Design and Testing of Electric Drive System for Maize Precision Seeder

Abstract

To improve the expandability, seeding accuracy, and operating speed range of the electric drive system (EDS) of precision seeders, this study constructed an EDS based on a controller area network (CAN) bus and designed a motor controller based on a field-orientated control (FOC) algorithm. Full-factorial bench and field tests based on seed spacing (0.1, 0.2, and 0.3 m) and operating speed (3, 6, 9, 12, and 15 km/h) were carried out to evaluate the performance of the EDS. The results of bench tests showed that seeding quality varied inversely with operating speed and positively with seed spacing. The average quality of feed index (QFI) at 0.1, 0.2, and 0.3 m seed spacing in bench tests was 88.38%, 96.67%, and 97.36%, with the average coefficient of variation (CV) being 20.13%, 16.27%, and 13.20%. Analysis of variance confirmed that both operating speed and seed spacing had a significant effect on QFI and CV (p < 0.001). The analysis of motor rotational speed accuracy showed that the relative error of motor rotational speed above 410 rpm did not exceed 2.24%, and the relative error had less influence on the seeding quality. The average QFI was 85.93%, 95.91%, and 96.24%, with the average CV being 21.12%, 15.50%, and 16.49% at 0.1, 0.2, and 0.3 m seed spacing in field tests. The methods and results of this study can provide a reference for the design and optimization of the EDS in a maize precision seeder and provide an effective solution for the improvement of maize yields.

1. Introduction

Land desertification and population growth have left most countries facing a severe food shortage in the future [1,2]. In this context, improving crop yields has become one of the important research hotspots at present [3,4,5]. Of the many ways to increase yields, improving seed spacing uniformity is especially critical [4,6,7]. The main reason for this is that seed spacing uniform effectively reduces competition for water, nutrients, and light during seed growth and development, which in turn results in higher yields and increased returns [8,9,10,11,12].

A precision seeder is the main carrier to achieve uniformity in seed spacing, and its seed spacing uniformity can effectively ensure a yield [13]. Traditional precision seeders use a mechanical drive system (MDS) to achieve the rotation of the seed meter plate. The MDS consists of a ground wheel, chain, and sprocket. Ground wheel slippage and chain jumping in MDS lead to fluctuations in the rotational speed of the seed meter plate, which in turn can lead to poor seeding quality in precision seeders [13,14,15]. In order to improve the impact of MDSs, EDSs with electric motors as drive elements have become a current research hotspot [2,16].

EDSs are common in developed countries [16]. The EDS developed by Horsch of Germany can achieve an operating speed of 15 km/h while ensuring seeding quality [17]. The vDrive motor developed by American Precision planting company can meet the operation demand of 16 km/h, which greatly improves the operation efficiency [18]. In Italy, the company Maschio launched the CHRONO high-speed seeder with an EDS that can achieve seeding in 15 km/h [19]. The above EDSs meet the high-speed requirements at the same time, with the use of a controller area network (CAN) bus based on the ISOBUS 11783 protocol, which can achieve the expansion of different rows of seeders, further improving the scope of their application. Meanwhile, the EDS with advanced motor control algorithm ensures seed spacing uniformity.

The widespread use of precision seeders with an MDS in developing countries is a serious constraint on increasing yields and is not conducive to improving farmers’ incomes and ameliorating possible future food shortages. Although the use of EDSs in developing countries in applications is relatively lacking, their superiority has attracted the attention of many researchers. Yang et al. [20] designed an EDS for a four-row precision seeder and carried out experiments, and the results of these experiments showed that under the three operating speeds of 9, 11, and 12 km/h, the quality of feed index (QFI) of the EDS was increased by an average of 4.7% compared with that of the MDS, and the miss index (MI) was reduced by an average of 3.54%, which proved that the EDS was better adapted to seeding. Cay et al. [21] found that the EDS had a lower fuel consumption, with a 22% year-over-year reduction in fuel consumption. In the above studies, EDSs were designed and found to be advantageous, but the expandability of the system was not considered, and the designed EDSs were only adapted to seeders with the corresponding number of rows. Therefore, He et al. [22], Ding et al. [23], and Yang et al. [24], on the other hand, designed an expandable EDS based on a CAN bus to adapt to different rows of seeders. However, they neglected the improvement of seeding quality by the control algorithm. Also, the operating speeds in the above EDS were within 12 km/h, and high-speed operating situations were not considered. Du et al. [25] designed an electric drive unit, and bench experiments were conducted to compare the effects of square wave control and field-orientated control (FOC) algorithms on motor control and seeding quality, and the results proved that the FOC algorithm could achieve a higher seeding quality by virtue of its good rotational speed accuracy. However, they did not apply the algorithm to the actual EDS, and there is still a large upside in terms of the seeding quality.

