Lineal Versus Transverse Scanned Edger Optimization

By Rod Nelson.   2001


Nelson Bros. Engineering advocates lineally scanned board edgers, so we are repeatedly asked why we think lineal is better than the more dominant transverse scanned board edgers. Our initial response is: Simplicity--Accuracy--Lower initial cost--Lower maintenance cost.


If this were true, one would think that mills would only buy lineally scanned systems; this is the case for many new mills, but is rarely the case for upgrades in existing mills. It is NBE's contention that there is never a valid reason to buy a board or gang edger with a transverse scanner and a positioning infeed table.


The most common arguments against lineal scanning are:

* It won't fit in the mill.
* It is not fast enough. 
* It is not accurate. 
* It is too expensive.


There are plenty of board edger configurations to consider:


* Top only transverse scanner 
* Top and bottom transverse scanner 
* Transverse scanner with point lasers, point laser plus light curtain or line lasers 
* Single zone lineal scanner 
* Multizone lineal scanners 
* Lineal scan zones with 2, 3 or 4 heads.


* Hand fed infeed table 
* Linebar infeed table 
* Centering infeed table 
* Positioning infeed table


* Conventional edger 
* Slewing edger 
* Flying edger


* Individually positioned finger picker 
* Stacked pop-up finger picker 
* Belt picker


How do you decide?

Compare a single zone lineally scanned system with a typical transverse scanned system.


First, a transverse scanned system measures the board on a scan deck as the board travels laterally toward the infeed table. The board is positioned over a feed chain to the desired skew angle and offset.


The board is lowered onto the feed chain, then carried thru the sawbox. Before the board enters the sawbox the saws are positioned to make the desired board sizes.


A lineal system scans the board after it is traveling lengthwise on a scan belt heading toward the edger. The saws or the saw arbor is set at an angle (usually in the range of +/- 2 ), then when the board enters the saws the saws are slewed left or right to make the desired skewed solution. The infeed for a lineal system can be as simple as a lug chain pushing the board onto a rollcase. The goal of the infeed is to keep the boards "kind of" straight and appropriately gapped. The major differences are in the scanner, infeed table, feedworks, sawbox and picker. The decks leading to the infeed table and the outfeed after the picker system are the same for either configuration.



The transverse scanner must span the maximum board length. For a simple 20 ft. top only system, the scanner would consist of ten 2 ft. long scan heads and a cabinet with scanner electronics.


The lineal scanner must span the board width (18-30 in.) and usually consists of two scan heads and a scanner cabinet with terminals and a network hub.


Both systems have the same computer configuration. Typical costs for an NBE scanner and optimizer: 20 ft. top only transverse--$150,000; 2 head lineal-- $75,000 (all figures in U.S. dollars).


Infeed Table

01jul Nelson 1The transverse system needs a positioning table to accurately position the board as directed by the optimizer. A 20 ft. table has four or five positioners. The two common positioning techniques are side clamping or top/bottom clamping the board, then positioning the clamps. Each positioner usually has three to five air actuators to help complicate matters. The table usually has plates that lower the board on the chain and overhead pressrolls to keep it from moving around too much. The feedchain has to serpentine around all the positioning devices and has to track straight. The need for speed and accuracy results in complicated positioning tables that are a maintenance nightmare.


The lineal system can be as simple as a lug pushing the board onto a canted rollcase, which runs the board to linebar, then onto a scan belt. The lineal infeed table does not need to accurately position the board, since the scanning is done later. To obtain better gap control, some systems push the board onto a straight rollcase with overhead pressrolls to accelerate the board onto the belt. The simplicity of the lineal infeed guarantees less maintenance and downtime than with a positioning table.


Typical costs: positioning table--$150,000; lineal infeed table--$35,000.



01jul Nelson 2The transverse system is closely coupled to the edger. The scanner feedchain usually extends 1 or 2 ft. past the zero lumber line, then it is into the sawbox.


The lineal system requires a scan belt to carry the board thru the scanner and into the sawbox. The length of this belt must be about two board lengths plus one second travel. For a 20 ft. system running 900 FPM, the length would be 55 ft. The belt must track correctly from the scan zone to the sawbox.

Typical costs: positioning system--zero; scan belt--$30,000



The lineal system requires a rotation of the saw guides or the saw arbor and guide assembly. Both systems require saw positioners, guides, lube, etc.


Typical costs: 4 saw conventional box--$130,000; 4 saw slewing saw box--$150,000.


Outfeed Picker

The finger picker with individually, positionable fingers is the most popular configuration on new systems. The finger picker with stacked pop-up fingers is harder to maintain and usually still requires a positioner to fine-tune the stack. Finger pickers usually require one more board length of runway behind the edger. When that length is not available, the mill can resort to a belt picker. Belt pickers are more expensive and require more maintenance.


