Copyright statement: This document is a detailed introduction to the "CNC DC Electronic Load" entry of micepring in the "GD32 Cup Fifth Lichuang Electronics Competition". Commercial use is prohibited! Please indicate the original link and author ID when reprinting.

Project selection has ended and the comment area is open. Since the project is locked and cannot be changed, all changes are written here, please pay attention before making

Issues to note:

1. The firmware currently provided is based on the 8M crystal oscillator , not the 25M crystal oscillator on the schematic diagram. If you use a 25M crystal oscillator, you can modify the relevant settings in the code yourself.
2. Because this project uses SPI to drive the TFT screen, the frequency is set to 50MHz to ensure smoothness. If you need to modify the PCB, please pay attention to the routing of the LCD signal line. The distance from the MCU should be as short as possible, otherwise the screen may be blank or painted.
3. The current fan speed adjustment is designed for 4-wire fans . If you use 2-wire fans, you need to modify the PWM frequency in the code.
4. Please use 24C02 for EEPROM

some advices:

• LM324 can be replaced by OPA2274, which can further reduce the error between the load current set value and the actual value after testing.

• If there is self-excitation, the reserved resistor-capacitor components around the LM324 amplification feedback circuit can be adjusted according to the actual situation.

• Under normal circumstances, C23, C37 and R23 do not need to be welded

• • -

1. Project details

Electronic load is a device that consumes electrical energy. Its main function is to test the power supply. Compared with traditional passive loads such as high-power adjustable resistors or electric furnace wires, electronic loads have many advantages such as adjustable parameters and ease of use. Whether you are a professional electronic engineering project developer or an amateur electronics enthusiast, the electronic load meter is one of the necessary equipment. For example, the directional theme of this competition is "DC Adjustable Power Supply Design". If you have an electronic load, it will be more convenient to debug the power supply and accelerate the development of the project.

Electronic loads can be divided into AC electronic loads and DC electronic loads based on the type of test power supply. In terms of function, there are four common types: constant current, constant voltage, constant resistance, and constant power. Because most of our common power supplies are constant voltage DC power supplies. When testing this type of power supply, the main test is its current output capability. Therefore, in most application scenarios, DC constant current electronic loads are the most common type. In terms of control methods, electronic loads can be divided into two types: numerical control and analog. Compared with electronic loads controlled by pure analog circuits, CNC electronic loads use digital control, which is more intuitive in parameter adjustment, has rich functions, is simple to expand, and can also easily realize test automation.

Based on the above analysis, I decided to make a CNC DC (constant current) electronic load. Since the CNC electronic load can realize the function of a pseudo-constant power electronic load through simulation, the word "constant current" is not explicitly stated in the name, but it is essentially a constant current electronic load.

When the project was established, the key technical parameters I set were:

1. Maximum test current: 10A
2. Maximum test input voltage: 36V
3. Maximum power: 120W
4. Current control range: 0.1~10A, control accuracy: 0.1A

The above target parameters and functions of the final product have been achieved.

2. Challenges faced by the project and problems solved

There are already many professional or non-professional electronic loads on the market, both analog and CNC. Most of the professional ones are very expensive, which is not cost-effective for individual enthusiasts. Most of the non-professional ones are too simple: the displays are mainly LED digital tubes and monochrome LCDs, and most of the functions are relatively simple, and some lack some key protection functions. In addition to commercial products, there are also many public electronic load designs on the Internet, but most of them use non-standard and non-universal devices and modules, and the quality of the final product fluctuates greatly. Based on this, this project will mainly solve the following problems

1. High cost problem: try to use common and economical components to achieve functions
2. Simple function problem: In addition to the basic constant current load function, the finished product will also implement more useful functions
3. Problems with crude display and control: Use a large-screen TFT color screen as the display device, a rotary encoder as the input device, and use mainstream graphics libraries to implement a modern UI interface.
4. Difficult production issues: Reduce the tools and steps required for production and improve the success rate of production.

