Category: Boom Sprayers

Main category for sprayers with horizontal booms

How Cameras, Computers and Cake Recipes will Drive the Future of Weed Recognition

The idea of controlling weeds individually instead of treating the whole field uniformly makes a lot of sense. Why waste herbicides, till soil unnecessarily or use other weed control methods on areas without weeds? Besides reducing unintended environmental impacts, it means cost savings for the farmer, reduced crop stress and the opportunity to concentrate efforts where they are needed.

In this era of widespread herbicide resistance, the latter opens the door for new weed control tools – such as lasers or electrical weeding (Carbon Robotics as one example). Unfortunately, all the above absolutely relies on the ability to reliably recognize weeds in all manner of crop-weed conditions. Enter cameras, computers and cake recipes.

Besides alliteration, they are all connected by their role in real-time, in-crop weed recognition. So why isn’t weed recognition already widely available and how will this trio change that? Well, the way I look at it, if our faces were weeds, this problem would have been solved a long time ago. Said differently: new technologies mean that this challenge can be addressed with the necessary focus, investment, and research.

In the last few years, there have been step changes in research and development for real-time weed recognition, which are driving rapid gains for in-crop site-specific weed control. We are seeing this with the emergence of image-based green-on-green (GoG) see-and spray systems globally, many of which are listed in the table below.

Sensor	Location
AutoWeed	Australia
Agtecnic SenseSpray	Austraila
Bilberry	France / Australia
Carbon Bee – SmartStriker	France
DeepAgro	Argentina
EXXACT Robotics	France
GreenEye	Israel / USA
John Deere / BlueRiver	USA
OpenWeedLocator (OWL) – developed by the author as a DIY, open-source weed detection system.	Australia
Xarvio / Bosch / BASF	Canada / Europe

Yet, as the last 50 years of plant detection, identification and recognition research have shown, reliable weed recognition is a challenging problem to target. The aim of this article is to take you on the journey of weed recognition – from simple plant detection for thinning in the 1970s to every metre of a boom equipped with camera-computer-cake recipe combinations. Fortunately, we are well on this path toward more effective weed recognition.

Green-on-brown weed detection

As far as the available research shows, the first attempts at plant detection were made for thinning sugar beets in the early 1970s. The method is impressive in its simplicity – two sensors (photodiodes) that generate a signal based on incoming light intensity, are each covered by a filter that only allows specific wavelengths of light through. By knowing the reflectance spectrum for plants and comparing the ratio/output of these two sensors means you can detect if a plant has entered the field of view, but not necessarily exactly where it is.

In this case the weed detection ‘algorithm’ is a ratio of sensor values and some predefined threshold, which can be adjusted as a form of ‘sensitivity’. The concept is largely the basis of WEEDits and WeedSeekers today. This system has all the principles of advanced image-based systems: (1) data stream from a sensor/camera + (2) computer running a weed recognition algorithm + (3) some form of actionable output (e.g. turn on a nozzle).

Image-based weed recognition

With the fundamentals of SSWC largely consistent between sensor and image-based systems, the interesting details and drivers for GoG technology emerge if we dive into the data stream (images) from the camera and the algorithm running on a computer.

At its most basic, a digital colour camera is a sensor that generates signals based on incoming light intensity. The difference with the photodiodes is that a camera records this reflectance intensity information for every pixel in the camera across the red, green and blue (RGB) parts of the spectrum. For example, a 12-megapixel camera has 12 million pixels reporting reflectance intensity for each RGB channel. That means 36 million individual numbers generated for every photo. Learn more about the basics of digital imaging, here.

When you bring this together in an image, you have information on object relationships in space, providing not just a ‘spectral’ dimension (RGB) but also a ‘spatial’ dimension. The use of computers to understand image content is known as computer vision. Having all this data (colour and spatial information) means there is a lot more to work with when differentiating two plants, increasing your chances of success. The downside of having more to work with, is having to work with more! In this case, the computer needs to deal with the 36 million numbers it receives 30 times per second.

The next part of this weed recognition puzzle is the computer and associated weed recognition algorithms, which receive the incoming images and determine if there is a weed in the image. In the case of ‘conventional’ or non-convolutional neural network (CNN) methods (we’ll get to those later), this analysis process is largely formed of four stages shown in Figure 2 – (1) pre-processing, (2) segmentation, (3) feature extraction and (4) classification. If you’re interested in the details on the volumes of research done in this space, I’d highly recommend this review by Wang et al., 2019.

Figure 2 Overview of the image analysis process including convolutional neural networks (CNNs) that automate much of what is done by hand in the more conventional methods with manual feature extraction. Adapted from Wang et al. 2019.

In the case of a simple colour-based detection system that just needs to find green plants in fallow, like our DIY weed detector the OpenWeedLocator (OWL), the algorithm is largely a threshold on the green colour channel using the RGB colour space. This carries some risk – for example if the lighting changes substantially or if weeds aren’t green the method can break down. Yet we went down this path because of the simplicity and speed with which it can be used on the rather computationally constrained Raspberry Pi.

Our field testing also showed acceptable levels of performance in variable fallow conditions. We managed these issues by combining multiple colour-based algorithms; relying on greenness in the RGB colour space from the ‘excess green’ (ExG) vegetation index, combined with thresholds in the hue, saturation, and value (equivalent to brightness) (HSV) colour space to avoid false detections on over/underexposed regions, which often occur in stubble. Even with these adjustments, the system is prone to errors, but the benefit of using cameras is that they can be levelled up for in-crop detection to more advanced algorithms.

