New Machine Day: Another F-ing Speedio

So this is all familiar from all the other Speedio threads, the big silver box, the cool space ship bag, yadda yadda.

Andy (2of3) put up with literally 2 years of bullshit questions from me. He and Yamazen have been as top-notch as everyone says. Ngon was my installer and he was great!

Some notes:

- This machine is absurdly well built. Literally every fastener was paint marked. The fit and finish is flawless. Hose routing, layout, wiring, etc are all just extremely well done.

- Yamazen is now distributing Metrol probes as part of their tooling division. I was a bit hesitant, but it's all backed by Yamazen and the pricing brings fitting a Speedio with probing to Haas WIPS level. So far I've been very impressed with it (the Metrol is absurdly well built). Yamazen supplies their own macro package that is as fully functioned as the ones from Renishaw or Blum. We'll see how it works out in the long term, but my initial impressions, and after probing a bunch of stuff for a week, I'm quite happy with Metrol.

- After seeing Tonyda36 continually praise them, I went with the Sankyo RCC170 for the 4th axis. It has the same rotational speed as the competing non-DDR tables, but uses a unique roller drive mechanism with zero backlash and absolutely no need to lock/unlock. I considered a DDR, but the way I want to run this is with a quick-change short tombstone full of parts that can come in/out easily... DDR tables *suck* for rigidity. The Sankyo gives up almost nothing in speed for one of the most rigid 4th axis drives made. It isn't slow:

@lumalabs on Instagram: “Sankyo very much! Not as fast as a DDR table (75RPM), but roller lock design means no .7sec clamp and .7sec unclamp, making the Sankyo…”


- I put two Lang 96mm Quick Point systems in the machine; one on the right side of the table, the other on the Sankyo. I considered 5th Axis, but they would have had to build a custom unit for the Sankyo (Lang simply bored an existing rotary face plate, as is their standard for this kinda situation). What put my decision between the two over the top is that Lang 20mm studs are $75, 5th Axis ones are almost double the price. One issue is that Lang plates are ground, but not concentric to the grid pattern. I tried cooking up an alignment dingus of my own, but could only get the rotary concentric within 0.0005". I'm new to 4th axis work, so I want to eliminate any fudge variables as possible... I think I can get it dialed in better than that, but I don't really know. The Lang stuff is absurdly well thought out and beautifully built.

- Eric at Orange customized a 6x17.5" vise for me with the standard 96mm grid pattern so I could plug it into the Lang on the table. My first machine was a Robodrill sold to me by Milacron, with one of Eric's very first 6x16" vises. I had updated it with some Gen II CarveSmart jaws, but Eric's latest product is just a world away from where he was just a couple of years ago. The modularity and flexibility of Orange vises is readily noticed from the marketing material, but with the Speedio, I've got a bunch of other highest-end kit in the door (Lang, Schunk, Sandvik). Orange is putting out product that is very apparently on the same quality level as these guys. For example:

@lumalabs on Instagram: “@orangeviseofficial 17.5x6” vise that Eric customized with a 96mm grid on the bottom. It’s sitting on a @langtechnik raster plate that I…”

- Finally, the Speedio itself is just silly gods damn fast. I poured over every Speedio video for years, and the first time I got to see one running absolutely balls-out has been my own machine and it's simply fucking amazing. It's machine that's gone to plad.

@lumalabs on Instagram: “My god, it’s full of stars here...”

- Unlike Robodrills, you can hit the ATC button and have the tool in the spindle get placed in the turret, and manually spin that turret around. Or basically take the tool out of the spindle directly and take it out without ever making it go for a turret spin. This is super nice!

- I don't know if this is how the rest of the world works, but G53 Z0 on this machine is the table. "Home" for the machine is really Z18.89 (or something that probably is way cleaner in metric).

- The control is way nicer than the old Fanuc. Only issue I have right now is that I can't call a custom macro as a G-code from the MDI (which the manual says it should be able to, Yamazen has yet to get back to me on this). C-00 is obviously not some great paradigm shift in controller human interfaces (lots of controls are slicker), but a better description for it would be if a Fanuc control was made by actual humans. They are very similar, but everything about the C-00 is just a lot more logically laid out and makes sense.

