Please read this document for more information about the two methods for determining trait detection within a commodity utilizing the F-750 Produce Quality Meter.
Several factors should be viewed when assessing the quality of a model. Read more here.
NIRS stands for Near Infrared Spectroscopy—the science of how light (visible and just outside of the visible region) interact with matter.
As noted by https://support.microsoft.com/en-us/kb/2571899 clickonce deployment is not compatible with roaming profiles. (roaming profiles are often used in universities where it is nice to have application data follow you from PC to PC)
When installing the software it is useful to create a shortcut from the ".application" file in the installation folder to a location available to all users; such as /user/public or /users/default
Yes! Starter models can be found on the F-750 product page, under the 'Starter Models' tab.
The Model Fit Parameter is a numerical indicator of usefulness of the current model. This number is located on the sample scan screen below the model name on the upper right corner of the screen. Numbers larger than 0 indicate that the sample does not fit well into the currently selected model, and thus the predicted variables should be viewed with skepticism. Variable prediction improves the closer this number gets to 0. The Model Fit Parameter is calculated based on the deviation of spectra signal from the training set range. When recording training set measurements, the F-750 internally tracks the high and low signal extremes for each wavelength. Upon taking a sample scan using these training sets, any point at which a sample spectrum deviates from the training set range triggers an addition to the model fit number. This addition is weighted based on the “wavelength influence” regression coefficient multiplied by the degree of deviation from the established range. Thus, a sample scan that fits perfectly within the training set range will yield a model fit number equal to 0, while subsequent deviations will scale according to the wavelength influence and magnitude of deviation.
For more information, please view our Model Fit presentation here.
The most typical cause of this error is a mismatch between the model builder file type and the firmware of the F-750. To correct this error:
Place the newly saved model in your ‘model’s folder on the F-750 SD card
If you would like to enable the F-750 to report averages across samples of fruit, you first need to add the feature to the device view in the Model Builder software. Second, you need to properly configure the F-750 under ‘setup instrument’ to set the number of samples you would like averaged. Details on how to perform each of these steps can be found below.
Reporting Lot Averaging in Device View
Lot averaging is not something that is included in the standard device view. You will need to first add this feature in Model Builder by following the steps below.
Assigning Increment Logic on the F-750
The next step is to assign the proper logic so that the lot averaging is applied to the right number of samples.
The Starter Models should not be used for accurate predictions without tailoring or verification with your commodity. The provide a foundation to build off or a starting model which you tailor to fit your specific commodity and desired predictions through the collection of additional specimen scans and correlated reference values. Felix Instruments has a standard operating procedure for tailoring a Starter Model to your commodity. Please see the “Tailoring Models Guide”.
The training materials below are to help one get started in their use of the F-750 Produce Quality Meter and the Model Builder software. Following the standard model building process, the SOP for reference value collection using the example of mangos should be reviewed first, followed by the instructional video for how to import and use the Model Builder software and finally the videos addressing interpretation of the results provided in Model Builder.
The Mango Data Collection Standard Operating Procedure should be used as a guide for appropriate training set creation and corresponding reference value collection during the model building process.
The F-750 Demo Reference Values and Demo Training Set files can be used in conjunction with the Interactive Model Builder Demo video. Watch the video and then download the reference values and training set files to try out the software for yourself.
For a brief introduction to the statistical results shown in the model performance analysis tab in model builder, review the videos below. Start with the Introduction to Model Analysis and then reference the following clips on each of the statistical graphs shown in Model Builder.
You can troubleshoot / enable / disable your devices GPS by navigating to the following menu:
Main Menu > Setup Instrument > GPS Receiver
If your GPS state is set to ‘Disabled’ you will want to toggle that on to ‘Enabled.’ If it is already listed as ‘Enabled,’ navigate to the View Status menu.
Here you will find a full diagnostic of your GPS module, including current location, number of visible satellites and a timestamp of the last location update.
Typically, if your instrument is functioning as intended, the ‘Updated’ option under ‘View Status’ will list a current timestamp. If not, we would expect there to be an issue with your device.