The high cost of products of developed countries has prevented EDSs from spreading rapidly in developing countries. Moreover, the EDS in developing countries is deficient in terms of expandability, seeding accuracy, and operating speed. Therefore, the adaptability and high efficiency of the EDS cannot be further improved, which could be the main reason limiting the rapid promotion and application of EDSs in developing countries.

In order to solve the above problems, this study aims to achieve the following objectives: (1) construct an expandable EDS based on a CAN bus to adapt to maize seeders with different row units; (2) design a motor controller based on the FOC algorithm for precise control of the motor rotational speed; (3) conduct an testing study to validate the system performance.

2. Materials and Methods

2.1. System Components

The EDS structure is shown in Figure 1. The EDS consists of a master control unit, a seeder lift status detection unit, and several seeding control units. The master control unit includes an Android terminal (Nongxin Technology Co., Ltd., Beijing, China), a GNSS antenna, and a 4G antenna, which is mainly responsible for receiving satellite signals, network signals and human–machine interaction. The seeder lift status detection unit includes a seeder lift status detection controller and a travel switch, which is mainly responsible for detecting the lift and fall status of the seeder. The seeding control unit includes a permanent magnet synchronous motor (PMSM) (Nongxin Technology Co., Ltd., Beijing, China), a motor controller, and an air-suction seed meter (Shandong Dahua Machinery Co., Ltd., Jining, China), and the PMSM is coaxially connected to the seed meter. The motor controller is installed at the end of the motor. Communication between the motor controllers and the terminal via the CAN bus allows the EDS to be adapted to precision seeders with different row units. The 120 Ω termination resistors at both ends of the CAN bus are used for impedance matching to improve the CAN bus interference immunity.

When the motor rotates, the seed meter plate connected to the motor rotates synchronously. The reduction ratio between the two is 32, and the diameter of the seed meter plate is 220 mm. The seeds are adsorbed on the seed meter plate (4.5 mm in diameter) under the action of negative pressure (4 to 6 kPa) and then pass through the double-sided seed-cleaning knives to scrape off the excess seeds. The seeds are finally transported to a seed feeding point, which operates without negative pressure, and placed into the seed tube under the influence of gravity and initial speed.

2.2. Working Principle

The working principle of the EDS is shown in Figure 2. The seeder lift status detection controller obtains the status of the tractor’s three-point suspension bar according to the travel switch and then identifies the seeder lift status and sends it to the CAN bus. The Android terminal, as the main controller, acquires the operational speed of the seeder through the built-in board, external GNSS antenna, and 4G antenna in network RTK mode and calculates the target rotational speed of the motor according to the seeding parameters based on Equation (1). At the same time, the terminal also receives the seeder lift status, and, in combination with the motor start/stop status input by the user, it determines the enable state of each motor and sends it, together with the motor target rotational speed, to the CAN bus.

$$R_m={\displaystyle \frac{50\nu i}{3n_{\mathfrak{p}}d}}$$

(1)

where R_m is the motor target rotational speed, rpm; v is the operating speed, km/h; i is the reduction ratio between the motor and the seed meter, 32; n_p is the number of holes in the seed meter plate, 26; and d is the seed spacing, m.