The ‘green detection only’ approach without machine learning that effectively exits Figure 2 at the segmentation stage is also likely in the initial launch of John Deere’s green-on-brown See & Spray Select ™. Probably for the reasons above, they also warn against use close to sunrise and sunset where lighting is changing rapidly. As expected, the system pivoted to in-crop detection with deep learning in early 2022 because images and embedded processors allow software-only changes for green-on-green.

If weed species classification or crop-weed discrimination is needed for green-on-green use, then the remaining two stages of feature extraction and classification are required. In the conventional process, someone selected which plant attributes (known as image features) you wanted to use, trained an algorithm on those features and then ran it in the field, a method generally known as machine learning. In spite of this more advanced approach the performance drop between the test dataset and the variable field conditions meant the method was still commercially unusable in large-scale systems. So, what has changed?

Well, in 2012 a research group managed to substantially outperform all these other methods with an algorithm known as a convolutional neural network (CNN). Instead of an ‘expert’ identifying which plant attributes in an image were important, the algorithm itself could select and learn which features were most important, making it more robust. The CNN effectively skips all the steps in the conventional process (Figure 2), instead replacing them with having large quantities of training images with weeds manually highlighted – a newfound bottleneck itself, but not insurmountable.

Part of CNNs robustness comes from the algorithms being capable of analyzing many dozens of features and combinations of features that wouldn’t necessarily be obvious to humans. In the training process, it tests one combination of features before correcting itself based on the training dataset you’ve provided and testing another. One of the biggest improvements is that by doing this feature extraction and selection process automatically, it removes the slow and somewhat qualitative process of deciding which features/plant attributes were most important. Besides this, the algorithms are incredibly large. Some of the modern CNNs have over 100 million (!) different dials that can be automatically tuned to learn the patterns of a weed.

Combined with these algorithms, we now also have low-cost (<US$150) credit-card sized computers such as the Raspberry Pi (when it has additional support) and Jetson Nano that can run these algorithms real time, or around 15 – 20 frames analyzed per second. Even when processing millions of pixels through CNNs with over a 100 million parameters 15 times per second, they consume very little power and can fit easily on agricultural equipment. Genuinely mind boggling every time I think of the sheer scale of it.

Figure 3 The fully assembled OpenWeedLocator without the cover showing all the parts necessary for a site-specific weed control system. Camera at the front, a credit card-sized computer in the middle (Raspberry Pi) and a relay control board at the back to activate solenoids for spot spraying. The OWL is an open-source, DIY weed detection system and can be accessed **here**.

That leaves us with the cake recipe – how does that fit? Well, the step-changes in technology, particularly deep learning for image analysis can be largely attributed to the use of open-source software, data and hardware. Accessible datasets gave rise to the first effective CNN; open-source deep learning libraries (e.g. Tensorflow and Pytorch) to widespread adoption and development and open-source and/or low-cost hardware to field-scale implementations of the work. The best analogy I have heard used to describe an open-source approach is that it’s like sharing the recipe for a cake – except the code/assembly guide are the recipe and the ingredient list all the tools/languages/packages/components required to make it work.

Even though I could make an average chocolate cake with the ingredients in my pantry, I’ll still go and buy one for many different reasons – quality, convenience or support/returns in case it doesn’t quite live up to standards. The emphasis in this approach is on the quality of the entire product experience not necessarily a secret combination or method of combining ingredients. Plus, it means everyone with the basic tools can try making the cake or training the algorithm, discovering opportunities for fixes, optimization or low-hanging fruit that may change its use case entirely.

I mean, Australians took sponge cake and made lamingtons! In my own experience with the OpenWeedLocator, we built a device for detecting green weeds in large-scale fallow situations. But in true open-source fashion, this has now been used for site-specific fungicide sprays, desiccant application and under trees for weed control. A Canadian innovation – AgOpenGPS – developed by Brian Tischler is an open-source GPS steering system for tractors, similarly enabling farmer-driven development. The examples of different uses are quite extraordinary.

One of the main tenets of open-source technology is that by allowing people to see the details of software and hardware, a larger and more diverse array of people can examine the code and any inefficiencies and errors can be picked up faster. Besides this, it makes research and development accessible to those that might need the technology – the farmers – instead of it being locked away in large companies with inaccessible customer support. Farmer-driven innovation has a long and successful history and open-source development facilitates this continuing to occur in the era of agritech.

Over these last 50 years of development, the cameras, the computers and the open-source recipes have each contributed at different points to site-specific weed control. It seems that now they are converging in agriculture in a storm of interest and development for weed recognition and targeted application.

