[HawkJET]

datac,

This is good information about FlashCut. My first impression (from what I saw - I didn't actually try anything) was not positive. Years ago, I was at a trade show and their booth was demonstrating several Sureline tabletop 'toys' and I assumed they were just that. I see their web site also has round column mill/drills and this only reinforces my previous impression. Plus, the packaging of their electronics appears 'hobbyish' to me (I can see this may not be fair since Kflop doesn't include an enclosure at all - 'hobbyish' or not).

Since my first impression, I have not had reason to look any closer. Perhaps having a way to test things as a 'demo' such as MACH3 and EMC2 may have given them a chance to be considered, but I can see how that is difficult in their situation.

Based on your comments about your experience, my first impression seems inaccurate. So it is good to hear your experience.

One of the main attractions for me with MACH3 is the ability to make custom screen sets. I have made several custom features that are non-standard and have really streamlined my setup and use of MACH3. Are FlashCut screens customizable?

My next upgrade is likely to be Kflop since I can use it in conjunction with my custom MACH3 screens.

[RCaffin]

Gee, it seems so easy to stir things up, doesn't it? Some of the replies sound a bit defensive though.

USB: I am willing to believe that some USB-based systems work. There will be errors due to noise, but SMART software can correct for erros. The FTDI chip does not have error-correcting smart SW, and this may be much of the problem. Fair enough.

> Mach3 and LinuxCNC in my opinion, are hobbyware.
Correct. That's where they started. But commercial users are starting to use them to make a living.
Have you ever noticed that so many things which start out as 'hobby-ware' end up displacing the 'professional' stuff in due course. I could cite the PC overtaking the mini-computer as an obvious example. Pity they never rewrote the core of Windows though - a real pity. But they are rewriting Mach4. And how many cars have a throttle retard lever these days?

> what do you expect from software that seemingly does everything. 3,4,5,6 axis and be great at doing them all?
Quite a lot, actually. And it delivers - for a reason.
With the older style of controllers (ie big boxes with lots of PCBs and hundred of buttons), adding a synchronised 4th axis means adding a lot of electronics, and it works out expensive. To add a 4th axis to a small HAAS seems to cost about $1.3k.
But when it is done in SW on a modern PC, synchronising 6 axes is just the same as sync'ing 3 axes. Yes, literally. Adding a 4th axis to a Mach machine costs ... well, zero $ (apart from the power amp or stepper driver). OK, it actually costs me about $1 worth of wire for the hook-up. Literally - I have just done it.

> Mach3 has an overly crowded UI, if all you have is a 3 axis mill.
Possibly. Perhaps.
But unlike the enormously complex front panels you get on dinosaur controllers, the screen on Mach is infinitely tunable. If there's stuff you don't need, you can remove it. If there are things you need that aren't there, you can add them. Useful. If you can't be bothered, no sweat. I will add that you can get a whole range of alternate screens, many of them free.

> you can't beat software that was specifically written for the intended purpose or machine
Oh, I agree. But then, both Mach and EMC2 and ... were 'specifically written for the intended purpose' of controlling a CNC.

Hi Chris
> Gee, I SEE pricing shown... They show ranges for each different class.
OK, so where are they? URL please! I would love to see them.
Contrary to what you might think, I am NOT wedded to any one SW package, although I would certainly prefer that it be able to talk to MY hardware via MY Ethernet SS. The box is built, after all.

Cheers
Roger

[LeeWay]

You guys are destroying my serenity. I like Mach 3.

[SCzEngrgGroup]

There will be errors due to noise, but SMART software can correct for erros. The FTDI chip does not have error-correcting smart SW, and this may be much of the problem.

You should read up on how USB actually works.... USB, when used in "bulk" mode as any machine controller will use it, guarantees correct data delivery at the hardware level. Data is CRC'd, and, if it arrives with any corruption, will be automatically re-sent. So, in pracice, the data either arrives completely intact, or not at all. About the only way it will not arrive at all is if the connection is lost. You will never see minor bit errors.