2.3. Design of Communication Protocol

The communication protocol of the EDS is developed with reference to the ISO 11783 protocol. The ISO 11783 protocol is an international standard for communication between agricultural machinery and equipment [26]. The protocol data unit (PDU) is the basic unit of transmission in ISO 11783 [24]. The PDU consists of seven parts including the priority (P), reserved bit (R), data page (DP), protocol data unit format (PF), protocol specific data unit (PS), source address (SA), and data domain [27]. With reference to ISO 11783 regulations and the EDS architecture, the PDU information of the EDS is shown in

Table 1. PDU information.

Equipment	P	R	DP	PF	PS	SA	Parameter Group PCN	PDU Identification
Controller	6	0	0	0xFF	0x22	0x80	00FF22	18FF2280
Android terminal	2	0	0	0xFF	0x23	0x26	00FF23	18FF2326

.

The data domain hosts the actual application data. According to the PDU identification, the EDS uses single-frame mode to transmit data, and the protocol is shown in

Table 2. Data domain protocols.

Device	Byte1	Byte2	Byte3	Byte4	Byte5	Byte6	Byte7	Byte8
Controller	/	Row number	Motor/seeder lift status	Motor actual rotational speed		Motor target rotational speed		/
Android terminal	/	Row number	Motor status	Motor target rotational speed		/	/	/

. Among them, the row number is used to distinguish the type of controller; ‘0’ means the seeder lift status detection sensor, ‘1’ means the motor controller in the first row, ‘2’ means the motor controller in the second row, and other sequential analogous types of controller.

2.4. Field-Orientated Control Algorithm

The schematic diagram of the FOC algorithm is shown in Figure 3. The currents I_a and I_b of PMSM are obtained by a current sampling process, and the current I_c can be obtained based on Kirchhoff’s Law, as shown in Equation (2).

$$I_c=-(I_a+I_b)$$

(2)

The two currents are transformed by Clark’s transformation to obtain the orthogonal axis currents I_α and I_β for the α-β coordinate system, and I_α, I_β are transformed by Park transformation to obtain the orthogonal axis currents I_q and I_d for the q-d coordinate system. I_q is the torque current component, and I_d is the excitation current component. Therefore, the target value of I_d, I_{d_ref}, is usually set to zero for torque maximization. The difference between the actual rotational speed, speed_act, obtained by the sensor and the target rotational speed, speed_ref, is fed into the PI controller to find the quadrature axis current I_q reference value I_{q_ref}. The actual (I_{q_act}, I_{d_act}) and reference (I_{q_ref}, I_{d_ref}) values of I_q and I_d are fed into the PI controller to find the quadrature axis voltages U_q and U_d, respectively. U_q and U_d are subjected to inverse Park transformation to obtain U_α and U_β for the α-β coordinate system, which are subjected to space voltage vector pulse width modulation (SVPWM) to obtain target voltage vectors (U_a, U_b, U_c) to achieve the FOC of the PMSM.

2.4.1. Coordinate Transformation

Coordinate transformation is used to solve the problem of the difficult calculation of AC quantities such as the voltage, current, and magnetic chain in the stator. The currents (I_a, I_b, I_c) in a three-phase stationary coordinate system (abc) are first projectively transformed into the two-phase orthogonal currents I_α and I_β in the two-phase stationary coordinate system (α-β) [25], as shown in Figure 4.

The relationship between the currents is shown in Equation (3).

$$\left[\begin{array}{c}{I}_{\alpha }\\ I_{\beta }\end{array}\right]={\displaystyle \frac{2}{3}}\left[\begin{array}{ccc}1& -{\displaystyle \frac{1}{2}}& -{\displaystyle \frac{1}{2}}\\ 0& {\displaystyle \frac{\sqrt{3}}{2}}& -{\displaystyle \frac{\sqrt{3}}{2}}\end{array}\right]\left[\begin{array}{c}{I}_a\\ I_b\\ I_c\end{array}\right]$$

(3)

For the convenience of describing the motion of the rotor, I_α and I_β are projected onto the two-phase rotating coordinate system (q-d), as shown in Figure 5. Where θ is the rotation angle of the rotor, its value can be obtained by the sensor.

