Troubleshooting¶

Symptoms you may see while bringing up an ADS7952, their causes and the fix for each.

Start with RawManualFrames(): it shows the raw replies, so it tells bus problems (no address, wrong address) apart from analog problems (right address, wrong count). API guide shows the call.

Bus and protocol¶

Symptom	Cause	Fix
`ReadChannel()` returns `Error::Timeout` on every channel	The reply address never matches: wrong SPI mode, CS released between the two bytes, MISO not connected	SPI mode 0; one CS cycle per 2-byte `transfer()`; check SDO with a logic analyser
Every reply is `0x0000` or `0xFFFF`	No supply on +VBD, MISO floating, or CS never goes low	Measure +VBD (1.7 V to +VA); check CS on the scope
Addresses look right but shifted by one frame	Your port splits or merges frames, so the chip and the driver count frames differently	Exactly 16 clocks per CS low
`ReadAllChannels()` returns `Error::Timeout` with some channels missing	Auto-1 mask on the chip differs from the driver's (chip reset or brown-out since `EnsureInitialized()`)	`EnsureInitialized(true)` after any reset; check `valid_mask`
Reads return `Error::SpiError`, `bool` calls return `false`	The port's `transfer()` returned `false`: the SDK reported a bus error	Check the SPI host, pins and bus lock in the port; see Porting
Every read returns `Error::NotInitialized`	`EnsureInitialized()` was never called on this object	Call it once after power-up

Readings¶

Symptom	Cause	Fix
Every count is 0	MXO not connected to AINP, AINP floating, or no reference	Short MXO (pin 7) to AINP (pin 8); AINM to AGND; check REFP
Counts stuck at 4095	Input above the range	Use `Range::TwoVref` (needs 2 × VREF ≤ +VA) or scale the input
Every channel reads about the same count	Source impedance too high: the sampling capacitor keeps the previous channel's charge	50 Ω or less into each input, or about 150 pF at each input; or lower SCLK
Voltages off by a fixed factor	`vref` passed to the driver is not the voltage at REFP, or it was clamped (outside 2.0 V to 3.0 V)	Measure REFP; `SetVref()`; see Configuration
Voltages off by exactly 2 after `SetRange()`	`SetRange()` returned `false` (bus error) and the chip kept the old range, or GPIO2 is configured as the range input and its pin decides the range	Check the return value; with `gpio2_as_range_input`, drive GPIO2 to match `SetRange()`
The first conversion after power-up is wrong	The chip's first conversion is invalid by design	`EnsureInitialized()` discards it; after a reset call `EnsureInitialized(true)`

Alarms and GPIOs¶

Symptom	Cause	Fix
Alarm pin never changes	GPIO0/GPIO1 not assigned an alarm function	`ProgramGPIO()` with a `GPIO01AlarmMode` other than `GPIO`
Alarm fires at the wrong voltage after a range change	Thresholds are stored as counts	Call `ProgramAlarmVoltage()` again after `SetRange()` or `SetVref()`
Deprecation warning for `GPIO1_HighAlarm`	It was misnamed: its code (DI06:04 = `100`) makes GPIO1 the low alarm output, and the chip has no GPIO1 high alarm	Use `GPIO1_LowAlarm` (same code); see API guide
GPIO outputs do not change	GPIO not configured as output, or assigned an alarm, range or power-down function	Set the bit in `direction_mask` and leave the pin's function as GPIO in `ProgramGPIO()`
Only GPIO0 responds	VQFN package: GPIO1 to GPIO3 do not exist	Use the TSSOP package for four GPIOs

Build¶

Symptom	Cause	Fix
`'ads7952.ipp' file not found`	`src/` is not on the include path	Link the `hf_ads7952` target, or add both `inc/` and `src/`
`hf::ads7952` not found	`add_subdirectory()` missing, or `find_package(hf_ads7952 CONFIG)` without an install	See Installation
Errors about C++20	Compiler older than GCC 10 / Clang 12, or the standard forced lower	The target requires `cxx_std_20`; do not override it