Guide: Why Etch A PCB When You Can Mill? | Hackaday

I recalled the idea that I took electronics seriously from the beginning. Although I was very excited, I was scared when faced with the two main obstacles faced by amateurs and even professionals: making your own PCB and fiddling less Surface mount footprint. Any resistance to the latter proved to be futile, expensive, and frankly, a bit stupid in retrospect. Cheap SMD tools make it very easy to store, place and weld all SMD items.

Once all hobbyist designs/experiments are limited to SMD, how will you produce the PCB needed for the prototype? I personally don't like to etch my own boards. This process is laborious and involves messy chemicals and particularly sensitized PCBs-I am not interested. I've only done it a few times and promised myself that I won't do it again. Professional but inexpensive PCB manufacturing is more like a board assembly service like OSH Park makes this easy and affordable-if you can wait for the turnaround.

So what are the options? If you really want to do rapid prototyping in your own laboratory, then I made a case of milling your own PCB. In the process of continuing reading, I will guide you through the typical workflow from design to prototype, and convince you to accept the relatively high start-up cost of buying a PCB factory.

In recent years, the terrible cost of purchasing CNC machines with sufficient rigidity and precision to meet the needs of the work has been greatly reduced. A quick search for "engraving CNC machine" on eBay and Aliexpress will return some interesting and affordable clicks. The general-purpose machine floating on these websites is called CNC 3020, which represents a 30cm x 20cm working area and costs about 250-300 pounds (350-400 US dollars). Other models are 3040 and 6040. These machines all seem to be based on a single design (or even a manufacturer) and are well structured. The aluminum-based structure is sturdy, and of course various materials can be processed under the correct spindle.

The cheapest machine is often equipped with a simple DC 50W electric spindle, which is reliable enough and, more importantly, cheap to replace. It is suitable for engraving and cutting wood at a moderate feed rate, nothing more. The newer machine is equipped with a 200W water-cooled spindle, which is undoubtedly a better choice. You can also use them to cut moderately thick aluminum.

I bought one of the early versions of CNC-3020 quite cheaply. The machine has no end stops, has an old parallel port-based interface, and does not install most of the PCB functions, such as automatic spindle control. We wrote articles for a while [whitequark]

 Upgrade the hardware. I also made the necessary modifications, just connect the 3 end stops to the CNC, and then weld them to the surface probe connector that was not originally installed. It is wise that many new machines are now pre-installed with these machines.

After preparing the machine, we need to obtain the correct tools for milling PCBs. The most important one is the engraving position, which is responsible for submitting all actual traces to the PCB. The bit needs to be sharp, so actually cut rather than tear the copper, and of course need to be fine enough to mill any tiny footprints or marks. After a lot of experimentation, I got the best result

. These are cheap and reliable enough for about 8-10 medium-scale jobs. However, the V profile does present a challenge when specifying its cutting width in the software, because the deeper the cut, the wider the cutting range. This can be adjusted by trial and error.

Next, we need some drill bits to drill through-hole component mounts and through-holes. Each different hole you specify requires a manual tool change, so it is best to stick to a single 0.8 mm – 1 mm drill bit, which should be suitable for most situations. I recommend buying

.

The last one we need is an end mill, which is responsible for cutting out our board from the larger PCB overlay. 1-2 mm end mills can do this job well.

In addition to the above points, we also need the actual PCB overlay. I found the universal, cheap, single-sided overlays on eBay are very bad, and they collapse like cookies when cut. The better ones seem to be two-sided, in both cases I have to try some different suppliers first

. Throughout Hackaday's history, purchasing high-quality copper cladding seems to be an unresolved topic.

In addition to the important items listed above, we also need the following:

The biggest challenge in the DIY process is to obtain the correct and consistent cutting depth. The deeper the cuts produced by these V bits, the deeper you go, effectively depriving you of precious engraving resolution. If you do not adjust the correct depth, some traces will become too thin and fragile. The key to success is to try some cutting depths, and the most important thing is to probe the PCB to compensate for any changes in the overall cladding height.

In order to adjust the correct cutting depth, I compiled a simple test design, which includes a breakout board for ultra-small size 

. The trace size on this board is as small as my design. Next, I tried several different cutting depths and finally found a cutting depth with reliable results for these fine traces. I use 0.8mm cutting depth with 0.1mm V position.

The automatic leveling process is set in the software and only includes the detection program before milling. What we need to do is to solder the scrap wire to the corner of the PCB cladding and use a lot of double-sided tape to fix it on the ideal flat scrap wood.

After completing this operation, fix the wood to the CNC work area and fix the milling cutter in the chuck. A pair of alligator clips are fixed on the wires soldered on the PCB and the milling cutter to complete the detection setup. The idea is to create a PCB height change map, if the PCB is not completely level, the rolling mill will actively compensate for the error.