The U_q and U_d in the qd coordinate system can be calculated by Equation (4), which is also known as the Park transformation.

$$\left[\begin{array}{c}{U}_d\\ U_q\end{array}\right]=\left[\begin{array}{cc}\cos \theta & \sin \theta \\ -\sin \theta & \cos \theta \end{array}\right]\left[\begin{array}{c}{U}_a\\ U_{\beta }\end{array}\right]$$

(4)

2.4.2. Space Voltage Vector Pulse Width Modulation

SVPWM outputs different space voltage vectors to drive the motor rotation by controlling the conduction of the MOSFET (Q1~Q6) in the three-phase inverter circuit (Figure 6) [28]. The three-phase inverter circuit consists of three sets of bridge arms, each of which contains two MOSFET arranged in upper and lower positions. There are eight combinations of upper and lower bridge arm switching, corresponding to eight space voltage vectors, including six non-zero vectors, U_l, U₂, U₃, U₄, U₅, and U₆, and two zero vectors, U₀ and U₇.

The magnitudes and positions of the eight space voltage vectors in the stationary coordinate system are shown in Figure 7, where six of the vectors divide the plane into six sectors (Ⅰ, Ⅱ, Ⅲ, Ⅳ, Ⅴ, and Ⅵ).

In each sector, two adjacent voltage vectors, as well as the zero vector, are selected to synthesize an arbitrary voltage vector according to the principle of volt–second balancing [29], as shown in Equation (5).

$$U_{\begin{array}{c}ref\end{array}}*T=U_{\begin{array}{c}x\end{array}}*T_{\begin{array}{c}x\end{array}}+U_{\begin{array}{c}y\end{array}}*T_{\begin{array}{c}y\end{array}}+U_{\begin{array}{c}0\end{array}}*T_{\begin{array}{c}0\end{array}}$$

(5)

where U_ref is the desired voltage vector, V; T is the PWM period, s; U_x and U_y are used to synthesize the two spatial voltage vectors of U_ref, corresponding to two of the six spatial voltage vectors mentioned above; and U₀ refers to the two zero vectors U₀ and U₇; and T_x, T_y, and T₀ are the action times corresponding to U_x, U_y, and U₀ in one PWM period, respectively.

The SVPWM allows the synthesis of space vectors of arbitrary direction and magnitude to achieve the FOC of PMSM [28].

2.5. Design of Motor Controller

The motor controller, tailored for the FOC in PMSM, comprises key components, as depicted in Figure 8. It incorporates two power supply modules (SCT2400, Silicon Content Technology Co., Ltd., Beijing, China) regulated to output 10 V (for gate drivers) and 3.3 V (for the MCU, magnetic encoder, and CAN transceiver). The heart of the system is an MCU (GD32C103CBT6, Zhaoyi Innovation Technology Group Co., Ltd., Beijing, China), leveraging USART, SWDIO for debugging, CAN (SN65HVD230, Texas Instruments Inc., Dallas, TX, USA) for communicating with Android terminal, TIMER0 for PWM generation, ADC for phase current sensing from operational amplifiers (OP-amp) (LVM324IPWR, Texas Instruments Inc., Dallas, TX, USA), and SPI for rotor angle acquisition from the magnetic encoder (MT6825, Magtek Inc., Shanghai, China). The PWMs were generated by the MCU drive gate drivers (FD6288Q, Texas Instruments Inc., Dallas, TX, USA), adjusting the MOSFETs (BSC0702LS, Infineon Technologies, Neubiberg, Germany) of the three-phase inverter circuit to achieve PMSM control.

The FOC process involves the MCU acquiring target speeds, rotor angle, and phase currents via CAN, SPI, and ADC. Based on these and the FOC algorithm (Figure 3), TIMER0 generates PWM signals, which, through SVPWM and gate drivers, produce a three-phase voltage to precisely control the PMSM.

It is worth mentioning that, in order to reduce the development period, the seeder lift status detection controller follows the motor controller, and the seeder lift status detection controller detects the triggering signal of the travel switch through the IO port and then obtains the lift status of the seeder.