3. Key points involved in the hardware and software parts of the project

Hardware part (analog circuit part)

Core principles

The core circuit of the constant current electronic load is a negative feedback control circuit composed of an operational amplifier, power MOS tube, sampling resistor, and reference voltage: the current flowing through the sampling resistor will produce a voltage drop. When the voltage drop VRS is less than the reference voltage REF_VOL, The op amp controls the MOS tube to turn on. At this time, the current flowing through RS begins to rise. When the voltage drop generated on RS is higher than the reference voltage, the op amp turns off the MOS tube and the current flowing through RS begins to decrease. This cycle repeats so that the MOS tube works in a certain position in the linear region to control the load current. The final constant current value of the load is determined by the sampling resistor RS and the reference voltage REF_VOL.

Power ground, analog ground and digital ground

The entire system should be grounded, which is the basis for circuit operation. However, the ground planes of different logic circuits cannot be directly connected together through copper laying, otherwise there will be crosstalk between high-frequency digital circuits, high-current power circuits and control/sampling analog circuits. The best way is to make a single point connection between the three grounds. In this project, a 0 ohm resistor is used to connect the input power ground and analog ground. Due to the design of this project, the digital ground and analog ground are originally on two different PCBs, and then connected through pin headers, which itself is a good isolation measure. In this way, the connection between the three reference ground planes is achieved and the interference between the three logic circuits is minimized.

power supply

The entire system of this project requires 3 different voltages, 12V is used to power the fans of the cooling system. 9V is used to supply the main op amp to obtain a larger dynamic range and drive the power MOSFET well. 3.3V is used for power supply of precision op amps, ADC/DAC, and digital circuits. Because the power consumption of the entire system is not large, and the entire system has relatively high requirements for power supply purity, the design here uses a 12V power input, and uses a linear regulator to obtain 9V and 3.3V voltages. In addition, a relatively high voltage is added. More capacitors for filtering.

Constant current control and reference voltage acquisition

It can be seen from the basic principles that the reference voltage and the size of the sampling resistor can be adjusted to control the size of the load current. But controlling the size of the resistor is obviously much more difficult than controlling the level of the voltage. So we choose to set the reference voltage to control the load current. For traditional analog electronic loads, the reference voltage here is generally obtained through a potentiometer connected between the voltage reference and ground. If you want to achieve digital control, you only need to replace this potentiometer with a digital potentiometer. Considering that the accuracy of digital potentiometers is generally not high, a better choice is to use a DAC, that is, a digital-to-analog conversion circuit to generate the reference voltage.

There are many ready-made ICs available for DAC. Here, I chose the MCP4725 DAC. This DAC has a 12-bit resolution, which means that it can ultimately achieve 4095 current levels. In addition, MCP4725 communicates with the MCU through the I2C interface, which can simplify PCB wiring difficulty.

Since the driving capability of the direct output of the DAC is weak, the output voltage is easily affected by the load. In order to stabilize the output voltage, an operational amplifier is used here as a voltage follower to ensure that the output voltage of the DAC has sufficient driving capability.

After passing through the op amp, in order to facilitate the adaptation of the feedback resistor, a precision potentiometer R21 is added here before being used as a reference voltage input to the feedback system. In this way, the output voltage range of MCP4725 changes from 0~VCC to 0~any value, which can facilitate us to choose the appropriate feedback resistor. In addition, two resistor pads R15 and R17 are reserved on both sides of the potentiometer. After the final debugging is completed and the value of the potentiometer is determined, the potentiometer can be replaced with a fixed resistor, which can increase stability and further save money. cost

Selection of sampling resistor

由于采样电阻最大的设计通过电流为10A。如果使用大阻值采样电阻，比如使用1欧姆的电阻时，电阻上的功率将达到100W，那么采样电阻就会因为发热而烧毁，或者因为发热引起的温漂导致电流的测量或控制出现偏差。但是如果直接使用小阻值的采样电阻，如1毫欧的电阻，那么最大电流时压降也不过10mV，这一方面使得整个电路对干扰及其敏感，另一方面由于运放的精度限制，电流将难以控制。综合考虑后，我使用了2个6毫欧2W的电阻（R8和R10）并联获得3毫欧的采样电阻阻值，同时，增加一路前置放大电路，将采样电阻上的电压经过放大后送入主反馈电路。这样一方面兼顾了采样电阻的功率，另一方面也解决了反馈系统信号太小导致容易干扰而且难以控制的问题。