July 7, 2022

Methods for Testing Nozzle Flow Rate
Calibration should be a regular practice for every operation that uses a sprayer. Part of that process is confirming that each nozzle is operating within the manufacturer’s specifications. This is a must for researchers that adhere to Good Laboratory Practices and for custom operators that sell their services. But we didn’t just fall off the turnip truck… we know nobody else does it. In fact, we’re surprised when we hear an operator HAS checked their nozzle flow.
And we get it. It can be awkward and time consuming. A field sprayer with 72 nozzle bodies and three nozzles in each position has a whopping 216 nozzles. A tower-style or wrap-around airblast sprayer has fewer nozzles, but the operator needs a ladder to reach them all and they don’t point straight down, so a tube must be used to guide the spray into a collection vessel.
And when pressed, any operator that does not regularly check their nozzles counters by saying “my tank empties in the same place every time, so why check them?” Even if the sprayer does start to go further on a tank, the operator can speed up or adjust the rate controller to drop the pressure a little.
Fair enough. This isn’t a hill we choose to die on.
But we will say that nozzles worn by even a few percent don’t only cause a change in flow rate, but may indicate a deteriorating spray quality and spray geometry. And, when one (or a few) nozzles are worn and others are not, it’s the same as when a single nozzle is plugged – the operator won’t be able to tell from the cab because the rate controller tends to mask the problem. And, if using PWM to apply a simultaneously reduced broadcast rate, perhaps the issue is amplified? All of this impacts coverage uniformity.
We’ll get off our soapbox now.
Over the years we’ve encountered many methods for determining a nozzle’s flow rate. We wanted to try each of them and characterize their accuracy, precision, time required, and ease of use. This is not a ranking where we wanted to find “The Best” method. The best method depends on your situation. If you’re a researcher, then accuracy and precision may trump time and expense. If you’re a custom applicator, then perhaps time is the critical factor. And if it’s your own operation, perhaps expense matters most. It’s up to you.
Method
The following tests were performed on a spray patternator table. A single nozzle was operated by a ShurFlo 2088-594-154 positive displacement pump. Pressure was set using a bypass regulator and an analog pressure gauge, confirmed by a SprayX digital manometer positioned under the nozzle body via a splitter. Room temperature water was used.
“Patty” the spray patternator table. Designed and constructed by Mohawk College, Brantford, Ontario.
Digital manometer on a splitter parallel with the test nozzle.
We tested ten nozzle flow rate measurement systems. There are others out there, but we limited the selection to farmer-oriented systems and not those used in mandatory government inspections.
1. Billericay Flowcheck
2. Delavan Calibration Cup
3. Graduated Cylinder
4. Greenleaf Calibration Pitcher
5. SprayX SprayFlow Turbo
6. SpotOn SC-1
7. SpotOn SC-2
8. SpotOn SC-4
9. Weighed Output
10. Lurmark McKenzie Calibrator
Three samples were taken from a new TeeJet XR8004 at ~40 psi and three samples taken from a new TeeJet AIXR11004 at ~70 psi. An exception was made for the Billericay Flowcheck which specified 43.5 psi (3 bar) for all sampling. All systems were emptied or dried as much as their design permitted between samples.
All data was converted to gallons per minute and the flow measured was compared to the calculated flow for the nozzle and pressure used. For example, if the manometer read 38 psi for the 8004, then the formula 0.4 x (38 psi ÷ 40) ^0.5gives us a calculated flow of 0.39 gpm. If the method reported 0.41 gpm, then it would be off by +5.1%.
Results
Consider the accuracy and the precision of each system when you review the results. Remember that precision means you get the same result with very little variation while accuracy means that on average you get the correct result. And, for context, remember that most recommend changing a nozzle when it is 10% more than the ideal flow rate. We prefer 5%, and if three or more nozzles are off spec, replace them all as a batch because they’re likely all very close. Compared to most spraying costs, a set of nozzles is not worth quibbling about. Some operators just change them annually and don’t bother with testing at all.
Billericay Flowcheck: This is a passive measurement system. You must select the nozzle size on the bottom of the collector and suspend the unit from the nozzle body. It’s designed for a horizontal boom and you’d have trouble using it with any other sprayer. You also have to set the pressure to exactly 43.5 psi (3 bar). While fairly accurate, it spanned about +/-2.5% off ideal. You have to read from the right scale, which in this case was red and rather difficult to read because of the low contrast. It took about two minutes to reach equilibrium for each reading and a lot of liquid is lost during the process.
We attempted to keep the unit plumb so the meniscus and scale aligned correctly. We found it difficult to read the ’04 scale because of low contrast. Pictured is 1.53 lpm.
Delavan Calibration Cup: This small, one-handed plastic cup had a scale printed on the outside. We were limited to a 15 second collection because of how quickly it filled. Some spray was lost to mist and bounce and we used the “Fluid Oz” scale to get the highest resolution from the measurement. It took less than a minute to collect and read from the cup, but had the lowest precision and accuracy.
Graduated Cylinder: There was little or no mist or bounce from escaping spray during collection. Our 1,000 ml graduated cylinder took 30 seconds at 40 psi and 20 seconds at 70 psi to fill making it roughly one minute per reading. A few light taps removed bubbles and once the liquid settled we could read the level. This must be performed on a level surface (in our case we used the digital level app on our iPhone). This was a very precise method, varying by less than 2%, but it wasn’t very accurate. We may have introduced error when reading the meniscus (always read from the centre) or perhaps the plastic distorted over time and affected accuracy. It may be difficult for most people to get a high quality, scientific-grade graduated cylinder.
