Fixed Gantry Router with HSK-E25 ATC for under $25K

[ishi]

The THK website is actually quite helpful and provides pricing upon registration.

After some research, the best I could find is this:

https://www.thkstore.com/products/li...k_preload=7123

With a height of 21mm, it's really low profile. 54mm x 59mm. And with 12.8kN of static loading, it should be plenty enough for everything I want to do. I am planning to use these for all axes.

Unfortunately, I could not find any block with more than light pre-loading on their website. Any idea why?

[ishi]

I spent some time with the THK folks, and here is what they recommended:

https://www.thkstore.com/products/li...k_preload=7123

HRW27CR1SSC1, 27mm x 62mm x 72.8mm.

The HRW model is more suitable than the SHW one that I had selected earlier, because it is full ball type instead of caged ball, making it better with respect to contamination. Additional seals can be added as well (still trying to figure out whether this is needed or not).

We considered using 3 blocks in 21mm instead of 2 blocks in 27mm, but the support engineer indicated that a pair of 27mm blocks will provide more rigidity than a trio of 21mm ones. They're not exactly cheap ($2,000 per axis essentially), but they're in line with the rest of the build.

Next, I'll review the ball screw assemblies.

[ishi]

Here is a model of a linear guide block. It's actually for an HRW-CA model with flanges, 27mm x 80mm x 72.8mm.

https://www.thkstore.com/products/li...k_preload=7123

This will allow blocks to be mounted from the top or from the bottom. I suspect that all of them will be top-mounted, but I want to be safe just in case.

[ishi]

Since we are going to use THK linear guide blocks, we might as well go for THK ball screws. These will be the precision ball screw non backlash (BIF):

https://www.thkstore.com/products/fe...s/bif-set.html

[ishi]

I quickly browsed through the 500+ pages of THK's ball screw catalog, and I am starting to get a better picture for the whole thing. I still have a long way to go, but so far, it seems that I should go for finished end shafts, using the non-backlash BNK line:

https://www.thkstore.com/products/fe.../bnk-set3.html

To mount the ballscrews, I'll use BF20, BK20, and MC2010 parts.

Moving to these parts, I have to increase pitch from 5mm to 10mm, which my servos can handle. As a result, rapids will be twice at fast, clocking a respectable 1,181 ipm. And of the 1.1kW servos for X and Y are not sufficient, I can upgrade to 1.9kW models without changing anything else in the design.

With such parts, every axis will cost about $3,500, including ball screws, linear blocks, rails, and mounting accessories. This is clearly a lot of money, but it seems in line with the frame. In fact, when I break down the whole BOM, we get something like this:

- Base and table: 25%
- Linear motion components: 20%
- Controller, drivers, and motors: 30%
- Spindle and VFD: 20%
- Others: 5%

Somehow, that seems about right.

Note: The THK catalog and website are awesome! So much better than anything else out there...

[ger21]

When you look at the market for professional wood CNC routers, a growing chunk of the market is being addressed by kits.

That depends on how you define "professional". The vast majority of woodworking businesses that want to add a CNC, will buy a ready to run machine. They don't have time (or desire) to build a kit.
And when you consider performance level, the "kits" are a step or two below what I'd consider an entry level commercial machine. I've been running industrial CNC routers for over 20 years, and imo, industrial quality routers start at about $80k-100K.

[ishi]

STATUS CHECK

So far, I am pretty happy with our BOM:

https://docs.google.com/spreadsheets...SwA/edit#gid=0

The last items that need work are the VFD and the proximity switches.

After that, I'll build a detailed parts list for controller, driver, and motors, just in case we decide to go for a custom configuration.

[ishi]

That depends on how you define "professional". The vast majority of woodworking businesses that want to add a CNC, will buy a ready to run machine. They don't have time (or desire) to build a kit.
And when you consider performance level, the "kits" are a step or two below what I'd consider an entry level commercial machine. I've been running industrial CNC routers for over 20 years, and imo, industrial quality routers start at about $80k-100K.

ger21,

I agree with you, in this price range, you're better off buying a commercial machine.

But in the $10k to $25k range, I would argue that you're much better off building it yourself. CNCRouterParts or DamenCNC are living proofs of that fact. Their kits are first rate and are selling really well.