至此，我们可以通过MCU控制MCP4725来决定恒流电流的大小，也就实现了数控恒流负载的功能。

电流的采样

现在，我们可以控制负载电流的大小了，但是，我们还需要实时的测量电流的大小。测量电流的方法很简单，通过测量采样电阻上的压降，然后除以采样电阻的阻值即可得到电流的大小。因为要测量电压的大小，所以这时候我们就需要用到ADC，即模数转换电路了。与DAC相同，ADC有很多现成的IC可供选择。这里，经过综合评估后，我选择了ADS1115这款ADC。ADS1115拥有16bit的分辨率和4路模拟输入，且内置电压基准源，并同样使用I2C与MCU进行通讯。由于I2C总线的特性，ADS1115和MCP4725可以挂载在同一个I2C总线上。这就进一步降低了布线的难度。

因为ADC采样的精度很大程度上会受到电源的影响，所以ADS1115使用了LC滤波来保证ADS1115能有一个稳定的供电环境。

ADS1115的量程可配置，这里为了方便起见，使用了略高于供电电压3.3v的档位4.095V。因为使用的采样电阻很小，即使在最大电流时压降依然很低，无法充分利用ADS1115的量程，所以又增加一个INA199放大器来将采样电阻上的压降放大后采样。

输入电压采样和四线测量法

电压的测试方法很简单，两个电阻连接在输入端子的正极和地上组成分压电路，分压后输入ADC（ADS1115）采样即可，如下图中的R4和R5即分压电阻，分压比约为1：0.09，这样在输入36V时，分压后的电压约为3.27V，正好可以最大化的利用ADC的量程。

这里需要注意的时分压电阻的阻值，如果阻值太低，那么会带来额外的电流消耗，导致恒流控制不准。另一方面，这些电流消耗会导致分压电阻温度上升，带来温漂的同时还可能导致分压电阻烧毁！但是分压电阻太大也不好，太大的话一方面会导致系统抗干扰能力下降，因为环境中的微小干扰加在一个巨大阻值的电阻上时就能感应出很高的电压。另一方面，ADC是有输入阻抗的，如果分压电阻太大甚至接近了ADC的输入阻抗，那么分压电阻会和ADC的输入阻抗之间再次构成一个分压网络，导致ADC测得的电压比实际偏小。所以经过综合考虑，这里使用了100K的上臂电阻和10k的下臂电阻，可在两者之间取得平衡。

另一个需要注意的是电压的采样点，如果采样点距离输入端子比较远，那么由于PCB铜皮的电阻不为0，在电流较大时，PCB铜皮上也会产生可观的压降。这样，采样得到的电压就不是输入端子处的电压了。所以，这里电压的采样点一定是在靠近输入端子的地方，如下图：

这样就能比较精确的采集输入端子处的电压了。已经可以应付大多数的测量需求。但是，这里我们无论多么努力，都只能尽可能准确的测量输入端子处的电压。而从电源到输入端子处肯定是要连接导线的，而只要有导线就会产生压降，所以我们无法在有电流的情况下准确的测量电源的实际输出电压。在只需要考虑电流大小的测试场景下，这是无伤大雅的。但是如果是需要考虑功率而且对功率精度要求比较高的场合，比如测试电池的容量和能量时。因为通过测得的电压和电流计算出的消耗功率并不是实际电源的输出功率，所以测试出来的结果就会不准。为了解决这个问题，我们就必须将电压测试点一步到位，直接连接到电源的输出点上。但是这里还有个问题，那就是简单的将电压采样点引出一根线放到电源输出点上还是不行的，因为我们整个系统的地平面参考点是连接在输入端子的负极上的。而输入端子的负极到电源的负极之间也有导线，也存在压降。这里我们绝对不能直接将接地参考点引出连接到电源的负极上，这会导致电路设计难度增高，抗干扰能力下降，而且可能会导致严重的故障（想象一下如果使用时忘了连接接地参考线会怎样）。为了解决这个问题，就必须使用四线测量法了。