A lineal system does require a lineally positioned picker. Since nearly all new pickers are positionable, there is no additional hardware costs. When a lineal edger slews in the cut, the resulting boards are slightly skewed on the outfeed belt. On finger picker systems, the fingers must be slewed, just like the saws were slewed. This keeps the center of the finger in the center of the board along the board's entire length. For belt pickers, the belts are positioned at the board's center, one-third of the length from the lead end.


The main problem with all picker systems is that the lead end of the boards occasionally move left or right as the boards are sawn. NBE is developing a scanner that finds the lead ends then adjusts the pickers as necessary.


Costs: The additional programming for the pickers on a lineal board edger costs $1.98.



The complication of the positioning infeed table adds costs to the operators console and the amount of I/O in the PLC. Typically each servo axis adds $2,000 and positioning tables usually have four or five servo positioners.


Typical costs: PLC system and console for transverse system--$65,000; PLC system and console for lineal systems--$55,000.



The simplicity of the lineal system is best illustrated by the calibration procedure during installation. For the lineal system you align the saws to machine centerline, linebar or piano wire.


You calibrate the scanner to a linebar of fixed reference. After the belt is tracking, you run some boards, then adjust the saws (typically less than +/-0.10 in.) to make the cuts match the optimized solution. I had the first belt scanned system running in three hours after arriving on site with scanners in my car trunk. Everyone was amazed.


Transverse systems need to have the positioning table calibrated in addition to the scanner and saws. All the scanner installation and machine calibration activity requires the system to be locked out, thus interfering with the PLC controls startup. It normally takes two days before you get a good optimized board.


Complication takes time and time costs money: seven day startup of transverse system--$8,000; five day startup for lineal system--$6,000.



The main advantage of the lineal system is cut accuracy. The goal is to measure the board, find the most value board or combination of boards then make the cuts at the desired positions. Whatever system has the best scan accuracy and least cut positioning uncertainty is the ultimate winner. That and piece count are what give you the payback, the ROI and all those other accounting terms we use to measure success.


The advantages of a lineal scanner over the transverse scanner are data density and board stability. Transverse scanners typically collect 20 to 40 thickness measurements per inch across the face of the board at each laser position. When these scanners get to the wane, the data density decreases and when the steepness angle of the wane exceeds 75 , they usually quit getting data. They collect 100 to 200 data points on a consistent 5 in. face, but at the edges, where data is important, they quit getting data. To mask this problem, most transverse scanner systems add a bottom scanner or a light curtain to at least get the width right. This also adds $50,000 to the $150,000 top only scanner/optimizer and makes it even more complicated.


The scan heads on a lineally scan system can be directed at the edges. This gets better data density where it is needed and less data density on the sawn face.


Also, transverse systems are plagued with chain vibration, surging encoders and board vibration, all of which decrease system accuracy. The lineal scanner has none of these problems.


Cut Uncertainty

Assuming a good scan and that the optimizer comes up with the best solution, the board still has to be presented to saws correctly. For a transverse scanner system, the board must be positioned on the feed chain at the correct angle and correct offset. To do this, there must be a positioning table. I like to describe a positioning table as a mechanism that randomly prevents boards from getting sawn correctly.


Here's why: The board must first be registered against a reference. For a top/bottom clamping system the reference is usually the last hook stop or a stop pin. The optimizer guesses how the board will settle against this reference, then clamps the board accordingly. If the board does not settle against the reference exactly as predicted by the optimizer, you get positioning uncertainty. For side clamping systems, the reference is one of the clamp faces. This has even greater positioning uncertainty, since the edge crush is impossible to guess. Now the board has to be positioned, held by overhead press rolls, released by the clamps, lowered onto a chain going 1000 FPM.


Each step and device adds positioning uncertainty. To make matters worse, we want this entire cycle to take 1.5 seconds. The best maintained top/bottom clamping tables have a +/- 0.10 in. positioning uncertainty. Typical positioning tables have a +/- 0.20 in. uncertainty.


I can not emphasize enough how difficult it is to install, maintain and troubleshoot a positioning table. Just ask mills that have done both. Once you have done a lineal system, you will never choose a positioning table. The board is already stable on the belt with a lineal system. The belt must track correctly to the sawbox. The only added uncertainty in a lineal saw box is the timing of saw slewing. This requires the detection of the lead end of the board and tracking the board to the saws. An abnormally large slew rate of 2 in. over 10 ft. of length would introduce a cut position error of +/- 0.02 in. if the lead end of the board is misjudged by 1 in. of travel. Since most boards require a smaller slew rate, the uncertainty due to slewing is usually negligible.



