help buying a good PCB mill

[G0HZU]

(ever watched a router bit cut it's way through 5 hours worth of milling and tool changes?)

I saw that happen to a colleague many, many years ago on a very large and complicated two layer board. On the older T-Tech machines the router bit can get very hot under certain circumstances and it can actually come loose in the chuck. This is a bad aspect of the design. So the tool can drop down into the PCB when the machine thinks it has lifted it. So the moment the machine moves the tool across the board it cuts a huge score across the PCB and wrecks it.

So I have learned from his pain and I always try and configure the router pattern such that it allows the tool to cool between long cuts and also I can control the order of cuts to prevent the machine crossing the board after a long (hot) cut.

Also, I think it really is a false economy to be milling boards that take a long time to mill. These machines are at their best when milling small RF boards or small piggy PCBs that are used to modify an existiing PCB. It can mill a smallish PCB in under half an hour. Or it can mill a laminated tool in even less.

My T Tech machines came with a service kit containing a spare motor, Z solenoid and other accessories and both are still on the original motor and solenoid. The machine I use most is over 20yrs old as I bought it used way back around 2001/2.

[mrpackethead]

(ever watched a router bit cut it's way through 5 hours worth of milling and tool changes?)

Yup. not always 5 hours worth though, sometimes it was just 3 hours..   Software just randomly thought it would go home in a straight line.. or something else dumb.    REally cool when it was a 2mm end mill and its just bored its way through the middle of your board.

Nuf said, order your boards in.  It seems a hard thing to do. Really its not.

[G0HZU]

Like I said earlier, at work, our LPKF machine must have produced a couple of thousand PCBs by now.

If you listened to the naysayers on here who reckon these things basically don't work then I guess you would conclude that I'm not telling the truth 

I'm aware of the history of four PCB mills dating back many years. The most unreliable one was the early T-Tech (1990? model) that ran from a PC running DOS and the controller was partly inside the PC in an ISA slot. This old machine proved to be very fussy wrt the PC that was used with it and it needed to be run from a (slow and basic) DOS PC built in the early 1990s. Otherwise it would behave erratically.

My main machine is a later T-Tech 7000S model with all the control done in an external controller box and it can tolerate a PC with a faster processor and it is controlled via RS-232 from the PC. But I still run it from an old and slow PC running Win98SE.  I have another T-Tech 7000S which I bought very cheaply, mainly intended as a (long term) spares donor.

If you want good results from these machines then you need to learn a few skills and show a bit of finesse when operating it and try and use a dedicated (stable) PC to operate it. It's very rare for me to trash a PCB or even to make a minor milling error.

The other thing is that you do have to service/maintain the machine or it will complain. I clean/oil/grease the linear ways and the lead screws regularly and I also use sympathetic settings for certain tools and this helps prolong the life of the motor and its bearings.

One option is to do what I did and buy a used one. Mine came with the Nilfisk extraction system and loads of tools and I got it for about £800. I reclaimed the cost within a few months by milling microwave filter PCBs on Rogers material for a local company.

Overall, I would recommend the LPKF C30S over my old T-Tech 7000S because it is easier to use, the SW is a lot better and is a lot safer for the operator in terms of exposure to moving parts. It also lives in a cabinet supplied by LPKF that reduces the noise a lot when the lid is closed over it.

[rich]

I was going to stay out of this as my home-brew setup won't offer useful insight for the OP, but I am curious to know what type of cutter gets used in these high end machines? - I've been experimenting with 0.2mm V cutters with mixed results.

[G0HZU]

I use lots of cutting tools here and I am also lucky in that I can salvage part worn tools from work. At work we dispense of tools when prompted by the software and I find they are still useful for non critical milling here at home. This helps me keep costs down at home. I have lots of new tools here at home but I only use these when absolutely necessary.

So I still get good results even with part worn pointed tools from LPKF. I also use various end mills from 0.005" through to 0.0371". The end mills are useful when milling precise circuits like microwave filters.

The pointed V tools go dull fairly quickly on materials like Rogers 4003.