四线测量法其实是额外引出了两根用于电压采样的线，这样与原来的两根连接线一起就组成了4线测试系统。额外引出的这两根电压测试线直接连接在电源的输出端子上，其中没有电流或者只有很微弱的电流，通过测量这两根线之间的电压，就可以得到准确的电源输出电压了。但是改如何实现这个功能呢，要知道电压测试线的负极对于这个系统来说，它的电位可是负的。而一般的ADC是无法测量负压的。所以我们要用到差分测量法，也就是对两根线上的电压做一个减法，来得到最终的电压。在电路的具体实现上，虽然ADS1115具有差分测量功能，但是使用这个功能会导致其模拟通道数量变为2，无法满足我们的需求（电流，电压，4线电压）。所以，这里采用了一个独立的运放组成一个差分放大电路，将电压测试线两端的电压差分、放大（衰减）后送入ADS1115采样。

运放的选型上，考虑到使用4线的场合都是对电压精度要求比较高的。所以这里采用了润石科技的精密、轨到轨单运算放大器RS8557XF，它仅20uV的输入失调电压和低至0.03uV/℃的温漂可以完美的实现我们的需求。

保护系统

除了软件上的保护外，硬件上为了安全起见是一定要增加保护的，具体实现为：

测试电源回路上的大功率防反接二极管

测试电源回路上的保险丝

ADC采样端的限压二极管

散热系统的设计

因为电子负载消耗的电能主要是转化为热能，所以必须要有一个稳定而且强大的散热系统，否则电子负载的功率就无法做大。这一方面要求散热片必须有足够的散热能力，可以将热量迅速散出；另一方面则要求功率管与电子负载有比较可靠的结合方式，否则热量如果无法及时从功率管上传输至散热片，那即使散热片散热效率再高，整个系统的功率也做不上去。

比较传统的做法都是直接使用预制铝制散热片配合风扇进行散热，然后功率管通过打孔的方式与散热片进行结合，此方法比较节省成本但是通用产品大多无法针对风扇进行针对性的风道调整，所以散热效率不佳，再者就是订制散热片，但是这样制作成本就太高了，对于非商业化的项目十分不划算。

经过考虑，最终选择了使用Intel 775规格的散热器，并使用类似CPU的压合方式与功率管进行结合。此为经过我考虑后的最佳解决方案，原因如下：

1. 775散热器是专业且成熟的标准，经过了市场的验证。
2. 有大量在售（或二手）的商业化产品可以购买，可选规格多，价格便宜。
3. Intel 775散热器的扣具安装十分简单，只需要在PCB上打4个孔即可，
4. 相较于最新的115X散热器扣具，775规格的安装孔孔距更小，可有效减少PCB板的变形
5. 压合的方式相对与打孔螺丝的结合方式更加方便

实际验证，表现超出预期，可以很轻松的处理功率管100W以上的发热。

另外，功率管附近我设计了一个热敏电阻来对功率管的温度进行采样，并根据采样获得的温度来对风扇的转速通过PWM调速进行控制，即节能静音，也可以在关键时刻如热失控时控制负载关闭来保护功率管防止烧毁。因为这套系统的作用，在整个制作过程中未发生一起MOS管烧毁事故:)

硬件部分（数字电路部分）

显示和控制部分

显示部分采用了2.4寸的6.5K色QVGA分辨率的TFT显示屏，比起小尺寸的OLED或者单色STN显示屏，彩色的TFT显示屏可以同屏显示更多内容(或更加明显的显示内容)，也可以提供更加优良的观感，并且为以后的功能扩充留下了更多的余地。并且，彩屏TFT并未明显增加成本。

