-
But I'm desperately trying to figure out what the tangible benefits to owning one is. Because clearly there are people who believe there is. But as yet, nobodys provided a 'full cycle' story about how it helps.
are you dead serious? where in the world cubicle are you live in? we have small project, or pcb that doesnt need 6 mils small trace, and we need it quick less than an hour, for prototype or a one-off, or an arduweeno project shield. how many times do people here have to explain? i dont do chemicals, or consumer market laser printer is not printing black enough toner for pcb transfer hence i over etch. drilling by hand using dremel is suck etc etc, hence i need a 3d drilling machine, pcb mill, or etch resist remover is another bonus for the machine. i also can make it to carve my project enclosure, or shape some plastic or wood or even light aluminium. etc..
just like how do you justify a sewing machine to your boss? everybody try to explain so you can knit your shirt easy fix this and that, and you said... i dont knit! you are doing it all wrong you should send it to tailor shop. you dont knit you dont buy sewing machine period, but many other people do knit. many people buy a car but some cant justify it since they work just the next block, same as the market where they buy stuffs etc and they will spend the rest of they entire life in the neighborhood. you should buy printer, no! whats the point if i can send it to the computer shop next door that can print at insanely cheaper price. everybody should have a dog, no! dog shits everywhere. etc etc. its a matter of interest, and as you said... workflow. nobody should be forced to follow your workflow. -
What is this machine useful for. how could i justify it to my boss.
If you can't think of any reasons then I don't think it's for you. I've contemplated this kind of machine even in the day and age of quick turn protos, oshpark, etc.
I'm just a one man shop but even I have those 3am "hm, I'd love to be able to test this out" moments. Making a couple of boards for a kid's project or quick hack/test. Being able to route out a quick board within hours rather than days. I don't even need through hole. I can think of lots of things to use this machine for.
As I mentioned earlier in this thread, I think he's even got the price point right. It's a bit of a tough sell for me, but because I'm a cheapskate, not because his price is wrong. -
What is this machine useful for. how could i justify it to my boss.
If you can't think of any reasons then I don't think it's for you. I've contemplated this kind of machine even in the day and age of quick turn protos, oshpark, etc.
I can think of multiple reasons why there is appeal to this machine. The lure of nearly instant results. the concept that you can have something much faster than you think you coudl previously. There was a time i thought this would be amazing.. Running lots of prototypes is expensive. ( lets exclude materials they actualy are not much more than a rounding error for lots of stuff compared to the cost of time ). When its easy to run prototypes ( like using a mill ) the tendency is that the quality of engineering goes down. The we'll work it out and see what happens approach...
Im convinced that spending more time on actual engineering and produce higher quality work acutally saves time and money overall. Rush jobs never quite hit the quality level, because they are just that.. rush jobs.
-
You are thinking businessmarket but this is consumer market machine.
For the hobbieist it is great to do some experimenting, they did this normally on breadboards and protoboards but with smt that is not easy. Using SMT-> TH adapters might also change the behavior of the circuit for instance with buck converters where you want every part (ic, caps, coil) close together and a firm groundplane. Yeah for >2 layers it is not going to work but there are few hobbieists that make 4 layer board due to cost and complication. So to test and improve a new design before sending it to the pcb fab this could be IMO be a worthwhile investment.
But for your boss and a company, naah unless they are a small time shop they have next day pcb delivery without extra cost contracts as long as they also let the same fab do the production run in the end. -
You're looking at it with a mindset of your own business. There are lots of other types of businesses with different constraints and requirements.What is this machine useful for. how could i justify it to my boss.
If you can't think of any reasons then I don't think it's for you. I've contemplated this kind of machine even in the day and age of quick turn protos, oshpark, etc.
I can think of multiple reasons why there is appeal to this machine. The lure of nearly instant results. the concept that you can have something much faster than you think you coudl previously. There was a time i thought this would be amazing.. Running lots of prototypes is expensive. ( lets exclude materials they actualy are not much more than a rounding error for lots of stuff compared to the cost of time ). When its easy to run prototypes ( like using a mill ) the tendency is that the quality of engineering goes down. The we'll work it out and see what happens approach...
