the x axis, each pixel is 11KHz, correct? The is nothing FRS can do about that, it's arithmetic. As you narrow the bandwidth the Hz/pixel ratio improves. That's the display. What frequency the slice is sitting on is absolute, I.e. 14,100,000Hz. And that assumes you have a well calibrated unit. That the fpga bins are structured differently is an idiosyncracy of the hardware, giving the spectrum geometry slightly different than the waterfall geometry. As Tim said, it's on their list.