Author Topic: runout question  (Read 4450 times)

0 Members and 1 Guest are viewing this topic.

Offline vivi-dTopic starter

  • Regular Contributor
  • *
  • Posts: 54
  • Country: us
runout question
« on: January 08, 2025, 07:46:18 pm »
This machine specifies its max runout is 0.05mm or about 2 mils. Could this mill 8 mil traces reliably?

https://www.nymolabs.com/products/nymolabs-6040-cnc-router-machine
No solder before coffee! Unless it's 0201...
 

Offline Smokey

  • Super Contributor
  • ***
  • Posts: 3216
  • Country: us
  • Not An Expert
Re: runout question
« Reply #1 on: January 08, 2025, 09:11:30 pm »
That machine doesn't look very rigid. 
Why not just get same week actual professional PCBs from JLCPCB/pcbway/etc? For $1800 you can afford the 2Day turn :)
 
The following users thanked this post: thm_w

Offline vivi-dTopic starter

  • Regular Contributor
  • *
  • Posts: 54
  • Country: us
Re: runout question
« Reply #2 on: January 08, 2025, 09:20:39 pm »
I could but that is not as fun. How can you tell it is not ridged?
No solder before coffee! Unless it's 0201...
 

Offline Smokey

  • Super Contributor
  • ***
  • Posts: 3216
  • Country: us
  • Not An Expert
Re: runout question
« Reply #3 on: January 08, 2025, 09:32:12 pm »
I could but that is not as fun. How can you tell it is not ridged?

Because it doesn't look like this :)
https://www.instagram.com/cnctechnikmueller/p/DDFDh_Jq4c-/?img_index=1
 

Offline thm_w

  • Super Contributor
  • ***
  • Posts: 7862
  • Country: ca
  • Non-expert
Re: runout question
« Reply #4 on: January 08, 2025, 09:50:56 pm »
I could but that is not as fun. How can you tell it is not ridged?

It will go from "fun" to tedious and annoying by the first or second board. Don't waste your money.
Profile -> Modify profile -> Look and Layout ->  Don't show users' signatures
 
The following users thanked this post: Smokey

Offline Gyro

  • Super Contributor
  • ***
  • Posts: 10313
  • Country: gb
Re: runout question
« Reply #5 on: January 09, 2025, 02:16:39 pm »
I just noticed online that the January issue of Circuit Cellar has an article on the design flow from KiCAD through Copper CAM and Candle to a similar one of those CNC machines.

I'm not commenting on the practicality, just for reference.
Best Regards, Chris
 
The following users thanked this post: vivi-d

Offline Doctorandus_P

  • Super Contributor
  • ***
  • Posts: 4029
  • Country: nl
Re: runout question
« Reply #6 on: January 09, 2025, 05:55:21 pm »
It's not a Mueller, but those are also in another price class.
For a completely built machine with MGN rails, the price is about right. Foxalien has also made some similar machines (look into the "upgrade kit" for the rails), I also like the Ratrig Stronghold Pro. and there are quite a lot of similar machines.

They're not the most rigid, but they are adequate for light work. If I was not planning to DIY a machine in the Mueller rigidity class, I already would have built or bought such a machine. That's also the reason I collected some info about building a machine from wood in your other thread:

https://www.eevblog.com/forum/mechanical-engineering/wood-cnc-machines/

I'd say avoid all the machines with the unsupported round shafts, even though they may still (sort of) work for PCB stuff, and other very light engraving. (A 0.5mm PCB routing bit has very light forces)

I also agree with the others here. If it's just for making PCB's, then you're probably better off with just ordering PCB's online, but if your interest in CNC is wider, then get (or build) a machine like this, and you can also use it for routing PCB's.

For PCB isolation routing single lip cutters are often used. As a result, runout is still a bit annoying, but it's not a disaster. By rotating the single lip cutter a bit in the toolholder, the effective result will be a wider or narrower cutting diameter. With very thin router bits, the clearance angle may become too small or negative and then it will not cut well.

(Wood) routers like that usually have an "acceptable" runout. They won't put a very low number in the specifications, because they can't really guarantee it, but it's quite likely that it's "good enough" for DIY. Overall stiffness is a much bigger concern.
« Last Edit: January 09, 2025, 06:11:43 pm by Doctorandus_P »
 
The following users thanked this post: vivi-d

Offline PlainName

  • Super Contributor
  • ***
  • Posts: 7634
  • Country: va
Re: runout question
« Reply #7 on: January 09, 2025, 08:26:35 pm »
I have a similar machine (but which moves the bed rather than the gantry for Y axis - photo is some random one off the web). I've used it (and still use it) for single-sided prototypes and think 8mil should be doable but you'd need a lot of practice to achieve it reliably. I second the other members who suggest JLCPCB for serious or non-prototype jobs.