The first step is to prepare the CAD files. What we need in our favorite PCB CAD software is 3 GERBER files: Top, Bottom and Drill. But before that, we need to ensure that all necessary precautions have been taken to maximize the chances of the factory copying the circuit.

As an example, let's talk about some things that I ensured when designing the following PCB that has been successfully milled many times. For anyone who is curious, this is the phase locked loop part of the DIY spectrum analyzer based on Analog Devices ADF4008 IC.

The smaller packaging used by milling in the later stages will certainly advance the functionality of the process. Since a smaller footprint also means smaller wiring, the reliability of the entire process is greatly reduced. For the same reason, I try to use 0805 or 0603 as passive components.

Compared with any professional manufacturing, these restrictions are definitely very strict, but as long as you pay a little price, you can produce practical and realistic circuit boards.

Once it is determined that all the above constraints are complied with, you need to export the GERBER files and then convert them into GCODE that can be run by CNC. To this end, we use excellent open source software:

, Specially designed for 2D PCB CAM process. It has a wide range of features, you should really check it out.

The tutorial on FlatCAM is beyond the scope of this article, please check

. In short, for each imported GERBER file, you can choose to generate various tool paths for different types of jobs, such as Cutout, Isolation milling, etc. All you have to do is to type in some parameters to specify the cutting depth and drill size.

For a single panel, we need 3 GCODE files from FlatCAM. Top layer, drilling and cutting. With these, we are almost ready to grind the board!

The final software step is to add an automatic leveling routine by post-processing the GCODE top-level file generated by FlatCAM. For this, we used another excellent software called

, which one

There are free and paid versions, and the free version lacks the ability to share detection data across sub-jobs. This is important because you can only probe the board once at the beginning of the job, that is, after you have cut off and therefore isolated part of the board from the PCB overlay, you cannot probe the drilling job! To be honest, I don’t have to buy the full version yet, because probing is only critical to the milling path.

The most important parameter in this process is the probe pitch, which determines the detection resolution. The smaller the probe spacing, the more points will be detected. Although this will take some time, it is worth at least 80-100 good results. After this file is generated, it will replace our previous "top Gcode" file, and we are ready to mill out the board!

At this time, we will provide 3 files for the single-layer board: the top file with detection, the drilling file, and finally the contour milling file. We grind the top file first. Start your favorite GCODE interpreter/CNC controller, I use MACH-3, but Linux CNC or most others are viable options.

We first reset all 3 axes, position the spindle about a few millimeters above the corner of the PCB, and then reset all 3 axes to zero. It is important not to lose the XY zero position between different jobs. Each new job requires re-zeroing on the Z axis, but the X and Y zero positions will remain unchanged. Make sure that the detection circuit is set up correctly, otherwise the milling machine will hit the PCB! Once we are engaged in this work, the mill should begin routine investigations. Once completed, the milling routine will begin, depending on your feed rate and number of passes, it usually takes a while. Once completed, despite the messy glass fiber dust, our circuit board is expected to be similar to our design.

Next, we continue to drill, this step does not require precise re-leveling. We only need to insert the drill bit and set the Z axis to zero so that the drill bit is flush with the PCB surface. After completing this step, all specified holes should be drilled in the board.

Last but not least, the board needs to be cut out from the remaining PCB overlay. The process is very similar to the previous step, but the position and feed rate are different. We zeroed the Z axis again and then run the job. If all goes well, we should have completed the PCB!

Now enter the more valuable part of the process. The milling process is not perfect, especially when the engraving head becomes blunt, it tends to tear rather than cut. To remove stray copper scraps, first scrub the PCB with a thick hard-bristled brush, and then scrub with a finer sandpaper. After wiping with alcohol at the end, we can finally weld the components!

Look at the final result. It looks good, although you can see some defects, such as residual copper traces between actual traces. Since we specified enough gaps between traces and pouring, these should not be a problem!

Now you may be wondering if this process applies to double panels? The answer is yes. Although aligning the sides may be a bit cumbersome, with the right exercises and fixtures, it is not too difficult.

In fact, FlatCAM provides a way to drill alignment holes for alignment with 2-layer designs. Finally, you may also want to know whether this well-designed workflow can produce consistent results for demanding footprints. The answer is yes. I have produced multiple boards for DFN-8 (3mm * 3mm * 8 pins!) packages at once! Check out the 1 GHz mixer based on the LTC5560 IC in DFN-8.

What does SMD have to do with it? Of course you can use the same method to create circuit boards for through-hole components!

I am kind of like someone told me that I should use SMD, just like someone told me that I should use Metric. They have many advantages and are correct in a perfect world, but they are totally opposed to other people's personal situation may be different from their own.