Greenleaf Calibration Pitcher: The pitcher had multiple scales but once again we used fluid ounces because it had the highest resolution. With the highest capacity, we were able to collect for an entire minute. Despite holding the vessel at different angles and distances, we lost a lot of spray to mist and bounce and the nozzle body was beaded with water at the end of each trial. After tapping the vessel to remove bubbles and reading on a level surface, it took about 1.5 minutes per sample and averaged an average 3% more than the calculated ideal flow rate.
Innoquest Spot On Digital Calibrator: We’ll discuss all three Spot Ons together. The Spot Ons were a game-changer in North America when they first came out. You can read a peer-reviewed article about the SC-1 by Dr. Bob Wolf et al. published in 2015 in the Journal of Pesticide Safety here. The SC-2 is a new version of the SC-1 with added digital features that allow the user to calculate gallons per acre and it indicates tip wear based on the 10% industry standard . The most important improvements were a reduced sensitivity to foam and a thicker foam diffuser to reduce the chance of errors. The SC-4 works exactly like the SC-1, but has a larger capacity intended for high flow rate nozzles (e.g. hollow cones on an airblast sprayer). In each case, the Spot On will report in several units, and must be held steady under the nozzle flow (i.e. not moved during reading). The SC-1 and 2 took less than 12 seconds for each reading and the SC-4 took closer to 30. The SC-1 and SC-2 were relatively precise but read consistently higher than the calculated flow rate. This may be an artefact given that the units only read to 2 decimal places and this may have exaggerated any error. The SC-4 was the least accurate and precise of the three. The Spot Ons were the fastest and easiest to read of the methods used.
Weighed Output: This method is based on the fact that 1 ml of water weighs one gram. Spray was collected for 30 seconds and weighed on a new, $25 CAD digital kitchen scale, which was tared (i.e. the weight of the vessel subtracted from the overall weight). While subject to errors from manual timing, it has the merit of removing the challenge of reading a meniscus and it’s relatively inexpensive. This method was precise and relatively accurate compared to the other methods used. It took about a minute per sample.
SprayX SprayFlow Turbo: This was the most sophisticated method we used. The kit comes with a digital manometer, a flowmeter and a digital scale. It works though a smartphone app (screenshot below). You first have to set up a virtual sprayer, informing the app how many sections and nozzles will be tested. Then you must calibrate the flow sensor by taking three measurements versus a weighed output to eliminate possible variations caused by the nozzle, pressure, temperature, and the density of the liquid. The app walks the user through each step. This method took the most time to set up (easily 10 minutes). However, once it was set up, each nozzle could be tested in less than 30 seconds apiece. This method was the most accurate and precise, but the price may place it out of reach for the typical user.
Screenshot from the SprayX SprayFlow app.
Lurmark McKenzie Calibrator: This method is not reported in the box-and-whisker plot because there were significant problems that prevented accurate readings. It was difficult to get a seal over the nozzle and the floater ball would either stick or fluctuate. After several attempts, this method was abandoned.
Conclusion
In order to test if a process, or a thing, is occurring or produced within acceptable limits, we need a detection system with a high enough resolution. In manufacturing (e.g. factory production) this is an essential requirement in quality assurance. Let’s consider a +10% deviation from the nozzle’s ideal flow rate to be our indication that a nozzle needs to be replaced. We need a measurement system with an appropriate scale and one with sufficient precision to ensure we don’t get a false reading. Based on our data, I would suggest all systems reviewed, save the measurement cup, are viable. Even if we elect to use a more stringent rejection threshold of 5%, some systems are more precise than others (i.e. less variability), but all but the cup should still be sufficient.
What’s the penalty for not testing, assuming we’re not talking about significantly deviant nozzles? Let’s say, for example, we are not using a rate controller and we are applying 20 US gpa at 12 mph using 72 nozzles on 20″ centres. Our boom would have to spray 58.2 gpm, which means each nozzle would have to emit 0.81 gpm. If those nozzles sprayed 5% more than intended, we’d be spraying 21 gpa instead of 20 gpa. That means for a 1,200 gallon sprayer, you’d do 57 acres instead of 60. We would have the same result if we dropped from 12 mph to 11.45 mph, which is about 5% slower. Maybe that’s a big deal for your operation, or maybe not. For most, 5% is well within the typical error inherent to spraying. Then again, perhaps it’s more important to know that each nozzle is performing in a manner similar to its neighbours to ensure the highest degree of coverage uniformity.
Ultimately, it is important to ensure you’re as efficient as possible, and that means understanding what your nozzles are doing so you can decide if-and-when it’s time to replace them. Pick whichever method makes it easiest for you to justify testing your nozzles and do it at least once a year when you take your sprayer out of long term storage.
Thanks to all the companies that donated or loaned their calibrators to make this article possible.
July 5, 2022
Herbicides in Asparagus – A creative solution
In 2016, an asparagus grower in southern Ontario picked up a used De Cloet Hi-Boy originally used to spray tobacco. His vision was to create a three-row herbicide sprayer for asparagus and we were invited to participate. His concept was to design shrouds that would contain the herbicide, but not snag the asparagus or drag heavily on the ground. This article follows the development of the sprayer from concept to testing to final product.
