I checked.
On purpose I let the resolution filters overlap somewhat? so this helps the interpolation.
There is no I/Q signal and FFT but 57 resolution filters ranging 2.4kHz to 600kHz and then a log? detector chip delivering a single dBm value.
Have a look at the technical desciption in the wiki:?

That little bit of code I put here only really helps if the carrier of interest is somehow spread across 2 or 3 spot frequencies (bins in the FFT world). Otherwise it has little to no effect at improving the indicated frequency. It's main use is with FFT's, but I thought might help here too.

I haven't looked at the receiver architect in your TinySA, so I don't yet know how you detect the amplitude of the carriers as you scan across the frequency range. Do you just have a single real value given to you from the detector chip, or do you have an I/Q pair which you use to create the power spectrum lines I wonder ?

Is there a schematic available ?

Mines been working OK, though have noticed a couple of firmware bugs at times. I really like the unit, extremely handy for RF sniffing while out in the car, interesting what you come across at times ;)

On Sat, Aug 15, 2020 at 05:41 AM, Erik Kaashoek wrote:

Integrated including using the calculated frequency i.s.o. the frequency array when printing
Will be in next FW release

FW released, version v1.0-52

On Sat, Aug 15, 2020 at 04:26 AM, OneOfEleven wrote:

You can get a better frequency estimate for a marker that is set to "Tracking the peak". I modified this bit with some extra code to estimate the actual carrier frequency .

Integrated including using the calculated frequency i.s.o. the frequency array when printing
Will be in next FW release

Line should be ..

if (idx > 0 && idx < (sweep_points - 1))

You can get a better frequency estimate for a marker that is set to "Tracking the peak". I modified this bit with some extra code to estimate the actual carrier frequency ..

=========
sa_core.c line 2070

????? while (m < MARKERS_MAX)
????? {
??????? if (markers[m].enabled && markers[m].mtype & M_TRACKING)
??????? {?? // Available marker found
????????? markers[m].index = max_index[i];
????????? markers[m].frequency = frequencies[markers[m].index];

????????? { // calculate the predicted frequency - OneOfEleven .. parabolic method (their are other methods)
??????????? const int idx????????? = markers[m].index;
??????????? if (idx > 0 && idx < sweep_points)
??????????? {
????????????? const float y1???????? = actual_t[idx - 1];
????????????? const float y2???????? = actual_t[idx + 0];
????????????? const float y3???????? = actual_t[idx + 1];
????????????? const float d????????? = 0.5f * (y1 - y3) / ((y1 - (2 * y2) + y3) + 1e-12f);
????????????? //const float bin????? = (float)idx + d;
????????????? const int32_t delta_Hz = abs((int64_t)frequencies[idx + 0] - frequencies[idx + 1]);
????????????? markers[m].frequency?? += (int32_t)(delta_Hz * d);
??????????? }
????????? }

????????? m++;
????????? break;????????????????????????? // Next maximum
??????? }
??????? m++;????????????????????????????? // Try next marker
????? }
????? i++;

=========

It doesn't yet use the calculated carrier frequency on my TinySA that this calculates as the display is still getting it from the frequencies[] array rather than the marker[].frequency value. I'll leave that to Erik if he so wishes to use the code.

But this would display a more accurate frequency that simply using the spot frequency from the frequencies[] array.

On Fri, Aug 14, 2020 at 01:52 PM, gary wrote:

What does the menu item 'correct frequency' under the 'expert config' menu do? The frequency displayed by the marker is 159.861 MHz for an actual input frequency of 160 MHz. Under the same 'expert config' there is an 'actual power' item. What does it do? The level calibration seems to be a fixed table with interpolation. Is it possible to modify that? The level is consistently off by 0.5 to 1dB.

Herb, Thanks for the good explanation.

Gary,
Thanks for find the ACTUAL FREQUENCY was missing in the wiki. Its added now:?
I added the frequency uncertainty in the FAQ :?

On Fri, Aug 14, 2020 at 01:52 PM, gary wrote:

1. What does the menu item 'correct frequency' under the 'expert config' menu do?
? ??
? ?[I haven't used that setting yet, so I'll leave it to Erik to answer.? I'm guessing it isn't for correcting a marker frequency at full span.]

2. The frequency displayed by the marker is 159.861 MHz for an actual input frequency of 160 MHz.

? [The frequency resolution of the display is dependent on the frequency span and and the display points.? From your marker reading it appears you are measuring the 160MHz input using a 0 - 350MHz span. The tinySA has 290 display points, so for a frequency span of 350MHz each display point is:?350e6/290 = approx. 1.2MHz apart.? The closest points to 160MHz using that increment are 159.861MHz and 161.072MHz.? If you need finer frequency resolution you need to use the 290 display points over a narrower span (i.e center frequency = 160MHz and span = 10MHz).]