In my case, it will be a garage filled with through-hole parts boxes and local hardware stores. These warehouses can carry almost every type of fastener, cutting and measuring tool required by the imperial system, but if the price is too high , You can only choose a handful of options and none at all.

It is not really possible to add anything to this particular conversation. Annoying, cut it out!

The circuit board I made here is for RF mixers> 1GHz. Try to do this with THT. In principle, enjoy the ease of THT as much as I can, but limit yourself, which prevents you from doing many things, such as RF.

IMO smd is much easier than THT, no need to constantly flip the board

As long as all your components are on one side...

Not only that, insert the component, flip the board, solder, cut the wire, flip again and repeat. Patching is much easier. Just place and solder, no need to cut lines and flip the board. Double-sided board does not matter, every side is fine.

Flip chip-welding bottom component-flip chip-welding top component.

Flip twice.

I only flipped twice. I pulled all the through-hole components to one side of the board and bend them to fix them while inserting the axial wires. Turn over the solder of all components on the board, trim the legs and finish. But I also do SMD for high density boards.

Ask you to accept:

Indeed, the main function of through-hole components that cannot be achieved in RF is ampere: all other functions are no big deal, because they are already quite lossy in nature. It's just more expensive.

Lol. Therefore, people have given reasons to use surface mount. I will repeat.

"They have many advantages and they are right in a perfect world, but they totally disagree that other people's personal situation may be different from their own."

So... I will not try a through-hole construction> 1GHz. So? But this does not want me to throw away a box of through-hole components and buy all new SMD products. Hmm...unless I find out that all the products I build are> 1GHz. I don't think this will happen, but at the same time, I have no reason to waste it.

However, for me, the more important point is:. I can perform CNC machining on a board, whether it is surface mount or through hole mount. So, why does an article about etching a board on a CNC controller even have to start a meaningless argument?

Why are you so angry?

It may be because you can paste breadboards on through-hole components, but not on SMDs (without breakout boards), so the main obstacle to using components provided only in SMD packages is the inability to quickly assemble prototype circuits. So... this is the way to solve this problem without having to deal with annoying chemicals*.

*Except for inhaled dust that can cause cancer

If we compare the workflows, then the feeling of people inserting through-hole components into a solderless breadboard will not replace the step of "using the breadboard out of the box and stealing the circuit board" for designing and manufacturing a prototype PCB. Moved by the convenience brought. A few jumpers, and then put them on the table. "

If you are sure that the circuit you are designing is a product, it will be beneficial to start with SMD, especially if you want to implement another standard and do not need to build a proof of concept first. But I think most people have not built their own circuits. Maybe it will eventually become a product, but it may not.

The complaint is not that someone is using SMD, but that many of them then turn to people who use THT and tell us that we did something wrong. It's not.

SMD nonsense is just that; meaningless parades serve as information. What I like the most is my opinion; what you like the most is not my opinion, just nonsense.

The actual theme of this story is not about SMD vs THT, but about using a mill to make PCBs. For people who use THT, it is as useful as people who perform SMD. After all, they did make SMD prototype boards, and SMD people actually didn't need to design PCBs to achieve this goal.

The only "about" and "about" in the story are "milling", "etching" and "ordering".

*I use emacs for programming and vim for system management, which is why I am an expert.

If you want to use double-sided overlays for through-hole machining, you need to mill both sides. As the article points out, it is feasible, but it adds a lot of pain and complexity. Therefore, SMD is much easier, and you can use the other side for the RF ground plane. :)

If single-sided copper is used, through holes are easy. But you cannot get SMD density.

In my garage, most of the boxes filled with through-holes disappeared because almost all manufacturing processes were handed over to SMD. Of course, there is still something that needs to be used/through holes, but it almost disappears. In the past, you could pick up a lot of extras through the holes in the hamfests, but now it's mainly SMD reels.

Not to mention that the complete E24 resistor/capacitor classification can be placed in a box of 20cmx30cmx2cm.

"You are milling on both sides. As the article points out, it is feasible, but it brings a lot of pain and complexity."

So... I guess you created a double-sided board but no vias, so each side looks like its own isolated single-layer PCB? After adding vias, you need good top/bottom alignment, which is the same as using a via plate.

As for the ground plane, even if every project I have ever worked on involves radio frequency, I can still use the old through-hole power supply to build power supplies, control boards, and any other products that do not require a ground plane.

If I want to compete with portable commercial products in size and weight, SMD density really matters. My things are probably disposable items that I live at home or in the car.

"You used to be able to pick up a lot of extras through holes through hamfests, but now it's mainly SMD reels."

Wow. We will never participate in the same carnival! In my area, they are still through-hole materials. This is how I originally used a garage full of it!

"In my garage, most of the boxes filled with through holes disappeared because almost all manufacturing processes have been transferred to SMD"

My garage is more like a playground for hobbyists than part of manufacturing.

