Clamping of printed circuit boards while milling tracks follow up

Previously on Woody’s Workshop … I had spent time trying to get a consistently level PCB blank clamped to the tooling table ready to mill the traces using CNC.   The results were not too bad but being anally fussy it left a bit to be desired, particularly if the board had a large area which magnified the variations.  

Any variation of the level of the PCB top surface will produce variable width cuts when using a V shaped cutter.   I had machined a clamping plate which was a simple open frame with a clamping step equal to the PCB thickness (1.6mm) and a sheet of MDF or hardboard as a sacrificial backing board. 

Despite having more clamping screw holes than a magazine burst from an AK47 I still ended with the corners of the PCB being a few thou lower.   Results were shall we say ‘variable’.   I had reason to run a new prototype board this week and once again hit the same frustration.   In the end it was a sit and look at it and have a think session. 

The resulting revelation was maybe the sharp edges on the step are applying too much pressure ?  What if I were to be more gentle with the clamping ?  

I cut some strips of thin foam rubber and put this into the step such as to push down on the PCB.   As a quick test I only fastened the frame down using the four corner holes. 

printed circuit board clamping for cnc milling
Sketches showing the two methods of clamping the printed circuit board ready for CNC milling

Absolute magic.   The PCB surface hardly moved the Hamer needle at any point on the surface.  Milling result was an artwork to be proud of.

Issues – the current step on the clamping frame is meant to clamp to a hard stop based on the sum of the PCB thickness and the sacrificial material thickness.  Adding the foam meant I had to do away with the sacrificial board.  The frame step therefore needs to be deeper.   The sacrificial material is essential to allow drilling to take place without breaking the drill as it runs into and potentially damages the tooling table. (For the board in question I drilled to only 1mm and then over drilled by hand off line to the mill).  

So a worthwhile bit of experimenting and hopefully a better result going forward  

Similar or related subjects : –

A Mini Vacuum Clamping Table for PCB Engraving

You know only too well how I keep on going on about FlatCam and milling printed circuit boards on the Tormach PCNC440.

You will also have read about my preoccupation with trying to hold the PCB material flat to avoid variations in milling depth.

I have got it to a reasonably repeatable process using mechanical clamping but you know when a perfectionist starts something it has to be as good as possible …. step forward the Vacuum Clamping Table.

The thinking for this followed on from the Rosebud Grate experiments on my live steam locomotive.   The grate consisted of a matrix of larger holes on the underside of the grate leading to a small bore hole on the top side of the grate.   The theory as I understand it was that the reduction in size creates a Venturi type effect and boosts the air stream into the fire.   I wondered therefore if I reversed the air flow i.e. sucked the air from the large hole into the small hole whether this would be beneficial in providing a boost of the suction.   It is a bit tenuous I must admit and I can’t point to lots of science to back this up, but certainly worth a play.

First stop was Fusion 360 and a two part plate was designed.   This consisted of a top and bottom part.   The bottom part is 15mm cast aluminium with a milled trough and the top plate is 10mm cast aluminium with 6.8mm holes (no science – this is tapping size for M8 that was already in a Tormach collet) on the top side that reduce down to 1.3mm holes (ditto also already in a collet) as breakthrough holes on the bottom surface.   Around the edges are M6 screw holes to clamp the two plates together and also M8 mounting holes to fasten the plate to the tooling plate on the Tormach. I didn’t quite think the suction connection fully.   After I had worked out the total area of the 1.3mm holes I realised that to accommodate this I needed a 16mm diameter hole for the air inlet.  This was not going to be possible to mount on the 25mm overall edge of the plate.   The solution was to 3D print a connecting pipe and mount this on the top surface.   This adapts to the vacuum cleaner pipe being used as the suction source.    The 3D printed adapter did not provide a good seal to the top plate so I had to fit a rubber gasket on it.  The parts were all put together as shown below.

Finished vacuum plate on test in the bench vice
Close up view of the 6.8mm blind holes leading to 1.3mm through holes

To my amazement it seems to work !

There does not seem to be leakage on the joint between the two plates and the vacuum pipe adapter with the rubber gasket seems to seal alright.   If I put a large piece of PCB material over all the holes it is very difficult to move it.  Single sided board is naturally bowed in the manufacturing lamination process and I can see it visibly jump flat when I turn on the vacuum.  If the PCB is smaller than the total area of suction holes it does not seem to matter about covering over the ‘non-used’ holes to maintain the grip.

Proof will be when I try to run a board.   

The milling process will not have major sideways pressure as the depth of milling is quite small so it should be fine. Clearly I can’t go drilling the component mounting holes in the PCB material with this holding technique but I can spot drill them to say 1mm depth and then finish them by hand having got a guide hole to start me off.

But all this will have to wait as the X axis limit switch has come apart on the Tormach and a spare has been ordered and is on its way.

