Writing BarCode With Plotter Pen Bit

[bstern]

I was wondering if the WidgetWorks Plotter Pen Bit is capable of writing a simple bar code quickly?

Parts labeling is one of the biggest time consumers in processing sheet goods. The most time is spent locating the part to put the label on it.

As I need lots of info on the label, its not practical to write it with a pen.
But, if I could quickly write a 2 to 4 digit bar code, that would suffice. It could be read with a scanner and produce the proper label at a later station.

I dont know where to begin researching. Anyone have any sugestions?

BTW, I was thinking of using an air drill actuator to engage the pen.

[harryball]

My first guess would be no since a barcode relies on some very small lines vs. large lines. If the pen is capable of drawing a narrow enough line then scaling all the others up from there... maybe. However, I don't know if it would be speed efficient enough for you with all the bars needed.

It's a curious enough question though... if I get a chance I might have to give it a try.

Another option would be to produce stick on labels and put them on the parts as they are removed from the table.

/RB

[bstern]

I produce stick on labels now. The part that is the biggest time waster is identifying where the labels go. I average from 5 to 15 parts per sheet. I was thinking if the pen could write, say a 2 digit code, it might not take too long.
Then after reading the code I could print the proper label one at a time.

Building up the line is what I was thinking.

[harryball]

It can definately write a couple of digits, just not quick at barcodes.

/RB

[richards]

Writing barcodes is not a problem. There are several free programs that can write barcodes. The problem is relating the barcode to the part. To do that, you need a database. Constructing a meaningful database is not trivial, but there are lots of freeware/shareware products that make it possible. All of the Linux type databases work on both Linux and Windows OSes (i.e. PostgreSQL, MySql, etc.). Basically, you need to know what each part is and where it goes. A label keeps everything straight.

Adding barcodes adds a layer of complexity to the system, but, if used correctly, the barcode will simplify production.

[Gary Campbell]

Bob...
You might want to take a look at the label software from Avery. They allow you to built a database, print labels... etc. from their free software. Its pretty easy to use and price is right. It also has barcodes.

Gary

[bstern]

Mike, The database is no problem. I wrote the CAM package that takes a cut list (generated by KCDW) nests it (using Cab Parts Pro) and generates the shopbot code. Currently after each sheet of a job is done, I have it print labels to be put on the parts.

Is there some code that I could use to generate "text" in a particular font? I think V caving code could work. Bar-code fonts are easy to find.
Finding one that is most easily "drawn" on the bot would be the challenge.

I guess the real equation is, can the bot write a simple bar-code quicker that is takes to find the part on a map and place the sticker on the part?

Scanning the bar-code, printing a new label and placing it on a part on down the line is extremely quick.

[bstern]

Mike, The database is no problem. I wrote the CAM package that takes a cut list (generated by KCDW) nests it (using Cab Parts Pro) and generates the shopbot code. Currently after each sheet of a job is done, I have it print labels to be put on the parts.

Is there some code that I could use to generate "text" in a particular font? I think V caving code could work. Bar-code fonts are easy to find.
Finding one that is most easily "drawn" on the bot would be the challenge.

I guess the real equation is, can the bot write a simple bar-code quicker that is takes to find the part on a map and place the sticker on the part?
(it takes longer than you think!)

Scanning the bar-code, printing a new label and placing it on a part on down the line is extremely quick. This actual label would contain my normal info: Job name,Room Name,Cab#,part Name and other info. This would be way too much to print on the part.

[jseiler]

I think I might be missing something here, but wouldn't it be faster to write down a item # on the item using a single line font rather than trying to lay down a multiple digit barcode? UV ink would be really cool. No label.

John

[bstern]

The objective it to get a full removable label on the part for shipping to the job site for assembly.

The way it is done now is to print a sheet or string of labels after each 4X8 sheet of parts is cut. Then before removing the parts you use a layout to match the labels to the part and place the labels on the parts. This hunting takes time and as my table it 60" wide (2 Zs) you cant reach all the parts from one side. You end up running around the machine.

If I could use a pen on the machine to print a bar-coded part number on the part, it could be read later and the proper label applied on down the line.

[bstern]

So I guess I have 3 questions.