3. Tests

In order to fully validate the seeding performance of the EDS, bench tests and field tests were carried out. The bench tests were used to evaluate the seed space uniformity of the EDS under a static environment and, at the same time, analyze the influence of the precision of motor rotational speed on seeding quality based on the relative error of the motor rotational speed. The field tests were used to validate the actual seeding performance of the EDS and to analyze the difference between bench tests and field tests regarding the seeding quality. NK815 maize seed was used in both bench tests and field tests. The seed had a half-horse tooth shape, and the 1000-seed weight was 312.05 g.

3.1. Bench Tests

Seed spacing and operating speed are important factors that affect seeding quality. Agronomy varies from region to region, and the seed spacing need is not consistent. In order to ensure the adaptability of the EDS, the range of seed spacing was set at 0.1–0.3 m. The operating speed greatly affects the operating efficiency, but due to the influence of the load of the seeder, the ground conditions, and the performance of the seed meter, the current seeder operating speed is generally within 15 km/h. Based on the range of seed spacing and operating speed, bench tests factor levels were set, as shown in

Table 3. Bench tests factor levels.

Test Factor	Level
Seed spacing/m	0.1, 0.2, 0.3
Operating speed/(km/h)	3, 6, 9, 12, 15

.

Bench tests were conducted in September 2023. The bench tests equipment is shown in Figure 9 and included a seed meter detector [30], an air-suction seed meter (Shandong Dahua Machinery Co., Ltd., Jining, China), a PMSM (Nongxin Technology Co., Ltd., Beijing, China), a motor controller, a CAN tool (CANalyst-II, Chuangxin Technology, Zhuhai, China), and a terminal (Nongxin Technology Co., Ltd., Beijing, China). The air-suction seed meter is installed on the seed meter detector, which provides negative pressure (6kPa) via the seed meter detector, and the PMSM is connected coaxially to the seed meter plate. The outer seed cleaning knife of the seed meter is adjusted to the 9th position (14 positions in total), and the inner seed cleaning knife is adjusted to the 6th position (9 positions in total). The CAN tool is connected to the CAN bus between the motor controller and the terminal and is used to read CAN bus data.

Before the tests, the seed meter detector and the terminal were configured with the same operating speed and seed spacing according to Table 3, respectively. The seeding quality was recorded by the seed meter detector, and this was repeated three times at each factor level. The target number of seeds detected by the seed meter detector was set to 1000. At the same time, the motor rotational speed returned by the motor controller (return frequency of 20 Hz) through the CAN bus was read in real time by the CAN tool during the process, which is used to analyze the precision of motor rotational speed control.

3.2. Field Tests

Field tests were conducted to further explore the actual seeding performance of the EDS. The same factor levels as those used in the bench tests were selected for the field tests (Table 3). The field tests were conducted in October 2023. Before the tests, the soil (brown loam) was tilled by a rotary tiller working at a depth of 0.15 m, to ensure that the soil was soft. The tests equipment is shown in Figure 10, including a 2104 tractor (Case IH, Racine, WI, USA), a six-row precision seeder (2BMYFQ-6D, Shandong Dahua Machinery Co., Ltd., Jining, China), and an EDS. The EDS was installed on the seeder. The double-sided seed cleaning knife configuration was consistent with that used in the bench tests. The parameters of the seeder are shown in

Table 4. Parameters of the seeder.

Name	Parameter
Overall dimensions/mm	2000 × 5900 × 1800
Weights/kg	1500
Working width/m	3.9
Number of rows	6
Row spacing/m	0.65
Seeding depth/mm	50
Fertilizer opener type	Double disk
Suppression wheel	V-shaped
Hitch way with tractor	Three-point suspension
Profiling mechanism	Machinery

.

Before the tests, six PMSMs were divided into three groups, and the three groups of motors were configured to 0.1, 0.2, and 0.3 m seed spacing, respectively. The seeding depth of each seeding unit was set at 0.05 m. The tractor rear hydraulic output was adjusted so that the rotational speed of the hydraulic motor of the seeder fan was increased as needed to ensure that the vacuum pressure of the seed meter was maintained at not less than 6.0 kPa. After the tests, 100 seed spacings per row were collected by digging out the seeds (Figure 11), translating to 200 seed spacings per group.