Regards,
Ray L.

[alan_3301]

> Mach3 and LinuxCNC in my opinion, are hobbyware.
Correct. That's where they started. But commercial users are starting to use them to make a living.

Not that it matters, but Linuxcnc is based on emc, which was developed by NIST, a division of the US dept. of commerce.
Definitely not hobby origins, but I would agree that most current users are probably in the hobby class.
I started using it because it was free, which I would think attracts most hobby users.
It has a nice advantage to companies that want to develop their own interface, being open source, that they have a solid foundation to customize to their own needs, without starting from scratch.

on a side note, count me in the list of people that tried mach3 but didn't get far due to the million button interface.

[datac]

One of the most important questions that should be asked is "what else have you really used" ?

It is amazing to be reading recent posts from people who finally stepped away from what they had been using to try something different, and low and behold, they are amazed at the differences. It is about SMART software. SMART methods and techniques to get things done without problems. It has been annoying to see all these people beating their heads against the walls trying to solve such simple "problems".

USB is not the problem... Ethernet automation is certainly an old long standing option, but has its own difficulties in a traditionally networked environment. Each has to be designed to work properly.

Bad gui design is a big, BIG part of a control to me, and both Mach and Linux in my opinion have problems in this area. Some of my recent posts at the zone have discussed debounce and start/end point arc tolerances. This, because recently, I cracked open LinuxCNC for the 3rd time, just to see where it is in the stream of development. What I find, is that indeed, it is "Customizable" as ever..... it is capable of doing whatever a person wants to do within its intended purposes..... But at what EXPENSE ?

It took me 3 days of digging in forums and asking questions in order to find out exactly what a person was required to manually add to a configuration text file in order to insert time adjustment to a single input line. WHY ? Because no one to date has felt it necessary to just put it in the/a damn gui and create/code the process for the text file to be written AUTOMATICALLY, and Correctly to the text file when adjusted or selected. THIS is not rocket science. And as an aside, I did have to reply thru the typical "you have noise", and you have "bad wiring methods" recommendations.... even though I dont, and even though I KNOW what my problem is and it was lack of debounce.

LinuxCNC fans can go to their documentation and read what it has to say about debounce...... Perhaps that is a no-brainer for a programmer.... But frankly, it is meaningless for an end user to figure out what you are supposed to do, and where you are supposed to do it. I've been "retrofitting" machines since 1998, and I had no clue.

Note too, that there is nothing written that I found that indicates that if you change such a setting, but use the gui they do have to adjust some other small thing, saving that change will OVERWRITE everything you added manually. Yep.... just brilliant ! I am not bashing linuxcnc at all..... it is incredible.... I know it is. But you have to be part geek and do your own legwork for this FREE program. Little is up front and streamlined to the point an end user can figure things out on their own. At least LCNC is free...... and your are really own your own to get what you want. On the other hand we have Mach, which has put a lot in their gui... and I mean A LOT. Sure, Ole Ger21 made a wonderful interface out of that default disaster, but people... please be fair when you tally costs as you spent money to GET that interface to make Mach tolerable..... I wont even go back into all the bugs of the program itself. I will say that a lot of people who used to use it no longer bother.

Why all this is frustrating is simply because because Flashcut had the simple ability to make such adjustments for eons ! Right there in the config gui, with easy to understand text and an excellent manual to give you pointers and tips.

Now here is another shocker. Over the last 20 years, I've used many different CAM programs. Some absolute garbage compared to todays offerings. Yet, Flashcut ran my files successfully for years without hiccup. But, LinuxCNC immediately kicks up an error claiming start and end points are off by .001". Well, back to the forums.... Well, Huh ! the immediate reactions are typical... "your cam program is bad - change it". The bigger question is WHERE is the ADJUSTMENT for start and end point tolerance ?

SURPRISE !!!! Its is hard coded ! You can not change it unless of course you are a member of the high order of programming geniuses. H'mm. Flashcut has had an adjustable tolerance in the gui for its entire life.