The typical transverse scanner requires both a thickness and an alignment calibration. The calibration bar is normally an 2x4 aluminum tube that is longer than the scanner zone. Two people are needed to handle the calibration bar. The thickness calibration can be done statically, but the alignment calibrate requires feeding the bar thru the scanner. The calibration process is complicated in that it is calibrating from 80 to 300 lasers plus 500 to 1000 light curtain sensors and requires stable transport thru the scanner.


Safe scanner calibration requires locking out the system to get the calibration bar, unlocking the system to run the calibration piece, then locking out the system again to store the calibration bar.


The positioning table must also be calibrated. Each servo positioner must be calibrated to a fixed reference and the reference stop for top/bottom clamping systems must likewise be calibrated. This always requires getting into unsafe and awkward locations.


The lineal scanner calibration is a simple static process. The belt is stopped, the 20 in. to 30 in. calibration piece is slid against a reference of mounted-on tooling pins. The calibration process is calibrating two or three scan heads each with one camera and one laser.


The saw calibration is the same for all types of systems, except the lineal system includes a rotation positioner.


After all the scanner, infeed and saw calibrations, you then need to do dynamic calibration. This is needed because there are too many things that influence the cut positions that can not easily be calibrated. Dynamic calibration consists of comparing the optimizer solutions to the final cuts. Simple offset calibration is easy. You look at 10 boards and if the cuts appear to be too far to the right you adjust the saws to left. You normally will be checking width sizes at the same time.


What if there is skewing error? First they are less obvious and if you have a positioning table the cause will likely be one of the four positioners, so the error will only show up on particular lengths.



The transverse scanner has 20 ft. of windows to keep clean. When a light curtain and/or bottom scanner is used a pressurized enclosure and blower is needed. Cleaning requires a lockout procedure and climbing into awkward and unsafe areas.


The lineal scan heads are to the side of the scan belt. They have two windows per head that can be accessed from the walkway.


A positioning table has four to five positioners, 20 to 30 air actuators, a serpentine chain, four to five pressrolls, seven to 12 sensors and a lube system. Because of the need for speed, many actuators have flow controls. It is reasonable to believe that there is always something not working correctly on a positioning table. No one wants to work on a positioning table, but every weekend someone is working on a positioning table.


The lineal system requires a rollcase and a 50 ft. scan belt. Need I say more.


The sawbox for a lineal system needs to handle slewing. This is accomplished by rotating the saw guides or rotating the entire saw arbor and guide assemble. Both require some additional maintenance over the traditional sawbox.


Actually saw rotation is not always needed. We have done two lineal systems with centering tables that have conventional guided saw sawboxes. With a centering infeed table, most of the recovery gain is realized with slew rates less than 1 in. in 10 ft. (i.e. one-half degree of saw rotation). These systems are slewing up to 3/4 in. in 10 ft. with minor increase in saw maintenance. On a centered system, most of the one board decisions do not require any slewing. Typically, the saws will slew on one out of 10 boards. But, a centering table can be as difficult to maintain as a positioning table.



The troubleshooting procedure is another revealing comparison. What do you do if the mill manager says, "It just isn't cutting right"? First you compare the board plots with the actual boards. This will let you know if you have a scanner problem.


Second you compare the optimizer solution with the board exiting the sawbox. If they are not the same solution, you have a data sequencing problem. For transverse systems, sequencing can get screwed up on the scan deck or at the positioning table and at the edger infeed. For a lineal system, there is rarely more than one board between the scanner and the sawbox, so sequencing problems are rarer.


Third, you check if the cut locations forecast by the optimizer match the actual cuts. On a lineal system the problem can be due to: poor scanner calibration; poor belt tracking; improper saw slewing.


For a transverse system the problem can be due to: 
* Poor scanner calibration 
* Poor scan due to board vibration or motion in scan zone 
* Poor registration at infeed at one or more stops 
* Poor positioning at one or more positioners 
* Poor infeed sequencing due to one or more air actuators 
* Board skidding left or right on a not-so-sharp chain 
* Misaligned pressroll or feed chain


The complicated positioning table leads to a common troubleshooting ailment. The millwrights blame it on the scanner, the electricians blame it on calibration, the QC people blame in on the mechanical items. The end result is nothing gets fixed.



If cost, simplicity, etc. are not enough to convince you that lineal is best, then consider the future. Future scanners will incorporate biological scanning such as Xray or high resolution color scanning. The cost of these scanners will prohibit their use on transverse systems for many years, while lineal biological scanning is already being used in planer mill application. It will not be long before split, rot and bark detection is added to lineal sawmill scanners.