3. Under the same 'expert config' there is an 'actual power' item. What does it do?

[Expert Config->Actual Power works like Config->Level Cal but allows you to use an external signal generator instead of the built in calibration output as your reference source.? If you want to perform a level cal on the HIGH INPUT then actual power using an external generator is the only way to do so.]

4. The level calibration seems to be a fixed table with interpolation. Is it possible to modify that?

[See the wiki topic???regarding modifying the correction table]??

5. The level is consistently off by 0.5 to 1dB.

?[From the CALIBRATE wiki, "?Directly after calibration the power level display error is expected to be below +/- 2dB".? If you are measuring 0.5 to 1dB off then your measurements are well within the expected display error.

As a comparison the display error for my Wavetek 8541 power meter is +/- 1dB, and I could purchase 15 tinySA's for what I paid for it and the accessory power sensors.? If I need to measure with better than 1 dB accuracy I substitute? a signal from my rf generator at the same level as the signal displayed on the tinySA and record my generator output as the measured level. ]

- Herb

What does the menu item 'correct frequency' under the 'expert config' menu do? The frequency displayed by the marker is 159.861 MHz for an actual input frequency of 160 MHz. Under the same 'expert config' there is an 'actual power' item. What does it do? The level calibration seems to be a fixed table with interpolation. Is it possible to modify that? The level is consistently off by 0.5 to 1dB.
Gary

On Fri, Aug 14, 2020 at 06:39 AM, W1PJE wrote:

One could solve that by perhaps putting a VBW entry in the Wiki and pointing them to the existing averaging feature?

Good suggestion

Thanks, Erik. ?No need to implement VBW - I can use the averaging functions. ?The only final comment is that if someone is coming in from a full-featured SA, they might wonder the same as I do. ?One could solve that by perhaps putting a VBW entry in the Wiki and pointing them to the existing averaging feature?

73
W1PJE

On Thu, Aug 13, 2020 at 08:02 AM, W1PJE wrote:

?Quick UI question: for a normal SA, there is the video bandwidth (VBW) setting independent of the RBW to average down the noise. ?Where is this equivalent function located??

I should have answered: Read the wiki:?

There is no left/right pixel averaging (or filtering) as in a regular sweeping SA but in the LEVEL/CALC menu you can choose AVERAGE 4 or AVERAGE 16
These are both running averaging filters, good are removing noise but they maintain the information of each pixel as these average over sweeps i.s.o. measurement points
It is possible to add a VBW setting but does this add so much value?

And yes, the CALC menu will move under the DISPLAY menu instead of the LEVEL menu in a next FW release

Hi all,

This is a _fantastic_ device. ?I'm having tons of fun. ?Quick UI question: for a normal SA, there is the video bandwidth (VBW) setting independent of the RBW to average down the noise. ?Where is this equivalent function located? ?I'm sure it is in an averaging setting, but having trouble finding it.

73
W1PJE

I did not develop the gui of tinySA-saver
tinySA-saver is the same code as nanoVNA-saver.
The only thing I changed was the hardware.py module where the data returned by the "data"? command is read
You are free to do what you want with tinySA-saver source code as the code is GPL license

tinySA.exe can not run native on linux as it is using dot.NET graph objects and these are not available for mono

开云体育

Hello Erik,
may I ask how you develop the gui of tinySA-Saver?
Did you made the GUI without a Python Gui-builder? I mean native programming...
Isn't it possible to reuse some modules of the tinySA-Saver code?
I would wish the tinySA.exe also for Linux because it is especially created for the tinySA.
Sorry but I love the nanoVNA-Saver for nanoVNA's but not for the tinySA.

best regards

Chris

---
'Everything should be made as simple as possible, but not simpler.' -- Albert Einstein

Dear Hugen and Erik,

Thank you for your reply.
I purchased it from the link on Aliexpress that you taught me.

Thank you for a great project.

Hi Everybody,

I've got the tinySA-saver and the nanoVNA-saver working from the menu on "Andy's Ham Radio Linux" (AHRL) v24c, based on Ubuntu 18.04.5.

I will be releasing this version of AHRL shortly.? For more information, please visit this website:
https://sourceforge.net/projects/kb1oiq-andysham/

Thanks again for all of your help!
73,
Andy
KB1OIQ

It passed all tests, just run them.

Tnx.

72

Dom
M1KTA