THIS is why Flashcut has no "thousands of forum posters" saying my spindle does not run,.... my machine crashed... my version is buggy.... What really is that worth to people ?

Oh, here is another. Flashcut pre-processes the entire file. makes sure its going to make it to the end.... I find linuxcnc wants to run half way thru an hour long job, then kick up and error that the "move will go outside of the machine envelope". Probably not a big deal for you guys with one machine, but I have a few. That's just plain crazy in this day and age.... What ? I cant Feedhold, jog around and then resume ? Oh boy.... but it is free. You can not argue that.

To be fair, the start/end issue was IMMEDIATELY attended to by forum members, taking it to the LinuxCNC "high priest" (where ever he is) and raising the question as to why it is hard coded and could it be put at least into an .INI file in future releases. You can not complain about the service...... meanwhile though, I couldn't run those older jobs now could I ? I think you will see that as an option in the near future and I did not even have to twist any arms.

Regards Flashcuts interfaces being manipulated. I believe there are some things a geek can do to the version 4 interfaces, and I know there have been provisions to allow other devices/ monitors/ Pc's to see or control certain aspects via API. But, frankly, I've never seen the need to change much of anything they already had because it works so darn well.

Version 5 is all new ground up stuff, and I am sure that they have taken account of this popular demand from end users. All along they have made OEM versions that you probably wouldn't even recognize as their product and this has been a large portion of their business.

Prices for flashcut are right there on their retrofit overview pages like this one:

Tabletop & Minimill / Lathe / Gantry Kits | FlashCut CNC

But people must recognize that this really is NOT a hobbydink control. There are a lot of variables that can be applied. I do believe though that the prices shown are for "turn key" products, and they always have offered just the software and signal generator separate, perhaps with some understandable reservations. Clearly, selling just those items so people can connect to whatever drivers and whatever motors to whatever flimsy system they think is good enough CAN insert a whole lot of SUPPORT demand.

Like I said, only they can determine their particular business direction, price structures and goals. It is in their right, and it is good business sense. But one thing it is not, is that they have a bad control. It is an EXCELLENT control and I can not think of a single typical machine that could not be fitted with it. Years ago, you could by the software and SG for something like $700. I believe that to be around $1250 or so these days but I'm not sure.

While that is a lot more than the $150 people are used to (and few run just that anymore with all the extra hardware they have to buy to get it all connected), it really comes down to what is your time worth to really have a control that is spectacular... completely non-invasive to a PC, smooth and easy to understand ?

To some, it is almost priceless !

Ooops... almost forgot... Roger, I know you want to use your ethernet smoothstepper.... but you see, even with all the "open source" hype that has encompassed the inexpensive control environment, the smooth stepper is not open source by any means. You are tied to continuously hoping that newer software will allow you to use it. You are pretty much tied to what you have, no different than those screaming that Flashcut is "closed source" and limited.

Flashcut works most excellently. They also have kept older users really happy by not throwing them over the cliff by demanding newer and newer hardware over the years. It was an awesome, well thought out development that with Version 5, seems will continue such trend.

I have a Flashcut SG here that is as old as the hills..... It is running the very latest software. What is that worth........

[kvom]

My first CNC mill experience was a school's Haas VF2. I wouldn't say that control's interface is any more intuitive and Mach. It's what you get used to.

[RCaffin]

Hi Ray

You should read up on how USB actually works....

Yeah, I should ... I have a PhD and 40+ years experience in the IT industry, all the way from writing device drivers to debugging compilers to creating database systems. Along the way a few real-time vision systems and robotic systems got delivered as well.

USB, when used in "bulk" mode as any machine controller will use it, guarantees correct data delivery at the hardware level. Data is CRC'd, and, if it arrives with any corruption, will be automatically re-sent. So, in pracice, the data either arrives completely intact, or not at all.