Im convinced that spending more time on actual engineering and produce higher quality work acutally saves time and money overall. Rush jobs never quite hit the quality level, because they are just that.. rush jobs.
People building 1-offs or small volumes on a tight schedule ( due to their customer, not their own bad planning) would benefit from anything that can speed up turnround.
For example I've done a job where there was less than a week from initially being called to shipping a couple of hundred tested boards with firmware, to rescue a time-critical installation for an exhibition.
The ability to verify a PCB before ordering 10 panels on an expensive 24h turnround could cover the cost of a mill in a single job.
The ability to quickly knock up a breakout board to evaluate an oddball part can also be very valuable - this is something I still occasionally use my PCB etch tanks for.
There isn't always something else you can be doing while waiting for PCBs to arrive. '
There is no such thing as "good" or "bad" engineering, only more or less appropriatefor a given situation.
If you can't see the potential benefit of a consistent and reliable way of doing PCBs in house, you're looking a things with too narrow a view.
-
I've said it many time on many threads here but these things are at their best (IMO) for doing RF research work on various exotic laminates. If you are into this kind of business the mill is a fabulous and very powerful tool.
But they aren't for everyone. Probably the worst thing you can do with a mill is make lots of large 2 layer digital boards with lots of 8 or 16 bit data bus lines strewn across the board and loads of vias to stitch/fit. The double whammy worse thing would be to try and do this with a 4 or 5 mil track width/space because you would probably wear out several 5 mil end mills by the time the board is finished. A large and complex digital board with lots of fine traces and a fair bit of rubout could take several hours to mill. The finer tools cost more and also wear out quicker. You get the board the same day but it will be expensive and time consuming to make. Plus the chances of mistakes goes up due to the complexity and you may trash the PCB at any time during the milling process. That's where the skill and experience of the operator really makes a difference. I rarely trash a board during milling but then I take extra time and I know how to set up the milling algorithm to get good results.
The newer machines may be better at all of this with less need for a skilled operator but you will still have the issue of rapid tool wear on the ultra fine end mills. At work we generally use the V or pointed tools as much as possible and try for a non critical 8-10mil trace width as a minimum. We only use finer traces where absolutely necessary. Any critical stuff gets done with an accurate end mill and then the rubout gets done with an end mill near the end of its tool life. By critical I mean traces or shapes that have to be accurate in terms of width or spacing. This would typically apply for filter design or oscillator design where the dimensions have to be right in order for the milled filter to be representative of the actual design itself. We have milled RF filters up beyond 10GHz with excellent results over the years.
But in a typical R&D environment there will be loads of uses for these machines. They are at their best doing small boards that can be milled quickly. So I don't see the small size of the Prometheus' work area to be a problem. Nearly every board I make could be done on Prometheus. My 7000S table area is 19" x 13" and I think I'd rather have the smaller 5000S version of my machine if I could swap. The 7000S is a very big mill and needs lots of bench area to operate in.
-
It's time to finish this thread, already advertised the obsolete machine, very few interests let's go forward,
-
It's time to finish this thread, already advertised the obsolete machine, very few interests let's go forward,
Just click off to another thread if you are bored? I won't miss you
-
What is this machine useful for. how could i justify it to my boss.
here again... https://www.eevblog.com/forum/reviews/prometheus-for-rapid-prototyping-forget-everything-you-know-about-pcb-milling/msg1189948/#msg1189948 and nobody is forcing you to justify it to your boss...
i just realized the OP claim 5 mils tolerance, i believe this is highly optimistic value. we need proof for that. in the video already showing inconsistencies in isolation/mill thickness, that i estimate between 0.5mm - 1mm milling. 20 mils (0.5mm) tolerance maybe believable/achievable. my 0.5mm drill bit may break at an instant.
