Git-Mirrors/portapack-mayhem
1
0
Fork 0
mirror of https://github.com/eried/portapack-mayhem.git synced 2025-08-02 11:36:38 -04:00
Code Issues Projects Releases Packages Wiki Activity

Oversample (#1336)

Browse source 
* WIP Oversample cleanup

* WIP

* WIP

* WIP dynamic interpolation

* WIP cleanup

* Fix math errors

* Add some optional assertions

* Add support for x32 interpolation

* Update proc_replay.cpp

Typo
This commit is contained in:
Kyle Reed 2023-08-02 12:59:26 -07:00 • committed by GitHub
parent e2ad0a1b1a
commit 37386c29cb
No known key found for this signature in database
GPG key ID: 4AEE18F83AFDEB23
20 changed files with 272 additions and 169 deletions
Download patch file Download diff file Expand all files Collapse all files

92
firmware/baseband/proc_replay.cpp
View file

@ -41,43 +41,75 @@ ReplayProcessor::ReplayProcessor() {
baseband_thread.start();
}
void ReplayProcessor::execute(const buffer_c8_t& buffer) {
/* 4MHz, 2048 samples */
// Change to 1 to enable buffer assertions in replay.
#define BUFFER_SIZE_ASSERT 0
void ReplayProcessor::execute(const buffer_c8_t& buffer) {
if (!configured || !stream) return;
buffer_c16_t iq_buffer{iq.data(), iq.size(), baseband_fs / 8};
// Because this is actually adding samples, alias
// oversample_rate so the math below is more clear.
const size_t interpolation_factor = toUType(oversample_rate);
// File data is in C16 format, we need C8
// File samplerate is 500kHz, we're at 4MHz
// iq_buffer can only be 512 C16 samples (RAM limitation)
// To fill up the 2048-sample C8 buffer, we need:
// 2048 samples * 2 bytes per sample = 4096 bytes
// Since we're oversampling by 4M/500k = 8, we only need 2048/8 = 256 samples from the file and duplicate them 8 times each
// So 256 * 4 bytes per sample (C16) = 1024 bytes from the file
const size_t bytes_to_read = sizeof(*buffer.p) * 2 * (buffer.count / 8); // *2 (C16), /8 (oversampling) should be == 1024
size_t bytes_read_this_iteration = stream->read(iq_buffer.p, bytes_to_read);
size_t oversamples_this_iteration = bytes_read_this_iteration * 8 / (sizeof(*buffer.p) * 2);
// Wrap the IQ data array in a buffer with the correct sample_rate.
buffer_c16_t iq_buffer{iq.data(), iq.size(), baseband_fs / interpolation_factor};
bytes_read += bytes_read_this_iteration;
// The IQ data in stream is C16 format and needs to be converted to C8 (N * 2).
// The data also needs to be interpolated so the effective sample rate is closer
// to 4Mhz. Because interpolation repeats a sample multiple times, fewer bytes
// are needed from the source stream in order to fill the buffer (count / oversample).
// Together the C16->C8 conversion and the interpolation give the number of
// bytes that need to be read from the source stream.
const size_t samples_to_read = buffer.count / interpolation_factor;
const size_t bytes_to_read = samples_to_read * sizeof(buffer_c16_t::Type);
// Fill and "stretch"
for (size_t i = 0; i < oversamples_this_iteration; i++) {
if (i & 7) {
buffer.p[i] = buffer.p[i - 1];
} else {
auto re_out = iq_buffer.p[i >> 3].real() >> 8;
auto im_out = iq_buffer.p[i >> 3].imag() >> 8;
buffer.p[i] = {(int8_t)re_out, (int8_t)im_out};
#if BUFFER_SIZE_ASSERT
// Verify the output buffer size is divisible by the interpolation factor.
if (samples_to_read * interpolation_factor != buffer.count)
chDbgPanic("Output not div.");
// Is the input smaple buffer big enough?
if (samples_to_read > iq_buffer.size())
chDbgPanic("IQ buf ovf.");
#endif
// Read the C16 IQ data from the source stream.
size_t current_bytes_read = stream->read(iq_buffer.p, bytes_to_read);
// Compute the number of samples were actually read from the source.
size_t samples_read = current_bytes_read / sizeof(buffer_c16_t::Type);
// Write converted source samples to the output buffer with interpolation.
for (auto i = 0u; i < samples_read; ++i) {
int8_t re_out = iq_buffer.p[i].real() >> 8;
int8_t im_out = iq_buffer.p[i].imag() >> 8;
auto out_value = buffer_c8_t::Type{re_out, im_out};
// Interpolate sample.
for (auto j = 0u; j < interpolation_factor; ++j) {
size_t index = i * interpolation_factor + j;
buffer.p[index] = out_value;
#if BUFFER_SIZE_ASSERT
// Verify the index is within bounds.
if (index >= buffer.count)
chDbgPanic("Output bounds");
#endif
}
}
spectrum_samples += oversamples_this_iteration;
// Update tracking stats.
bytes_read += current_bytes_read;
spectrum_samples += samples_read * interpolation_factor;
if (spectrum_samples >= spectrum_interval_samples) {
spectrum_samples -= spectrum_interval_samples;
channel_spectrum.feed(iq_buffer, channel_filter_low_f, channel_filter_high_f, channel_filter_transition);
channel_spectrum.feed(
iq_buffer, channel_filter_low_f,
channel_filter_high_f, channel_filter_transition);
txprogress_message.progress = bytes_read; // Inform UI about progress
// Inform UI about progress.
txprogress_message.progress = bytes_read;
txprogress_message.done = false;
shared_memory.application_queue.push(txprogress_message);
}
@ -90,8 +122,8 @@ void ReplayProcessor::on_message(const Message* const message) {
channel_spectrum.on_message(message);
break;
case Message::ID::SamplerateConfig:
samplerate_config(*reinterpret_cast<const SamplerateConfigMessage*>(message));
case Message::ID::SampleRateConfig:
sample_rate_config(*reinterpret_cast<const SampleRateConfigMessage*>(message));
break;
case Message::ID::ReplayConfig:
@ -110,9 +142,11 @@ void ReplayProcessor::on_message(const Message* const message) {
}
}
void ReplayProcessor::samplerate_config(const SamplerateConfigMessage& message) {
baseband_fs = message.sample_rate;
void ReplayProcessor::sample_rate_config(const SampleRateConfigMessage& message) {
baseband_fs = message.sample_rate * toUType(message.oversample_rate);
oversample_rate = message.oversample_rate;
baseband_thread.set_sampling_rate(baseband_fs);
spectrum_interval_samples = baseband_fs / spectrum_rate_hz;
}