We are preparing a cheap robot vacuum cleaner

Good day.
Today study giblets and budgetary scheme of the robot cleaner elektrovenik Agait EC of the MINI .
Given the primitive algorithm, it can be called a robot very conditionally.
A lot of text, pictures and a survey for a snack.

When buying, there were no bench trips, although I was pleasantly surprised - it works and even collects dust, maintaining cleanliness, increasing the intervals between ordinary “manual” cleaning.
Since this model is the youngest (it was bought in a chain store for the equivalent of $ 100), we will not discuss the functionality of this product “out of the box” but look at the scope for hand-ware and modernization.
I see this device as a ready-made platform for a DIY robot vacuum cleaner.
The first candidate for an upgrade is, of course, asking for a brain.
For starters, you can leave the power unit available, and use something from the Arduino clan as a controller.
There should be no problems with implantation into the current scheme, because in the original version, the microcontroller EM78P153K is used with 5V power in a 14-pin package (minus 2 pins of power) for a total of 12 pins for communication with the circuit.

General view of the insides

To plan the implantation of a new one, you first need to imagine what actuators and "sense organs" this "animal" has.

What does the manufacturer of this miracle offer us?

In this model of the vacuum cleaner there is no optical sensor on the suction port (IR LED + photodiode) and the corresponding part of the circuit. Although there are places in the case and wiring on the board, so you can add it if you wish.

Consider the electrical circuit

The circuit is copied from the board, so inaccuracies are possible.

Unnamed transistors are something petty SO23. Q1, Q9, Q10 with the Chinese marking Y1 - possibly SS8050, the rest with the marking CR - possibly 2SC945. Although to understand the logic of the circuit, this is not particularly necessary. Unnamed diodes are most likely 1N4148 in SMD performance, also their type is not particularly important.
There is no microcontroller harness. Absolutely not. So it is not in the diagram, there is a reference to the conclusions. He corny eats from + 5V and the remaining legs diverge according to the scheme.

Let's go through the main nodes

Q1, Q2, Q15 This is the battery charging key. I note that here it is charged just in time with the limitation of the maximum current through a 5-watt resistor R73. No control is provided, so looking at the circuit, I charge my battery with the IMAX clone with the end of the charge according to ΔV, it will live longer.

The 8.25V stabilizer on the MC34063 is depicted as a block, since the microcircuit is turned on according to the standard scheme. Rsc resistor (see datasheet) 0.22 Ohm. Those. there is a current limitation, not only to protect the chip itself, for which a little lower.
Wheel modules and side brush drives are powered from it.

The LM393 dual comparator controls the power drawdown of the wheel modules and side brushes (in case of jamming with foreign objects or mechanical malfunction) and battery discharge. These two conditions for the controller are one event.

The suction fan is turned on with the side brush drives by Q24. In this case, the fan is powered almost directly (minus the voltage drop across the D16 diode and the open transistor) from the battery. Acceleration, however :-) Side brushes opposite are powered by a low voltage of 8.25V minus a drop on 3 diodes and an open transistor.

Optocouplers JK1 and JK2 are slotted transistor. JK2 is normally dimmed (the cover is closed — the transistor is closed) and JK1 is normally illuminated (the bumper is not stuck anywhere — the transistor is open)

On the Q25 transistor, a switch is assembled for switching the power of the LEDs of the optocouplers and the entire node of the drop sensors. If there is 19V from the charger, it is closed, in all other cases it is open.

A transistor Q8 has a control circuit for the presence of 19V from the charger. The signal goes to 7 pins. microcontroller. The phototransistor of the optocoupler cap is connected there. Those. The connected charger and an open cover for the controller are one event. How does the controller distinguish when the cover is open, and when the charger is connected? By bumper sensor. When the charging phototransistor is connected, it will be dimmed due to the power off of the LED (key Q25). So if you open the lid and press the bumper with the charger turned off, the vacuum cleaner will think that it is charging, it should also stand on the surface so that the drop sensors do not work (when the charger is connected, they are disabled Q25). This is the reckoning for the utmost simplification of the scheme. The charge mode is indicated by a flashing green LED (in the lower left corner of the circuit). In order not to mislead (or maybe not scare) the user with a vacuum cleaner that shows charging without a charger, the designers simply do not let the LED blink thanks to the Q9 transistor, although a meander is coming from the first output of the microcontroller to the LED. Crutches-crutches.

Wheel Motor Drivers

Nothing remarkable stands out.
It works simply - at both inputs a logical 0 - we stand
We give 1 to one of the inputs - we go either forward or backward.
We give two 1-tsi sit + VCC Motor to the ground. There is no protection "from the fool", so either two 0 or one 1-tse.

Step protection sensors

Schematically, they are optical pairs of an LED-photodiode directed to the surface, while the photodiode is structurally more distant from the surface and can be partially covered by an adjustable curtain to select the response height (photo is at the beginning of the article). To decouple the level of illumination in the room, the LED is modulated at a certain frequency.
The scheme is approximate for understanding the principle of work. The selected part is individual for each channel, the generator on the first and the comparator on the last operationalist are common to all.
Logical 1 on diode D2 is
showing its activity. The triggering of the drop sensors and the emphasis of the bumper on the obstacle for the microcontroller is one event.

"Summary" of the legs of the microcontroller

In case of active 0 there will be a mark.
1 Green LED
2. The bumper has run into an obstacle or any fall sensor has triggered - active 0
3. Bazzer (tweeter)
4. + 5V
5. Power drawdown of brush and wheel motors and battery discharge.
6. Turn on the suction fan and side brushes.
7. The charger is connected or the cover is open - active 0
8. Left wheel
9. Left wheel
10. Turn on the charging key - active 0
11. GND
12. Right wheel
13. Right wheel
14. Red LED.

Total

In my opinion, the main drawback of the current circuit is the combined signals of the bumper and drop sensors, therefore, with the current algorithm, the vacuum cleaner encountering an obstacle when moving “directly” simply rotates 180 ° and travels away from it to another wall, and so several times along the same path . Therefore, it is very desirable to separate these signals, for a more adequate response to obstacles and the "edge of the earth."
It would also be nice to add intelligence to the charge circuit.

Now you can unsolder the native microcontroller, connect * uino, or whatever you like and invent your own algorithms, but this will already be in the second part.

Everything stated in this article is purely my conclusions and impressions and expresses my opinion on this issue, but how many people have so many opinions.