3.3. Evaluation Index

The quality of feed index (QFI), miss index (MI), multiple index (MUL), and coefficient of variation (CV) specified in ISO 7256-1:1984, “Sowing equipment—Test methods Part 1: Single seed drills” [31], were used as evaluation indexes in both the bench tests and the field tests. The difference is that the evaluation indexes in the bench tests were given by direct calculations from the seed meter detector, and the results of the field tests were manually calculated based on seed spacings.

4. Analysis and Discussion of Results

4.1. Analysis and Discussion of Bench Tests

4.1.1. Seeding Quality of Bench Tests

The seeding quality at three seed spacings (0.1, 0.2, and 0.3 m) and five operating speeds (3, 6, 9, 12, and 15 km/h) is shown in Figure 12. With the increase in operating speed, the overall trend of QFI decreased, and the trend of MI, MUL and CV increased; with the increase in seed spacing, the trend of QFI decreased, and the trend of CV decreased.

Table 5. Results of the bench tests.

Seed Spacing/m	Operating Speed/(km/h)	QFI/%	Average QFI/%	ED of QFI/%	CV/%	Average CV/%	ED of CV/%
0.1	3	98.30	88.38	24.20	13.00	20.13	11.00
	6	97.83			19.67
	9	88.90			22.00
	12	82.77			22.00
	15	74.10			24.00
0.2	3	99.00	96.67	6.13	11.33	16.27	10.67
	6	98.10			14.67
	9	97.67			16.67
	12	92.87			16.67
	15	95.70			22.00
0.3	3	98.07	97.36	2.23	10.33	13.20	4.67
	6	97.87			13.33
	9	98.13			15.00
	12	95.90			14.33
	15	96.83			13.00

shows the results of bench tests. In the range of 3–15 km/h, the QFI varied from 74.10% to 98.30%, 92.87% to 99.00%, and 95.90% to 98.07% at 0.1, 0.2, and 0.3 m seed spacing, respectively. The average QFI was 88.38%, 96.67%, and 97.36%, respectively; the extreme difference (ED) in the QFI was 24.20%, 6.13%, and 2.23%, respectively. The CV varied from 13.00% to 24.00%, 11.33% to 22.00%, and 10.33% to 15.00% at 0.1, 0.2, and 0.3 m seed spacing, respectively; the average CV was 20.13%, 16.27%, and 13.20%, respectively; and the ED in the CV was 11.00%, 10.67%, and 4.67%, respectively.

Based on the extreme difference data for QFI, it can be seen that the effect of increasing seed spacing on the decrease in QFI is not linear, but rather an accelerated deterioration process. The extreme difference in QFI at 0.1 m and 0.2 m seed spacing is 2.75 times and 10.84 times of that at 0.3 m seed spacing, respectively. The cause of this condition may be related to the rotational speed of seed meter plate and seed bouncing [2,16,32]. The decrease in seed spacing and the increase in operating speed lead to an increase in the rotational speed of the seed meter plate, which will result in less time for the seed adsorption and clearing process [16], thereby causing a decrease in QFI. Moreover, the increase in the rotational speed leads to an increase in the initial velocity of the seed when it is put into the seed tube, and the seed bouncing in the seed tube becomes violent, which leads to a decrease in QFI and an increase in CV.

4.1.2. Analysis of Variance

To further explore the effect of operating speed and seed spacing on seeding quality, an analysis of variance (ANOVA) was conducted, and the results of the ANOVA are shown in

Table 6. ANOVA of bench tests results.

Index	Factor	SS	Df	MS	F	p
Seed spacing	QFI	748.950	2	374.475	655.695	0.000
	CV	362.133	2	181.067	94.744	0.000
Operating speed	QFI	670.056	4	167.514	293.313	0.000
	CV	343.200	4	85.800	44.895	0.000

.

Table 6 shows that both operating speed and seed spacing had a significant effect on QFI and CV (p < 0.001). From the F-values, it can be seen that both seed spacing and operating speed had a greater effect on QFI than on CV, and seed spacing had a greater effect on QFI and CV than operating speed had on QFI and CV.