1. Can the WidgetWorks Plotter Pen be used to print or draw a bar Code?

2. How long would it take to draw a simple 2 to 4 digit bar-code? (it just needs to be big enough to be read by a scanner)

3.Is there code available to generate a tool path for text using any font? (Bar-Code is just a font)

[Brady Watson]

Bob,
If you are looking to do things right for high production, then look into a thermal printer (such as Zebra etc) to print out your labels. They are fast & reliable.

A barcode, is nothing more than a font. You will want to try using either '3 of 9' or '9 of 3' barcode font. You can easily see how the barcode font can work for you by typing a phrase or part number into MSWord or even notepad. Then add a * to the beginning of the part# and a * after it. Select the whole line & choose the barcode font. Print it out & you should be able to scan it and have it decode properly.

-B

[bstern]

I have really done a poor job of communicating what I am trying to accomplish.

I Currently have a Zebra printer.
I am trying to cut down the time it takes to apply labels on my parts.
Current Method:
Prior to production I print out a layout sheet or map of the parts. One for each 4X8 Sheet of parts. Average 8 parts per sheet.
After each 4X8 Sheet of parts is cut, I print out a string of labels on the Zebra printer.
I then have to play the match game. I take each label, look at the map of parts to find where it goes. Then go to the part and apply the label.
After all parts are labeled I then remove them from the machine.

I have a 60X96 machine with a 48X96 bed because I have 2 Zs. This means you can't reach all the parts to label them from one side. So you run around with a map and sting of labels.

This is a slow process. It is also prone to mistakes (blurry old eyes)

What I would like to do:
Have the bot use a pen to draw a bar-code on the part identifying the part.
After cutting the parts would be removed.
Parts are processed down a production line to insert connectors then edgeband and so-forth.
At the end of the line, I would like to scan the bar-code drawn by the bot and have the Zebra printer print the correct label (removable with lots of info) for applying to the part.

There is too much info to be written on the part by a pen on the machine and I simply would not like the much permanently on the parts.

Bottom line I would like to speed up this process and make it more accurate

[Brady Watson]

You could extend your 8' rails to 10' and get anywhere you need to with the 2nd Z. In my experience with barcodes in a production environment, writing directly on each part with a pen would be problematic, time consuming and not 100% reliable in terms of barcode integrity/readability.

Personally, I would look into combining the utility of a price/sticker gun and an el cheapo thermal printer. Then automate the 2nd Z to apply each label to the center of each part using a drilling routine (select any shape, tell it to drill & it will go to the center of each part) - just have the label printer 'drill' each part with a label.

These are just ideas as I doubt that anyone in this group has the ultimate solution you are looking for. Your 'R&D dept' will have to spend some time coming up with an affordable, reliable and fast solution. I know at one point or another there was talk about retrofitting an injet to the Bot to produce barcodes. The ball on that got dropped somewhere along the line & I would not be suprised if technical difficulties like sawdust and print heads not mixing, was the cause of the delay.

-B

[eaglesplsh]

Hey Bob - I'm Russ, the guy that makes the WidgetWorks Unlimited Plotter Pen Bits.

Sorry to join the discussion late, I can throw in a my $0.02 here and maybe that will help out...

I worked for several years in manufacturing, so I understand your desire to quickly and accurately identify your parts. I also worked in warehouse operations, so I have some experience with bar codes and bar code readers.

I'm fairly certain that the plotter pen will be able to produce a readable bar code. If you're drawing on a non-porous material, the pen ink can be removed with acetone, so you've got a permanent/removable option that may be useful in this application.

You can create a tool path to draw any font (including bar codes) using the same CAD/CAM software you use to cut your parts. Full directions on how to setup/draw are included with the pen kit.

To maximize speed, you want to limit the number of bars and the bar code's size - both height and width. Since this is for internal use, your bar code probably doesn't need to be too tall - check your code reader's specs. As for width, the thickness of the "thick" lines are all drawn in proportion to the thinnest line. If you match the thinnest line width to just over the pen's tip (0.015") that will dramatically reduce drawing time. If the ink bleeds when you draw on your material, you'll have to increase the width of your lines to stay within the accuracy/ledgability tolerances of your bar code reader. Your code reader may also have a minimum size spec to consider.

Your max feed and speed rates for drawing will depend on your machine's ability to accurately position itself for tiny work - ramping will probably prohibit you from moving a large/heavy gantry at top speed. You can always test this with some simple air cutting to find out how long it will actually take.