"Not to mention packing a complete E24 resistor/capacitor classification into a 20cmx30cmx2cm box."

good idea. But... all the salvaged mechanical parts really make my workshop messy. Although my components may take up more space than you, eliminating them will not really reduce costs.

good idea. I have a corner dedicated to components, and the rest of the mess is dead equipment (most of which are salvageable annual sales discoveries)

It seems that there are some good arguments for SMD in RF applications. It's kind of like metrics with some good arguments. Perhaps the complaint really stands out in a useful and constructive conversation?

Most amateurs need to be able to use metric and standard units. So someone tells you that a person did a mistake by leaning on a windmill.

Then someone said: "Oh my God, I don't like complaining, how dare you complain about others!" In my opinion, this is not consistent. The original complaint was not complaining about existence, but about other people's choices, and the response was complaining that the specific complaint made ("this should be metric") was inappropriate and impatient. Then, the third person said: "Maybe the complaining speech really annoys people." They didn't even realize that they didn't address what other people complained about, talked about different topics and didn't do anything. There is no substantive meaning at all.

Maybe you are right, but maybe you are wrong, so maybe you should flesh out your ideas and check their consistency?

Island routers are slow (for example, the equipment we purchased takes 3 to 5 hours per side), but they are useful for certain types of problems.

For example, ultra-thick high-power switch boards or tuned RF cables.

However, if there is no through-hole plated through-hole, it does not actually provide any other functions except the classic chemical etching process.

There was a time when the CNC process was used to deburr the RF wire, but most high-speed rolling mills usually add more potpourri than they remove.

I recommend using Chilipeppr or bCNC on the automatic leveler. Both can be automatically leveled, but they are free and well integrated.

+1

And check the woodpecker CNC controller.

If you are using Windows, openCNCpilot is a simple gcode transmitter with automatic level of grbl. There is no web interface like chilipeppr. bcnc is very good.

I am using mach3, but I lost the steps due to a problem on the parallel port card. Running arduino grgr on the parallel port input on the 3020 cnc runs more smoothly and reliably. It allows me to use usb. I use the same toolchain written by the author, such as eagle –> pcb-gcode –> autolevel –> mach3 –> cnc. Grbl has different detection methods, so the tool chain is now eagle –> pcb2gcode –> openCNCpilot (with built-in automatic level) –> Grbl –> CNC

hoo! I really like using these cheap CNC milling machines for PCB!

Buy used 1/32" end mills from eBay

They are so cheap that they won't even sweat to throw them away when they become dull.

I think that by using an end mill, you do not need to perform any automatic leveling. Moreover, you can use the same tools for trace cuts, circuit board outlines and drilling.

Make sure you buy FR-1/Bakelite board, not fiberglass. I just got it directly from Bantam

I used 2 inches wide double-sided crepe paper tape to press

. I think it is easier to flatten than plastic double-sided tape.

We have been using CNCJS to drive GRBL arduino to connect to the factory

I hope more people start to do this. Super fun!

Good tip, thanks

I have achieved great success in these areas

0.5mm end mill, two packages, very cheap

I agree to use an end mill for engraving and drilling. I did the opposite of the author. I drill first, then engrave.

However, even with an end mill, the result of leveling with AutoLeveler using a leveling instrument is better than "digging in", which just finds the higher part and over-engraves one area to ensure that the other area is engraved correctly. I did some retests on the board to compensate for the uneven copper cladding.

First drill a good hint, thank you!

Is there any benefit to drilling first?

I grind first. My drilling tips: 0.8mm drills sometimes move around (and when the cemented carbide then breaks;)), but the remaining 2-3mm fractured cemented carbide drills work well and no longer drift, thus forming very Clean hole. (Some of my milling can be seen in the YT link here: hackaday.io/project/24926-arduino-driven-led-display-with-bt-data-transfer)

Greetings

Is the hobbyist market so big, or is this small shop a one-off?

If your market is large, the time it takes can be prohibitive. In this case, this only applies to prototypes, and you will send the product to the production board.

Unfortunately, the CNC 3020 link (not the hackaday article link, but the link contained in the articel) is closed...

AliExpress/eBay is full of these 3020 (or larger/or smaller-numbers are size) machines. Just search for "CNC 3020".

The problem is that the machine usually has a curved main shaft, the main shaft has a large runout, and the shaft will bend under load -> poor results. Look for a variant where the main shaft is driven by a belt and is located in a suitable bearing.

Many of them also need some simple modifications to make it work with modern computers without classic parallel ports (electronic devices are just a few power transistors controlled through the parallel port, bringing all the negative effects).

The detection circuits required for leveling are also additional parts (not difficult) that you must build, such as sacrificial plates and parts that actually clamp the workpiece. The fixtures provided by some suppliers are generally unusable.

