Low power, long lasting, personal Geiger counter experiment

[madexp]

INTRODUCTION

Hello! Over the past few weeks, I've been working on how to efficiently generate the 400V bias required by Geiger tubes. This effort stems from my desire to build a counter powered by a 4xAA battery pack, aiming to maximize its longevity before needing a recharge.

SPECIFICATIONS

My design adheres to the KISS (Keep It Simple, Stupid) philosophy:

• Two buttons on the front panel: one for various functions through long/short presses and the other for turning the device on/off.
• A cost-effective, low-power MCU as the central processor.
• A single 16x2 character LCD, highly effective in low-light conditions.
• A durable, water-submersible enclosure made of cast polyurethane resin.

IMPLEMENTATION

The first component I developed was the HV converter, following the straightforward assembly of the MCU to LCD connections. I believe the solution I chose is intriguing and merits detailed discussion.

The HV converter's operation is straightforward: it uses a 3.3mH ferrite inductor to store magnetic energy when the transistor conducts current, then releases it as a voltage spike when the transistor turns off. This spike is rectified, and the voltage is increased using a classic voltage multiplier circuit.

What sets this design apart is the voltage stabilization control loop. Instead of employing resource-intensive stabilization methods with ADC sampling, the system uses a simple approach. The MCU generates a PWM signal that turns on when the 400V HV target has not been reached and turns off once it is. This process is managed through a straightforward check: the MCU monitors a pin in polling mode within the main loop. This pin connects to a voltage-sensitive chain comprising a 400V zener diode in series with a 5.1V protection zener diode. When the HV exceeds 400V, the zener diode conducts, setting the pin to a high logical value and thus halting PWM operation. If the HV drops below 400V, the pin remains low, prompting the PWM signal to resume driving the converter.



This schematic illustrates the HV converter, and the subsequent image displays the PWM trace that drives the MPSA42 transistor. It's straightforward to identify the moments when the PWM is deactivated.



I plan to continue describing the device in a subsequent post if there's interest in the topic.

[Dsl71]

Super, Oh yes there's interest. 

[madexp]

MEASURES

The MCU (Microcontroller Unit) + converter consumes between 3.5 and 5.5 mA at a 4.8V supply. The LCD display, without the backlight, adds an additional 3 mA. This means the complete counter circuit operates within a range of 6.5-8.5 mA. Excluding the unconventional implementation of the HV (High Voltage) control loop—specifically, with a 400V zener diode to ground—the consumption was 13.5 mA.

I'm using four AA NiMH (Nickel-Metal Hydride) batteries with a declared capacity of 2400 mAh but a measured capacity of 1950 mAh. With such a battery pack, the device can operate continuously for approximately 230 hours, which is slightly more than 9 days.

The LCD backlight can be turned on or off by the user. When in use, it requires an additional 12 mA.

I'm considering adding a Bluetooth module that would send data to a mobile phone, but it would require an additional 15 mA. With the Bluetooth active, I can estimate the operation time to be around 82 hours. This is not ideal, but it is also not disastrous.

ENCLOSURE

I've designed the plastic enclosure with ruggedness in mind: the walls are 4 mm thick, there are no screws, and it should be waterproof. The enclosure is divided into two halves: the front panel with two buttons and the display, and the main body, which only has a multipolar connector on the back for connecting to an external probe.



The PCB slides inside the main body; the battery pack is secured onto it. To change the batteries, the user must slide out the PCB. The main PCB is connected to the front panel via flexible cables.

Currently, I am awaiting the delivery of professionally printed master models for the front and body. Upon receiving them, I will create two flexible silicone molds using these models. I've 3D-printed the models to verify their accuracy and everything seems to be proceeding as planned.

For duplication of the pieces as needed, I will use resin casting. Polyurethane resin, known for its extreme strength and ease of casting in complex shapes, will be my choice. I can also incorporate fillers such as wood or marble into the resin to alter its characteristics, affecting not just the aesthetics but also the flexibility and toughness.

[Dsl71]

What MCU and LCD you are using?

We experimented with the Bosean FS-600 and the opensource firmware radpro and got following results in power consumption:

1,25 kHz / 3%: ohne Backlight 3,45 mA, mit Backlight 5,79 mA
2,5  kHz / 3%: ohne Backlight 3,17 mA, mit Backlight 5,52 mA

See https://www.geigerzaehlerforum.de/index.php/topic,1068.135.html post #145
and https://github.com/Gissio/radpro

[madexp]

Oh that's interesting. I see similar figures. What changes the game is the LCD backlight!
Actually I'm using the damn simple stupid 16x2 HD44780 compatible displays cause the philosopy KISS.
The MCU is what I've found in my lab: PIC16F917 but you can easily replace it with F84, F628A...
That specific MCU needs 0.8mA at 8Mhz, I can reduce clock frequency to try saving some energy.
At this point there is another thing that maybe could be fixed: LCD contrast trimmer.
I have the standard 10k trimmer wich waste 0.5mA there too. 

[Dsl71]

Okay, but the STM32G070CB used in the FS-600 is a much stronger MCU.

https://www.st.com/en/microcontrollers-microprocessors/stm32g070cb.html

The used LCD i didn't find out  yet.

Ah Gissio says: Bosean FS-600 128x64 LCD display (ST7567)

[madexp]

Ok but... I prefer to put the part that I have at hand or that will be fully used and that doesn't require aditional components (LDO). I've 4xAA there and I run everything beautifully unregulated

[Dsl71]

What is LDO?

[madexp]

LDO: low dropout voltage regulator.    I'm trying to keep things as simple as possible with the smaller part count possible. Probably I will chose to use also only THT components... it's an experiment to keep down complexity and make something very simple yet effective and reliable.

[Dsl71]

Ah, now i understand fully, then this is the way to go - simple yet effective and reliable is allways good.