
GR716-DS-UM, May 2019, Version 1.29
102
www.cobham.com/gaisler
GR716
Each
ADCx
control unit has access to external ADC pins via ADC converters and has a unique
AMBA address described in chapter 2.11. ADC control unit 0, 1, 2, 3, 4, 5, 6 and 7 have identical con-
figuration and status registers. Configuration and status registers are described in section 12.3.
The system can be configured to protect and restrict access to individual ADC units in the
MEM-
PROT
unit. See section 47 for more information.
12.2
Operation
12.2.1 System overview
Each ADC converter is a 11bit/200kSps SAR converter, and has an analog MUX in-front of it, which
means that one MUX channel at a time can be measured. The ADC can be programmed to single-
ended 11-bit range (0 V - 2.5 V) using one input pin per channel, or to differential-input 11-bit range
(-2.0V- 2.0V) using two input pins per channel. In-between the ADC and MUX, there is a fully differ-
ential pre-amplifier, which has three programmable gain-settings (x1, x2, x4). It is to be used together
with the fully-differential ADC setting. The amplifier input impedance is in the order of 5 to 20 kohm
(TBC). The amplifier can be by-passed by programming; then, the DC input impedance is high (dom-
inated by MUX leakage currents). These three blocks are supplied by VDDA_ADC and VSSA_ADC.
This supply is not the analog reference for the ADC measurements. However, it must still be really
well decoupled/filtered at high frequencies (>~1MHz) to not degrade the ADC performance.
The ADC supply ground, VSSA_ADC, should be hardwired to the same PCB ground point as
VSSA_REF, directly outside the Microcontroller package.
12.2.2 Detailed description
The on-chip ADC and pre-amplifier has a digital control and status interface accessible via register on
the APB bus. To support CPU offloading, autonomously ADC measurements and level detection a
digital interface has been implemented in the digital core of the microcontroller to support different
complex sampling modes over multiple ADC channels. The digital interface also supports sampling
modes to suppress noise and to increase the resolution and ENOB. Increasing the resolution is sup-
ported via oversampling and increasing the ENOB by using higher gain-settings in the pre-amplifier.
In order to make the oversampling effective the measured signal needs to be of AC signal type or a
DC signal with dithering i.e. introduce random noise in the analog input signal to the ADC. AC signal
is defined as a signal where the quantization error of 2 consequence samples are independent.
Single-ended or differential mode
is selected per ADC channel but will only be valid for ADC chan-
nel 0,2,4 and 6, since multiple channel inputs will be used when differential mode is enabled. Differ-
ential mode have the capability to measure more accurately and the input gain can be adjusted using
the on-chip amplifier. The on-chip amplifier gain can be configured individually for the differential
channels to x1, x2 or x4. To use just a pin when single-ended mode is selected, bypass mode should be
selected too. If bypass is not selected, the negative pin shall be grounded.
The digital control logic supports following
Sampling modes and configuration
:
•
Read
current value
i.e. via the ADC status register see section12.3for register description.
•
Oversampling and averaging
to extend the number of effective bits. The sampling rate and
number of samples accumulated is controlled via registers. The number of samples can be con-
figured in the range of 1 to 65535 samples. No other manipulation of the results is performed in
hardware. The accumulated samples can be consecutive or taken with a configurable distance in-
between. For each additional bit resolution, the signal must be oversampled by a factor of four.
For simplicity the hardware do not perform any truncation or division of the accumulated sam-
ples. To get a correct average sampled value the accumulated sample value shall be rounded and
divided by the number of samples used. For application where the truncation error is less import-