Therefore, you buy a machine that costs $800, but expect to make a lot of modifications before it can be used for this type of work.

I personally don’t think it makes much sense to mill this type of board-it requires a lot of preparation, the milling itself is slow, noisy (not good if you live in a flat apartment!) and very dusty (you need a vacuum cleaner) to remove Dust, do not grind FR4->Dust can cause cancer! ).

If I use conventional toner transfer or photolithography to complete the milling of the board, then I will have completed the same work and completed the assembly work-in the latter case, the resolution is also cheaper than these The milling machine can achieve better. The only problem is to drill holes in the through-hole parts, but a decent drill stand is much cheaper than a CNC milling machine, and there are very few holes in typical surface mount boards.

Like some microwave designs, milling requires precise surface dimensions. But for the general prototype? why? ? Some people are afraid to worry about chemical reactions, but of course I prefer to deal with things like ferric chloride instead of factory noise and dust, especially in apartments.

Oh, don’t forget, today you can get ten 100x100mm circuit boards for $2 in a week-two layers, solder mask, plated through holes, screen printing, and wiring in any shape you want. Then why should I polish it? I only make my own boards for the various adapters/branches I need now, otherwise it doesn't make sense.

***This is a special price of JLPCB, but even the normal US$20 is pretty good, especially considering the fast turnaround time.

See my comment below, the $230 all includes Pi (connected to the mill via USB) and limit switch.

When you do small SMD with small pitch, the milling effect is very good. I always try to etch away the traces and wait for completion elsewhere.

I agree that sending it out is a multiple method. But these are very useful for prototyping. Verify the design to avoid delays in Asian development boards. Yes, they are noisy, but like you, I can take out a plank, including drilling holes in less than 20 minutes. And I don’t have to explain to my spouse the small amount and disposal of chemical substances. Different tools can accomplish the same job. To everyone himself.

Although I would not use this setting on the PCB, I found this inspiration. I'm thinking about keyboard keycaps.

A small frame can place several keycaps flat on the side or side (for engraving on the front). Depending on the size and shape of the drill, you can even change the width of the font by deliberately changing the drilling depth. Fill the grooves with paint and/or resin to achieve a lasting effect.

I bought a similar factory at MicroCenter for about $200. It has a controller based on the woodpecker GRBL. There is no end stop switch, but it took several hours to reach the end.

I added Pi Zero W, which is completely independent of chillipeper. You can even drag and drop Eagle files directly on it, and it will spit out wooden boards.

Great tutorial. I used Bantam CNC (formerly known as Othermill) for this, which was originally designed and made the software easier. But considering that companies like OSHPark can manufacture PCBs for as little as $1 in a few days, I don't do much milling anymore. Why not just use OSHPark and get a double-sided circuit board with through holes in a few clicks, which is really difficult to do in the factory?

>However, given that companies like OSHPark can manufacture PCBs for as little as $1 in a few days, I don't do much milling anymore.

I never understand why people think that the price of OSHPark is so low, when it is not. The price of a two-layer board is $5 per square inch, plus shipping. Usually, transportation is the main cost...JLCPCB is the cheapest price I know, 10cm x 10cm is 2 dollars, but transportation is not free, and the top cost is about 10 dollars.

OSHpark 2-layer boards cost $5 per square inch, but you can get three of them. Therefore, the effective cost is $1.66 per square inch. Free shipping from OSHpark using regular post. Yes, if you want to speed up production and pay for faster delivery costs, you can do so, but it is not required.

Recently, I have been using dirtypcbs.com, and you can buy a standard 2-layer PCB at a price of US$10 x 10 cm (~10 copies) for only US$17, plus a Hong Kong Airlines shipping fee of US$9. According to my experience, there are several weeks. Compared with OSHpark, the price of four or more layers is not expensive. Therefore, the price of ten boards is 26 US dollars, and the area of ​​each board is about 4 square meters.

> Free shipping from OSHpark using regular post.

Only if you live in the United States (especially people who live in the United States tend to forget) in the United States, most people on the planet will not do this.

Dirtypcbs is a terrible service; low quality planks and slow shipping. However, even if you seek high-quality services with reasonably fast shipping (it only takes a few days worldwide), it will not damage your bank...

OSHpark is ok. For people living in the United States, it is undoubtedly one of the best choices in the local area, and the price is relatively cheap. Take Europe as an example. Although there must be very good and expensive PCB manufacturers, there are actually not so many *reasonable price PCB manufacturers.

Just correct this here: we (OSHPark) have free international postal mail shipping for most of the 4 years.

Hey, I am glad to hear that.