The tool I use the most is the universal/pointed cutter (tool #2) from LPKF in the pdf file below. I also use the T1 pointed tool from T-Tech.

http://www.lpkf.com/_mediafiles/1792-brochure-lpkf-drilling-milling-routing-tools-en.pdf

[rich]

Thanks for the pointers, I shall give the universal type a tryout next time and see if I get a cleaner finish.

[IconicPCB]

I have used both type 1 and 2 from LPKF.. great tools but horribly expensive.

I have not used any T-Tech tools. I had trialed Union tools, Kyocera a and a few others and settled on Precise Bits.

Excellent price compared to LPKF... Great tools. Sturdy, long life at half the LPKF price.

Included angle 10 degrees triple fluted clean cutting. This is second generation tool from Precise bits. First generation is still offered at about two third the price of second generation.

I also use TCI carbide endmills... get them from manufacturer directly , minimum order values in the order of a couple of thousand dollars.
Anybody needs square endmill and diamond cut router bits or drill bits.. drop me a line.

I also get the drill bits from the same source.

[mrpackethead]

Like I said earlier, at work, our LPKF machine must have produced a couple of thousand PCBs by now.

If you listened to the naysayers on here who reckon these things basically don't work then I guess you would conclude that I'm not telling the truth 

Well, then your calling me a liar.   The machine did'tn work the software was buggy as crap, oh and the suppllier and the agent coud'tn get it work either.   

Overall, I would recommend the LPKF C30S

Maybe LPFK was better to you than me, but i cant' offer any recommednation from them other than to say, i've not had any purchasing experiecnes go as bad as buying the S62.

[free_electron]

Like I said earlier, at work, our LPKF machine must have produced a couple of thousand PCBs by now.

good for you. i got none.

Here are my requirements :
100% IDENTICAL boards to what an etched board would look like. so no 'isolation miling only'. i needed full copper rubout.
smallest track : 4 mil .
smallest gap : 4 mils.
Smallest drill : 14mils
standard 62 mils board thickness
plated
board : PCI plug-in card : full height , short form factor ( so roughty 10cm by 14 cm )

milling : (lpkf95S)
the milling bits are V-tipped ( not flat) you set the width of the mill by pressing a plastic ring over the shaft of the milling bit. this controls how deep the bit bites.
if your target material is not perfectly flat , or not mounted perfectly flat ( the pea under the princesses matress problem ) the width fluctuates. the spindle floats on an air bearing and they claim they have precise distance control. they didn't. over a length of 200mm i saw the groove fluctuate from 3 mils to more than 5 mils. i wanted milling of a 4 mil isolation gap. milling is inconsistent.

the same issue happens with the drill bits. i can't remember how many times it broke the drill bit mid cycle. especially their double copper foil material was a disaster.

issues with the machine
- serial port problems. machine locks up milling head down , max rpm
- serial port problems. machine draws a line random through everytinhg on the board rendering entire board garbage. they swapped controllers 3 times. finally blamed the computer.
- serial port problems : during a comms failure while performing a tool change ( our machine had an automatic toolchanger) the machine ran the head into the tool tray bending the entire mechanism...

software lockups during parsing of the gerber.

Plating problems : the chemistry goes bad very quickly end is costly. getting rid of the chemicals is also a nightmare. they refused to give out what the chemicals exactly were so our environmental inspector had problems classifying that stuff.
getting the plating chemistry right was a nightmare. the copper was never shiny , non uniform , air bubbles in the tiny holes... yadda yadda. their board moving mechanism was basically a plastic frame on an eccentric motor. that drive bar broke after 1 hour ...

this was 2003...  maybe it's better now. we invested 40k$ in that setup. zero boards ! ZERO !


That stuff is like the hobby 3d printer : you spend more time dicking around with the machine itself than doing work with the machine.
send your boards to china. you can do multilayers, get soldermask and silkscreen , enig and 10% electrically tested. and while waiting for the boards you can do other useful things as opposed to babysitting the crappy mill.

[IconicPCB]

One would think the equipment would have improved in the past ten odd years.

I mean it is German engineering after all. Second to none.

[mrpackethead]

One would think the equipment would have improved in the past ten odd years.

I mean it is German engineering after all. Second to none.

My experience was 4 years ago, and it was still crap then.  Pretty much same issues as above.

[IconicPCB]

The fact You had a bad experience with equipment and the agent could not help means both You and agent were at similar level of need to train on this equipment.