UPDATE Feb 2021 – Flatcam and milling pcbs 2021 pdf download

Similar or related subjects : –

Stretching FlatCam and PCB milling on Tormach PCNC440

Tormach M Code Expansion Card

I have got a project underway to use the Tormach USB M Code expansion board in association with an A axis rotary table.   Details of this will follow in due course.   The expansion card when added to a Tormach mill allows the operator to embed M Codes in their CNC program which will operate up to 4 dry contact SPCO relays or accept 4 inputs as handshaking acks.

My Cabling Masterplan

As part of this project I need to have cables from the USB expansion board to various devices and in a rush of blood to the head decided to use standard readily available Micro USB cables for this. Now the cable connectors are pretty small and the PCB mating socket is even smaller with its 5 connections.   Glibly overlooking this I asked a colleague to produce a PCB layout for the connector to a breakout connector strip.   Dave duly produced a layout and a scratching of head resulted.   How was I going to produce the PCB and how was I going to solder to the connections assuming I could see them ….

FlatCam to the Rescue

Elsewhere in my blog there is mention of the use of FlatCam to create a CNC GCode listing from PCB Gerber and Excellon files.   This program works really well and many successful PCBs have been produced but I have never attempted to mill such fine PCB tracks. A number of problems needed to be addressed to make this successful.   The PCB sheet needed to be held very flat on the PCNC440 tooling table and the correct milling tool with its associated feeds and speeds needed to be chosen. In the past I have used strips of aluminium to fasten the PCB blank down on the tooling table.   This is never perfect and leads to variations in the pressure around the edges of the board.   With single sided PCB there is a natural curvature of the board material as a result of the surface tension of the laminating process.  A single sided blank has a concave surface on the copper side.  I needed to create something more repeatable.

Milling Window Restrictions

Before I bought the Tormach PCNC440 I had a discussion with John Saunders at NYC CNC and he recommended going for the biggest machine I could fit in my workshop.   I could have squeezed the 770 in at a push but I would have had to sell off my Myford VMB which I was reluctant to do.   My order therefore went through as a 440.   With hindsight this decision has been justified on two counts.   I rarely need a larger working area than the 440 offers and the VMB gets used very regularly for quick jobs that don’t justify CNC.   This project was an exception. I wanted to make a frame that would clamp the PCB blank down onto the tooling table.   In order to get the maximum working area for the PCB blank the clamping frame would have to sit outside the machining area.   How was I going to manufacture it ? Fortunately my tooling plate was designed to have a mix of M8 clamping holes and 3.7mm tooling holes and I was going to use this to advantage.   The clamping frame would be symmetrical.   By adding some matching tooling holes in the frame I could cut just over half of the frame and then flip it round 180 degrees and cut the second half. Here is a picture of the CAD showing half of the machining on what will be the underside of the plate when in use. The outer holes are for the M8 clamping to the table and the four smaller holes are the tooling holes.   Being tight with my materials I did not want to just mill out the centre of the plate and have a mountain of swarf (chips).  Instead I designed it with two slots as shown,  one for the clamping surface and one that almost cut through the stock.   The partial cut was to ensure the central piece did not flip out once cut free and damage my cutter. First one half was drilled and cut and then the plate was rotated 180 degrees and the second half cut.   This left the central island just held in place by less than 0.5mm of material.   This was easy to hand cut through to liberate the central area.   The plate was then turned over and the cut edge cleaned using the same tooling position and doing the same 180 degree rotation. To my surprise the rotation process on the tooling pins worked very well with only a minor step transition at the overlap point on all cuts.  This was probably more down to my 3.7mm tooling pins being not quite concentrically turned from 4mm silver steel. With this finished I now had a much more robust clamp for the PCB material.   I had made the clamping step 4mm deep so I could put sacrificial backing boards behind the PCB being run.   This would allow drilling through as needed.   Checking the flatness of a clamped PCB blank with my Haimer showed variation of a few thou in the top surface of the PCB Z position. The worst case variation in Z was at dead centre where the PCB’s natural bow was most dominant.

Tooling and Feeds and Speeds

The next problem was the milling tool and feeds and speeds.   I experimented with various V shaped routers but was not happy with the results.   The 5 thou tip on a 10 degree V tool was incredibly fragile.   Also because the tool was V shaped, any residual bow on the PCB surface lead to a variable width cut.   In the end I opted for Think & Tinkers 15 degree, 2 flute tapered stub (P/N EM2E8-0051-15VC).  This has a 5.1 thou cutting tip which is parallel for the first section so depth variations have no impact on the width of cut.  I ran the program at 10,000 RPM (PCNC 440 maximum) and at 150mm per minute feed rate. The PCB does not look particularly beautiful after milling as there are burrs and shavings present but a gentle rub over with a fine wet and dry removes this and leaves a remarkably clean cut tracking.   The images below show some of the results.   The fine tracking for the USB connector connections is shown on the microscope with a scale for reference.   This shows the five fingers occupying 120 thou with fairly similar track to gap widths of around 15 thou. So now I just have to solder the connectors in place …. I will let you know how it goes.