Chuckle. It'll be the day when a bit of hardware can guarrantee clean delivery of data. Until then, some SW is needed, and for some funny reason modern SW writers seem incapable of not creating bugs. Sometimes ... 'automatically' just doesn't happen.

About the only way it will not arrive at all is if the connection is lost. You will never see minor bit errors.

But you can see lost connections, packet corruption, and general stuff-ups. Happens in the field, even if theory says it shouldn't.

Cheers
Roger

[RCaffin]

Hi Chris

re Linux and so on - yeah, I know. Completely customisable, provided you have a C++ compiler. :-)

Yet, Flashcut ran my files successfully for years without hiccup. But, LinuxCNC immediately kicks up an error claiming start and end points are off by .001".

Um, well, should they be off by 0.001"? How come the CAM (or whatever) gave that sort of error? After all, a lot of machines have better accuracy than that these days, and parts specs are often that tight, so that sort of error could be a deal-breaker sometimes.
OK, it might be useful to be able to specfy the threshold - I agree there. It should NOT be hard-coded in!

Prices for flashcut are right there on their retrofit overview pages like this one:

All I could see was
'Prices start at $1560 for Stepper and $4670 for Servo.'
That's not really all that helpful.
I will add, that's a helluva difference for servo tho.

Roger, I know you want to use your ethernet smoothstepper.... but you see, even with all the "open source" hype that has encompassed the inexpensive control environment, the smooth stepper is not open source by any means. You are tied to continuously hoping that newer software will allow you to use it. You are pretty much tied to what you have, no different than those screaming that Flashcut is "closed source" and limited.

Um - but that's not entirely the full story, is it? Yes, the code in the SS itself is proprietary, but the hardware interface to the outside world is simply 2 or 3 LPT ports. That lets me connect a wide range of 3rd party interfaces. And that keeps the prices down.
On the other hand, imagining you could have a simple open interface between any CNC SW system and the outside world is a bit of a stretch. So claims of 'open source' are difficult, to say the least. I take your point.

Incidentally, do you know whether FlashCut could drive 3 DC servo motors with encoders, a spindle with index pulse, and a 4th axis with stepper motors, in one package? The retrofit kits seem all to be either pure stepper or pure servo, nothing mixed.

Cheers
Roger

[SCzEngrgGroup]

Chuckle. It'll be the day when a bit of hardware can guarrantee clean delivery of data. Until then, some SW is needed, and for some funny reason modern SW writers seem incapable of not creating bugs. Sometimes ... 'automatically' just doesn't happen.

I have personally designed numerous chips with integrated USB controllers, both host and device. With modern USB cores, by the time the software even knows data is available, the hardware has already ensured it has arrived intact. If the software mucks it up beyond that, you can hardly blame the USB device or the protocol. The fact is, bulk data will be delivered complete, correct, and intact, or not at all. Retries occur at the hardware level, without the software needing to do anything. If the connection is lost, the whole device is dead - it won't re-connect without unplugging and re-plugging the cable. So, there is no such thing as a "little" data loss. It either works, or it doesn't.

The "conventional wisdom" regarding the reliability of USB for communicating with a motion controller is a bunch of garbage. USB is as reliable as any other media, including Ethernet. If a given device has been proven unreliable, the problem lies with that device, not the protocol. I've never heard of any USB-related problems with KFlop, and they are all USB devices. My two KFlops have been dead-reliable since day one, where I had constant problems with multiple SmoothSteppers - buffer under-runs, timeouts, and all kinds of headaches. I did nothing but swap out the SmoothStepper for a KFlop, and have not had one single problem, not even a minor one, in over three years.

Regards,
Ray L.

[ger21]

All I could see was
'Prices start at $1560 for Stepper and $4670 for Servo.'

For the software and USB signal generator, prices start at $1295
USB Signal Generator | FlashCut CNC

Mach3, an Ethernet Smoothstepper, a PMDX-126 and my 2010 screenset is about $550.

That's the biggest reason that people are not using Flashcut, no matter how many times you tell them how good it is.