The sprayer itself was a classic three-wheel, self-propelled affair. The asparagus was planted on four foot centres, leaving a three foot alley. While the goal was to hang three shrouds off the boom, we started with one to work out the bugs.
This operation uses 2,4-D to control weeds in the alleys and while a little can hit the asparagus stem up to 12 inches (where the branching starts), we wanted to avoid contact at all costs. That led us to the TeeJet AI 95° flat fan nozzle, which produces a Very Coarse to Extremely Coarse spray quality. A single nozzle could be suspended to span the 3 foot width of the alley.
The first version of the shroud was suspended off the boom from four anchorage points. A certain amount of of play was allowed so the shroud would find plumb (i.e. hang vertically), even when the sprayer boom yawed or pitched over uneven ground.
The shroud was constructed of sheet metal, angled to reduce the potential for contact with the asparagus branches, and terminated in stiff, nylon brush-style mud flaps commonly seen on trucks. These brushes were cut to a few inches in length to span the distance between the side of the shroud and the ground. This would create a “seal” to prevent spray from escaping, maintaining some degree of contact with uneven ground.
We tested the first version by placing water sensitive paper in two positions on the ground, just inside the reach of the brushes. We had to be careful not to run them over with the centre wheel of the sprayer. We also adhered two papers to the angled inner walls to see how much, if any, spray was hitting the inside of the shroud.
Our first pass on June 16th was at 9:00 am, 19.1 ºC (66.4 ºF) with a cross wind of 5 to 7 km/h (3.1 – 4.3 mph). relative humidity was high at 85% and travel speed was slow at 3.2 km/h (2 mph). We started with the .06 AI tip at 50 psi, but we drenched all the targets with excessive coverage because we were travelling so slow. We also found the stiff brushes were creating furrows in the soil, as shown below.
For our second pass, we tried the .04 tip and raised the shroud while dropping the tip to keep it suspended 15 inches over the ground. We were still drenching the targets and noticed the shroud was hitting the asparagus spears, causing physical damage. The damage is shown below – note the dark green on the bent spear.
This led to a decision to flare the side walls more aggressively, bringing them further into the centre of the alley and away from the spears (shown later in the article). This had the added benefit of angling the brushes as well to get a maximum span for weed control in the alley. For the final coverage pass we used the AI .03 tip, which gave more than 45% coverage on the ground, with even distribution, and there was no indication of spray on the papers adhered to the inside of the shroud. This coverage is more than is likely required, and the operator should be able to spray up to 6.5 km/h (4 mph) without compromising coverage.
Since the coverage tests, the grower added additional sheet metal fenders to the the existing fenders, encasing the wheels and creating a smooth transition for the shroud to gently deflect the asparagus. The fenders were needed because the grower found the asparagus was being pushed out by the wheel fender only to bounce back in front of the shroud, which snagged the fern and damaged it. The additional fenders keep the fern spread and prevent it getting caught in front of the shrouds.
The grower was very happy with the sprayer’s performance and planed to build another. Why be satisfied with the status quo when you can tap into your creative side and be innovative? If you don’t think you’re imaginative enough to try upgrading equipment on your farm, here’s a simple test to prove that it’s in you. It’s easy to see the bird in the image below, but with a little concentration you’ll be rewarded with a ski-jumping rabbit.
Thanks to TeeJet for donating the nozzles and water-sensitive paper and to Ray and Brad Vogel of Lingwood Farms for inviting me to participate.
Learn more about spraying asparagus here.
June 6, 2022
Installing a Continuous Rinse System
Cleaning, flushing, triple-rinsing… whatever you call it, sprayer sanitation is a time-consuming and distasteful task.
Methods vary, but they generally span from the classic triple rinse (30-45 minutes) to a full tear-down and decontamination (many hours and likely an overnight soak). The operator decides how much time and effort to invest depending on the chemistry they’ve just used and the crop they intend to spray next. Learn more about the power of dilution in this article and in this article.
Unfortunately, two facts are certain:
1. At minimum, operators should rinse the sprayer at the end of each day… and they generally don’t.
2. It is only after spraying a sensitive crop that the operator truly knows whether the sprayer was cleaned sufficiently.
Continuous Rinsing
We’ve promoted Continuous Rinsing as a viable alternative to Triple Rinsing in previous articles (see here and here). Executed correctly, the method:
- greatly reduces the time required,
- is as effective,
- eliminates operator exposure, and
- reduces potential environmental contamination.
Continuous rinsing requires the installation of a dedicated “rinse pump” to transfer clean water to the product tank from the rinse tank via the wash-down nozzles. This permits the main product pump to operate simultaneously, emptying the product tank and spraying the rinsate out the boom.
Imagine your sprayer empties at the end of the row. You position the sprayer at a headland or a row you sprayed earlier. A toggle switch in the cab engages the rinse pump and the wash-down nozzles start spraying clean water into the product tank. You then resume driving and spray until the rinse tank is empty. During the process, any solution in the return/bypass line is quickly diluted, and any standing volume in the system is displaced by clean water.
It takes five minutes and you never left the cab.
Remember: Rinsing can dilute residue to ~2-5% in most of the sprayer plumbing, but it is not intended to replace the more rigorous decontamination process. Closed circuits, filters and dead-end plumbing can still harbour residue >15%.