Feel free to contact me if I can help (contact info is at WidgetWorksUnlimited.com) I'd love to see you pursue this new use of our plotter pens and report back to the forum with your results.

Russ

[zeykr]

No idea of the cost, but something like this sounds like just what you need. Pneumatic and prints on contact with the material. We used some similar products on a shipping line - adapts automatically to different height products which was a problem with some of the inkjet spray head type systems.

http://www.matthewsmarking.com/products/tech/reciprocating.php

http://www.matthewsmarking.com/products/productPage.php?modelName=CC100

http://www.matthewsmarking.com/products/productPage.php?modelName=DryDot

[dana_swift]

I have been resisting joining this discussion, but here goes.. there is more than one kind of bar code. The "width modulated codes" like UPC, 3of9 (usually the one used for product identification) and a host of others, are not optimally suited for "drawing" with a plotter (or cutting on a bot). However there are "length modulated codes" like "posnet" used by the post office which do work well for plotting or cutting with a bot. Then there are "holographic" (UPS for example) and "multi-column" codes. Many options to consider.

All of these have software to generate the information image, and scanners that read them. While 3of9 is clearly the "standard" for this application, it is not the only choice, or maybe even the best choice.

There is a lot to know about bar codes, with the 3of9 codes the ratio of wide to narrow spaces must be controlled, as well as the ration of wide to narrow lines, however the width of a narrow line is not the same as the width of a narrow space! The barcodes are designed to deal with a problem called "dot gain" where the ink that makes the lines "spreads out" over the spaces. Most readers are tolerant of out-of-spec bar codes, but that is hardly guaranteed, misreads will rise rapidly as the tolerances get pushed.

Just a heads up before you make an investment- I have to agree that a dedicated printer is a great way to go, they have already dealt with dot gain issues so the results meets specs. BarCode fonts usually do not deal with dot gain issues, as the font does not know about the properties of the print media. If the effect is small there is no problem, again counting on the tolerance of the reader.

I had to learn all this stuff once when somebody paid me to read bar-codes from faxed images which were often xerox copies of the original. It can be done, but all the tolerances are shot by then. It made me learn things I never set out to know. And I probably just said many things of no concern.

Nose now retracted from the topic--

D

[harryball]

As I feared... I tried to draw a barcode with my pen kit using my 3 of 9 font... my barcode reader would not read it. I didn't spend a lot of time with it but that's kind of the point. I just don't think you can control the width well enough with a plotter pen which will dry out too.

Why not adopt a visual code if all you want to do is match up labels...

Example, use 6 dots in a pattern like this

Full pattern
..
..
..

Variations
.
..
..

..
.
..

..
..
.

(note the forum software would not allow me to show the variations with the dots missing to the left as it strips out spaces when posting.)

and so on... these become easy to recognize and if the dots are on the part and on the label when you print them matching becomes a quick glance. You should be able to easily create 20 or so easy to recognize non-ambiguous dot patterns to suit your needs.

/RB

[bstern]

I have found a code I think will work.
Postal bard codes are made up of long and short lines. You need a 2D reader which is a little more expensive but not to bad. About $250.
I wrote some test code to write a 3 digit number.
The simulator said it would take 12 seconds.
That seems a little long. I will get back to my shop later today or tomorrow and air cut it to see how it works.

It sounds promising.
While it may be slower than I want, the fact that the machine is doing the work not me is a decent improvement. Accuarcy would improve as well.

RB - its funny that the postal code works much like the patters you sugested.

Here is a link to a Microsoft article about how it works. http://support.microsoft.com/default.aspx/kb/214181

[rhfurniture]

Fascinating. Below, my name in UK postal code. Is this the same as USA? (our post-codes have letters in them)

7709

[dana_swift]

RB- you are hitting on the problems with barcode fonts. Either they work for you and you think therefore they work for everybody, or the other way around.

Bob- it sounds like you have hit on an idea that is workable.

Ralph, the code you are showing is a length modulated code, and ideal for cutting or printing. With most of these fonts the gap between bars needs to be only roughly equal to the width of the line, and there is usually lots of tolerance on that point. The us post office uses "posnet", but the code you are showing may be some other specification. There are more than one of them. Thanks for showing that example.

D

[Brady Watson]

Dana,
Interesting points & insights. I know exactly what you mean about "learning things you never set out to know" !