Most (not all) Mach3 users are hobbyists, and Mach3 works fine for them. Any issues they run into are not worth the $750 premium for Flashcut.
While Mach3 didn't work for you, it does indeed work fine for the majority of users.

In another year or two, this argument will be over as Mach3 starts to fade away. It should have already been over, but Mach4's development is ridiculously behind where it should be.
I have a Mach4 license, but will not be buying any hardware until a solid, finished product has proven itself. Until that happens, I'll be exploring other options.

UCCNC shows a lot of promise, but it's a very new product, with a lot of development work still to come.
USBCNC looks like a very solid product, and the ability to customize it appears to rival Mach3's flexibility

[LeeWay]

I run 5 different machines with Mach 3 and they do exactly what I need them to. I consider linux to be a down grade for me. I did check it out when it was EMC2. Not for me.
I have two mills, a router, a lathe and a plasma cutter. Since they are all working well and doing the job that I need them to, any change for anything else will just cost more money and waste time learning it. My machines do very little for my hobbies. Mostly all production. Granted I do not sell cnctime to customers like some do. I do some R&D, but for the most part, I run tried and true Gcodes every week. The Plasma cutter cuts different parts each week, but Sheetcam handles that with ease. I have researched Flashcut before. Researched Kflop too. I am not willing to learn programming without an absolute need.
Soo, some are happy with Mach 3 and are not really hobbyists.

[RCaffin]

Hi Ray

With modern USB cores, by the time the software even knows data is available, the hardware has already ensured it has arrived intact. If the software mucks it up beyond that, you can hardly blame the USB device or the protocol. The fact is, bulk data will be delivered complete, correct, and intact, or not at all. Retries occur at the hardware level, without the software needing to do anything.

Well, sounds very good, but apparently even FlashCut don't believe it 100%. From the manual on their web site:
-----
If the FlashCut software loses communication with the Signal Generator, electrical noise may be the cause. To reduce electrical noise problems, try using a shorter USB cable, or attach one or more ferrite chokes to the USB cable. Toroid-shaped chokes are more effective than snap-on cylindrical chokes.
-----
I guess some of us have noise problems, while others don't. Wish I knew why.
USB: DC coupled and low-level 'logic' signals (~3 V?)
Ethernet: transformer-coupled, and 15 V signals (from memory)

Mind you, as a complete hand-hold, Flashcut may be quite a smart move IF you have the money but not the skills.

Cheers
Roger

[datac]

>> Um, well, should they be off by 0.001"? How come the CAM (or whatever) gave that sort of error?

SIGNLAB - That should be enough said, but I'll elaborate. Signlab from Cadlink is/was pretty much one of only one or two programs made primarily for signwork. I've always been told that it was really an offshoot of very early Corel Draw foundations, as people started to take HPGL plotters and convert them for "engraving". Wasn't long before the whole engraving machine market was born, and suddenly we had Dahlgrens, New Hermes, Mimaki, Western Digital and a whole gaggle of "computerized engravers" to choose from. ALL of them were hpgl code based, all of them somewhat limited, all of them somewhat painful. But, it still beat a pantograph !

Oh, to think back of the days of Font or Flourish Cartridges.... Sheez !

Bottom line, Signlab was what it was. They eventually created user adjustable posts for G&M and the horse race was on. Certainly not a Cad/Cam program as we view them today, but definitely way ahead of its time to the point that I used it as my primary 2.5D CAM program for years (and years). And, I have thousands of jobs that were created in it, many of which I still occasionally access and use. What I dont want it to have to regenerate the ones I open at a whim just because the control wont accept them..... Signlabs Cam was just a bit "sloppy"... you couldnt ever SEE .001" in what I was doing, but the control can. Give me the adjustment... its that simple. Lucky me, I have had that all these years.

Truth be told, when it comes to engraving fonts, paragraph texts etc, Cadlinks Signlab or Engravelab still beats pants off the stuff that's out there currently in many, MANY ways. I always gripe that some of the new developers fail to look at whats already been done, thinking what they have is really good or good enough. What a shame.