I'm happy to share pictures of the 5-mil isolation! Here are two (which you can also view on www.zippyrobotics.com):
http://www.zippyrobotics.com/wp-content/uploads/2017/03/IMG_0944.jpg . This entire board was done with the same 3.9-mil end mill.
and
http://www.zippyrobotics.com/wp-content/uploads/2017/03/1.png under microscope.
The gaps are cut to 5-mils, to within the tip diameter tolerance of the bit. The end mill used was 3.9-mil diameter, 15 degrees, fluted. 50,000 RPM and 7 IPM feed rate to make those examples. IMHO, Prometheus could probably support an even lower trace/space but the bits become more expensive and the feed rate must decrease even further.
20 mils is super-easy for Prometheus. 7-mils is routine even with square end mills (as opposed to angled ones). The trick is to use an accurate spindle that has really low runout ("wobble"). Large spindle runout is the major limiting factor that causes small tools to break instantly. Another reason is the runout of the bit itself, meaning how concentric the center of the cutting helix is to the center of the shank, so it is important to use high-quality bits when micro milling also. -
IMHO there is no reason to over-obsess about minimal widths. A board needs to be soldered as well and very small clearances are easy to bridge. When I etch a prototype I use minimal clearances of 0.25mm and preferably much more.
-
Thats quite impressive Rocco, given the price of the machine. nice work.
-
IMHO there is no reason to over-obsess about minimal widths. A board needs to be soldered as well and very small clearances are easy to bridge. When I etch a prototype I use minimal clearances of 0.25mm and preferably much more.
It's about margin - if a system will only just do 10mil/0.25mm, chances are it will be unreliable long-term if you expect it to do dense boards with those design rules.
However if it will comfortably do half that, chances are you will get a good yield from more relaxed design rules, and the capability to go more dense when necessary.
-
IMHO there is no reason to over-obsess about minimal widths. A board needs to be soldered as well and very small clearances are easy to bridge. When I etch a prototype I use minimal clearances of 0.25mm and preferably much more.
I agree with that in general - it's always good to design with larger trace/space if you can do it. Though minimal trace and space can become very important if you select components that have smaller clearances between their pads (or if they only come in smaller packages). For example, many QFPs have a clearance between pads in their recommended footprints of only .2 mm (8 mils) or less, so in those cases if your machine only supports 10-mil spaces you won't be able to do it at all.
For the rest of the board, you could always rub out a larger clearance around the traces using a larger and more durable bit or just remove all of the unwanted copper completely. -
IMHO there is no reason to over-obsess about minimal widths. A board needs to be soldered as well and very small clearances are easy to bridge. When I etch a prototype I use minimal clearances of 0.25mm and preferably much more.
It's about margin - if a system will only just do 10mil/0.25mm, chances are it will be unreliable long-term if you expect it to do dense boards with those design rules.
However if it will comfortably do half that, chances are you will get a good yield from more relaxed design rules, and the capability to go more dense when necessary.
Yup. -
In case of prototypes you'll always have to be creative and tailor the PCB design to the process used to create the prototype. A QFP has a 0.5mm pitch so if you have a space of 0.25 between the pads there is still 0.25mm of pad width. Not a real problem and the bigger spacing will probably improve the solderability of the board as well (*). Don't mix absolute accuracy and minimum clearance here.IMHO there is no reason to over-obsess about minimal widths. A board needs to be soldered as well and very small clearances are easy to bridge. When I etch a prototype I use minimal clearances of 0.25mm and preferably much more.
I agree with that in general - it's always good to design with larger trace/space if you can do it. Though minimal trace and space can become very important if you select components that have smaller clearances between their pads (or if they only come in smaller packages). For example, many QFPs have a clearance between pads in their recommended footprints of only .2 mm (8 mils) or less, so in those cases if your machine only supports 10-mil spaces you won't be able to do it at all.
