32 Channel Analog IC Switch Control with STM32F7 SPI

[A.Ersoz]

Hello Guys,

On my custom design board, there are a STM32F767ZI microprocessor and a HV2901 Analog Switch IC. My plan is that with controlling HV2901, already generated signal should flow the other components on my board. I have two questions. One of them is defining variables, so how can I fix the errors in this part. The other one is that I define SPI as full duplex master. How can I write codes for receiving data?

Here is my code;

const uint8_t testdata_out; // Main signal which will flow through ON-switch to other part of the circuit
const uint32_t switch_output={1,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0}; // I try to define which switch should ON. Based on data // sheet, for first switch, DO should 1.
uint16_t SW_Out_1; // In CubeMX, there is only option for setting 16-bit data as maximum
uint16_t SW_Out_2; // In CubeMX, there is only option for setting 16-bit data as maximum
uint32_t SW_Out = (SW_Out_1>>16) | (SW_Out_2<<16); // HV2901 needs 32 bit data length for controlling switch positions, so I merged previous //variables as 32-bit.   

// Latch or SYNC Pin needs to LOW
HAL_GPIO_WritePin(LE_Pin_GPIO_Port, LE_Pin_Pin, GPIO_PIN_RESET);
// Write data with SPI
SW_Out = switch_output<<1 | testdata_out<<31;
HAL_SPI_Transmit(&hspi4,(uint8_t *)&SW_Out,32,100);

// When data is ready, LATCH will high
HAL_GPIO_WritePin(LE_Pin_GPIO_Port, LE_Pin_Pin, GPIO_PIN_SET);

[A.Ersoz]

I modified a little bit of my code. But still the first switch is not ON.

Variables:

uint16_t Switch_Select1st = 0x8000;// I try to define that the first switch should ON. Based on datasheet, for first switch, DO should 1.
uint16_t Switch_Select2nd = 0x0000; // Rest of the 16 bit

// SPI4 Communication between microprocessor (master) to DAC (slave)
HAL_GPIO_WritePin(AnalogSWIC_LE_GPIO_Port, AnalogSWIC_LE_Pin, GPIO_PIN_RESET); // LE Pin set as low
HAL_GPIO_WritePin(AnalogSWIC_CLR_GPIO_Port, AnalogSWIC_CLR_Pin, GPIO_PIN_RESET); // CLR pin set as low
HAL_SPI_Transmit(&hspi6,(uint8_t*)&Switch_Select1st,2,100);
HAL_SPI_Transmit(&hspi6,(uint8_t*)&Switch_Select2nd,2,100);
HAL_GPIO_WritePin(AnalogSWIC_LE_GPIO_Port, AnalogSWIC_LE_Pin, GPIO_PIN_SET); // LE pins set as high
HAL_GPIO_WritePin(AnalogSWIC_CLR_GPIO_Port, AnalogSWIC_CLR_Pin, GPIO_PIN_SET); // CLR pin set as high

[newbrain]

I modified a little bit of my code. But still the first switch is not ON.

Variables:

uint16_t Switch_Select1st = 0x8000;// I try to define that the first switch should ON. Based on datasheet, for first switch, DO should 1.
uint16_t Switch_Select2nd = 0x0000; // Rest of the 16 bit

// SPI4 Communication between microprocessor (master) to DAC (slave)
HAL_GPIO_WritePin(AnalogSWIC_LE_GPIO_Port, AnalogSWIC_LE_Pin, GPIO_PIN_RESET); // LE Pin set as low
HAL_GPIO_WritePin(AnalogSWIC_CLR_GPIO_Port, AnalogSWIC_CLR_Pin, GPIO_PIN_RESET); // CLR pin set as low
HAL_SPI_Transmit(&hspi6,(uint8_t*)&Switch_Select1st,2,100);
HAL_SPI_Transmit(&hspi6,(uint8_t*)&Switch_Select2nd,2,100);
HAL_GPIO_WritePin(AnalogSWIC_LE_GPIO_Port, AnalogSWIC_LE_Pin, GPIO_PIN_SET); // LE pins set as high
HAL_GPIO_WritePin(AnalogSWIC_CLR_GPIO_Port, AnalogSWIC_CLR_Pin, GPIO_PIN_SET); // CLR pin set as high

Use the code quote (the # button) tags, please.

In any case, this looks much better than the first attempt!
But we still have some problems.

The major one is that you are using SPI6, and I assume the configuration is the same as in your other question.
If that's the case, you must have it configured for low polarity and clocking on the second (so, falling) edge in order to use that DAC:

Code: [Select]

  hspi6.Init.CLKPolarity = SPI_POLARITY_LOW;
  hspi6.Init.CLKPhase = SPI_PHASE_2EDGE;

This switch, as described in and just above the picture at page 7 in the DS, clocks data into the shift register at the raising edge of the clock.
So there's the need to reconfigure the SPI when switching peripherals (or simply use another SPI!).
I would advise against doing that through the HAL_SPI_Init(), as that's really shooting a mosquito with a nuclear warhead.
You can directly manipulate the registers or use a function provided by the LL library, taking care that all trasmissions have finished; this is not a problem if you are using the blocking functions.
According the DS, it would seem that it's not even needed to disable the peripheral.

