legacy-api gain

[raccoman]

Hi, inside the legacy-api how do I specify the gainToUse?
Cause seems to be set by default on UNKNOWN but I need to change the LNA, TIA and PGA gains.

LMS_APIDevice(lime::SDRDevice* device)
    : device(device)
    , lastSavedStreamConfig()
    , lastSavedLPFValue()
    , statsDeltas()
    , moduleIndex(0)
    , deviceInfo(nullptr)
    , rxGain(lime::eGainTypes::UNKNOWN)
    , txGain(lime::eGainTypes::UNKNOWN)
{
}

API_EXPORT int CALL_CONV LMS_SetGaindB(lms_device_t* device, bool dir_tx, size_t chan, unsigned gain)
{
    LMS_APIDevice* apiDevice = CheckDevice(device, chan);
    if (apiDevice == nullptr)
    {
        return -1;
    }

    const lime::TRXDir direction = DirFromBool(dir_tx);
    auto gainToUse = dir_tx ? apiDevice->txGain : apiDevice->rxGain;

    const OpStatus status = apiDevice->device->SetGain(apiDevice->moduleIndex, direction, chan, gainToUse, gain);
    return OpStatusToReturnCode(status);
}

[rjonaitis]

legacy API provides only single value for setting gain, it internally translates it into a combination of LNA/TIA/PGA values (

LimeSuiteNG/src/boards/LMS7002M_SDRDevice.cpp

Lines 508 to 545 in f8fa599

	OpStatus LMS7002M_SDRDevice::SetGenericRxGain(lime::LMS7002M& chip, LMS7002M::Channel channel, double value)
	{
	LMS7002M::ChannelScope scope(&chip, channel);
	value = std::clamp(static_cast<int>(value + 12), 0, MAXIMUM_GAIN_VALUE - 1);
	unsigned int lna = LNATable.at(std::lround(value));
	unsigned int pga = PGATable.at(std::lround(value));
	unsigned int tia = 0;
	#ifdef NEW_GAIN_BEHAVIOUR
	if (value > 0)
	{
	tia = 1;
	}
	#else
	// TIA table
	if (value > 51)
	{
	tia = 2;
	}
	else if (value > 42)
	{
	tia = 1;
	}
	#endif
	int rcc_ctl_pga_rbb = (430 * (pow(0.65, pga / 10.0)) - 110.35) / 20.4516 + 16; // From data sheet
	// TODO: optimize into single write batch
	if ((chip.Modify_SPI_Reg_bits(LMS7002MCSR::G_LNA_RFE, lna + 1) != OpStatus::Success) ||
	(chip.Modify_SPI_Reg_bits(LMS7002MCSR::G_TIA_RFE, tia + 1) != OpStatus::Success) ||
	(chip.Modify_SPI_Reg_bits(LMS7002MCSR::G_PGA_RBB, pga) != OpStatus::Success) ||
	(chip.Modify_SPI_Reg_bits(LMS7002MCSR::RCC_CTL_PGA_RBB, rcc_ctl_pga_rbb) != OpStatus::Success))
	{
	return OpStatus::IOFailure;
	}
	return OpStatus::Success;
	}

)

With the legacy API a fine grained control of each individual gain could be achieved by directly writing values to the LMS7002M chip registers. For example:
LMS_WriteParam(device, LMS7_G_PGA_RBB, value)
LMS_WriteParam(device, LMS7_G_LNA_RFE, value)
LMS_WriteParam(device, LMS7_G_TIA_RFE, value)

[raccoman]

Thanks!

[raccoman]

Hi,

I understood that I can directly write values to the LMS7002M chip registers. But I don't think is a good solution...

lime_Result lms7002m_set_rfelna_db(lms7002m_context* self, const struct lms7002m_decibel value, const enum lms7002m_channel channel)
{
    enum lms7002m_channel savedChannel = lms7002m_set_active_channel_readback(self, channel);

    const int32_t gmax = 30;
    int32_t val = decibel_int(value) - gmax;

    int g_lna_rfe = 1;
    if (val >= 0)
        g_lna_rfe = 15;
    else if (val >= -1)
        g_lna_rfe = 14;
    else if (val >= -2)
        g_lna_rfe = 13;
    else if (val >= -3)
        g_lna_rfe = 12;
    else if (val >= -4)
        g_lna_rfe = 11;
    else if (val >= -5)
        g_lna_rfe = 10;
    else if (val >= -6)
        g_lna_rfe = 9;
    else if (val >= -9)
        g_lna_rfe = 8;
    else if (val >= -12)
        g_lna_rfe = 7;
    else if (val >= -15)
        g_lna_rfe = 6;
    else if (val >= -18)
        g_lna_rfe = 5;
    else if (val >= -21)
        g_lna_rfe = 4;
    else if (val >= -24)
        g_lna_rfe = 3;
    else if (val >= -27)
        g_lna_rfe = 2;

    lime_Result ret = lms7002m_spi_modify_csr(self, LMS7002M_G_LNA_RFE, g_lna_rfe);
    lms7002m_set_active_channel(self, savedChannel);
    return ret;
}

This is the internal function that receive a gain as double and converts it automatically by following it's internal logic. Since there is a user interface, the user inputs a gain value directly (as a double). It wouldn't be correct or ideal to duplicate the internal logic from the lms7002m_set_rfelna_db function in my own code just to convert that double value back to the corresponding register value. Doing so would lead to unnecessary code duplication and could cause inconsistencies if the internal logic ever changes.

What do you think? Thanks

[rjonaitis]

It all depends if you really need to use the legacy API. It is not going to be extended with additional functions.
You could access individual gains through C++ API.
the implementation of functions for each individual gain are not going to change, as it's just basically a lookup table, to get the nearest register value.

[raccoman]

It's possible somehow to access that function? lms7002m_set_rfelna_db