>>>> All I could see was Prices start at $1560 for Stepper and $4670 for Servo.' That's not really all that helpful.

Yep.. I know... but like I said, there are plenty of options that can change things up. They have built a rather modular methodology.... You can buy your analog output ability separately.... you can buy a limit switch kit separately.... an E-stop assembly separately..... and it all plug and plays with excellence..... but what if you dont need those things or you have those things already in place ? No problem..... you just do not have to buy them. I do think the prices shown are for rather complete turn key controls......


>>>> Um - but that's not entirely the full story, is it? Yes, the code in the SS itself is proprietary, but the hardware interface to the outside world is simply 2 or 3 LPT ports. That lets me connect a wide range of 3rd party interfaces. And that keeps the prices down.
On the other hand, imagining you could have a simple open interface between any CNC SW system and the outside world is a bit of a stretch. So claims of 'open source' are difficult, to say the least. I take your point.

Probably the closest thing this market is starting to see that might be as open source as possible is this new Arduino / GRBL / TinyG control craze created by the 3d printing explosion. Unfortunately the open source aspect is simply that they have made available all of the control specifications and requirements to the point that again, a true geek programmer could make whatever interface/ gui he stinking wanted to, and it would control..... it would control ????

Well, nothing more that once again, some PC board, that someone made, with what they thought it needed AT THE TIME they made it...... And poor old saps like us that would buy into it, would find that 6 months after we bought it, there is a new "version" board that is all the craze... and we WANT it !!!

Yep, the whole "open source" thing is rather flimsy when you actually get into it.

To me, the scary part of this has been the massive push to bring a lot of hardware/accessories to market for Mach 3. Frankly, a lot of it never really worked.... but people sure paid for it !

>>> Incidentally, do you know whether FlashCut could drive 3 DC servo motors with encoders,

Yes, it should be able to do that, but this too is along the "quasi" servo aspect built upon the world of step and direction... as far as I am aware.....
I have to be blatantly honest in that I do not have a Flashcut Servo anything.... I have not found the need ! I understand the need, I see people griping about the need and frankly they provided. You ought to see me defend steppers when someone says "steppers are junk" and "you must have servos"!! LOL !!!

Everything has its place.... there are places and benefits to either and to both.

>>> a spindle with index pulse,

Came across this when I did my Denford lathe..... They don't really use a single index pulse. They require an encoder. Funny thing was when I first fired up the Denford, I was hoping I could use the slotted wheel that was originally on the spindle. Trouble was, I think the number of holes in that disc was something like 146.... Well, the DIALOG box where you would put in the pulses per revolution would not allow me to type anything that small in there !

So, being a customer service orientated company, I called them an asked them about that. They pretty much replied that any REAL machine would use a regular encoder...... and they threw out some numbers like anything from 500 to 10,000 pulses per rev.

But, you know what they offered ? They said, we simply have checks in place on that field that stopped people from putting in some crazy number like 146... they certainly could OPEN that field up in a version, send it to me so I could see exactly what it would do !!! And, I had it by the next day as I recall. Now will they offer such service to just anyone ? Well, darned if I know... I do have a bit of an advantage by using their products for so many years.

Nonetheless, I tried it and I did cut some threads. But you know, I was curious... now that the field was open, what if I did hang a real encoder on that axis... what would it do. It did sound different, and there is a reason why. This might sound crazy, but when I fired that lathe up for the first time, that old main spindle belt had been sitting a box, all flattened out and such, that I noticed at slow speeds, the motor was actually speeding up and slowing down as that belt started to heat up. And what I was actually hearing was the control making CHANGES to the travel speeds much faster because the new encoder gave multiple more times feedback than that lowly 146 holes in a disc.