With that out of the way, a couple of other points:

• Have you studied the DS? On page 7 it is also shown clearly that when you have CLR high, all the switches will be off.
  You are setting CLR high after every write in your code...
• The interface is very SPI like, but the shifting happens even with LE high! This will disrupt any data in the switch if you use the same SPI for other tasks and result in many glitches when updating it unless you:
  1. always keep LE high so as to maintain the latched output configuration, independently from the shift register data.
  2. only briefly pulse it (H->L, L->H) after shifting in the whole 32 bits needed.
  This also is hinted in the DS, in the first page:
  "..., the latch enable bar should be left high until all bits are clocked in."
  though they suggest this in order to reduce clock feed through noise.
  

[A.Ersoz]

So there's the need to reconfigure the SPI when switching peripherals (or simply use another SPI!)

I reserved in to SPI4 for communication, and I selected transmit only master.

Code: [Select]

// SPI4 Communication between microprocessor (master) to DAC (slave)
HAL_GPIO_WritePin(AnalogSWIC_LE_GPIO_Port, AnalogSWIC_LE_Pin, GPIO_PIN_RESET); // LE Pin set as low
HAL_GPIO_WritePin(AnalogSWIC_CLR_GPIO_Port, AnalogSWIC_CLR_Pin, GPIO_PIN_RESET); // CLR pin set as low
        HAL_SPI_Transmit(&hspi4,(uint8_t*)&Switch_Select1st,2,100);
HAL_Delay(1);
HAL_SPI_Transmit(&hspi4,(uint8_t*)&Switch_Select2nd,2,100);
I keep LE and CLR pins as low, so I deactivated about shifting high level of the signals.

Actually, I didn't understand briefly pulse.

Hence, I didn't switch the the first latch yet.

[newbrain]

I keep LE and CLR pins as low, so I deactivated about shifting high level of the signals.

Actually, I didn't understand briefly pulse.

Hence, I didn't switch the the first latch yet.

Good, so you are now on a separate SPI, I would imagine you've set it up as shown in the datasheet: clock polarity high and second edge matches the picture, but from the description also clock polarity low and first edge should work (only the L to H transitions are important).

CLR is fine, it can stay low as you've set it. No need to change it.

As for the /LE: the shift register and the output latch are independent.
The SR is always active, and will shift its content at every clock tick.
The output latch is transparent if /LE is low, so the switches will directly follow the values in the SR.
When /LE goes high, the last configuration is frozen in the output latch.

My advice, since you were using SPI6 in the  code (same as for the DAC), was to do something like:

Code: [Select]

HAL_GPIO_WritePin(AnalogSWIC_LE_GPIO_Port, AnalogSWIC_LE_Pin, GPIO_PIN_SET); // LE Pin set as high (normal condition)
HAL_GPIO_WritePin(AnalogSWIC_CLR_GPIO_Port, AnalogSWIC_CLR_Pin, GPIO_PIN_RESET); // CLR pin set as low
HAL_SPI_Transmit(&hspi4,(uint8_t*)&Switch_Select1st,2,100);
HAL_SPI_Transmit(&hspi4,(uint8_t*)&Switch_Select2nd,2,100);
HAL_GPIO_WritePin(AnalogSWIC_LE_GPIO_Port, AnalogSWIC_LE_Pin, GPIO_PIN_RESET); // Give a brief pulse ( \_/ )
HAL_GPIO_WritePin(AnalogSWIC_LE_GPIO_Port, AnalogSWIC_LE_Pin, GPIO_PIN_SET);   // to LE in order to latch the SR content

As suggested in the datasheet. This is OK in any case, and avoid the switch configuration changing while shifting.

If the switch is correctly powered (at least +/-40V for the HV side and 3-5.5V for the LV side), connections are correct and signal integrity is decent, I see no reason why it should not work.

[A.Ersoz]

If the switch is correctly powered (at least +/-40V for the HV side and 3-5.5V for the LV side), connections are correct and signal integrity is decent, I see no reason why it should not work.

Actually, I connected VNN as -5V; VPP as 5V; VDD as 5V.

I got around 1V constant signal from DOUT, but I couldn't see clock signal. Maybe, I make a mistake setting parameters of SPI4?

I set in CubeMX SPI4 Configuration;

Transmit Only Master
NSS is disable
Motorola
16 bits
MSB
8 (Prescaler)
13.5 Mbit/s
Low
2 Edge

[newbrain]

If the switch is correctly powered (at least +/-40V for the HV side and 3-5.5V for the LV side), connections are correct and signal integrity is decent, I see no reason why it should not work.

Actually, I connected VNN as -5V; VPP as 5V; VDD as 5V.

I got around 1V constant signal from DOUT, but I couldn't see clock signal. Maybe, I make a mistake setting parameters of SPI4?

I set in CubeMX SPI4 Configuration;

Transmit Only Master
NSS is disable
Motorola
16 bits
MSB
8 (Prescaler)
13.5 Mbit/s
Low
2 Edge

First of all a disclaimer: my advice is worth exactly what you pay for, I try to be correct and check my posts, but I might be and have been wrong.

The datasheet for that part provides quite scant information.
But from my reading, the operating conditions for that chip call for at least V_PP = +40V and V_NN = -40V.
There is no characterization for values outside that range.
Will it work with +/-5V?
I can't say, as the DS does not say, but it's definitely out of the recommended operating conditions.
My default answer would then be "no".

DOUT at 1V is not good. It should be much closer to either GND or V_DD, unless it's heavily loaded (it should not be).

Why second edge? With low clock polarity it should be first edge:
(Or dually, why low clock polarity? With second edge it should be high.)

Code: [Select]

2. Serial data is clocked in on the L to H transition of the CLK.
Pardon my bluntness, but I start to have the feeling that you're trying to run a marathon with a 2000m training 
Is this a hobby project, a study assignment, a task for your job?