Overview shot of the clamping plate in position on the my tooling plate on the PCNC440
Finished clamping plate in position on the Tormach PCNC440 holding down a 6″ square piece of single sided PCB.
Tracking on the USB micro connector mounting
Zoom shot on the USB connector tracks with the graticule giving an idea of scale (small divisions are 0.5mm)

Similar or related subjects : –

Milling Circuit Boards Update

I have made some good progress on taking a PCB design Gerber copper and Excellon drilling files into CNC.  I think it is worthy of a full write up but while that gets put together here are some comments.

First of all the conversion process using FlatCAM is very straightforward and I like the fact that you can default save your GCode startup and end routines along with other default settings.  Note that I had to scale the drilling data by a factor of 10.  Apparently this is not unusual.

The fun starts once you have code ready to run on the CNC.  The board design I was working on was a single sided copper design.  Single sided board tends to always have a curvature with the copper on the inside of the curve and the fibre glass outside (if you see what I mean).  This is probably the manufacturing process with the copper and its adhesive ‘pulling’ the board.

Double sided PCB tends not to be so bad in this respect and the effect is balanced out by the two coatings.   My board was therefore much more bowed than a double sided one.  (Incidentally FlatCAM allows for double sided board designs).

If you think about the geometry of what is going on it is critical to make sure the PCB material is flat on the milling table.   The greater the included angle of the milling cutter tip the worse things get if there are variations in surface height.  A height variation equates to a widening of the tool cut.  See the image below. (Not to scale).

I initially used 6mm MDF as my sacrificial backing board to protect my tooling table.   When I checked the MDF for flatness with my Haimer I was disappointed with the result.  Increasing the MDF to 12mm made a huge difference and good enough for the purpose.   This could have been a different manufactured MDF so the change of size is not definitive.

Initially I clamped the PCB to the MDF with a number of woodscrews around the periphery.   On checking with the Haimer this was not good with visible variations that I could impact by pressing on the PCB surface.

Next step was to replace the woodscrews with strips of 10mm square aluminium with a 1.5mm step on one edge.  These were screwed to the MDF on all 4 sides of the PCB blank and this dramatically improved the flatness to a point were it was adequate.  Pressing the board surface did not change the Haimer readings.

Flatness having been solved I addressed the cutter problem.   I had ordered some 10 degree included angle cutters from China but while they were in transit I got to talking with Think & Tinker in the US.   They were incredibly helpful and suggested that I should consider a 60 degree included angle cutter with a 5 thou tip.   They also suggested I try their lubrication to improve the cut quality and to also help protect the tool from wear.   Their tools also come with a fixed collar which means you can change out the cutter without having to reset your Z zero.

This 60 degree cutter worked a treat and the results were startlingly good.   I did not use the lubrication from T&T but instead used my normal FogBuster fluid (QualiChem ExtremeCut 250C) on a gentle repetitive puff.  This seemed to work and kept the dust damped as well as improving the cut.

While I could run the spindle at up to 10,000 RPM, I kept it down at 6,000RPM with a cutting speed of 3″ per minute (75mm).  I had a Z clearance of 0.1″ and depth of cut of 0.005″.  (Sorry for the mixed dimension standards but PCBs tend to be designed in Imperial but I prefer to work in Metric).

After the milling of the copper was complete I drilled all holes at 0.6mm (24 thou) using a carbide drill sourced from Drill Services of Horley (UK).  This was simply a change of tool, registering the tool length and loading the drilling GCode produced by FlatCAM.   The drilled holes were spot on dead centre in the copper lands.

In closing I would like to say how impressed I have been with the Tormach.   I had milled the copper one day and switched off for the night.  Next day I switched on the mill and absolute referenced XYZ and put the drilling tool in the spindle and hit go.  The holes were smack on dead centre in the lands without having to tweak anything.

It has been an interesting challenge that my friend had set me and he has gone away with a good looking PCB and my knowledge base has improved which is what it is all about.

        

A more detailed write up to follow.

Similar or related subjects : –

FlatCAM Update and experimental copper cutting

I’m quite pleased with progress converting GSpark PCB gerber files to CNC with FlatCAM.   The conversion process is straightforward and the resulting GCode looks OK.

I have tried cutting copper using the cutters I have to hand but quickly realised I need to get some much finer ones in carbide.  My tests with modified dental burrs does not work or at least not for very long before the burr goes blunt and the cut width degrades.

I have ordered some 10 degree included angle 0.1mm wide cutters but they won’t be here until February.

The picture shows two runs.   The left hand run used 30 degree cone shaped burr and the right hand run used a modified teardrop burr.   The initial cut on both was where the stars are.

The left hand cut left severe copper burrs which were easily removed using a scalpel blade flat to the surface of the board.

The right hand side was clean of all burrs but gradually degraded in quality as the cutter became blunt.

The teardrop burr was ground to half diameter in an attempt simulate a more normal engraving cutter profile.  I must have drawn the temper in the grinding process.

More experimentation needed once the better cutters arrive.

Similar or related subjects : –

Verified by ExactMetrics