For non-ferrous metal machines, my hypothesis is that the same is true. In the $250k to $1M range, go for a commercial machine. But in the $50k to $150k range, there is a gap. This is where I want to play, because this is where there is a massive untapped market that will open once deep-pocketed tinkerers realize that 3D printing will remain limited to a narrow range of applications for a long, long time, until someone figures out how to combine additive and subtractive processes together in a productive fashion.

I also believe that subtractive manufacturing remains too complex and out of reach for many people who could benefit from it. Things have improved considerably for wood cutting over the past 20 years, but the same did not happen for metal somehow. I think it's about time that someone fixes that. My goal is to become a little part of that solution.

[ishi]

ger21,

In order to better understand what I am trying to achieve, I invite you to take a look at this website:

Discommon

These guys produce really cool low-volume products and manage to sell them with very nice profit margins. Most of them are made out of aluminum, and I am willing to bet that they are using CNC machines that retail for anywhere between $250K to $500K (if not more).

My hypothesis is that one can build a similar machine out of parts worth $50K to $75K in low volume (single prototype), and about $35K to $50K in higher volumes (100 or more). If you can sell a DIY kit like that for $75K and a fully-assembled machine for $100K, you create a whole new market, and there is no way to tell how big this market could be.

Another hypothesis is that if you make the machine small, simple, and open (open as in no enclosure), you make it a lot more suitable for robotic automation. I invite you to watch some videos of factories churning plastic products with hundreds of cheap 3D printers. Now, imagine a factory with hundreds of $50k non-ferrous CNC machines with a cubic meter footprint. With that kind of setup, you can disrupt an entire industry. But for that to happen, you need precision, as in the precision found in a machine that can make an iPhone or MacBook shell.

This is the kind of machine that I want to build.

I know that I don't have the skills to do it on my own, but I also know that the skills needed to build it are to be found within your community, and that is why I am hanging out here.

Again, thank you so very much for your patience and attention.

PS: Please accept my apology for not having explained all this background upfront. It is quite challenging to dump so much information all at once. And I am learning as I go, while following a rather non-conventional path. I trust that you can see the full picture past my naiveté and unbridled enthusiasm.

[ishi]

Here is some clarification regarding the servos. Somehow, I got confused between two models before. The models that are currently included in the retrofit kit that I am planning to use are the Siemens 1FK7060-5AF71, similar to that one:

https://mall.industry.siemens.com/ma...060-5AF71-1PG0

The last 4 characters of the model number are subject to change (with or without 12 bit multi-turn encoder option), with or without feather key and keyway, with or without brake. The motors on all three axes will be the same, with the only exception that the motor on the Z axis will have a brake. All three motors come with the AM20DQI encoder: absolute encoder 20 bits (resolution 1048576, encoder-internal 512 S/R) + 12 bits multi-turn (traversing range 4096 revolutions).

Basic characteristics are as follows:

1.5kW
6Nm
3000rpm
63mm shaft height
7.1kg without brake, 8.5kg with brake

Dimensions in attachment.

[ishi]

I am really struggling with the selection of ball screws. I would like to go with THK, because they have a wide range of options with online prices, but a few things remain unclear. My assumption is that I should go for the BIF set (Precision Ground Ball Screw-Non Backlash):

https://www.thkstore.com/products/fe...s/bif-set.html

And for reference purposes, the 3 axes of my machine are as follows:

# X axis:
Orientation: Wallmount
Stroke: 1000mm
Weight: 125kg
Rapids: 1,200ipm

# Y axis:
Orientation: Horizontal
Stroke: 600mm
Weight: 250kg
Rapids: 1,200ipm

# Z axis:
Orientation: Vertical
Stroke: 300mm
Weight: 75kg
Rapids: 1,200ipm

I went for 32mm shafts because this is what DamenCNC recommends for 1,500W servos:

https://www.damencnc.com/products/me..._314_2329_GB_1

Is that the right way of thinking about it? Most importantly, isn't it oversized for the Z axis? Ideally, I would like to use the same shaft diameter on all ballscrews in order to use the same mounts and brackets for all axes, but I want to make sure that oversizing a ball screw for a particular axis won't lead to a loss of precision.

In order to get 1,200ipm with 3,000rpm servos, I am planning to use screws with 10mm pitch. Would I get more accuracy with 5mm pitch models (in which case the rapids would go down to 600ipm of course)?

For the X axis, it seems that I would have to buy the 1,430mm version. And for the Z axis, the 730mm version. This seems like a massive waste of precision-grounded shaft. Am I missing something?