Installation
Working with GreenLea Ag Center in 2017, we installed a Continuous Rinse system on a Case IH Patriot 4440. It has a 1,200 gal. product tank, a 140 gal. rinse tank and a 120 foot boom. A parts/price list for the Patriot installation appears at the end of this article.
Additionally, we have included the parts/price list from our 2016 HJV Equipment installation on a RoGator 700, which had a 700 gal. product tank, 50 gal. rinse tank and a 90 foot boom.
Still further, we have included three homegrown solutions from operators that developed their own continuous rinse systems.
Sizing the Rinse Pump
It is very important that the rinse pump has the capacity to operate the wash-down nozzles and still supply clean water at a rate approximately equal to the rate at the boom. Basically, “in must equal out”. If the rinse pump supplies too much clean water, the volume rises in the product tank and efficiency is reduced. If it cannot supply enough, the main product pump will lose suction and not function correctly.
We installed a Hypro 9303C-HM1C centrifugal pump (max flow rate of 114 gpm at 130 psi), matching the make and model of the exiting product pump. A length of channel was installed on the chassis to mount the pump and close-coupled hydraulic rinse pump motor, and a valve block.
Really, electric pump installation is easiest. An alternate pump that has been used is this one from Pattison Liquid. For added benefit, it’s a chem transfer pump that can handle the pesticide formulations. If the pump doesn’t give enough flow, a second one can be installed parallel to double the flow.
Hydraulics
Let’s being with advising caution: If you are uncertain about your hydraulic capacity (and tightly designed systems rarely have extra) then consult with a manufacturer-certified service technician, or consider an electrical alternative.
For the Patriot, the auxiliary hydraulic circuit was used to drive the hydraulic rinse pump; we piggy-backed off of that existing system. In this case, Continuous Rinsing increased the load on the auxiliary hydraulic circuit, but only marginally, so performance was acceptable.
We drew that hydraulic flow directly from the auxiliary pump output using a ‘T’ piece to ensure full pressure was available when needed. Then we broke into a common low pressure return manifold using another ‘T’ piece to provide the return flow.
Originally, we were concerned that robbing too much hydraulic flow could compromise sprayer operations. We therefore exchanged the hydraulic motor that came with the pump for one that required less hydraulic flow. However, the pump operated at such a high speed that the rinse tank was drained in two minutes! We felt this would not give the operator enough time to make minor adjustments (see the “Avoid Airlock” section later in the article). We also felt the rinsate would not have enough time to hyrdate any residue in the tank and lines. We therefore returned to the motor that came with the pump, slowing the pump and bringing our rinse time to approximately five minutes.
We installed an on/off hydraulic control valve block and solenoid controlled by a toggle switch in the cab. When the rinse switch was engaged, 12 volt DC opened the solenoid, allowing hydraulic oil from the auxiliary pump to turn the rinse motor, which in turn powered the rinse product pump.
Avoid Airlock by Balancing Flow
While Continuous Rinsing works well with an unbroken stream of clean water, there is demonstrated benefit to allowing the pump to draw a small amount of air. The bubbles are reputed to scrub the lines more effectively than water alone. It is possible that the new Hypro 9307 series centrifugal pump, which claims to eliminate dry run, would facilitate this.
However, avoid excessive cavitation or airlock of the main product pump. This will damage the pump seals and interfere with pump suction. If the main product pump is a piston-diaphragm pump, avoid losing the prime by letting a small volume of rinse water build up in the product tank before spraying the rinsate.
Maintaining the balance between the supply from the rinse pump, and demand by the product pump, will take careful trial and error. If the sprayer employs a rate controller, speeding up or slowing down travel speed is a means for making adjustments to match the two flows. Alternately, an operator can adjust the pressure regulator manually. Remember, the nozzles won’t need to work optimally so you have the option to use fairly low pressures to match flows.
In the case of an operator applying 28-0-0 using dribble bars or fertilizer nozzles, there is likely too much flow at the boom for the rinse pump to keep up. While we have not tried it, but as long as there was sufficient volume in the clean water tank, it might be possible to rinse the boom section by section, starting with the outside sections and moving in towards the centre.
Lessons Learned
The installation was a learning process, during which we noted the following:
- At first, the rinse tank slowly emptied through the rinse pump, even when it wasn’t in use. We prevented this by installing a 10 psi check valve between the pump and main tank.
- The rinse pump ran dry and burned the seals when the operator forgot to turn it off after the rinse tank was empty. We considered a timer or alarms to prevent this, but chose to install a level sensor (essentially a float) which would cut the 12 Volt DC feed to the on /off solenoid, effectively turning the system off when the rinse tank was empty. Note: the sensor is not in the parts list – it was purchased for ~$10.00 CAD from Amazon.
- When deciding where to draw hydraulic flow to run the rinse pump, we first tied into the main hydraulic circuit (i.e. not the auxiliary). This negatively affected both steering and boom control. Beware drawing flow from critical safety systems such as steering.
Future Development and Other Advantages