However, there is a certain amount of satisfaction in DIY PCBs that come out close to expectations, and just as some people use wood chisels and lathes to turn doorknobs instead of just running down the shop and buying a proper one for peanuts, if you want to make your own PCBs then don't listen to the killjoys who will never know the fun of doing so - you don't need a reason or excuse to do it, and they should be ashamed at always trying to shame others into not doing it.

One aspect often missed is that the mill can be used for other stuff too. I just recently cut some LED filter panels from 2mm red perspex (let's see JLCPCB supply that!) and, luckily, I didn't waste loadsadosh sending the drawings off somewhere to be done because I got a hole positioned slightly wrong. No problem - design fixed and another panel in my hands within 15 mins. If you don't have one of these mills around you'll never find a job for it to do, but if it's there you can do many useful things with it.

Re PCB specifically, there are lots of blogs explaining how to cope with warped boards and stuff, specialised height adjusters, etc. This, which I've also mentioned elsewhere on EEVBlog, works for me (and I've tried a lot of things) but YMMV:

1. Get a half-inch (12mm) perspex slab and bolt it to the bed with recessed bolts. Then fly cut it so it is perfectly flat and at right angles to the Z axis. This is both your flat base and sacrificial base board (which you don't need now). Although you don't want to, it's OK to drill too far through the board and into this, OK to mill an outline a bit deep, etc. Photo shows a section of mine - it's perfectly OK and doesn't need surfacing yet.

2. Use 2" (50mm) carpet tape to stick the PCBs down. Carpet tape is very thin, double-sided and, more importantly, consistent thickness. That will keep them flat and in place without the need for clamps or any sensors or anything. Unless your PCB copper thickness is really terribly inconsistent, you'll get a consistent depth of cut.

3. Use 60 or 90 degree cutters from LPKF. Seems counter-intuitive since the slightest change of depth is amplified in width, but it works fine if you've got the PCB stuck down flat. LPKF are expensive but you can sometimes find them on Ebay, and treated right they last a long time (see 4). The photo shows LPKF on the left, typical 'CNC PCB' cutter on the right. Forget the right one or anything like it.

4. Smear some oil over the PCB (I use 3in1 but only because that's what's to hand - anything of similar consistency would be OK). You don't want to be cutting dry copper or FR4 because your bit will last approx no time at all. With oil you'll be wondering why they haven't worn out yet. A bonus is that the oil will trap the debris into nice piles alongside the tracks, so it's easy to clean up and doesn't go down your nose (or require extractor fans). Doesn't need to be deep, but keep it wet (might need to refresh as the job goes along). You shouldn't need to worry about oil being sprayed everywhere because the centrifugal force at the cutting tip will be quite low (see 5).

5. Use a high speed motor. You need to cut the copper within a certain range of speeds (surface feet per minute), which is the rate at which the cutting edge bites into the copper. That speed is a function of the tool diameter and rotational speed, so with a really tiny tip it amounts to not very much at all for normal drill speed and 20krpm isn't very high for this. The speed at which the cutter is being pushed along the track determines how wide the strip of copper being cut is. You don't want it too wide, so slow rotational speed means slow axis movements too.
 
The following users thanked this post: vivi-d

Offline jpanhalt

  • Super Contributor
  • ***
  • Posts: 4199
  • Country: us
Re: runout question
« Reply #8 on: January 09, 2025, 09:06:55 pm »
You're are making a PCB, not ball bearings.  +/- 0.001" (0.002" total runout) shouldn't matter on the traces, if it is really that accurate over a few inches.

EDIT:  Runout applies to the spindle itself, not to the accuracy of the whole gadget.  As others have pointed out, the lack of rigidity is a concern, but it is still probably OK for the sort of PCB's you might otherwise make with toner transfer.
« Last Edit: January 09, 2025, 09:20:45 pm by jpanhalt »
 
The following users thanked this post: vivi-d

Offline PlainName

  • Super Contributor
  • ***
  • Posts: 7634
  • Country: va
Re: runout question
« Reply #9 on: January 09, 2025, 09:54:23 pm »
Quote
This machine specifies its max runout is 0.05mm or about 2 mils

Runout is a function of the chuck (which is the business end of this motor). Worst case, you can replace it with something better. I got mine without a motor and put a Proxxon job on there. Not as fast as that one but it is very well made and runout is very good.