Again, OSH Park has saved me countless hours of frustration-when I sent you the Gerber files produced by KiCad, I got exactly what I expected, and for me, it was worth the wait. I am actually like a purple solder mask. Apart from our enthusiasts, no one even knows what it means to see a purple board-everyone else says: "Oh, pretty".

Is there an ulp or other simple way to make gcode burn paint from the gerber file? Most of the things I found are used to isolate routing.

The Gerber format was developed to drive the Gerber photoelectric plotter. It is basically a 2D CNC machine, so it is basically a list of XY movements. I think it is a very simple matter to convert each line in the Gerber file into one or several lines of G code. It does use "holes" for the pads, so there is no need to specify a path for each individual pad, but again, each hole translates into many discrete movements. But no, I don't have the name of any software that performs this operation. I think the simpler and perhaps faster way is to generate a raster file (.png format), you can use several programs (I use gerbv for final verification, and do this), and etch it as a picture.

Since you are talking about ulp, I assume you are using Eagle. I just asked Eagle to export the bitmap (BMP) at the resolution of my laser cutter (500 x 500 dpi-full spectrum laser), and then only used the Laser program to rasterize the bitmap. The traces are not as good as those obtained by grinding or photoresist. Enough for 15 million traces and spacing, but don't try to use 0.5mm spacing SMD for this operation-at least I did not succeed. If anyone else has achieved greater success, please leave a comment and let us know.

As you have to wait three, four or more weeks to reach the board of directors, this dispels people's mood and destroys the concentration of the development cycle. .Shipping from OSHPark to all over the world is not so fast (and not so cheap)

+1 I like OSH Park, Seeed (for white mask, no code) and other suppliers for making beautiful wooden boards, and I like the milling that I want to complete the project today.

You can buy a lot of express delivery for $300...

Since most Chinese express freight rates have fallen to around US$50, this is a good thing for the couple.

I have tried this method, not just laser engraving PCB, but so far, the easiest method I found is to spray bare copper, use the laser to ablate the paint only where it needs to be etched, and then use hydrochloric acid Mix it with hydrogen peroxide (you must pay attention to the mixing ratio), and then wipe it with MEK after etching to remove the remaining paint. It is possible to use a router, but I found that the cheaper machines are cumbersome and the installation time is longer than just using laser and acid etching.

To try the laser method at home:

What power laser do you use?

What is your minimum trace width?

Does the paint type matter?

I am also interested in it. This sounds like the easiest way. I don't have a laser cutting machine, but if it doesn't require much power (for example, more than 5 watts), you can place a laser module on my CNC engraving machine, which is very useful. All other methods have many ways to solve these problems.

In my opinion, it all depends on the density and size of the traces or the overall complexity of the circuit board. I can see milling simple boards, but I don't think it will be great to mill complex boards. I also agree that it will be more difficult to process double-sided boards with a mill. One reason is that photoetching has been around for so long.

I can't see that the registration of the second side in the mill is much more difficult than the registration of photolithography. This is indeed the least difficult.

Does the through hole in this design have a purpose?

I originally thought they were fenced vias, but they have not yet been connected to the (presumed) bottom ground plane.

In fact, these may be fenced through holes. If you choose to plate through holes, you can keep these through holes, which are preferred in the milled version shown here.

Can anyone calculate the cost of a 100mm x 100mm board, including all consumables (engraving machine drill bits), just for comparison with outsourcing? I saw the 20-minute figure, just to compare the cost of using this and the cost.

Apart from being able to iterate faster, I don't see any monetary advantage.

If you have time, please waste it, if you want to be faster, please do it yourself :)

I recently converted my old Ordbot Hadron 3D printer to a PCB board, and I am not excited about the best stitch length I can make. Now, I want to perform a mixture of drilling and milling first, and then toner transfer and etching. We will see how it works.

For the actual PCB, I usually order from OSH Park. Because when I need something quick and dirty, I often use perfume. And since less than 10 cents per 5×7 perforated board is not cheap enough for me, I often take apart eh Dremel and cut off the unused parts for later use.

"Each 5×7 perforated board is less than 10 cents"...Unless you need a spacing of less than 2.54mm, it will either become very expensive (I found that the cheapest price for a 1.27mm perforated board is a small piece of 10 U.S. dollars), or completely impossible. I have never found such a perforated board, it can meet TSOP components (most large-capacity RAM chips need) pitch 0.8mm or QFP (most modern programmable logic devices, such as CPLD or FPGA required pitch) 0.5mm spacing. . You can* try to use a breakout board, but the signal may not work well in the gigahertz range of hundreds of megahertz or lower.

I tried FlatCAM, it is difficult to run under OSX. I used Carbide3D's network tools to play a lot, and its quality left a deep impression on people. In addition to the unexpected "zero offset" in the generated GCode, it also works perfectly.

It's nice to read the article about this technology. I started to work on the entire design/printing/etching method, but I was tired of all the etching steps and personal contradictions.