An agent is not an expert on every piece of gear they rep.

I saw LPKF equipment mid 2007  at an exhibition having corresponded with LPKF Germany in the couple of months leading up to the exhibition.

All this negativity flies in the face of personal good experience with this technology.  May be its because I don't mind getting wet during sex.

[G0HZU]

One trick I can offer wrt drilling with fine drills is to prick the PCB first using the pointed tool. This helps to prevent the drill tip skating on the copper surface and breaking the drill bit.

This can be done by simply running the drill layer as a pointed tool and this puts a tiny little surface dimple into the PCB that helps to lock the drill in place.

However, I don't use drill bits below 0.5mm diameter and this is the smallest drill bit in the drill kit supplied to me by LPKF. The finest end mills I have are 0.0055" and these can be used to mill very accurately and this is the finest pitch I normally use. This is much more precise that using the pointed tool and the cutting width is effectively inde[endent of tool depth with the end mill.

The finest work I've done with mine was to mill a flexible ribbon cable using double sided flexi PCB material that is about the same thickness overall as printer paper. The PCB looked and felt like copper paper. Obviously, this took a lot of preparation and some skill and patience to achieve but I managed to mill the material without perforating through to the other copper layer. I used the pointed tool here and LPKF supplied me with some hard and smooth backing material that was as smooth and flat as glass.

Obviously, I'm not an LPKF engineer/rep but I would suggest that all of the erratic milling behaviour experienced would be due to the PC controlling the machine. It's debateable whether this is still LPKF's problem but it must be getting harder and harder to get reliable operation with modern PCs running a modern OS and all the baggage that goes with a modern PC in terms of security/AV/networking/ports etc.

[nctnico]

IMHO you must invest time to get a good grip on any technology used to create PCBs by yourself. The milling machines are sold as being an easy solution but I think that is a "salesmen's truth" (IOW: an outright lie). I think the biggest question you must ask yourself is: how quick do I really need a board? And can I make a prototype in a way it is easy to modify / patch so a prototype doesn't need a respin at the first time of problems?

About once or twice a year I etch boards myself which I need really quick but in most of the cases I just plan my projects in a way I can do something different while waiting for a board to get produced by a professional PCB maker. So far I'm not convinced milling is a good technology for anything else than creating RF prototypes. Doing multilayer boards and/or plated through holes in a company lab or your shed is just crazy because it will be hard to get right and also due to the toxic chemicals involved and their disposal issues. At some point you have to leave a job in the hands of a professional who has all the right equipment and does the work day in day out.

[mrpackethead]

The fact You had a bad experience with equipment and the agent could not help means both You and agent were at similar level of need to train on this equipment.

For clarity, both the agent and the manufacturer were unable to resolve the issues, which was software related.  No amount of "training" would have resolved the issue.  I guess i could have requested the source code and debugged it... but...

I guess its why in the end they took the machine back and refunded the money, becuase it did'tn work

[G0HZU]

No amount of "training" would have resolved the issue.

It looks like you were let down by poorly trained reps. Considering what these things cost new I'm surprised they didn't supply you with a known working PC to resolve the issue. But even then, it may fail again if the PC OS gets updated.

Back in the days when I used to see the LPKF rep fairly regularly he always brought his own laptop with him and I suspect that this was because it was known to work reliably with all of their machines. These days he wouldn't be allowed to bring a laptop to my place of work for various security reasons and I suspect that this may be the same for other companies too.

Maybe a stable version of Linux would be a better option for the OS?

[rx8pilot]

FWIW, I had the same fiddly experiences with my CNC milling machines when I first purchased them. Computer and control problems on top of the learning curve of the process itself let me down a very challenging learning curve. Eventually, I learned how to work around and/or fix the various issues but not before I broke a few things.

I think the mistake LPKF may be making is that they market the machine as a PCB printer where it appears to be in fact a delicate task requiring significant training and experience. Of course I should say I have zero actual experience with these machines. Only went through the pre-sale research.