What you should be more worried about is backlash, which is the play in the axis. 'Normal' milling operations can compensate for that to some extent by always approaching a cut from the same direction on an axis. However, typical track cutting just follows where the track goes in a single operation, so if there is any backlash at all it will show up and affect the final track width.
 
The following users thanked this post: vivi-d

Offline Smokey

  • Super Contributor
  • ***
  • Posts: 3216
  • Country: us
  • Not An Expert
Re: runout question
« Reply #10 on: January 10, 2025, 01:26:08 am »
This is sort of like 3D printing....  The question you need to ask is.. Is the thing you actually want PCBs or a new PCB making hobby. 
If you want PCBs = JLCPCB (or equivalent).
If you want a new PCB making hobby = Any PCB mill, but especially cheap desktop routers.
 
The following users thanked this post: thm_w

Offline vivi-dTopic starter

  • Regular Contributor
  • *
  • Posts: 54
  • Country: us
Re: runout question
« Reply #11 on: January 10, 2025, 01:55:55 pm »
How about both!
No solder before coffee! Unless it's 0201...
 

Offline Stringwinder

  • Contributor
  • Posts: 34
  • Country: se
Re: runout question
« Reply #12 on: January 10, 2025, 03:29:16 pm »
The max size of PCB's for milling is important and also what components will be used,
hole mounted or SMD (and the corresponding track widths). I only make tiny PCB's so
a small bench-sized setup for PCB, wood and plastics is my choise. A larger handcranked
mill is also available in my garden shed.

To make a long story short I use a Proxxon KT70 coordinate table and a Proxxon TBM220 bench drill
with bells, whistles and motors added.

The TBM220 has the axle replaced with DIY one having a good quality ER11 fixed collet chuck
attached by heat shrinking so no grub screws or anything to ruin the runout.
The tapered inside of the chuck has a runout of a few micrometers but the big
problem is finding collets of the same quality. ER11 standard collets are 15 µm
and precision 5 µm runout. The dirt cheap ones from Chinesia are worthless in my opinion.
Of two complete sets no collets match the 15 µm and some are really bad with faulty
machining and probably were "rescued" from the reject bin to end up on Ali-whatever.
Looking for a swedish supplier of single unit precision collets.

Make some tests with challenging patterns first and iron out the imperfections before
starting on the "real" PCB. Cutting overlapping circles shows lots of things in one go.
The mill in this case is 0,35 mm with a point angle of 130 degrees, also DIY.

If 50 µm runout is acceptable depends on what you want to machine and if other factors
make it even worse like bad collets, sloppy or inaccurate fixing of the chuck to the axle.
 
The following users thanked this post: vivi-d

Offline vivi-dTopic starter

  • Regular Contributor
  • *
  • Posts: 54
  • Country: us
Re: runout question
« Reply #13 on: January 10, 2025, 04:14:52 pm »
Wow very cool! The Proxxon KT70 looks very nice. Could be used for many DIY projects.
 
Besides PCBs, I've always wanted to make a tiny keychain vice grips for fun. I'm sure with CNC something close to that could be made. Maybe a set of lockpicks? I'm pretty noobish now with machining, but there are lots of possibilities!
No solder before coffee! Unless it's 0201...
 

Offline abeyer

  • Frequent Contributor
  • **
  • Posts: 492
  • Country: us
Re: runout question
« Reply #14 on: January 10, 2025, 10:14:44 pm »
If you're looking at proxxon, there's also the MF 70 which has the table and er collet spindle mill all prebuilt. Picked one up new in box off ebay very cheap recently.
 

Offline Stringwinder

  • Contributor
  • Posts: 34
  • Country: se
Re: runout question
« Reply #15 on: January 11, 2025, 02:43:10 pm »
The standard version of MF70 uses MICROMOT collets that are less accurate than ER11 collets.
Maybe there are "add ons" that convert to ER11 but the threaded part of the axle is still
there and is not the best precision. MICROMOT collets also come in a limited range of sizes.