Now, if I need to do something, I can do the layout, print it out on plain paper, and then tie it to some copper cladding with a dremel with a cut-off wheel to cut the traces. It's not pretty, but it works. The smallest component I need is SOT23-6, so it is not small enough that I need more detailed wiring.

I have never liked milling to satisfy the hobby of PCBs, because it is a hobby in itself. As mentioned above, they are getting better and better. It is under hacking, so you can conduct electronic hacking.

I stay away from photoresist to meet the needs of amateur printed circuit boards, because it has become a complete hobby in itself, and progress seems to have stalled. It is under hacking, so you can conduct electronic hacking.

I bought an inkjet printer for PCB only. It is certain that it has been seen by HAD for a long time. The special anti-corrosion cartridge can work and needs improvement, but it can work. It's almost here... just failed.

At this level, we have been stagnant for many years, but the progress made here and there for such examples is very interesting.

I'm still using a laser printer for transmission, and for small quick things, I need to make a sharpener by hand. There are still some datak plugins that work properly, they will corrupt themselves, but they will get old.

Maybe you can drag several garbage laser printers together and build a double-sided laser scanner to directly expose the pre-coated photoresist plate... But again, another hobby that can be developed on its own.

Thirty years ago, we believed that we were on the verge of some new hobby toys that could solve this problem for us. Many people like this are okay, but, guys...we have been trapped in sand and spinning wheels. Some people have proposed to get rich, but time is running out and things have become so small.

The most needed part of the hobby is more than 20 years old. Nothing can be resolved yet. Almost, but no ringtones.

Until someone creates a home chip technology factory...just like Jeri Ellsworth tried to do.

Rinse/repeat ha ha, hope th can be realized soon

"Until someone creates a home chip technology factory"-until?

Describes a manufacturing system that can be used to produce 200nm-scale semiconductors on 4-inch wafers using a desktop device, while the manufacturing cost of the desktop device is less than $400.

Julian: I haven't walked through the paywall, but the $400 seems to cover only the UV illuminator needed to expose the photoresist. This is just the simplest of the many equipment required to actually produce the chip. You need to be able to produce a mask and reduce it to the actual chip size, and copy the pattern of each layer at precise intervals. You need something to align the mask so that the layers can be perfectly aligned (in Tens of nanometers), you need to be able to spin-coat the wafer with photoresist, you need to use a furnace to diffuse the dopant into the crystal lattice, you need to score the wafer and then break it into chips, and you need a Method Bonding wires to pads on the chip and then bonding to the lead frame or PCB. About fifteen years ago, this is what I learned from the loose connection with the manufacturing process (I did laser trimming and test development, so it was processed when I got the wafer).

Jeri Ellesworth (Jeri Ellesworth) proved by making a huge MOSFET that the real wafer processing steps are achievable for enthusiasts, but this is at the level of masking with tape. Not at the level of chips used in production.

Again, this may become a hobby, and it will be a good choice for hobby communities like CNC machining and 3D printing, but I think the whole process does not exist yet. Someone please prove me wrong.

Now that I have read what I wrote, there are actually not many devices. Among them, the wafer scoring step may be completed by a hobby-level CNC engraving machine. And someone outside may figure out how to do wire bonding. This is also within the accuracy of amateur CNC equipment. Mask alignment? Don't know which one. I know that by combining a mechanical stepper and galvanometer with a computer vision system (video microscope and pattern recognition software), you can reduce to the 1 micron range, but even this is not suitable for the faint-hearted.

Hi Biomed, I have a 102% consensus with you on this point – it’s fun to try something non-stereotypical, such as making a PCB with an engraving machine, like trying to make a resist layer with a laser printer, and like trying Using a pen plotter is as fun as using a fine-pointed Sharpie pen to draw circuits. Even trying to use "professional" methods to make masks, sensitize the plates, expose and develop them, etc., all have a good size learning curve. However, when I try to do other things (such as building a circuit), these things are no longer interesting. I am the opposite of ADHD. I hate having to change the task on the fly, because when I return to the original task, I have lost any thoughts in progress.

The method described above is the method described by Cody, which uses paint peeled off with a laser cutter, which sounds like the easiest to correct-I have never encountered troublesome etching parts. The question has always been how to obtain a resist pattern on copper.

I feel very lucky because KiCad has reached the level of development comparable to the professional systems used in the past, and now provides services such as Seeed Studio, DirtyPCB and OSH Park, so I don’t have to be confused. I have not developed a reliable process. The K-child. In my last three projects that required PCBs, I went to OSH Park, and in all cases, I got my design completely. In other words, any problems I encounter are design problems, not PCB fab problems. It took nearly two weeks to get the board, but it was because I spent two weeks doing other things instead of two weeks, and wasted a lot of materials trying to make the process work properly.