- Finally, the brass tacks. I expected a few weeks of downtime as I moved out the Robodrill, moved in the Speedio, re-did all my workholding and machining strategies, etc. We had an unanticipated rush on one product, so I decided to knock 45 of them out on the old fixture, with old strategies, and the same tooling I had on the Robodrill... The fixture is a progressive (multi-op) affair that takes one piece of stock for Op1, makes 2 parts, and takes them through 2 more stations for Op2, and a final drilling station for Op3. 40 operations in total, with 12 tools. One complete, finished set of parts every cycle start, and after a little tweaking across the run of parts:

Robodrill: 8:55
Speedio: 4:58

In every way, the Speedio is a total fucking win.

Over the next couple of weeks, I'll be building a little sub-plate to move the Sankyo closer to the edge of the table and towards the door (not hanging it 5" off the way some folks have!), and relocating the Metrol tool setter behind it and to the Y axis edge.

gkoenig said:

This machine is absurdly well built.
In every way, the Speedio is a total fucking win.

Click to expand...

Amen Brother! No pun intended. Congrats and really nice set up there.

gkoenig said:

Robodrill: 8:55
Speedio: 4:58

In every way, the Speedio is a total fucking win.

Click to expand...

Congrats!!!
My experience as well. Wait until you have a 450 or 650 with the quick table. You'll have all sorts of first world problems.......

Agree on the control, too. Not quite as friendly as the Haas stuff I had, but tons better than the Fanuc on the Robo.
All I have now is Brother on the mill side.

Super nice set up. If I only needed a new machine.....Nice dream anyway.

Metrol has been one of the standards for lathe and mill tool touch off sensors for at least 30 years. If their probe is of similar quality, it should be a win too.

BTW, beautiful parts on your Instagram page.

eaglemike said:

My experience as well. Wait until you have a 450 or 650 with the quick table. You'll have all sorts of first world problems.......

Click to expand...

It took some learning for me to see the real value in a R series machine, but yea...

The nearly $3k in Lang stuff is all about minimizing spindle downtime. 5x5" cube tombstone on the rotary, and/or a 10x12" pallet on the table, both in and out of the machine in 15 seconds each thanks to the Lang. With an R machine, you could both do away with all that and get the spindle time to something absurd like 99%. Rotary on the A side for Op1 hitting all the things, fixture plate on the B side for finishing it all off.

Quick change pallets are a wonderful hack, but you start getting into a lot of complex fuckery when trying to bring that concept to a 4th axis (every part we make has side work to some extent, so a 4th is a must-have). I did a whole lot of homework behind this, and it is looking like this will be extremely productive for us while also balancing the need for to do a lot of prototype iterations rapidly (at Speedio speeds and with our business model, we're about 30% production, 70% prototyping).

I have a bunch of neat stuff on the drawing board that is enabled with this thing, and a product line currently successful enough to justify it's purchase (barely). We'll see how it plays out, but my industry is like a lot of others in the consumer product space - any idea with traction gets ripped off by Hung Lo Dong Shenzhen Export Co in a nano second and is on Amazon a tick later at 1/8th the price. Try to fight that dynamic directly is a losing proposition. We need to focus on what we can do that they can't: make cool shit, get it on the internet, and repeat that cycle before Hung Lo Dong's OODA loop can even start. If I was betting real money on this (and I am!), the future is probably 5 axis...

I firmly believe you will be happy with that Sankyo 4th axis. I got about 40,000,000 indexes out of my first pair of them before anything went wrong (motor went out). You should be set for a while.

Congrats!

Thanks for the detail on the workholding setup. I am looking at either a Pierson pallet system or Orange vise and pallet system for my 700s. If I go Pierson I would mount some Kurts we already have to pallet plates so we could swap them out quickly. I'm not afraid of losing Z-height since there is only 12" of travel and spindle to table is 20", but I want a good solid return to zero system that will switch out quickly.

gkoenig said:

So this is all familiar from all the other Speedio threads, the big silver box, the cool space ship bag, yadda yadda.

Andy (2of3) put up with literally 2 years of bullshit questions from me. He and Yamazen have been as top-notch as everyone says. Ngon was my installer and he was great!

Some notes:

- This machine is absurdly well built. Literally every fastener was paint marked. The fit and finish is flawless. Hose routing, layout, wiring, etc are all just extremely well done.

- Yamazen is now distributing Metrol probes as part of their tooling division. I was a bit hesitant, but it's all backed by Yamazen and the pricing brings fitting a Speedio with probing to Haas WIPS level. So far I've been very impressed with it (the Metrol is absurdly well built). Yamazen supplies their own macro package that is as fully functioned as the ones from Renishaw or Blum. We'll see how it works out in the long term, but my initial impressions, and after probing a bunch of stuff for a week, I'm quite happy with Metrol.