4.1.3. Accuracy Analysis of Motor Rotational Speed

The decrease in seeding quality may be related to seed spacing and operating speed on the one hand, but there may also be a correlation with motor rotational speed accuracy [15,33]. In order to clarify the effect of motor rotational speed accuracy on seeding quality, the motor rotational speed accuracy in the tests was analyzed. The motor target rotational speed, relative error of motor rotational speed, QFI, and CV value correspondences during the tests are shown in Figure 13.

Figure 13 shows that as the motor rotational speed increases, the QFI and the relative error of the motor rotational speed gradually decrease and the CV gradually increases. Among them, the relative error of the motor rotational speed does not exceed 2.24% when the motor rotational speed is more than 410 rpm; when the motor rotational speed is less than 410 rpm, the relative error of the motor rotational speed does not exceed 11.34%.

The reason for this condition is that at low motor rotational speeds, the resolution of the magnetic encoder is insufficient to accurately measure speed changes [34]. At the same time, the phase currents are small and the current sampling process is susceptible to noise. However, the QFI and CV corresponding to the low motor rotational speeds are at a high level. It can be assumed that the relative error of the motor rotational speed is low, having little effect on the seeding quality. The EDS’s motor rotational speed accuracy can meet the demands for precision seeding of maize within the ranges of 3–15 km/h and 0.1–0.3 m seed spacing. Therefore, the reduction in QFI and the increase in CV are mainly due to the seed meter and seed bouncing. The seed adsorption and clearing time decreases at higher motor rotational speeds, leading to a reduction in QFI. At the same time, excessive rotational speed in the seed meter plate will make the seed bouncing in the seed tube become more violent, resulting in a decrease in the seeding quality and an increase in CV.

4.2. Analysis and Discussion of Field Tests

4.2.1. Seeding Quality of Field Tests

The seeding quality of field tests at three seed spacings (0.1, 0.2, and 0.3 m) and five operating speeds (3, 6, 9, 12, and 15 km/h) is shown in Figure 14. With the increase in operation speed, the overall trend of the QFI was decreasing, and the MI, MUL, and CV increased. With the increase in seed spacing, the QFI first increased rapidly and then stabilized, and the MI, MUL, and CV first decreased rapidly and then stabilized. The overall trend is similar to that of the indexes in the bench tests.

Table 7. Results of the field tests.

Seed Spacing/m	Operating Speed/(km/h)	QFI/%	Average QFI/%	ED of QFI	CV	Average CV	ED of CV
0.1	3	98.02	85.93	31.65	17.94	21.12	6.08
	6	96.59			19.05
	9	87.92			21.54
	12	80.77			23.05
	15	66.37			24.02
0.2	3	97.56	95.91	6.73	11.61	15.50	10.22
	6	98.52			10.43
	9	98			15.69
	12	91.79			20.65
	15	93.66			19.13
0.3	3	99.5	96.24	6.82	12.19	16.49	8.61
	6	96.45			14.49
	9	98.51			15.38
	12	92.68			20.8
	15	94.06			19.58

shows the results of field tests. In the range of 3–15 km/h, the QFI varied from 66.37% to 98.02%, 91.79% to 98.52%, and 92.68% to 99.50% at 0.1, 0.2, and 0.3 m seed spacing, respectively. The average QFI was 85.93%, 95.91%, and 96.24%, respectively; the extreme difference in the QFI was 31.65%, 6.73%, and 6.82%, respectively. The CV varied from 17.94% to 24.02%, 10.43% to 20.65%, and 12.19% to 20.8% at 0.1, 0.2, and 0.3 m seed spacing, respectively; the average CV was 21.12%, 15.50%, and 16.49%, respectively; and the extreme difference in the CV was 6.08%, 10.22%, and 8.61%, respectively.