用户控制部分使用了一个带中心按键的旋转编码器和两个按键，旋转编码器用于快速选择选项或者设定数值，两个按键则一个实现控制另一个作为功能键使用。

这里还提供了两个指示灯，一个作为负载运行的指示灯，另一个则作为电源指示灯。

主控的选择

主控选择上，因为使用了大屏TFT，所以小容量的单片机就无法满足要求了。而且大屏的内容绘制也更加复杂，仅使用底层的绘图法绘制内容将极大的增加编码的难度，所以需要专用的使用图形库来进行内容绘制。另外还有显示汉字需要存储的字库。这要求主控MCU的ROM容量和RAM大小都不能太小。而且因为TFT屏选择的是SPI接口的，所以选择的主控最好还能有高速的硬件SPI。

经过综合考量，最终选择的主控为意法半导体STM32系列的STM32F412RET6，这款MCU采用带单精度浮点运算器的Cortex-M4内核，工作在100MHz的频率。拥有512KB的ROM和256KB的大容量SRAM，后者是流畅运行图形库的基础。外设上，它拥有高达50MHz的硬件SPI，支持DMA，可以以极高的速度驱动TFT显示屏（实测最高刷新率可以达到40FPS以上）。并且与ADC、DAC等的通讯也可以使用纯硬件IIC进行，以达到最高的运行效率。

另外，虽然此MCU拥有内置的RC时钟电路，但是为了保证进行电池容量测试时数值累加的准确性，这里还是使用了外置的晶体振荡器来获得准确的时钟。

非易失型储存器

整个系统需要对校准数据和用户设计以及测试数据进行存储。虽然可以使用MCU内部的Flash ROM来解决，但是考虑到以后的功能扩充，MCU内部的Flash容量其实不算宽裕。另外将数据保存在MCU中会导致MCU内部Flash的磨损，当出现坏块时MCU就报废了，这是难以接受的。所以我们需要外部的非易失型储存器来进行数据的持久化存储。常用的非易失型储存器有EEPROM, NOR Flash，NAND Flash等。NAND Flash虽然有容量（非常）大且价格便宜的特点，但是它的控制和数据的读取/写入都比较麻烦，而且需要磨损均衡算法来规避坏块，况且大多数情况下我们都不需要用到这么大的容量，所以NAND Flash用在这里是不合适的。剩下的EEPROM和NOR FLASH虽然在功能上都能较好的满足需求，但是EEPROM大容量的很贵，NOR Flash写入前需要擦除，无法随机写入。考虑到随机存储需求，比如测试数据的存储（防止意外掉电丢失），最终本项目选择了同时使用一片小容量EEPROM和一片大容量的NOR Flash， EEPROM作为数据快速的随机存储使用，NOR Flash用来存储不要经常变动的那些数据（如校准数据）。

EEPROM的选择上使用了台湾合泰的HT24LC64，IIC接口，拥有8KB的容量。足以满足随机数据存储满足要求：

NOR Flash选择了兆易创新（GD）的GD25Q64CSIGR，拥有高达8MB的容量和Quad-SPI接口，数据擦写次数10万次，20年的数据持久性。可以在最大程度上满足校准数据的稳定存储。

软件部分

开发环境

项目使用Atollic TrueStudio 9.3.0 IDE进行开发，通过STM32 CubeMX生成初始化工程，开发语言为C语言，编译器为GCC。

图形库

图形库使用LittleVGL，一个开源、强大的嵌入式图形库。littlVGL提供了丰富的控件，还包括了一个软件任务系统和内置的内存管理器，这些都可以极大的降低开发的难度。

LittleVGL需要一个用于存储控件属性的堆空间，和一到二个图形缓冲区。本项目中，为LittleVGL分配了32KB的堆空间和50KB的图形缓冲区，并使用DMA进行屏幕数据的刷新。 LittleVGL主页：LittleVGL< br> < br>

代码结构

代码主要分为3层：硬件驱动层，硬件抽象层和用户交互层

硬件驱动层用于底层硬件的沟通和交流，比如ADC数据的采集，DAC控制，负载开关等等。硬件抽象层则用来作为用户层代码和硬件层代码之间的桥梁，它屏蔽了底层硬件的细节，只向用户态代码提供加工好的数据或者提供硬件控制的高级接口。用户交互层则直接负责界面的绘制,功能逻辑的实现等等。