If the ‘Updated’ record is current but you are still not able to receive a GPS lock, we would recommend the following power / hardware cycling:
Windows 7, 8, 8.1 or 10. Standard SD cards require SD card reader or USB 2.0 port and compatible adapter. Wireless SD cards require WiFi access. No internet connection required.
** Please note: user accounts on the Windows platform with a space in their name may experience issues using the ModelBuilder software **
Note: Due to limitations of Microsoft Translate, manuals greater than 50 pages in length currently do not support dynamic translation.
Models are stored on the SD-card of your F-750. To add or remove models, navigate to the /models/ folder on the SD-card using an PC. From here, you can drag and drop in any additional models you may wish to use, or delete those you would like to remove. Once complete, you may eject the SD-card from your PC and reinsert it in the F-750. Once powered on, you can find your model available under Main Menu > Select Model.
A few things to note:
If your F-750 keeps indicating that it is "checking SD card" or is creating 0 byte length files, this is likely an indicator of a corrupt SD card. Remove the SD card and insert it into a computer.
Copy the "Models" folder and .dat files on the SD card to your PC where you can find them later. Do not have Windows scan and repair the SD card. Rather, use the following SD card manufacturers tool: https://www.sdcard.org/downloads/formatter_4/index.html
Copy the "Models" folder back onto the SD card. The .dat files are your previous data and do not need to be copied back to the SD card. Then re-install the SD card into the F-750 and turn the power on.
Always wait for the F-750 to completely power off and shut down before removing the SD card.
If you experience any issues with the firmware updating process, “Bootloader Assist” mode can be accessed by pressing and holding the ‘down’ arrow while turning the device on. Follow the instructions displayed on screen, and continue the firmware updating process as usual.
Models have been created and validated for many types of fruit, including mangoes, cherries, grapes, apples, persimmons and pears. Please see the application notes for more details on validated proof of concept models: https://www.felixinstruments.com/applications/use-case/
This message is displayed if the calibration information for your device has been deleted. These settings are necessary for you device to function correctly, and can be re-entered under Setup Instrument > Advanced Settings. At the back of the user manual which was included with your device, you will find Pixel Coefficients, Reference Voltage, and ADC Gain specific to your device. Please email us at firstname.lastname@example.org with your device serial number located on the back of your F-750, and we will happily walk you through the process of re-calibrating your device.
Yes, press and hold the power button until you hear a beep. This will cause the instrument to reset. If you have held the power button for longer than ten 10 seconds, and the instrument will not restart, you may need to remove, then re-insert the batteries.
To recharge the batteries, peel off the rubber base from the F-750. Then twist to remove the battery compartment cap (located on the bottom of the F-750). Next, remove the batteries from the F-750 to place them in the battery charger. Charged batteries should be inserted into the F-750 positive (+) side first.
As a general rule, we recommend 100-200 specimen to build a model. For a proof of concept model, 20-30 specimen should be sufficient. The specimen must include samples across the entire range of the quality or trait you are trying to predict.
The F-750 Produce Quality Meter is typically used to measure the concentration of dry matter, total soluble solids, chlorophyll, titratable acids, etc. in fruits and other produce. However, the F-750 can be used to measure any substance with a distinct correlation to light in the 285-1070nm range. This includes everything from steel quality to the liquid content of fish or dry matter in cheese.
If the software suffers from drawing or display issues with Windows 10, verify that the display settings have a zoom of 100%. Windows 10 may adjust the display settings to have a zoom of 125% by default.
This message is displayed if the spectrometer ribbon cable becomes disconnected. This can sometimes happen during travel or transport, especially if the instrument is jostled or roughly handled. Please contact a technical support representative located in your country, or email us at email@example.com
This is essentially changing the path length and thus the effective interaction volume. This violates one of the fundamental assumptions of the PLSR math by introducing some nonlinear experimental sources of error. As such, it will increase your prediction error or require additional principle components (latent variables) to explain. In Felix Instruments testing experience, the variation of fruit shape does not dramatically increase the prediction error of the model. More significant error comes from mistakes during collection of reference data. If the sample population is large enough, the affect can be effectively excluded by the math. Version 2 of the Model Builder software will support standard normal variance and multiplicative scatter correction. These methods can mathematically remove the negative affects of hardware and sample presentation.