4.2.2. Comparison of Bench and Field Tests Results

Based on the extreme difference data of QFI in field tests, it is clear that smaller seed spacing has a significant effect on QFI. The extreme difference in QFI at 0.1 m seed spacing was 4.64 times that at 0.3 m seed spacing. In contrast, the extreme differences in QFI at 0.2 m and 0.3 m were similar, which may be related to the method used to obtain the seeding quality. Seeds were monitored in the bench tests by a monitoring sensor mounted on the seed tube, while, in the field tests, seeds were dug out manually. There was relatively little seed bouncing within the seed tube in the bench tests, whereas in the field tests seed bouncing was present throughout the seed tube and in the seed furrow. Seed bouncing affects seed spacing [35]; the effect of seed bouncing is more pronounced when seed spacing is small and less pronounced when seed spacing is large. As a result, the extreme differences in QFI are close at seed spacings of 0.2 and 0.3 m.

More seed bouncing in the field tests resulted in an overall decrease in seeding quality, with the average QFI at 0.1 m, 0.2 m, and 0.3 m seed spacing decreasing by 2.45%, 0.76%, and 1.12% compared to the average QFI in the bench tests and the average CV increasing by 0.98%, −0.77%, and 3.29% compared to the increase in the bench tests.

Overall, the EDS can meet the needs of a wide operating speed range (3–15 km/h) and a wide seed spacing range (0.1–0.3 m), and the motor controller designed based on the FOC algorithm can achieve high control accuracy. However, the seed meter performance and seed bouncing are important constraints that could limit further improvement in seeding quality, especially at high operating speeds. Therefore, it is necessary to optimize the seed meter based on the demand for high speeds and to improve seed bouncing in the seed tube and the furrow, which will help improve the yield in the future.

5. Conclusions

In this study, an EDS for a maize precision seeder based on a CAN bus was designed and a seeding controller based on FOC algorithm was developed to achieve the FOC of the PMSM. The seeding quality change rule for the EDS under different seed spacings and operating speeds was studied to determine the seeding performance of the system. Specific conclusions are as follows:

• To improve the expandability, seeding accuracy, and operating speed range, an EDS for a maize precision seeder was designed based on the CAN bus and FOC algorithm. A CAN bus communication protocol designed based on ISO 11783 standard can be applied to different row seeders. The seeding controller based on the FOC algorithm can effectively ensure the seeding accuracy and speed range.
• To explore the performance of the EDS and the change rule for seeding quality, bench tests were carried out. The results of the bench tests showed that seeding quality varied inversely with operating speed and positively with seed spacing. over a range of seed spacings (0.1, 0.2, and 0.3 m) and operating speeds (3, 6, 9, 12, and 15 km/h). The average QFI at 0.1, 0.2, and 0.3 m seed spacing in bench tests was 88.38%, 96.67%, and 97.36%, with the average CV being 20.13%, 16.27%, and 13.20%.
• To explore the effect of tests factors on seeding quality, ANOVA and rotational speed accuracy were conducted based on bench tests. ANOVA showed that both operating speed and seed spacing have a significant effect on QFI and CV (p < 0.001). Both seed spacing and operating speed have a greater effect on QFI than on CV, and the effect of seed spacing on QFI and CV is greater than the effect of operating speed on QFI and CV. The analysis of motor rotational speed accuracy showed that the relative error of motor rotational speed above 410 rpm does not exceed 2.24% and the rotational speed control error has less influence on the seeding quality.
• To further determine the performance of the system, field tests were conducted. The results of the field tests showed that the average QFI was 85.93%, 95.91%, and 96.24% at 0.1, 0.2, and 0.3 m seed spacing, and the average CV was 21.12%, 15.50%, and 16.49% in the range of operating speeds of 3, 6, 9, 12, and 15 km/h. Compared with the bench tests, the average QFI at 0.1 m, 0.2 m, and 0.3 m seed spacing decreasing by 2.45%, 0.76%, and 1.12%, and the average CV increased by 0.98%, −0.77%, and 3.29%, respectively.

Synthesizing the results of the above tests, the EDS meets the demand for expandability, seeding accuracy, and operating speed range. The seed meter performance and seed bouncing can limit seeding quality at high operating speeds, and further optimization of the seed meter and seed guide mechanism will be needed in the future.