Load Detect (or how to detect vandalism)

In this text, I wrote how to diagnose and test an electrical circuit such as an H-bridge. Programming the H-bridge is easy. It is much more difficult to determine Fault (s) in this notorious H-bridge in Run-Time (e) in order to turn off the keys in time and put the entire circuit in a safe state, thereby saving expensive equipment from burning.

A bit of theory

In electronics, there is a classic electrical circuit called the H-bridge.

pic 1
pic 1

This circuit allows you to pass current through the load in 2 directions: along or across the load. If a DC motor is connected to this circuit, it will spin either clockwise (CW) or counterclockwise (CCW). You can also modulate the current value if you switch the keys on a PWM signal

H-bridges are actively used in devices that turn stepper motors in CNC machines in glass lift controllers (spinner firmware), to control car headlights, and so on.

There is a problem. Rather, there are 31 problems. That is, there can be 31 types of failures in H-bridges, which will lead to the fact that the H-bridge simply will no longer work. And 14 cases will lead to the fact that sparks will simply fall from the H-bridge. Here is a list of those failures.

pic 2
pic 2

How to detect crashes in the H-bridge at runtime? We need to develop a mechanism that can be called Load Detect.

It is clear that the H-bridge must be somehow observed. In electronics, there are only 2 ways to monitor anything: GPIOs working as an input and ADC. Here is a possible circuit design that will allow you to diagnose the condition of the H-bridge. Here, an analog-to-digital converter (ADC) is connected to each arm of the H-bridge and there is also a key that controls the voltage pull-up to the power supply.

pbc 3. H-bridge monitoring scheme

pbc 3. H-bridge monitoring scheme

Each microcontroller has analog pins that are connected to an internal ADC converter. The voltage pull-ups to the power supply can be controlled via GPIO or using a key and a high-ohm resistor or using special source chips. wetting current.

pic 4
pic 4

It is obvious that some algorithm or a high-level software state machine that will report when an error occurs in the H-bridge. Ideally, I would like a separate ASIC to do this. But you can also sprinkle a purely software solution.

The development of a state machine is just a mechanical operation. As you know, any finite state machines are designed in 7 phases:

Phase

State machine development stage

1–

define exits

2–

define states

3–

define actions

four–

define inputs

five–

compose table transitions

6–

draw graph transitions

7–

to embody state machine in code

1– Here exits state machine diagnostics of the H-bridge. These are just those high-level events that the user wants to hear (circuit designer, programmer, technician, integrator, driver)

No.

FSM Output

1

left shoulder short GND

2

left shoulder short

3

right shoulder short GND

four

right shoulder short

five

open load

2– states state machine can be determined by the state of the power pullups. It is important to install braces exactly in gray code. This will reduce the time of the transient process of installing suspenders and reduce the power consumption of the device, and it’s easier to debug on the oscilloscope.

No.

left shoulder

right shoulder

1

Pull None

Pull None

2

Pull None

pull up

3

pull up

pull up

four

pull up

Pull None

3– What will this state machine do at a low (hardware) level? Or what he will do systemic actions?

No.

Action

H-bridge arm

1

Set up a power brace on the left shoulder

left

2

Install a power brace on the right shoulder

right

3

Disable the power pull-up on the left shoulder

left

four

Disable the power pull-up on the right shoulder

right

five

Change the voltage reading on the left shoulder

left

6

Change the voltage reading on the right shoulder

right

7

Calculate analytics

—