KE9NS v3.0 Revision History

Nov 12 2026 v3.0.0.13: (by KE9NS) Update 3D display. Change from isometric to "Forced Perspective" and changed the code from simple array's to a ring buffer. Gradient colors match the gradient Pan option that was already present. A work in progress

Nov 10 2026 v3.0.0.12: (by KE9NS) Update 3D display by allowing you to set a display threshold to help clean up a noise display. Also 3D X,Y,Z, threshold now saved to database. A work in progress.

Nov 10 2026 v3.0.0.11: (by PY2ECM, testingPY5VB) Fix issue with NNR RX2 settings were applied to RX1 as well.

Nov 6 2026 v3.0.0.10 (by PY2ECM, testingPY5VB). Upgrade to latest WDSP 2.11 NNR model 1 x86 with automatic voice/CW models for NR5/NR6 and the existing post-DSP RX1 neural CW audio tap.

NR1 (ANR — Adaptive Least Mean Squares):
Minimum CPU usage, fast. Can introduce artifacts on voice, including hollowing artifacts.

NR2 (Spectral Noise Reduction / EMNR):
Spectral subtraction or estimation MMSE-LSA or OSMS in freq domain. uses FFT to estimate noise floor in bins.
Effective for SSB and AM voice and cleaner than LMS NR1


NR3 / RN2 (RNNoise-based Neural Denoiser):
open source neural network from rnnoise trained on human speech patterns. recognizes human voice.

NR4 / Specbleach (Spectral Bleach / Statistical Denoiser):
spectral subtraction libspecbleach apply fine-grained smoothing and noise rescaling. Favored for SSB and CW on HF. Strips Hash and avoids artifacts.

Warren Pratts NNR (Neural Noise Reduction) NR5, NR6 is machine learning using deep neural network trained on audio datasets to dynamically distinguish human speech patterns from background noise, static, and interfearance. NNR generate a continuous spectral mask. It supresses non-voice frequencies while preserving natural speech formats. NNR excels at removing impulse noise, power-line buzz, lightning crashes, and atmospheric hash. NNR requires higher CPU usage.

NR5 NNR Standard Model (WDSP Deep Neural Network):
WDSP native neural network. Deep learning models to differentiate signals from noise. Offers massive improvements in weak-signal intelligibility with min distortion.

NR6 NNR Premium Model (and CW-Specific Variants):
higher capacity deep neural network specialized CW premium models with deeper layers, more complex weight matrices tailored for rigorous voice recovery or CW tone extraction.

CAT ZZNE (RX1/RX-S) and ZZNF (physical RX2), values 0–6. Legacy NR/ZZNR behavior retained.
Restore the selected NR after DIGU/DIGL, including DIGU → DIGL transitions. Temporary suspension memory does not add cross-restart persistence in digital mode.
Route TrackSun/VOACAP through the existing registered worker/exception boundary and guard the standalone NOAA entry point. Disable VOACAP through a queued UI callback on detected missing network or data/worker failure, with an English status message and shutdown-map.log details. Re-enable manually to retry.

Nov 5 2026 v3.0.0.9: (by KE9NS) Update "3D" panadapter display code and allow for a larger display.
( Setup->Appearance->General->3D x,y,z (change the shape and size), and ->Apearance->Display->3DAlpha and 3DPan color)

Oct 31 2026 v3.0.0.8: (by KE9NS) Add Beam heading to Pan Spots option. (note: you must have PowerSDR logged into LoTW to get the Colors)
If you have PowerSDR console in focus and move your mouse pointer over the callsign and hit the CTRL button, you will move the VFOA to the DX spot.
But if you move your mouse over the last letter of the callsign and hit the CTRL button, your antenna will move to the beam heading show below the callsign (if you have Setup->CAT Control->Rotor Control COM port activated):

Oct 28 2026 v3.0.0.8: (by KE9NS) Fix issue with Real-Time lightning map download.

Oct 27 2026 v3.0.0.7: (by PY2ECM, testingPY5VB). Add WDSP FFT Preperation progress window during initialization. Fixes to BandStack and BandText for IARU2 and Austrialia.

Oct 27 2026 v3.0.0.6: (by KE9NS) Cleaned up Spotter screen layout and added NASA world precipitation map. (Same as v2.8.0.0.340)

Oct 19 2026 v3.0.0.5: (by PY2ECM, testingPY5VB) Upgraded to WDSP 2.10. Add NR5-6 and CW Decoder (under the CWX panel). NOTE: There is a 1 minute initialization after Starting PowerSDR the very first time (PowerSDR will appear frozen for 1 minute).

Oct 15 2026 v3.0.0.4: Fix VOACAP shutdown issue.(by PY2ECM, testingPY5VB)

Oct 8, 2026 v3.0.0.3: Fix issue with RX2 VFOB spoof CAT port not returning TX back to VFOA when done transmitting on VFOB

Oct 8, 2026 v3.0.0.2: (by PY2ECM, testingPY5VB)Fix receive delay after Transmit.

Oct 5, 2026 v3.0.0.1: (by KE9NS) Improve speed of dragging the Pan with VFO slide rule enabled. Disable auto Water/Pan adjust when in Diversity mode.

Oct 4, 2026 v3.0.0.1: Fix issues with the VFO Slide Rule. You can now Left Click and Drag on the Slide Rule to move the VFO. Left Click and Drag on the Pandapter will move the VFO more smoothly. See Setup->Appearance->Display->Display VFO Slide Rule to activate. This works best when you move the Meters to the Top of PowerSDR. See Setup->Appearance->Meter->Meters on Top

Oct 1st, 2026, v3.0.0.0: Based on PowerSDR ke9ns v2.8.0.338 as a starting point + integration of WDSP and DirectX by PY2ECM (testing by PY5VB). Eric PY2ECM has implemented a partial integration of WDSP 1.3 code to take the place of some of the functions of the original DttSP code.
Features: independent WDSP receive channels for RX1, RX-S and physical RX2;
NR1/NR2/NR3/NR4, NB1/NB2 without SNB; shared NR/NB parameter pages;
independent RX2 NR selection and RX1/RX2 NR persistence; AGC/pan/gain,
notch and EQ bridge fixes; complex-IQ diversity before RX1 processing;
Direct2D traces/fills/waterfall with GDI+ compatibility and fallback;
display diagnostics, pacing and audio-progress-based startup FPS guard.

Click this link to view the PowerDSR ke9ns v2.8.0 revision history