Using wood as a design material for small CNC mills is doable. I use 16 mm MDF to make a
a cube-shaped box with top and front missing. Stabilty depends on both materials used and
mechanical design. Find out what forces are involved and add support in those directions.
MDF (and HDF) is not always stronger than plywood but less flexible (brittle) and does not
encourage vibrations as much. Try testing it the same way as luthiers test violin plates by
tapping a plate and see (hear) what happens. Internal friction is another name for this
propery, low in plywood and high in MDF.
 

Offline abeyer

  • Frequent Contributor
  • **
  • Posts: 492
  • Country: us
Re: runout question
« Reply #16 on: January 12, 2025, 04:40:03 am »
The standard version of MF70 uses MICROMOT collets that are less accurate than ER11 collets.
Maybe there are "add ons" that convert to ER11 but the threaded part of the axle is still
there and is not the best precision. MICROMOT collets also come in a limited range of sizes.

Ah, that's right, I'd noticed that when I unpacked it and did some test cuts, but had forgotten... some of their bigger machines are ER, but yeah, that one is their own collet. I suspect still might be easier than converting the drill press, though depends on your needs.
 

Offline jwet

  • Frequent Contributor
  • **
  • Posts: 725
  • Country: us
Re: runout question
« Reply #17 on: February 12, 2025, 03:22:39 am »
One thing that I haven't seen mentioned in this thread is a used LPKF machine.  It would be a bit of a project but you'd get something really workable.  All of the C and S series machines are kind of orphanware and sell cheap on Ebay.  The mechanicals of these things is really solid, the table is a big piece of cast tooling plate, the gantry is cast, riding on linear guide bearings- smooth.  The head is solenoid operated- you don't want or need a z axis lead screws- SLOW.  They have a reliable  leveling system to cut at consitent depth.  All the lead screws are precision ball screws- there is zero backlash.    The brushless spindles turn at 50K rpm with really good runout- they use a lever operated collet.  The cutters are one lengh and the drills and routers (thru) are a bit longer- will mill tracks and wide areas with larger tools, drill holes and route with 1 or 2 mm router at nice speeds.  They can also be used for aluminum front panel engraving.
One tip- if the motor has a black heat sink on it, its a brushed motor- limited life- stay away- silver cap is BLDC but verify.  LPKF isn't supporting these older machines- inquiries get a "buy a new $25k machine- these are old" basically.

Gut the internals, put a couple of stepper drivers in, some kind of controller, many options and a BLDC driver for the spindle and GRBL or whatever you like.  There were enough of these sold in 90's, that this could almost be a business.  AND you could sell the razor blades too- the tool bits.  Might be a nice business.

I have an aging LPKF C30S- I bought it new in 95 or so and its really been very good.  The basic software will run in W10 compatibility mode but soon, I'll have to maintain an old machine for it.   Its nice to have around for instant gratification and adapter boards, etc.  It can comfortable  12/12 rules.  I think over the years I've pushed it to 8 mills.  The S series will do 6 mils.

I am a hobby machinist and I played with a 3018 machine with a teenage nephew, they're one step up from toys.  They have no rigidity and no mass, threaded rod lead screws= backlash.  What they are good for is to learn G code and the design flow or machining but they won't cut anything beyond plastic or wood really.  1/8 aluminum even is not possible- just doodle bugs around with depth of cut of 5 thou even.   Don't believe any specs or promises.  To a mechanically inclined person, it seems like they should work better than they do but they don't.

 

Offline benseno

  • Regular Contributor
  • *
  • Posts: 75
  • Country: tr
Re: runout question
« Reply #18 on: February 12, 2025, 05:26:46 am »
Note that spindle runout is specified at the edge of the collet not on the tip of milling bit where it gets bigger. And the longer the tip hangs from the collet edge the bigger runout.
Certainly 0.05mm runout is quite high for such spindles. I have couple of them. Worst one has 0.035mm.
imo it can cause problem when using 0.1mm diameter bits as the bit itself is of eccentric type and its cutting edge will be shifted.

It can be possible to solve the spindle runout if its source mostly located on the badly machined collet sitting by grinding the sit on the machine itself using small grinder attached to the table - it is done on machining center which has stiff metal frame. Not sure if that router is stiff enough to do this.

Yet, the machine for pcb milling should implement pcb surface landscape height measurement to correct the height of the bit during milling.
 


Share me

Digg  Facebook  SlashDot  Delicious  Technorati  Twitter  Google  Yahoo
Smf