The configuration dialog does have a checkbox to indicate whether one has an "index channel".... I'd have to review the whats and whys to that checkbox as well as go look if I have it on or off at this point. All I know is I can thread stuff ! But I also recall you can NOT thread with a single index pulse. I've also read some of the Mach to LinuxCNC converts moving away from Mach specifically because they could not get accurate threading with single pulse monitoring.

>>> and a 4th axis with stepper motors, in one package? The retrofit kits seem all to be either pure stepper or pure servo, nothing mixed.

And I believe that this too is absolutely possible again because of the whole ordeal essentially built on top of step/dir. Its just a good thing that step/dir servo drives have come as far as they have to make servo even possible at this level and $$$ amount.

I know that I had actually asked about that many years ago when I was considering a servo setup for a 3 axis router. Something I found out that I happened to really like being able to do, is manually notch up or down my Z axis on some jobs. The best example is lets say you are cutting multiple small parts out of a larger sheet of material, and you want to leave as little as possible on the bottom to keep the parts "connected".. so you can "snap" them out.

Sorry, "bridges" are great cam features, but snapping a part out and essentially being DONE with it beats filing off bridges. I'd set my Z Zero height, start the job.... Well, if that first part looks like I could make that bottom web a bit thinner... all I needed to do was turn a knob on my z screw when the z axis was not running (like when at depth running a profile) to adjust up or down. No need to feed hold.... I could SEE what I wanted on the fly.

Its all fun stuff Roger.... Dont take ANY of my Text methods the wrong way... We all just have our own observations and passions... Flashcut happens to have made me as happy as one could be about a control, and it keeps doing it year after year after year !

[datac]

How are the "ethernet" hardware devices dealing with typical network communication ? If you run any monitoring software on a network of even a few computers, you quickly see that network cards rattle on and on like old wash women (not that there is anything wrong with old wash women).

I understand that Ethernet based control from years ago was strictly for a CLOSED network..... Something they could control. Are people using to NIC cards in a PC ? How are they doing it with a laptop ? Machine control ala wired connection and internet per say via wireless ?

I hope that people running "ethernet" motion hardware these days are not doing it such that the computer is not also on the regular network... Transferring files via usb thumb drive or not being able to just google something from a control in MY world would drive me nuts ! For one or two of my machines, I dont keep a dedicated control PC next to them because I dont run them all the time.. I just pop over there with my laptop, fire up Flashcut, open the setup file for that machine and have at it.... My daily laptop..... Network... heck, Netflix ! lol !

[RCaffin]

Hi Chris

as people started to take HPGL plotters and convert them for "engraving".

I still have an HP7475 plotter ... but I use a laser printer these days instead.

this new Arduino / GRBL / TinyG control craze created by the 3d printing explosion.

I remember some sneers when I had some early prototypes done from an STL model using laser-fused plastic powder. Sneer sneer sneer, you will have to get a real mold made ... we never did. We kept tweaking the design and making a few more - and putting them into the field. The mold costs for all those iterations would have been horrendous.
Yes, the Arduino or Raspberry Pi craze is all hobbyest at the moment, except that I do see them creeping into serious field use at 1/10 (1/100th?) the budget cost. And they do work.

the massive push to bring a lot of hardware/accessories to market for Mach 3.

I may or may not believe, but when the market goes in a certain direction, I notice.

But I also recall you can NOT thread with a single index pulse. I've also read some of the Mach to LinuxCNC converts moving away from Mach specifically because they could not get accurate threading with single pulse monitoring.

Oh, some people will say anything when they fail to follow the instructions. Doesn't mean they are right though.

Yes, you can thread on a CNC lathe with just an index pulse - because I have been doing it for years. What a lot of the failures do not realise is that you have to TELL Mach to use the index pulse to sync for threading. It's a config option.

Well, if that first part looks like I could make that bottom web a bit thinner... all I needed to do was turn a knob on my z screw when the z axis was not running

Ouch! Loss of Zero! Yeah, can do, but ... :-)
I write my own parametric code, and altering a web thickness simply means altering one of the defined parameters. That gets done a LOT of the time!