GreenLea was exploring an option to use the rinse pump to bypass the product tank, and flow directly to the boom. This can be accomplished by teeing an electrical 3-way ball valve just after the pump to allow flow directly from the rinse tank (see dashed line in the flow schematic shown earlier in the article). Imagine being rained out, or having excess mix left in the tank at the end of the day. This system would allow the dilution of any corrosive chemical from a sensitive precision application system without losing or contaminating the spray tank. It should be noted, however, that high precision spray systems (e.g. AIM Command, Pro and Flex) would still require the operator to open the boom flush valves manually to allow the boom purge.
Growers have suggested the system might also be used to get a sprayer to end of a row if it threatens to run empty before completing the pass. A small volume of clean water added to the tank would displace the 15-30 gallons of unusable volume and stretch the application. Be aware that this would also dilute the product due to the agitation/bypass and should only be considered when a minute or less of additional spray is required.
Homegrown Solutions
Tyler Patriot (Electrical)
David Kucher (@DavidKucher) from Saskatchewan installed Continuous Rinse on his Tyler Patriot (75 foot boom, 800 gal. product tank).
Here’s what he had to say:
The rinse system I was using on my sprayer previously involved a lot of time and effort. Plus, the quality of job it did was sometimes imperfect (I keep pictures on my phone of a canola crop that was damaged because of a poor rinse job from a few years ago). The old system used the main product pump to rinse, so there was a bunch of valves under the sprayer that needed to be turned, and the pump had to reprime for each rinse. It was tedious.
Uncertain about the hydraulics, David elected to use an electrical pump, but had difficulty finding one that would produce enough pressure and flow. Most electric pumps were too small and it would have taken more than one, plumbed in parallel, to achieve the volume numbers required. However, David found a high-flow 489G-95 AMT High Head Washdown Pump (1 HP, 1-1/4×1 IN/OUT, 12 VDC,Cast Iron,Buna-N) which he got from the US for about $1,200.00 CAD. Max flow was 56 gpm.
Note: In 2020 this pump model changed to the 12DC-95.
He removed the majority of plumbing, valves, and related complexity from the old rinse system. The Continuous Rinse was comparably simpler and isolated from the rest of the sprayer plumbing. It just involved a fill line from his two clean water tanks, the new rinse pump, and the existing rinse nozzle inside the product tank.
When the product tank empties, David holds down a push button dead-man switch he installed to activate the rinse pump. If he wants to do a more thorough job, he flushes the product tank and plumbing for about two minutes, then stops, gets out and opens the boom end valves. Then he climbs back in and does another one minute flush.
Approximately 30 gallons of water go through on each flush and my only issue is that I waited so long to install the system.
Author’s note: Positive displacement electrical pumps (which have zero risk of seal loss) are reasonable alternatives to centrifugal pumps. Depending on the size of the sprayer, multiple pumps plumbed in parallel can provide sufficient flow. We elected to use two Hypro electric roller pumps (model 4101 N-H) for the 2016 RoGator 700 installation. Cheaper, low amperage 12V diaphragm pumps from Delevan and FLOJET with capacities of 5-8 gpm are also available.
John Deere 4830 (Hydraulic)
Russ Enns (@EnnsFarms) from Saskatchewan installed a Delavan HD Magnum 125 hydraulic driven pump (1-1/4” suction, 1” discharge, 5-7 gpm of hydraulic flow). He mounted it on the same mounting plate as the main product pump, just on the opposite side, using the same bolt holes.
It was tied hydraulically to the main product pump, so the rinse pump could only run when the product pump was operating. The hydraulic supply from the sprayer went through an electric/hydraulic block via a solenoid resting in the closed position. A rocker switch in the cab used 12V to activate the rinse pump from the cab. Return hydraulic pressure from the rinse pump was tee’d into the main solution pump hydraulic return.
The clean water intake for the rinse pump was tee’d into the factory rinse tank. The discharge side of the rinse pump was plumbed to a check valve and tee’d into factory tank rinse system. Here’s the discharge line, check valve and tee into factory rinse (below).
Russ mounted a large pressure gauge on front right axle to monitor rinse pressure. It’s easy to see from the cab, and easy to tell from the pressure when the rinse tank is empty.
In this case, Continuous Rinse is used in tandem with an Accu-volume tank gauge so Russ could monitor the level in the main product tank from the cab. Depending on the nozzles being used, he found that the rinse pump supplied clean water faster than the rinsate could be sprayed.
So, after finishing a field (or changing chemical, etc.) Russ turned on the rinse system while spraying the rinsate out on the field. The Accu-volume alerted him if clean water was accumulating in the product tank. If it got to ~20 gallons, he would briefly suspend the rinse pump while spraying to allow the level to drop. Then, he would start the rinse pump back up. He repeated this process until the clean water tank was empty.
Russ had many of the main components on hand, but estimates replacement value at ~$1,200.00 CAD. He noted that while installation was straight-forward, he originally piggy-backed the rinse pump’s hydraulic supply off the main solution pump, and it didn’t work correctly. We did that too, Russ 🙁
“Time savings and environmental considerations are the biggest benefit of this system to me. Being able to finish spraying a field, and immediately start rinsing and spraying the diluted solution is a huge time saver. I feel it’s a far more thorough rinse and a better/quicker dilution rate compared to how I previously handled rinsing and spraying out the diluted solution. Another benefit is that even though it’s plumbed into the factory rinse, the factory rinse system can still be used normally if for some reason the continuous rinse pump quits.”
Gregson Trailed (Electrical)
Continuous Rinse isn’t only for grains and beans. Matthew Droogendyk installed two 12v pumps on his trailed vegetable sprayer that matched the flow of the main pump. They had an electrician install a box for switching the the pumps and two solenoid valves on at the same time.