Also, what kind of service provider should I work with to finish the shafts for using BF20 and BK20 mounts in a supported/fixed configuration? Should I work with THK directly for the finishing, or am I better off working with a local shop?

Finally, where could I find a nut mounting bracket? The only one I could find is for the BNK line:

https://www.thkstore.com/products/fe...screws/mc.html

I am really sorry for all the newbie questions, but no amount of catalog review seems to provide answers to these.

[ger21]

I am planning to use screws with 10mm pitch. Would I get more accuracy with 5mm pitch models

You'd have higher resolution.
Accuracy and resolution are two different things.

Accuracy relies on all the components in the machine, and how they are assembled.
If you have backlash in the screw, it affects accuracy.
If anything flexes, it affects accuracy.
If anything is assembled out of square, it affects accuracy.

[ishi]

ger21,

Thanks a lot for the clarification. I will go for 5mm then. 600ipm rapids are plenty enough.

[ishi]

The two attached pictures show the internal structure of the Hwacheon SIRIUS-UX Vertical Machining Center for Die and Mold:

Milling Center - Hwacheon Corporation

I stumbled upon these pictures while doing some research and found them interesting for the following reasons: the relative proportions of my design are getting very close to the ones of this machine. This is especially true for the following elements:

- Vertical spacing of the two rails on the girder (X axis)
- Length of the X axis in relation to the Y axis
- Wide table (longer alongside the X axis than the Y axis)
- Width of the carriage assembly
- Deep canal on girder for motor and ball screw assembly
- Very beefy head assembly
- Direct coupling of motors to ball screws on all 3 axes
- Single motor and ball screw for Y axis of table
- Three pairs of linear blocks for Z axis (or so it seems)

The second picture seems to be one for an earlier design of the machine. On the new design, the motor for the X axis went from the right to the left, and the motor for the Y axis from the front to the rear. My design does exactly the same as their new design.

What this design tells me is that:

1. We should increase the height of the bridge's pillars from 12" to 14", 16" or even 18" to make room for the moving table and increase stroke on the Z axis from 10" to 12" or 14".
2. The spindle plate should really become a spindle box in order to increase its rigidity.
3. The table must be made much thicker than originally thought (as recommended by pippin88 in an earlier post).

Going a step further, we might want to consider building the carriage plate and spindle box using cast iron instead of steel plate. This would further increase rigidity and would make for a fun learning exercise (working with a third-party of course).

Now, here is something really interesting: the servos for X and Y are rated at 1.6kW, but the one for Z is rated at 3kW. This is quite confusing, because the Z axis servo is moving something that is lighter than the X axis one. Of course, it's doing it vertically instead of horizontally, which means that it's fighting against gravity. I wonder if the designers used a more powerful servo for Z in order to avoid a Z-axis counterbalance. If that is the case, I'd much rather use a bigger servo as well (much simpler).

Bottomline: the high-level design isn't too far off from something that could work. Of course, accuracy will come from having the right components working well together, and that will require a lot more work.

[mactec54]

I am really struggling with the selection of ball screws. I would like to go with THK, because they have a wide range of options with online prices, but a few things remain unclear. My assumption is that I should go for the BIF set (Precision Ground Ball Screw-Non Backlash):

https://www.thkstore.com/products/fe...s/bif-set.html

And for reference purposes, the 3 axes of my machine are as follows:

# X axis:
Orientation: Wallmount
Stroke: 1000mm
Weight: 125kg
Rapids: 1,200ipm

# Y axis:
Orientation: Horizontal
Stroke: 600mm
Weight: 250kg
Rapids: 1,200ipm

# Z axis:
Orientation: Vertical
Stroke: 300mm
Weight: 75kg
Rapids: 1,200ipm

I went for 32mm shafts because this is what DamenCNC recommends for 1,500W servos:

https://www.damencnc.com/products/me..._314_2329_GB_1

Is that the right way of thinking about it? Most importantly, isn't it oversized for the Z axis? Ideally, I would like to use the same shaft diameter on all ballscrews in order to use the same mounts and brackets for all axes, but I want to make sure that oversizing a ball screw for a particular axis won't lead to a loss of precision.

In order to get 1,200ipm with 3,000rpm servos, I am planning to use screws with 10mm pitch. Would I get more accuracy with 5mm pitch models (in which case the rapids would go down to 600ipm of course)?

For the X axis, it seems that I would have to buy the 1,430mm version. And for the Z axis, the 730mm version. This seems like a massive waste of precision-grounded shaft. Am I missing something?

