Embedded Files
History
History
Being involved in testing and building of light electric vehicles quite early it came out that some form of energy tracking and monitoring was necessary. When I was using SLA (sealed lead acid batteries aka ‘gels’) the problem was not that acute since acid batteries can be discharged deeply without significant degradation. However when I finally switched to LiPo batteries it came out really quickly that without maintaining strict operating limits they can degrade quickly or, what worse can even explode. YouTube is full of videos showing explosions and flames coming out of mistreated LiPo cells… Detailed justification for creation of the device you can find here. Anyway, to get full grip of the discharge process I started looking for ways to visualize and monitor the process. Conceptually the idea is simple: integrate current over the time to get consumed charge. Add voltage monitoring to get power. Monitor temperatures. Store peak and aggregated values. Present the data to user. It happened to be more complicated than this though. After numerous trials and errors I learned how to use various current sensors, how to handle calibration issues and cope with aspects I’d never imagine they could pose any issues at all. The first working prototypes proved to be not only very useful but to my surprise quite unique devices. In practice the only good alternative is Cycle Analyst device. The only problem for me was it could not be extended in a way I’d like. I wanted to cover wide range of applications I worked on from converted bikes through larger recumbent electric vehicles to cars. So I needed something scalable capable of measuring from a few tents of amps to more than many hundreds. Also voltage range should cover ranges from zero (technically speaking around 12V which for EV applications is WELL below zero) to more than 165V to begin with.
Being involved in testing and building of light electric vehicles quite early it came out that some form of energy tracking and monitoring was necessary. When I was using SLA (sealed lead acid batteries aka ‘gels’) the problem was not that acute since acid batteries can be discharged deeply without significant degradation. However when I finally switched to LiPo batteries it came out really quickly that without maintaining strict operating limits they can degrade quickly or, what worse can even explode. YouTube is full of videos showing explosions and flames coming out of mistreated LiPo cells… Detailed justification for creation of the device you can find here. Anyway, to get full grip of the discharge process I started looking for ways to visualize and monitor the process. Conceptually the idea is simple: integrate current over the time to get consumed charge. Add voltage monitoring to get power. Monitor temperatures. Store peak and aggregated values. Present the data to user. It happened to be more complicated than this though. After numerous trials and errors I learned how to use various current sensors, how to handle calibration issues and cope with aspects I’d never imagine they could pose any issues at all. The first working prototypes proved to be not only very useful but to my surprise quite unique devices. In practice the only good alternative is Cycle Analyst device. The only problem for me was it could not be extended in a way I’d like. I wanted to cover wide range of applications I worked on from converted bikes through larger recumbent electric vehicles to cars. So I needed something scalable capable of measuring from a few tents of amps to more than many hundreds. Also voltage range should cover ranges from zero (technically speaking around 12V which for EV applications is WELL below zero) to more than 165V to begin with.
Early prototype of Vacek. Fully functional and accurate. Above Watts-up meter (from hobbycity.com) is visible. The differences being shown are irrelevant and it's not possible to say which one showed more 'real' figures.
Development
Development
Various attempts to get satisfactory results have been attempted. Choosing right combination of sensors, microcontroller, pcb design, features and LCDs took long months of trials and errors. The biggest challenge was to find a way to scale measurement range so that broad scale of currents and voltages met in EV solutions could be handled by single, not expensive device.
Various attempts to get satisfactory results have been attempted. Choosing right combination of sensors, microcontroller, pcb design, features and LCDs took long months of trials and errors. The biggest challenge was to find a way to scale measurement range so that broad scale of currents and voltages met in EV solutions could be handled by single, not expensive device.
Technology
Technology
Microcontroller
Microcontroller
Cypress (8C29 series) has been choosen as the best choice for the design. It proved to be far more flexible than ‘classic’ microcontrollers like PIC (18F series) or Atmel’s ATMEGA series. It is a 8bit (M8C Harvard architecture), 2 MIPS (at 12MHz) microcontroller with good quality 14bit ADC. It is a System On Chip (PSOC) type of microcontroller which means it can be internally reconfigured. It is something in the middle between classic non-reconfigurable microcontrollers and FPGAs.
Cypress (8C29 series) has been choosen as the best choice for the design. It proved to be far more flexible than ‘classic’ microcontrollers like PIC (18F series) or Atmel’s ATMEGA series. It is a 8bit (M8C Harvard architecture), 2 MIPS (at 12MHz) microcontroller with good quality 14bit ADC. It is a System On Chip (PSOC) type of microcontroller which means it can be internally reconfigured. It is something in the middle between classic non-reconfigurable microcontrollers and FPGAs.