(*) Not every contract assembler is happy with the manufacturer's PCB footprints! -
Quote
It's about margin - if a system will only just do 10mil/0.25mm, chances are it will be unreliable long-term if you expect it to do dense boards with those design rules. However if it will comfortably do half that, chances are you will get a good yield from more relaxed design rules, and the capability to go more dense when necessary.
Yes, I'd agree with that. I think most/all of the high end machines from T-Tech and LPKF advertise 4 mil trace capability, even on old machines like my 7000S from the mid 1990s. But I think it is worth mentioning that things get very expensive very quickly if an attempt is made to mill using a 4 or 5 mil end mill to get a tiny trace and width across loads of address and data lines on a large and complex CPU design for example. You might end up using three tools to finish the PCB.
The price of the 4 mil end mills (eg $320 for 10) is nearly three times that of a 31mil end mill and about double the cost of a 15 mil end mill from T-Tech and these tiny 4 mil end mills have a much reduced life. Things get much worse on materials like Rogers 4003C or 4350 if you use fine end mills because the wear rate is much higher still. I think the top grade LPKF tools are even more expensive but at my place of work this isn't an issue. We just buy what we need. The bonus for me is that I can salvage the part worn tools from work for free so I rarely buy new tools here at home. I think the LPKF tool timer is very aggressive so I often get tools with loads of life left in them and I get them at a rate that far exceeds my usage at home. The only tools I won't salvage are drills and contour routers. I always buy these new and they are generally quite cheap anyway.
-
In case of prototypes you'll always have to be creative. A QFP has a 0.5mm pitch so if you have a space of 0.25 between the pads there is still 0.25mm of pad width. Not a real problem and the bigger spacing will probably improve the solderability of the board as well (*). Don't mix absolute accuracy and minimum clearance here.IMHO there is no reason to over-obsess about minimal widths. A board needs to be soldered as well and very small clearances are easy to bridge. When I etch a prototype I use minimal clearances of 0.25mm and preferably much more.
I agree with that in general - it's always good to design with larger trace/space if you can do it. Though minimal trace and space can become very important if you select components that have smaller clearances between their pads (or if they only come in smaller packages). For example, many QFPs have a clearance between pads in their recommended footprints of only .2 mm (8 mils) or less, so in those cases if your machine only supports 10-mil spaces you won't be able to do it at all.
(*) Not every contract assembler is happy with the manufacturer's PCB footprints!
Finer trace/space support gives you more freedom. Period.
Also, .4 mm-pitch QFPs exist and have a *recommended* pad width of .3 mm. A .25 mm end mill would destroy ALL of your pads in that case. There'd be nothing left to solder to.
Who's mixing accuracy and minimum clearance? You're saying that the recommended clearance isn't really the minimum. That's fine, it's good to get creative if you have to. I truly applaud it. There are reasons to do use a larger bit and not to obey recommended footprints, like faster milling speed and cheaper bits. But I know you're not asking me to apologize for making a more capable tool that has the ability to do fine trace and space if you need it. It's not like I'm charging more money for it! Right?? It's only an option, there if you need it.
-
The other thing is that it's worth trying to find the optimal spindle motor spin speed and tool travel speed for each type of board material and tool type.
T-Tech gave me this data many years ago although I was scared away by the very high tool travel speeds they were suggesting. A case of my ignorance of what is OK against what seems 'not OK' when you watch the machine going that fast. So I tend to run my machine slower. Also it's worth having a fast/variable motor speed. One limitation of my old machine is that it is stuck at 23k rpm max.
Also, if you want to produce some nice marketing samples then try doing some pretty looking RF filters on Rogers 4003C as it looks very professional once milled. The white substrate against the clean copper looks great especially if the cuts are clean. I often feel a bit sad when soldering parts to the board because it spoils the beauty of a freshly milled board on 4003C
-
Yes, this is true. Small bits can get expensive.QuoteIt's about margin - if a system will only just do 10mil/0.25mm, chances are it will be unreliable long-term if you expect it to do dense boards with those design rules. However if it will comfortably do half that, chances are you will get a good yield from more relaxed design rules, and the capability to go more dense when necessary.