Also, what kind of service provider should I work with to finish the shafts for using BF20 and BK20 mounts in a supported/fixed configuration? Should I work with THK directly for the finishing, or am I better off working with a local shop?

Finally, where could I find a nut mounting bracket? The only one I could find is for the BNK line:

https://www.thkstore.com/products/fe...screws/mc.html

I am really sorry for all the newbie questions, but no amount of catalog review seems to provide answers to these.

If you do build this machine then it would be best to have THK do the complete Ballscrew, having it done from a local shop would cost you more to get the ends ground to your spec's, just use the THK s standard ends, Ballscrew end mounting then there would be no extra cost, and the Ballscrews would be there standard off the shelf, go with 6mm to 8mm pitch, your choice C5 grade 32mm is what most machines of this size use, so very easy to choose the Ballscrew diameter, you only have to figure out how you are mounting them and if you want to use them in tension, which is best, you need very rigid mounting for this

I have to ask this, have you ever built a machine before, you choice of control and servos are not for a first time build, and a bad choice, just remember the higher you make the Gantry above the table the support structure has to be very substantial I don't believe the machines in your snips are a good example of this

[TTalma]

Catching up with a few comments I have. We use THK pre-loaded rails at work, we use from 20mm to 60mm. They are extremely well made, we have tried other suppliers but we have a very small tolerance, and the THK's are always well within our tolerance. It's very rare that we have to send one back, and when we do it's usually due to damage in shipping.

Same with the ball screw We use THK's and they are always good. We use very few that are under 40mm though. It's generally cheaper to use a linear motor on the smaller parts.

For positioning we use both encoders and glass scales, using scales on an axis feeds back actual position, reading encoder pulses is theoretical position (An encoder only reads how much a motor turned and does not take into account play in a system)

You are hoping your design will be built by others, because you see that something is missing. At the price point you are talking about I would guess 99.9% of the people building this would be commercial companies. And they most likely wouldn't do it simply because of OSHA, and the headaches involved. Also every manufacturing companies I've worked at's biggest expense is time. If they need it they need it to make a product to make money. Paying someone to build a machine when they can buy one doesn't make sense (even though the machine they need may have a 6 month lead time).

We make machines that make parts for something the size of a matchbox car to parts used in container ships. But we have holes in our product line, not because we can't make the machines, we would if ordered, but there is simply no market. I think the reason no machine like yours exists is because very few people need the precision your machine will be capable of for products the size yours will make. They can just buy a HAAS, that comes with a warranty, has a proven design and can be running in a few weeks.

And I don't mean to sound rude, but to think that your first machine design will be good enough to be used in a production environment is very hopeful. For a machine this size to be commercially viable you would plan on 5-10 people working for 2-3 years before it's ready for market. For one person to do this in their garage with internet help, you should plan on about 10 years, and at least one complete redesign. If my goal was to make this something others would use I would have to be ready to turn my $10k of granite into a garden planter.

[ishi]

mactec54,

Thank you so much for your confirmation that THK is the way to go. It really feels like the best option at this point, and I'll work with them to get complete ball screw assemblies.

Yes, this is my first build, and yes, SINUMERIK is certainly overkill for a machine of that kind, and certainly a bad choice for a first build in general. Nevertheless, I have some good reasons for going with this platform, which were explained in earlier posts.

[ishi]

Catching up with a few comments I have. We use THK pre-loaded rails at work, we use from 20mm to 60mm. They are extremely well made, we have tried other suppliers but we have a very small tolerance, and the THK's are always well within our tolerance. It's very rare that we have to send one back, and when we do it's usually due to damage in shipping.

Same with the ball screw We use THK's and they are always good. We use very few that are under 40mm though. It's generally cheaper to use a linear motor on the smaller parts.

For positioning we use both encoders and glass scales, using scales on an axis feeds back actual position, reading encoder pulses is theoretical position (An encoder only reads how much a motor turned and does not take into account play in a system)

You are hoping your design will be built by others, because you see that something is missing. At the price point you are talking about I would guess 99.9% of the people building this would be commercial companies. And they most likely wouldn't do it simply because of OSHA, and the headaches involved. Also every manufacturing companies I've worked at's biggest expense is time. If they need it they need it to make a product to make money. Paying someone to build a machine when they can buy one doesn't make sense (even though the machine they need may have a 6 month lead time).