I don't want to discourage anyone from doing their own R&D, but my thought is that 'writing' on the parts is going to slow down production - not speed it up. It isn't a reliable solution in the long run, as pens dry out, streak and even bleed.

Personally I think that the absolute fastest and most flexible solution is to come up with a long wand that draws a small vacuum at the tip to hold a pre-peeled barcode label. Then slap it on any of the parts from in front of the machine, following a screenshot printout of your CAD and part layout. It should take no more than 10 seconds a label. Labels could be peeled on the corner and temporarily applied to the part 'map' to eliminate confusion in the shop & cut down on 'collation' errors. The other advantage that may not be readily apparent is that, you may be cutting a ton of melamine today, but what happens if you cut birch ply? MDF? OSB? Flakeboard? An ink pen will bleed all over these materials.

-B

[knight_toolworks]

why not just use numbers or letters on the parts? then the same numbers on the labels use the second z to draw the numbers. but if you have to change bits there is no real time saved.

[bcammack]

I'm seeing a lot of laser projectors like we use for positioning our stone countertops for machining coming up for auction as countertop fabricators go out of business. Price of admission for a new one is roughly $20k, but used, at auction, it's more like $3-4k.

I'd use one to project the DXF the toolpaths were created from onto the sheet that's loaded on the ShopBot and pre-place the labels before cutting.

Here's what we use (z-laser), but there are others: http://www.carterproducts.com/

I understand as well that using a single gun video projector mounted above the machine bed is a new, alternate method of projecting the layout. I've not researched those, however. They're supposed to be less expensive.

[dana_swift]

Brady, the bar code question is really a very interesting one. For me the question is can the barcode be "cut" into a part so it uses a (presumably small) bit just like anything else. I can imagine a .125 bit cutting a length modulated code quickly and easily into a part. Perfect application of a CNC router.

Then the question is how to read the code. Normal scanners expect contrast modulation and don't understand this Z modulation, but there is no reason the code could not be read by a hand-held scanner with a little adapting. I can imagine a modified scanner where the laser scanner is offset from the reader photo-diode so it would always see the shadows (it would really be the occlusions) cast by the line cuts in the part.

Perhaps products exist that do this.

Then the next question is how to generate bar codes that are unique for each part without re-generating an SBP file per part. The SBP "language" is nice for what it can do, and stunning for how many programming concepts it does not support. Local variables and true function calls for instance... that could make this task fairly easy, but those necessary features are missing.

Asking for such features may seem unreasonable, and often I find that is my lot. Then:

"A reasonable man adapts himself to his environment. An unreasonable man persists in attempting to adapt his environment to suit himself. Therefore, all progress depends on the unreasonable man." -- George Bernard Shaw

Bob- Thanks for asking the original question, its been a fascinating topic for thought. I see the value and the possibilities, and very much understand the issues involved. Putting barcodes on parts is just something a shopbot should "just do".

Then again I am being unreasonable-

D

[eaglesplsh]

I know this discussion has gotten away from the original concept/question, but as long as we're getting creative...

One could make their own code and reader - a combination of Rob's dot idea and Dana's thoughts...

You could draw dots and read them visually like a scantron machine does w/one photo-diode over each dot location. Similarly, you could also cut holes/dimples. A "dimple scanner" could have pins that align with the hole locations - pin drops into a hole = 0, pin touching surface completes elec. circuit = 1.

At first glance these solutions might require less precision than actual bar codes.

[rb99]

could you drill Braille?

How many different parts do you have to label?

R

[henrik_o]

Maybe an integrated printer/applicator for use directly on a cnc machine exists, but I haven't seen it. I've been to some advanced plants, the kind where each part is barcoded and then as the part goes through the various cells the barcode tells each machine what it shall do (i.e it downloads it from the machine LAN) and so on, but in every case the actual application of the barcode label on the part as it came out of the first machine (typically a huge panel saw) was done manually.

There is another way of looking at this, however. While it is true that you need a barcode to contain the amount of indo needed and we run into problems printing this with enough resolution etc, and it is also true that the combination of parts and labels as a sum total is for practical purposes infinite, this is not the case if we reduce the problem to the individual cycle. You can have millions of combinations of barcodes/parts, but there will always be a finite –and luckily for us very small—number of them for each sheet.