系统功能层级

系统功能层级如下图所示

有关更详细的功能介绍，请参考项目视频。

校准功能的实现

为了消除硬件上的偏差导致的测量误差，本项目设计了软件校准功能。校准的基本思路是使用多段拟合的方法，在整个的电压/电流的量程上设置多个校准点，记录下各个采样点的电压/电流值和对应的实际ADC值。测量时将采样获得的ADC值与这些校准点进行比对，找到对应的区段。然后在区段内采用线性插值的计算方法获取最终的电压/电流值。校准点越多，最终的结果就会越精确，但校准所需要记录的数据就越大。 经过综合的考量，最终电压校准点步进为为0.2V，电流为0.1A 以下为获取电压值的关键代码：

uint16_t GetTestInputVoltage(){
    uint16_t value = GetTestInputVoltageRawValue();

    // 校准数据不存在的情况
    if(voltageCalData == NULL){
        double volPerBit = 1.379282892314428;
        return value*volPerBit;
    }

    // ADC采样值小于零点值的情况
    if(value <     = voltageCalData->data[0]){
        return 0;
    }

    // 找到对应的区段，并进行线性插值，计算实际的电压值
    uint16_t voltage = 0;
    for(int i = 1; i <      voltageCalData->stepCounts; i++){
        if(value <     = voltageCalData->data[i]){
            voltage = (voltageCalData->stepVoltage) * (i - 1)
                    + (value - voltageCalData->data[i - 1]) * (voltageCalData->stepVoltage) / (voltageCalData->data[i] - voltageCalData->data[i - 1]);
            return voltage;
        }
    }

    voltage = (voltageCalData->stepCounts - 1) * voltageCalData->stepVoltage * value / voltageCalData->data[voltageCalData->stepCounts - 1];

    return voltage;
}

恒流控制

因为DAC的输出值和最终的恒流值之间是存在一定的误差的，而且由于采样电阻的温漂原因，此误差会不停的变化。考虑到代码实现的复杂度和功能需求，此处使用软件校准法。即在使用前进行一次校准，通过记录下不同的恒流值对应的DAC设定值，然后在设定电流时，根据此校准表查找到对应的DAC值并设定。在实际验证时，恒流值误差在±2%以内，基本满足使用需求。

*** NOTE：如果需要更加精细的电流控制，此处可以使用PID控制。此功能可以作为本项目以后的改进点之一。

其他功能的实现关键点

恒功率电子负载

恒功率电子负载功能是通过纯软件实现的，通过不停的采样当前的电压，并根据设定的功率计算出应该设定的电流，然后通过恒流控制功能重新设定负载电流，以此来实现恒功率负载功能。可见，此功能无法应付电压高频变动的情况。

电池容量测试

电池的容量（I·t）和能量（p·t），通过定时器每隔100ms采样一次电流值和功率值，并进行累加计算得出。这些数据被实时的存储在EEPROM中，这样，即使意外断电测试数据也不会丢失。

四、项目材料清单

控制板材料清单

模拟板材料清单

五、项目图片

含有大赛LOGO特写图片

其他图片

正面

侧面

控制板和模拟板分离照

实际运行图1

实际运行图2

性能展示——电压准确度

性能展示——电流准确度

性能展示——最大负载功率和四线电压

六、演示视频链接

自我介绍

90后男生，目前从事服务业工作。业余电子爱好者。

Video link: https://www.bilibili.com/video/bv1Y5411Y7TS

Project Description

CNC DC electronic load can be used to test constant voltage power supplies. With energy statistics and battery capacity testing functions.

Video link: https://www.bilibili.com/video/BV1Aa4y1a7RQ/

function display

Introduction to the current progress of the project and display of completed functions:

Video link: https://www.bilibili.com/video/BV17f4y1Q7cg/ < br> < br> Demonstrate the function of electronic load through examples:

Video link: https://www.bilibili.com/video/BV15D4y1m7JF/

7. Other descriptions of the project

This project is released to the public for the first time

This project has open sourced all the code and the code has been uploaded to the attachment.

The binary file suitable for STM32F412RE has also been uploaded to the attachment and can be directly burned and used.