- After seeing Tonyda36 continually praise them, I went with the Sankyo RCC170 for the 4th axis. It has the same rotational speed as the competing non-DDR tables, but uses a unique roller drive mechanism with zero backlash and absolutely no need to lock/unlock. I considered a DDR, but the way I want to run this is with a quick-change short tombstone full of parts that can come in/out easily... DDR tables *suck* for rigidity. The Sankyo gives up almost nothing in speed for one of the most rigid 4th axis drives made. It isn't slow:

@lumalabs on Instagram: “Sankyo very much! Not as fast as a DDR table (75RPM), but roller lock design means no .7sec clamp and .7sec unclamp, making the Sankyo…”


- I put two Lang 96mm Quick Point systems in the machine; one on the right side of the table, the other on the Sankyo. I considered 5th Axis, but they would have had to build a custom unit for the Sankyo (Lang simply bored an existing rotary face plate, as is their standard for this kinda situation). What put my decision between the two over the top is that Lang 20mm studs are $75, 5th Axis ones are almost double the price. One issue is that Lang plates are ground, but not concentric to the grid pattern. I tried cooking up an alignment dingus of my own, but could only get the rotary concentric within 0.0005". I'm new to 4th axis work, so I want to eliminate any fudge variables as possible... I think I can get it dialed in better than that, but I don't really know. The Lang stuff is absurdly well thought out and beautifully built.

- Eric at Orange customized a 6x17.5" vise for me with the standard 96mm grid pattern so I could plug it into the Lang on the table. My first machine was a Robodrill sold to me by Milacron, with one of Eric's very first 6x16" vises. I had updated it with some Gen II CarveSmart jaws, but Eric's latest product is just a world away from where he was just a couple of years ago. The modularity and flexibility of Orange vises is readily noticed from the marketing material, but with the Speedio, I've got a bunch of other highest-end kit in the door (Lang, Schunk, Sandvik). Orange is putting out product that is very apparently on the same quality level as these guys. For example:

@lumalabs on Instagram: “@orangeviseofficial 17.5x6” vise that Eric customized with a 96mm grid on the bottom. It’s sitting on a @langtechnik raster plate that I…”

- Finally, the Speedio itself is just silly gods damn fast. I poured over every Speedio video for years, and the first time I got to see one running absolutely balls-out has been my own machine and it's simply fucking amazing. It's machine that's gone to plad.

@lumalabs on Instagram: “My god, it’s full of stars here...”

- Unlike Robodrills, you can hit the ATC button and have the tool in the spindle get placed in the turret, and manually spin that turret around. Or basically take the tool out of the spindle directly and take it out without ever making it go for a turret spin. This is super nice!

- I don't know if this is how the rest of the world works, but G53 Z0 on this machine is the table. "Home" for the machine is really Z18.89 (or something that probably is way cleaner in metric).

- The control is way nicer than the old Fanuc. Only issue I have right now is that I can't call a custom macro as a G-code from the MDI (which the manual says it should be able to, Yamazen has yet to get back to me on this). C-00 is obviously not some great paradigm shift in controller human interfaces (lots of controls are slicker), but a better description for it would be if a Fanuc control was made by actual humans. They are very similar, but everything about the C-00 is just a lot more logically laid out and makes sense.

- Finally, the brass tacks. I expected a few weeks of downtime as I moved out the Robodrill, moved in the Speedio, re-did all my workholding and machining strategies, etc. We had an unanticipated rush on one product, so I decided to knock 45 of them out on the old fixture, with old strategies, and the same tooling I had on the Robodrill... The fixture is a progressive (multi-op) affair that takes one piece of stock for Op1, makes 2 parts, and takes them through 2 more stations for Op2, and a final drilling station for Op3. 40 operations in total, with 12 tools. One complete, finished set of parts every cycle start, and after a little tweaking across the run of parts:

Robodrill: 8:55
Speedio: 4:58

In every way, the Speedio is a total fucking win.

Over the next couple of weeks, I'll be building a little sub-plate to move the Sankyo closer to the edge of the table and towards the door (not hanging it 5" off the way some folks have!), and relocating the Metrol tool setter behind it and to the Y axis edge.

Click to expand...