Can we reduce complexity to the single cycle? Yes, if we add another data layer to the process, a data layer which is specific to one (any) sheet and as such has very low demands on data containership (logic) and thus resolution (in the mechanical application). We cheat the problem, since the problem is containership and thus resolution.

We don’t need to guess where each barcode label goes on a sheet if each part and each label is printed with a matching marking: then we just have to match the part with the correct marking to the correctly marked barcode label. We can do this off the machine as we stack the parts, for example. Since the amount of data contained in the marking is very low and it is not machine-read, we can apply it to the part rapidly and with low tech on the cnc, such as some form of marking pen, ink blot or otherwise – no need for machine resolution specs. A very basic setup could be to use a marking set of 1-99 (i.e the permitted values are digits 1 to 99). We could probably make it even simpler, but reading digits is something we are well trained to do and do automatically so using digits for the values seems logical.

Practically speaking, let’s say we use a filt tip pen attachment to print the sheet of parts starting with “1” , then 2, 3, 4, 5 etc in the same run as the parts are cut. After the parts are evacuated from the table, you attach the barcode label marked “1” to the part marked “1” as they are stacked, since we don’t need a permanent marking you could wipe off non-permanent ink as you attach the label. This should work well for pre-finished non-porous material. If you need to work with other materials, then it could perhaps be modified to print the marking on the scrap “skeleton” if you evacuate your panels as one piece rather than picking off parts one by one.

Ok, maybe this is making it more complex than it needs to be, but I guess if I was looking into this I’d sit down and think a bit about doing it this way, or something similar. It doesn’t solve the problem of completely automatic labeling, but it kind of cheats the problem in a way I think might well work quite smoothly in a production setting.

To ask a completely different question, how do you clean the table between cycles today?

[dvanr]

Going off of Henriks idea (ink blot or otherwise)

http://www.rubberstampchamp.com

There are automatic numbering machines availble to rubber stamp serial numbers starting at 6 digit and $50USD

[henrik_o]

Dick van Randen,

That's one approach, but I do believe the pen attachment from WidgetWorks is ideal for this task.

The issue, as I see it, is that we do not want to ‘waste’ a precious axis for this task.

Putting the pen attachment on a pneumatic actuator controlled by one of the output channels on the bot is probably the easiest way of doing it properly, ideally with a slight downward puff of air as it engages to clear the writing area of dust/debris. There would certainly be some work designing and getting such a unit to work both in the hardware and software department, but it is not a forbidding task if one sees this as a serious productivity enhancement.

My personal crusade is to automate the cleaning of the table between cycles. I hate, hate, hate all the stupid brooming and vacuuming I do, hopefully I will get some time this holiday season to finally finalize one of my (several) candidate designs and confine this mindless task to the annals of pre-robotic slavery…

[bstern]

Great Discussion! Thanks for all the response.
Ive been out of pocket and now it looks like I'm overloaded till after Christmas. Nobody want to make a decision and then they all want it done before the end of the year! I will have to table my research till next year.

Here are my conclusions to date:

Postal code looks the most promising.

I need to look at the British as it looks to be more compact. 3 different "characters" as opposed to 2 with the US.

It looks doable at a cost of about $1500 to $2000 in equipment. That at the top end but still would have a reasonable payback for me.

Henrik and I agree that the shorter the code the better. 2 Digits should be fine.

My CAM package can write the "Data" layer into shopbot code. I have already done this with all the label info I print now.

Henrik,
I want to hear about your cleaning solution! My current cycle time for a 4X8 sheet is about 15 minutes. Of this, only about 7 minutes is actual cutting, The rest is loading, labeling, unloading, and cleaning. Cleaning is about 2 min.

My improvement plan is this: Load, Cut parts and print bar-codes on the parts. Then Use a "Sweeping Arm" with a bar and a vacuum slot ,table width, to push the parts to an "unloading table" while vacuuming the table at the same time. Cleaning and unloading in one sweep.

What are you thinking??

[henrik_o]

Bob,

Great minds, eh?

Yes, I'm also looking at 'de-snaggifying' the table, have an outfeed bar which during the evacuation sequence vacuums behind the panel, then do a full retract to the end of the table during which pressurized air helps blow the last remaining dust into the vacuum hood.

From some preliminary tests I believe the latter is necessary to perform a complete cleaning.

I am also experimenting with a pressurized air bar to clean the underside of the panel as it is loaded onto the table.

The problem, as always, is time.