We make machines that make parts for something the size of a matchbox car to parts used in container ships. But we have holes in our product line, not because we can't make the machines, we would if ordered, but there is simply no market. I think the reason no machine like yours exists is because very few people need the precision your machine will be capable of for products the size yours will make. They can just buy a HAAS, that comes with a warranty, has a proven design and can be running in a few weeks.

And I don't mean to sound rude, but to think that your first machine design will be good enough to be used in a production environment is very hopeful. For a machine this size to be commercially viable you would plan on 5-10 people working for 2-3 years before it's ready for market. For one person to do this in their garage with internet help, you should plan on about 10 years, and at least one complete redesign. If my goal was to make this something others would use I would have to be ready to turn my $10k of granite into a garden planter.

TTalma,

Thank you so very much for your super constructive feedback. I would *love* to go for glass scales at some point, but they would add significant complexity and cost, so I'll stick with encoders for now. Much like I would love to go for linear motors and torque motors for my second build, but I want some experience with servos for my first build. I am crazy, but not *that* crazy.

And yes, it might take me 5 or 10 years to actually get something up and running with the level of precision and accuracy that I am looking for. And it might take a full team. And two or three full redesigns. And it certainly won't be built in my garage anymore (I'm currently looking for an industrial place in the area I live in). And yes, I am prepared to lose the investment on the granite base. If nothing works, I'll just return the base to its manufacturer for "recycling", and I'll be grateful for the learning (and hard lessons) that I received.

Your comment on OSHA is a really important one: my machine is initially targeted at individual entrepreneurs who have no employees. As a result, OSHA does not fully apply (as far as I can tell). And this is the primary reason why the price point ($50K to $75K) is so important. It's in the same range as the kind of car that Uber Black drivers buy. The whole model is predicated upon the hypothesis that:

1. Demand for low-volume manufacturing exceeds supply
2. Low-volume manufacturing usually requires more precision
3. Low-volume manufacturing comes in a wide range of sizes
4. Many people would rather make parts than drive a car all day
5. Affordable training is available from local community colleges

With that in mind, things like machine downtime is something that the entrepreneur will have to deal with directly, much like car maintenance for an Uber driver. If you drive your car with Uber and it breaks down, you won't be able to work. You won't go out and rent another car for making your shift. As a result, Uber drivers tend to take good care of their car. In the case of the CNC machine, I believe that taking care of the machine is going to be a lot easier if you assembled it yourself. But the assembly required will be limited to putting together the macro components listed there:

https://docs.google.com/spreadsheets...SwA/edit#gid=0

In fact, it is very likely that linear rails for the X and Y axes would come pre-installed on the granite base. And most of the work that will be put into the design will be done in such a way as to simplify this assembly. For example, the most critical component (as far as I can tell) will be the spindle plate/box. It should be designed in such a way that squaring is simplified. It's also very likely that a professional machine builder will have to help the customer on site for a few days, and the price of the machine will have to take that into account. By the way, the very same is done when someone builds a kit plane.

[ger21]

my machine is initially targeted at individual entrepreneurs who have no employees.

Possibly even more so than large companies, you'll find that these individuals don't have the time to build a machine. Building a complete machine can take hundreds of hours, during which you aren't making any money.
I've seen many members here point out this very issue.

And yes, I am prepared to lose the investment on the granite base. If nothing works, I'll just return the base to its manufacturer for "recycling", and I'll be grateful for the learning (and hard lessons) that I received.

You can learn the same amount with a $10K investment. Possibly even more, as buying cheaper individual components may require a larger learning investment to get everything up and running.

A smaller, much less expensive investment would greatly increase your chances of success with the expensive machine.

[ishi]

ger21,

I agree with you, and I am quite likely to buy this kit just to get started:

Benchtop PRO 2436 2' x 3' CNC Machine Kit | CNCRouterParts

But I'll keep working on the large machine in parallel, because as you (and several others) have pointed out, it will take much longer than I expect to get it done properly.

What I am trying to figure out is how short the build process could be once a few machines have been built and all the tooling and jigs have been developed. What I mean by that is that the kit would come with macro-components pre-assembled:

- Granite base and bridge with rails for X and Y
- Controller enclosure (with all the electricals)
- Ball screw assemblies
- Head assembly (with the spindle)

With that in mind, let's assume that a skilled machine builder shows up on site with all necessary tools to do the final assembly. How long would it take to be done? If it can be done in a matter of days, we have a winning formula.