Looks like a first class setup! Lang makes great stuff.....and it is the only German company that I know of that is customer focused and reasonably priced considering what you get.

We have a couple of Speedios here. They work great but the only thing is they are slow in profiling. We do some contouring on G10 knife handles and our Okuma MB46 is way faster.

Have fun! Let me know if you have any questions as you get farther into it.

The air compressor ... I hope yours lasts longer than mine did. My Speedio consumes a surprising amount of air. The 2 HP California Air that I was hoping would be enough was, technically, but it ran almost continuously and died. I finally got a screw compressor and life is good .

Regards.

Mike

I love new Brother machine day threads!

If you reduced almost in half your cycle time from a Robodrill, I'm wondering what I will reduce from a MiniMill, LOL!

Very interesting set-up, and beautiful parts in your Instagram. I really enjoy your posts.

As Mike said, I will consider another compressor. I recently upgraded to a screw style and it was the best investment I ever made (But only until my new S700 shows up!).

Enjoy it and have fun!

Great setup. Nice looking indexer.

As far as the machine using lots of air it is probably the air purge for the spindle. On my machines I turn it way down. The machine tool builders dont care how loud it is and how much it makes your air compresser work, they just want to protect the spindle from getting coolant inside. But sometimes I think they are a little extreme. Anyway I wouldn't recommend doing it if you want to keep your warantee valid. I know my Quaser had a special tamper proof seal on the regulator ( that I cut off).

All I can say is thank you.

G-Auto said:

We have a couple of Speedios here. They work great but the only thing is they are slow in profiling. We do some contouring on G10 knife handles and our Okuma MB46 is way faster.

Click to expand...

Yea, I saw you make a similar comment on another thread and was very curious...

Brother has a demo video of a 27k Speedio feeding 3D paths quite beautifully, up to ~300IPM:

Speedio High Speed - YouTube

I will say, one thing that has hit me by surprise is that the manual is a bit bullshitty on the high accuracy modes. They have this pretty chart with an X form that shows how each mode balances face accuracy with shape accuracy. So I turned on Mode B's "high speed" setting (M285) for some 2D adaptive toolpath feeding at 300IPM. The test cycle went from 14 seconds to 20!

All the M28X settings are just pre-configured, you can go into Data Bank / User Parameters / High Accuracy to fiddle with them. Got to the parameters page and guess what?

Every Mode B setting was configured 100% identically!

I think the only exception was M280 which had Smoothing turned to 3 (out of 0-5), the rest were set at 1.

I reconfigured M285 to turn arc and radius deceleration way the hell down. It's a funky system. Setting to 0 turns it off. 10 is maximum deceleration, 999 is minimum. It's almost like the Brother C-00 development team let a Fanuc engineer sneak in to set that up. Rejiggering these settings brought my M285 test cycle down to 17 seconds. I think some more tuning might get us an even better trade off.

For roughing, I now see why Drinkwater often just lets the machine rip without any HA stuff turned on. If you're clipping, just increase stock to leave a bit to go fast. Second solution is to turn on HA and fiddle with it to balance accuracy versus speed.

I also just turn off HA for roughing and get after it, leaving .03". Then, with M280 on, take a high-speed pre-finish contour pass with the roughing tool leaving .01" to level out the HSM loops, then finish.

Regards.

I've been using a vari-drive screw compressor (3 phase) for a few years now. IMHO it's the way to go.

Brother Frank showed me a few setups of the pallet machines with the 4th and 5th on there, so It's pretty easily done.

A few people in the US have sent some of my parts to China, and copied them, some even copied here in the States. Fortunately, the quality isn't as good.....

Enjoy your new machine! I hope it makes life a LOT easier for you!
all the best,
Mike

gkoenig said:

Yea, I saw you make a similar comment on another thread and was very curious...

Brother has a demo video of a 27k Speedio feeding 3D paths quite beautifully, up to ~300IPM:

Speedio High Speed - YouTube

I will say, one thing that has hit me by surprise is that the manual is a bit bullshitty on the high accuracy modes. They have this pretty chart with an X form that shows how each mode balances face accuracy with shape accuracy. So I turned on Mode B's "high speed" setting (M285) for some 2D adaptive toolpath feeding at 300IPM. The test cycle went from 14 seconds to 20!

All the M28X settings are just pre-configured, you can go into Data Bank / User Parameters / High Accuracy to fiddle with them. Got to the parameters page and guess what?