Yes, I'd agree with that. I think most/all of the high end machines from T-Tech and LPKF advertise 4 mil trace capability, even on old machines like my 7000S from the mid 1990s. But I think it is worth mentioning that things get very expensive very quickly if an attempt is made to mill using a 4 or 5 mil end mill to get a tiny trace and width across loads of address and data lines on a large and complex CPU design for example. You might end up using three tools to finish the PCB.QuoteThe price of the 4 mil end mills (eg $320 for 10) is nearly three times that of a 31mil end mill and about double the cost of a 15 mil end mill from T-Tech and these tiny 4 mil end mills have a much reduced life. Things get much worse on materials like Rogers 4003C or 4350 if you use fine end mills because the wear rate is much higher still. I think the top grade LPKF tools are even more expensive but at my place of work this isn't an issue. We just buy what we need. The bonus for me is that I can salvage the part worn tools from work for free so I rarely buy new tools here at home. I think the LPKF tool timer is very aggressive so I often get tools with loads of life left in them and I get them at a rate that far exceeds my usage at home. The only tools I won't salvage are drills and contour routers. I always buy these new and they are generally quite cheap anyway.
Whoa! Our bits won't cost that much. I'm not saying they're cheap, but there's a significant difference. I'm estimating we're going to retail the 3.9-mil for less than 200 USD for 10, certainly no more than 230 USD. The reason I don't have a firm price yet is because I'm waiting on one of my manufacturers to quote me (I work with several). I do have solid pricing for the 5-mil ones. A 10-pack of those will be 185 USD. 31% cheaper than their 6-mil 10-pack. -
Thats quite impressive Rocco, given the price of the machine. nice work.
Thank you! -
The other thing is that it's worth trying to find the optimal spindle motor spin speed and tool travel speed for each type of board material and tool type.
Yes, that info is coded into ProCAM. You just select the bit you're using from the drop-down menu. You don't have to manually punch in any feeds/speeds.
If you're curious, here are the current parameters:
Square profile
7-mil: 20 IPM (inches/minute)
10-mil: 40 IPM
30-mil: 150 IPM
15 deg. fluted profile (these last significantly longer than the above and go much faster given their size).
3.9-mil: 7 IPM
5-mil: 40 IPM
7-mil: 85 IPM (this is what you see in the video)
60 deg. and 90 deg. engraver
13 IPM - slowest but cheapest. 120 USD for 10. Last longest. Bottoms out at 8-mil space though.QuoteT-Tech gave me this data many years ago although I was scared away by the very high tool travel speeds they were suggesting. A case of my ignorance of what is OK against what seems 'not OK' when you watch the machine going that fast. So I tend to run my machine slower. Also it's worth having a fast/variable motor speed. One limitation of my old machine is that it is stuck at 23k rpm max.
Just one tip if I may - and this might not be your problem but make sure you're not going too slow in terms of feed rate, because there comes a point where you're not creating chips with each rotation and instead rubbing (burnishing) the end mill teeth wearing them down faster. This could cause you to break bits at a higher frequency than if you went to a higher feed rate.QuoteAlso, if you want to produce some nice marketing samples then try doing some pretty looking RF filters on Rogers 4003C as it looks very professional once milled. The white substrate against the clean copper looks great especially if the cuts are clean. I often feel a bit sad when soldering parts to the board because it spoils the beauty of a freshly milled board on 4003C
I did this with Rogers 4350 to show a client! It does look pretty. It was all done with the 3.9-mil bit. I wish I could show it, but it's covered under NDA. If you have some small designs you don't mind me milling and posting pictures of on here, you can message me or get in touch through the contact form at www.zippyrobotics.com and I'll do a test. Probably have to wait a week or so because I'm super busy with package testing right now, but I'd get to it. I would need Gerbers or preferably the project file so I can tweak the Gerber output if necessary. -
Who's mixing accuracy and minimum clearance? You're saying that the recommended clearance isn't really the minimum. That's fine, it's good to get creative if you have to. I truly applaud it. There are reasons to do use a larger bit and not to obey recommended footprints, like faster milling speed and cheaper bits. But I know you're not asking me to apologize for making a more capable tool that has the ability to do fine trace and space if you need it. It's not like I'm charging more money for it! Right?? It's only an option, there if you need it.