Great article, thank you. I have some small questions:

\ cite {I use a 0.1mm V-drill with a cutting depth of 0.8mm. }-Is this a typo or is it really cut by nearly a millimeter? There are many methods. Depending on the cutter head, 0.1-0.2mm (I use a 45° engraving cutter head) is enough to cut through the copper wire reliably (also depends on the stiffness of the Z axis).

The cardboard cutting board on your picture looks terrible-is the cutting knife dull? When cutting FR4 boards with a simple 2 or 4 edge milling cutter head, I got clean cutting edges.

Many Chinese factories are offering 10 pieces of 100mm x 100mm or less circuit boards with green solder masks priced at $5. 2 layers, double-sided screen printing and solder mask. With the DHL Shenzhen shipping fee of about US$17, I can reliably get the board within 7 or 8 working days from the start of the order. The circuit board in each hand is about $2.20.

For schools that "but I can never use too many boards", for some reason they *must* buy 10 boards, even if their out-of-pocket costs are lower than buying fewer boards from other sources, they still feel torn, you You can also order 5 yuan for the same price of 5 yuan. Hehe, how is this saved? You save transportation costs.

I have printed Cyclone PCB Factory and it is very close to milling my first board. However, I was having trouble centering the chuck on the spindle. Looks like yours, two hex nuts. The adjustment is very sensitive and I always feel a little swayed. The video here is bad:

.

Is there a tip?

These spindles are notorious for being eccentric. My condition was good at first, but as the bearings/bearings wear out, it gets worse. The spindle seems to be malfunctioning.

I have done it all: CNC milling PCB and chemical etching. I don't understand why people claim that you have to deal with "nasty chemicals"? Ferric chloride is very safe and can even be used for water treatment. If it enters the environment, it's no big deal. It does stain your clothes and many other things, but I used it when I was a kid with chemical devices. Even the photoresist developer is quite safe, and sodium hydroxide is a waste. Sodium carbonate is heated baking soda (yes, it can be found in food). Don't think they are "nasty" and try it. Both methods have advantages and disadvantages. When I need to drill holes for through-holes or many through-holes, if very fine pitch is required, I use cnc milling bars. Oh, to answer other people’s questions, you have to drill before milling because you want to prevent the drill from moving.

I have read that ground PCBs can cause lung cancer because if tiny particles are released in the air. Does any of you consider it?

Where did you read it? ? ?

Well, where did you read it-are there any details? First, how about allergic dermatitis? If you are worried about dust, please try wet grinding?

It is dangerous to drill or grind (or mill) glass fiber here. One solution is to use paper/phenolic paperboard (FR-1 or FR-2).

In any case, these should be enough for side projects. I found that even the cheapest circuit board manufacturers, you can't beat their quality. Isolation routing is suitable for testing and prototyping to confirm whether the design is correct. But in the end it makes you want.

Bart, yes... Some kind of vacuum/filter system that sucks in particles during the grinding process would be very good-not only good for health, but also reducing the "wood dust" manually vacuumed after the cutting process. Although I really like PCB milling. If the user has the appropriate system (software, hardware, measurement/alignment/adjustment/setting method), this particular technique is certainly feasible. For the home environment, it is definitely better than handling chemicals. But...in any case... it works for each of us... Very good and, as always, safety is important.... Therefore, it is best to wear masks and goggles, and consider how to set up the system to avoid health problems.

If you have sharper tools and proper settings, you don't have to sand your planks, don't sand for high-speed steel, and don't go for cemented carbide. Unless you plan to cut hundreds of plates, they will not wear out, and they can increase the feed speed without heating and cracking. Heat is your worst enemy to keep your tools sharp during machining.

Now, as long as you have the right tools, you should use a very high rpm, preferably 30000 or higher.

Then keep the feed speed at 50-100 mm/min.

Keep the depth around -0.2 to 0.3 mm. If you go deeper, the track will be wider, and the shallower will be burrs.

Soak the boards with oil. This must be used: First, it keeps the tip of the knife cool, thereby keeping it sharp and avoiding burrs.

Second, it will absorb all the dust, so you don't need to use a vacuum cleaner or your lungs to suck it up, just wipe it when you are done. As an added benefit, it can also suppress noise because it can reduce vibration.

It took me several hours of thinking and testing to reach this conclusion, but it was worth it. When typing this command, there is a cheap $200 CNC on the right side of the keyboard, which is running the PCB. I don't even have to watch it at runtime, just play the game and read it at the same time. I will never use acid to make PCB:s anymore, which is time-consuming and laborious.

Please be kind and respectful to help make the comment section great. (

)

The site uses Akismet to reduce spam.

By using our website and services, you expressly agree to our placement of performance, functionality and advertising cookies.