[G0HZU]

There's a version history of LPKF Circuitcam SW here:

http://www.circuitcam.com/history

Although I don't have an LPKF machine here at home I reverse engineered part of the main exe file of CCam and modded it such that it would produce mill files supported by my old T-Tech machine. The LPKF SW is much better than T-Tech in this respect as it uses much better isolation algorithms. So I was determined to find a way to use this SW with my T-Tech machine at home. I'm still using version 4.0 of CCam so this dates right back to 2001 when I must have bought the T-Tech 7000S machine. However, CCam can't actually control my mill. I just use it to generate isolation files.

This is the same version of CCam we used with the LPKF C30S when it was new. I'm not sure what version we are using today at work but it could still be V4.0.

Because my T-Tech 7000S is an old (ish) model it wasn't supported by the T-Tech/Isopro milling SW and therefore I started out having to use some pretty hideous DOS SW from T-Tech to drive it. So around the same time I hacked about with CCam I reverse engineered the (then) current Isopro SW and worked out how to modify the exe file to add milling support for my older machine. Without this the SW would just crash every time it tried to control the motor because it sent commands my controller didn't support.

I've been using these modified versions ever since and they have been flawless. Note: There are known bugs in the old versions of isopro SW (especially wrt the rubout algorithms) but I don't use this part of the SW. There are control bugs too but I know how to workaround them and avoid problems.

[mrpackethead]

It looks like you were let down by poorly trained reps. Considering what these things cost new I'm surprised they didn't supply you with a known working PC to resolve the issue. But even then, it may fail again if the PC OS gets updated.

In deed. Poor reps, but I was also direclty working with teh engineers in Germany. They coudnt resolve it.  Two differnet machiens, built exactly to their specs, o/s versions and patches.   Nothing else running on the machine.

[free_electron]

this was not a rep... this was their main application engineer. he got flown in specifically for 3 days to come and set the machine up properly ... zilch.

they always had an excuse. it was the pc's serial port. it was this, it was unstable power supply ( we ran off a hulking big UPS ).
they swapped the electronics box twice and shipped new eproms once. it once even ground its spindle to pieces.

like i said : these machines are like the cheapo 3d printers : you spend more time futzing with the machine than producing things with it. it may work for some small proto's when babysat by someone.

at the price you can get boards in china these days ... for 40K$ i can get a LOT of boards from iteadstudio or PCBway at layerstacks and resolutions these machines can not handle. they have outlived there usability . the world has moved on.

[nctnico]

this was not a rep... this was their main application engineer. he got flown in specifically for 3 days to come and set the machine up properly ... zilch.

they always had an excuse. it was the pc's serial port. it was this, it was unstable power supply ( we ran off a hulking big UPS ).
they swapped the electronics box twice and shipped new eproms once. it once even ground its spindle to pieces.

Perhaps they didn't had it tested for EMC compliance. Such tests can reveal all kinds of hidden features of circuits causing all kinds of mayhem. If every machine would behave like you described it they would have been out of business years ago.

[free_electron]

could very well be. anyway. after spending 40K$ and havving nothing but trouble : i am not touching those machines EVER again. there is no convincing me that this stuff works right. it may work for little things but it is too much of a hassle. machinery needs to be fire and forget. the idea is to gain time , not have to babysit it ( same reason i don't like those hobby 3d printers. )

[rx8pilot]

$40k is no joke. The story coming out is tough to hear, I feel lucky that I did not get so far.

The crazy thing is that they are a tough justification even if they worked perfectly and were very easy. It does not take much trouble to blow the idea out of the water.

[IconicPCB]

"... the idea is to gain time , not have to babysit it..."

Quite so. That is why quick turn prototypes are required. Keep high cost crew ( engineers ) productive in development and low cost crew ( support technicians ) in support manufacturing prototype PCBs.

[tonyarkles]

This may be me whinging a little bit, but the most amazing thing in the world for me would being able to get prototype boards at a reasonable price at about the same speed Digikey can get parts to my house.

I mean, having to work out a schedule for various projects (schematic for project A, layout for project A, order boards for A, schematic B, layout B, order boards B, receive boards A, test boards A, schematic C...) isn't the worst thing in the world, the context switch is a bit painful.

But... cost-wise, I don't yet have enough projects going through to justify the cost (both training and equipment-wise). The boards from seeed come fast enough and are cheap enough.

As a curiosity, has anyone tried the Othermill for doing prototype boards?