No need to get upset. A big part of running is business is managing the customer's expectations. Ofcourse you are proud of what you have made (as an engineer with appreciation for mechanical stuff I think you should be) but IMHO you are way too focussed on the narrow clearances which will lead to dissapointed customers because of the high wear on milling bits and slow speed. Customers really don't care that you devoted a lot of time and effort in getting it the way it is especially if the machine is slow and eating tools. This is probably one of the things LPKF has done wrong so don't repeat that mistake. A better approach would be to paint a more balanced picture where potential customers can get a good feel on the tradeoffs between speed, tool wear and minimum clearances. -
Who's mixing accuracy and minimum clearance? You're saying that the recommended clearance isn't really the minimum. That's fine, it's good to get creative if you have to. I truly applaud it. There are reasons to do use a larger bit and not to obey recommended footprints, like faster milling speed and cheaper bits. But I know you're not asking me to apologize for making a more capable tool that has the ability to do fine trace and space if you need it. It's not like I'm charging more money for it! Right?? It's only an option, there if you need it.
No need to get upset. A big part of running is business is managing the customer's expectations. Ofcourse you are proud of what you have made (as an engineer with appreciation for mechanical stuff I think you should be) but IMHO you are way too focussed on the narrow clearances which will lead to dissapointed customers because of the high wear on milling bits and slow speed. Customers really don't care that you devoted a lot of time and effort in getting it the way it is especially if the machine is slow and eating tools. This is probably one of the things LPKF has done wrong so don't repeat that mistake.
Not upset - the exclamation point was just to emphasize my point... which is that it's there if you need it, unlike some other machines. I didn't think I was really focused on the narrow clearances... the pictures were just a response to the poster who was skeptical (and rightfully so) of the 5-mil spaces and the discussion went from there. I just wanted to make clear that it can absolutely do it.QuoteA better approach would be to paint a more balanced picture where potential customers can get a good feel on the tradeoffs between speed, tool wear and minimum clearances.
I think you make a good point here about familiarizing potential users with the machine and what to expect. Thank you for that. The way I see it, it will take time to educate the market. What I had in mind was to post a series of videos as often as I can (maybe once a week) to show Prometheus making different boards. Now that you mentioned the expectation part of it, I'll keep it in mind to show the slower bits and things to give people a full picture. A full picture is something that will just take time to convey though, I think.
-
Customers really don't care that you devoted a lot of time and effort in getting it the way it is especially if the machine is slow and eating tools. This is probably one of the things LPKF has done wrong so don't repeat that mistake. A better approach would be to paint a more balanced picture where potential customers can get a good feel on the tradeoffs between speed, tool wear and minimum clearances.
Ok, now that you made me think of this, another idea is that when I do these videos I can keep making copies of the board being demoed until the bit breaks this way I can assign a bit depreciation cost to each of the boards I demo. That will probably give people the best sense of what to expect if they see a variety of boards.
-
Thats quite impressive Rocco, given the price of the machine. nice work.
yup +1.I'm happy to share pictures of the 5-mil isolation! Here are two (which you can also view on www.zippyrobotics.com):
impressive compared to my $200 machine + the blunt mill (scratching) tip
. you have very rigid and low runout machine there, well worth the price imho. but well i rarely needs that thin trace i only expect 1.27mm pitch soic and will limit myself to that since smaller pitch i will have trouble hand soldering even on the masked board. smaller pitch board should be solder pasted and baked for proper result anyway, more $$$ for another machine. precise machine like this the user will have a peace of mind and confidence in doing slightly larger tolerance. you've proven that.