How are the "ethernet" hardware devices dealing with typical network communication ? If you run any monitoring software on a network of even a few computers, you quickly see that network cards rattle on and on like old wash women

A lot of that is completely invisible to the server and client. It's just 'keep-alive stuff continuously monitoring the links. Ignore.

I hope that people running "ethernet" motion hardware these days are not doing it such that the computer is not also on the regular network...

FIRST rule for a production machine: do not use it for web browsing!!!!! Sorry, BUT!

Sure, a modern PC may have performance to spare, but we keep that ethernet link CLEAR! Two NICs? Dunno, maybe.
Yes, I may be doing some CAD or CNC programming while the machine is running, but that does not interfere with the network stack.
Look, the CNC may be worth anywhere from $10 to $50k: it's worth dedicating a <$1k PC to running it imho. So we do.
Yes, USB keys are great, especially as the machine is a bit too far from my design office. You could probably run FO or even an old serial link the distance ... but why bother?

Cheers
Roger

[samco]

The next release version of linuxcnc (2.7) has support for real-time Ethernet communication. I have been testing it for the last year or so. Pretty exciting stuff.

You do need a dedicated nic to interface with the hardware. And I do other things on the computer also (wireless nic connected to the net or dual Ethernet....) (The hole point of the realtime setup is the machine specific operations take priority over everything else)

The Ethernet devices in linuxcnc are not buffered like say the ess. It is communicating with the device every ms (or .5ms or whatever you set)

This is one of the tests I was running - Reading the step/dir signals from mach3 with a mesa 7i80 interface card. (Ethernet)
https://www.youtube.com/watch?v=HfU4uyGgLZw

I actually started it to test the new trajectory planner (also in linuxcnc 2.7) to make sure it wasn't violating acc/vel constraints and thought - Hey I could test other systems...

As always - you still need a system / nic that plays well with realtime.

sam

[BanduraMaker]

datac,
My next upgrade is likely to be Kflop since I can use it in conjunction with my custom MACH3 screens.

I know I'm late to this party but, using a Kflop as a plug in to Mach 3 isn't really going to get much benefit. As Ger mentioned in an earlier post, one of the biggest issues with Mach3 is that it violates acceleration settings at seemingly random (but reproducible) points when running in CV mode. That bug is in the trajectory planner and since the Kflop plugin uses Mach 3's trajectory planner, it will still exhibit the same behavior.

I have a Kflop sitting on the shelf but haven't gotten it up and running yet because I have to build a whole new control box for it. I got sick and tired of ruining parts due to Mach 3s deficiencies* so I picked up a UC100 USB to Parallel port motion controller and.....same problems with violating acceleration parameters

The good news is that UCCNC makes their own control software which although pretty basic, has a much better trajectory planner than Mach 3. I've done some testing with it and although we did find a major bug, UCCNC fixed in in about 2 days and now that it's fixed, it runs circles around Mach 3 in terms of machine performance. I have a little video I made for another thread here. I'm dedicating today to re-writing Ger's tool change macros so that I can put it into service hopefully by the end of the day.

*Mach 3's bugs in the trajectory planner are real and if you're pushing your machine, you will find them and it will cost you money. True, the vast majority of Mach 3 users won't have these problems as they're running their machines at rather sedate feedrates but once you start pushing it in order to save time (and make more money) these ugly problems will rear their heads.

[SCzEngrgGroup]

KFlop also has it's own controller software - KMotionCNC, which works perfectly.

Regards,
Ray L.

[HawkJET]

That bug is in the trajectory planner and since the Kflop plugin uses Mach 3's trajectory planner, it will still exhibit the same behavior.

Thanks for clearing this up. I had suspected it may be the case. However I am not one of the users that have had any issues with MACH3 so I am assuming that I will be okay. If I do run into any issues, as Ray points out, KMotionCNC works perfectly. I would then have to give up my existing MACH3 screenset and possibly make custom screens for KmotionCNC as Ray has done (although it is definitely more cumbersome to do).