They noted an issue when trying to prime the main pump after emptying the tank. If the tank was sprayed completely empty, the main pump took time to get primed again. This affected rinsing time as well as the balance between supply and demand. Through trial and error they determined that running the rinse pumps for 1 minute (~15 gal) gave enough time to rid the main pump of air. Then the flows matched at about 15 gpa. Re-priming took about 5 minutes, and then an additional 2 or 3 to rinse using about 45 gallons of clean water. They found there was no need to replace their original tank rinse nozzles.
Tank Rinse Nozzles
One of the challenges of installing continuous rinse is ensuring the tank rinse nozzles are capable of rinsing the entire solution tank interior at potentially low pressure and low flow. In 2019, Lechler released the ContiCleaner range of rinse nozzle. Four ISO colour-coded nozzles capable of operating from 2-5 bar (29-72.5 psi), with flows from 6.5-32.3 L/min. (1.7-8.5 gpm). This will enable operators to better match the rinse nozzle(s) to the clean water pump. Be aware they are very difficult to source in North America. We tried and weren’t able to get them.
Parts / Price List
The following two parts/price lists are in Canadian Dollars. They do not include tax or labour and prices change depending on where and when parts are purchased. As you have read from the operators that installed their own Continuous Rinse systems, there are many possible solutions, so these lists are provided only for reference. Click the link to download a PDF.
- Parts/Costs List – Patriot 4440
- Parts/Costs List – RoGator 700
Learn More
So far we’re aware of two Ontario companies and one Belgian company with experience installing the system. We will expand this list over time.
Before contacting them, have the following information on hand:
- Sprayer tank volume (both product and rinse, if applicable)
- Boom length
- Nozzle spacing
- Largest nozzles mounted/used on the sprayer (excluding fertilizer nozzles)
- Power available on sprayer (e.g. 12V available? Max amp? Hydraulic capacity?)
Thanks to Russ Enns, David Kucher and Matthew Droogendyk for sharing their install stories. Thanks to Adam Beaumont and Ehrin Frid for the Case IH and RoGator installations, and to Mike Cowbrough (@cowbrough) of OMAFRA and the Ontario Soil and Crop Improvement Association for collaborative support.
May 19, 2022
Drop Hoses Improve Coverage in Field Peppers
In early July 2016, a farm supplier contacted us on behalf of a client with a history of disease control issues in his field pepper operation. He wanted us to calibrate their sprayer and diagnose spray coverage to see if there was room for improvement. Improved coverage doesn’t necessarily mean improved efficacy, but generally it’s a reliable indicator. When we arrived at the field the winds were gusting over 15 km/h, which had the potential to create a massive drift issue. We were only spraying water, so it was decided that if we managed decent coverage in those conditions, there would be no need to worry on an acceptable spray day.
Field pepper in Southern Ontario in mid-July
The grower traditionally ran two different settings on his sprayer. They were relatively low volumes for a vegetable operation, but the crop was still small at this stage, so we did not propose raising the volume:
1. TeeJet AITX 11008’s on 50 cm (20″) centres at 11.25 kmh (7 mph) and 3.44 bar (50 psi). That’s 3.35 L/min (0.89 gpm) per nozzle for a total rate of 350 L/ha (37.5 gpa).
2. TeeJet ConeJet TXVK18’s on 50 cm (20″) centres at 7 kmh (4.5 mph) and 3.44 bar (80 psi). That’s 1.6 L/min (0.42 gpm) per nozzle for a total rate of 275 L/ha (29.5 gpa).
To test the coverage with these settings, we folded a piece of water-sensitive paper over a leaf to cover both surfaces, and wrapped one around a hollow tube to mimic a plant stem (see figure). Three plants were papered for each sprayer pass. Papers were collected, digitized and analysed for percent-coverage and droplet density. When diagnosing coverage for a horticultural crop, a distribution of 85 medium deposits/cm² and 10-15% coverage is a reasonable standard for most applications.
Location of water-sensitive papers in situ.
The first condition (the AITX tips) averaged 17% coverage on upper leaf surfaces (37 deposits/cm²). These were coarser droplets at relatively low volume, so it was no surprise that we didn’t achieve 85 deposit/cm² target. When using such large droplets, it is more important to achieve an even distribution and the 10-15% surface coverage (we achieved 17%). There were no deposits on the underside of the leaves (See figure 1), but that was also expected as coarser droplets tend to follow a downward vector that is not conductive to under-leaf coverage.
Figure 1 – Water-sensitive papers from three plants sprayed in Condition 1. Percent coverage and droplet density are calculated for the leaves, and a visual inspection is made of the stems.
The second condition (the ConeJets) provided better coverage. The fine droplets produced covered an average 17.5% coverage with a distribution of 99 deposits/cm² on upper surfaces, and 23% coverage with a distribution of 185 deposits/cm² on lower surfaces. Panoramic stem coverage was improved as well (see figure 2). This is excellent coverage, but the finer droplets were highly prone to drift (see below). With no form of drift control, this set up is undesirable.
Figure 2 – Water-sensitive papers from three plants sprayed in Condition 2. Percent coverage and droplet density are calculated for the leaves, and a visual inspection is made of the stems.
With no form of drift control, the finer droplets produced by hollow cones create unacceptable spray drift, even in moderate wind conditions.
This led us to propose a more directed boom arrangement: We set up a hollow cone over the row (the grower’s original ConeJet) and a drop hose suspended in each alley with two TeeJet XR 8004 flat fans positioned on an angle (i.e. not vertical or horizontal to ground). This gave sufficient height to span the canopy with as little direct waste on the ground as possible. As the crop grows, the nozzles would need to be twisted into a more vertical alignment.
ConeJet TXVK18’s alternating with drop hoses with TeeJet XR 8004’s.
We did not use an air induction fan to avoid the Very Coarse spray quality and we used 80° instead of 110° to ensure the spray did not overshoot or undershoot the plant. Here are the details of the third set up:
3. TeeJet ConeJet TXVK-18’s on 100 cm (40″) centres at 7 kmh (4.5 mph) and 3.44 bar (80 psi). That’s 1.6 L/min (0.42 gpm) per nozzle. Also, two TeeJet XR 8004’s per drop on 100 cm (40″) centres at 7 kmh (4.5 mph) and 3.44 bar (80 psi). That’s ~4.5 L/min (1.2 gpm) per drop hose. Together, set of nozzle for a total rate of 523 L/ha (56 gpa).
This set up raised the volume considerably and aimed spray directly at the sides of the plant. Coverage was excessive and in a few cases exceeded what the diagnostic software could reliably resolve (see figure 3). Since the plants were still small at this stage, it was decided we would let them “grow into the volume” and come back to check coverage once they were at full size.
Figure 3 – Water-sensitive papers from three plants sprayed in Condition 3. Percent coverage and droplet density are calculated for the leaves, and a visual inspection is made of the stems.
When we returned in mid-August the plants had reached full maturity. In this final coverage trial, we added a second water-sensitive paper to each plant to span the height of the crop canopy, which had grown considerably.
The same pepper plants ~5 weeks later had more than doubled in size.
Coverage was reduced compared to how we left things in July, but appeared to be sufficient on key surfaces (see figure 4). The papers showed upper leaf-surface coverage of 63%-to-offscale and deposit distribution of 137 deposits/cm²-to-offscale. Coverage on the lower leaf surfaces was greatly reduced to 4-4.5% and 36-90 deposits/cm². Panoramic stem coverage was present, but minimal. Applying higher volumes would likely have improved matters.
Figure 4 – Water-sensitive papers from three plants sprayed in Condition 3, ~5 weeks later. Percent coverage and deposit density are calculated for the leaves, and a visual inspection is made of the stems.
When asked about the drop hoses, the grower reported “They are a bit of a nuisance because they take extra time to put on, and they get caught in the bush at the back of the field. But if they increase our coverage, then they’re worth the extra effort.”
Final thoughts
Adding drop hoses to a vegetable sprayer may be unconventional, but if fungicide coverage is a concern, and the drops will fit between rows, they might be worth a try. Carefully consider the volumes you use because they should reflect the size of the plant canopy you are trying to protect. Finally, water-sensitive paper provides excellent feedback to help you decide if your field volume, nozzle rates and nozzle positions are providing acceptable coverage.
April 22, 2022