Memory
Memory
I2C EEPROM (24C08) memory has been used to store cumulative values. It is small yet prone to wear (1000000 cycles) due to write operations. Data are being logged every 30 seconds so it gives 11.57 months of guaranteed non-stop operation. In practice in non-military applications J it yields many years of reliable service. Before any problems occur though user will be able to switch to FRAM memory which is a type of non-volitile ferromagnetic memory pin-compatible with “classic” flash serial memories.
I2C EEPROM (24C08) memory has been used to store cumulative values. It is small yet prone to wear (1000000 cycles) due to write operations. Data are being logged every 30 seconds so it gives 11.57 months of guaranteed non-stop operation. In practice in non-military applications J it yields many years of reliable service. Before any problems occur though user will be able to switch to FRAM memory which is a type of non-volitile ferromagnetic memory pin-compatible with “classic” flash serial memories.
Sensors
Sensors
Hall sensors
Hall sensors
Various types of Hall sensors has been tested and evaluated. I tried LEM’s HTA-300 and 400 series, Allegro’s ACS750 series as well as numerous 3-pin Hall sensors in TO-92 packages. Here’s brief description of each with my evaluation results as far as EV applications are concerned.
LEM HTA sensors are BIG and heavy. They require bipolar supply for operation since they have operational amplifier built in which output can swing in both directions according to current direction. Good for measuring large currents – many hundreds amperes.
Advantages
1. Accurate measurment of large currents
2. Robust casing
3. Trimmers (Potentiometers) for offset and gain.
4. Very good temperature stability.
5. Contactless measurements. Just put cable through it.
Disadvantages
1. Cost. They are rather expensive. Compensated models can be 2 or 3 times more expensive than non compensated ones.
2. Bipolar supply needed. This complicates design.
3. Large current sensors like LEM’s HTA series are heavy.
4. Sensitive to external magnetic fields including Earth’s geomagnetic field. This poses a problem when measuring small currents (around 1A or smaller).
ACS750 series from Allegromicro. Small, cheaper than LEM’s transducers.
Advantages
1. Very small and very light
2. Single supply needed (5V)
3. Not susceptible to Earth’s magnetic field yet still external magnetic sources can influence its operation.
Disadvantages
1. You have to break your power line in order to use it.
2. Not that cheap as well. Cheaper than LEMs though.
TO-92 Hall sensors. Various manufacturers.
These are the cheapest ones. Unfortunately one has to build special casing for it consisting of a toroid made of soft ferromagnetic with a slit for a sensor in it. Very vulnerable to temperature changes. Using these sensors is like reinventing wheel. I lost too much time trying to achieve satisfactory results with such sensors.
Shunt resistors. These are typically cheper than LEM type and Allegro’s ACS sensors but much more expensive than TO-92 ones. The most frequently chosen sensors for high current measurements in light Evs so far. They shouldn’t be used in high current applications but for most typical cases they are just fine. E.g: typical shunt sensor has 60mV voltage drop. Conduction of 300 A current make it radiate 18W of energy in form of heat. The more current – the bigger energy loss. However in practice losses are not significant for light EVs and this makes such sensors reasonable choice for the application. Sometimes the topology used may pose problems. This is related to the fact that “low side” configuration raises ground level by level depending on current flowing through the shunt resistor. Some devices may not function properly if they use ground as voltage reference.
[This work has not been finished yet]
Versions
Versions
This section describe the differences brought about after each new release.
This section describe the differences brought about after each new release.
Version 1.0
Version 1.0
Basic functionality. 4x16 LCD, LEM and ACS sensors support. This is development release for internal and external field tests with various configurations of sensors.
Version 1.1
Version 1.1
Major PCB improvments. Corrected pin headers locations. 4x16 LCD now is controlled via 3-wire interface saving 5 microcontroller pins for future applications. PCB is thicker and rigid. Many minor changes in firmware. Switching to 24c08 flash memory.
Version 1.2
Major redesign due to problems with isolation in bigger vehicles. Version 1.1 will probably stay as it is on its own.
Here Vacek is visible in aluminim CNC machined case. LCD is almost invisible in direct sun light. It has been later replaced by non inverted type and this flaw's been repaired.
Here Vacek is visible in aluminim CNC machined case. LCD is almost invisible in direct sun light. It has been later replaced by non inverted type and this flaw's been repaired.
Page updated
Google Sites
Report abuse