Every Mode B setting was configured 100% identically!

I think the only exception was M280 which had Smoothing turned to 3 (out of 0-5), the rest were set at 1.

I reconfigured M285 to turn arc and radius deceleration way the hell down. It's a funky system. Setting to 0 turns it off. 10 is maximum deceleration, 999 is minimum. It's almost like the Brother C-00 development team let a Fanuc engineer sneak in to set that up. Rejiggering these settings brought my M285 test cycle down to 17 seconds. I think some more tuning might get us an even better trade off.

For roughing, I now see why Drinkwater often just lets the machine rip without any HA stuff turned on. If you're clipping, just increase stock to leave a bit to go fast. Second solution is to turn on HA and fiddle with it to balance accuracy versus speed.

Click to expand...

Ours is config'd that way too! The default settings are absolute trash. And I don't know what I'm doing to mess with them. I'm used to the Hi-Cut Pro on the Okuma which is leaps and bounds ahead of the HA stuff.
Even with HA off, and leaving .05" stk I was still clipping corners feeding at 460ipm at times around the outside of a 4x4" block of alum.
So, running a semi finish pass and then a finish pass totally defeats the purpose of High Accuracy.

May I ask what your settings are? Can you take a pic of them?
Also, they claim a max feedrate of 1181ipm, but ours was locked to 787ipm.
AND if you turn on HA mode, the max feedrate is LOWER, go figure. I found it went down to 500ipm max feedrate, and that's no good when doing air cuts in an HSM path.
Overall we're not too happy with the accuracy of the Brother at the moment. Hopefully once we get someone in here who knows the HA stuff that will change.

Mtndew said:

So, running a semi finish pass and then a finish pass totally defeats the purpose of High Accuracy.

Click to expand...

No it doesn't. You are only zipping around the final contour (+.01") one time around, at 288 or 384 or whatever IPM you want, with HA on. That only takes a few seconds, and the HA only takes substantive time away at the corners which for most prismatic parts are few. Granted, if you are doing a complex organic shape, there are a zillion "corners," and then the HA tradeoff gets more complicated.

One other important reason (for me at least) to take a pre-finish pass is to level the little loops that HSM (aka Dynamic Mill aka Volumill etc.) leaves on the walls, especially inside pockets -- I have seen these "shine through" to the finished surface. You can't measure them after the finish pass, but you can see them, since the finish tool is taking a variable depth cut as it clips the little loops.

Regards.

Mike

Finegrain said:

No it doesn't.

Click to expand...

Sure it does. You're adding another cut to the cycle.
The same program has zero issues on our Okuma leaving .005" stk at 460ipm.
1 finish pass.
On these parts, the only time saved is from the pallet changer and tool changer.
That's it.
I even had to lessen the d.o.c on one of the toolpaths because the spindle couldn't handle it, that only added 10 seconds, but on 1,000+ parts multiple times a year, we want these running as fast as possible.

Mtndew said:

Sure it does. You're adding another cut to the cycle.
The same program has zero issues on our Okuma leaving .005" stk at 460ipm.
1 finish pass.
On these parts, the only time saved is from the pallet changer and tool changer.
That's it.
I even had to lessen the d.o.c on one of the toolpaths because the spindle couldn't handle it, that only added 10 seconds, but on 1,000+ parts multiple times a year, we want these running as fast as possible.

Click to expand...

I’m going to be doing some testing in the next couple of days to find the sweet spot of settings for Mode B roughing. Unless your Okuma has SuperNURBS, the math both controls are running is quite similar. Mode B should have no trouble keeping up with the Okuma’s feed rate.

The Brother factory clearly could do a better job of pre-configuring these this out of the box to provide a baseline. Again, they come with ZERO configuration, despite what the manual says...

On the other hand- you’re comparing Okuma’s premier heavy duty/high accuracy beast of a VMC against a Speedio where (ignoring the QT table’s efficiency gains), you’re putting a Porsche GT3 up against a McLaren P1. The fact that the Speedio is even in this conversation (over an issue I think could have the gap substantially narrowed or even closed) is a testimant to my good taste in machinery purchase!

gkoenig said:

Robodrill: 8:55
Speedio: 4:58

Click to expand...

New Brother.

How old was the Fanuc?
I wonder what a Brother of the same year would have been cycle wise?

I could compare a new Kia to a '54 Desoto and say how wonderfull it was too...


----------------

Think Snow Eh!
Ox