What's new
What's new

Do we have a repair option?

I have used these guys on several occasions, amp, servo motor, spindle drive, and even power supply for my 20 yr old faro arm. Some times turn around time was a little slower than I would have liked, but over all I have been very pleased with the repairs Ive had done. My experience with them has been positive.
 
Several months ago I could have bought a vf2 without the control boards for $2000. It was cold and snowy and I had enough projects so I said not now. One of these days I will get one and do a Centroid retrofit. I will share all the plc programs and wiring instructions to anyone. The only problem it wont be a real Haas anymore. The operation and programming will be different.
 
I am thinking along your lines. I know someday my '91 VF-1 will kick the bucket and $15K is way too much to upgrade. It would need new everything since nothing is compatible.
There are FPGA boards that can handle the real time operations and SOC boards that can handle the I/O. Considering my processor is 20MHz and a BeagleBone is 1 Ghz, I think it could handle it. Plus, there are LinuxCNC programs that have all the trajectory software done. What we need is an interface program that works with the Haas keypad, servo drives and I/O board.
The existing LinuxCNC programs are made to work with a PC, mouse and keyboard, all unsuitable to our application. Nobody wants to replace their controls with something that works completely different to what they are used to.
My project is to replace the computer stack with an SoC and FPGA board on a new board with all the existing connectors. Everything works the same plus upgrades such as rigid tapping and programmable coolant nozzle. And, of course, change the screen to an LCD.
A plug and machine upgrade.
 
I am thinking along your lines. I know someday my '91 VF-1 will kick the bucket and $15K is way too much to upgrade. It would need new everything since nothing is compatible.
There are FPGA boards that can handle the real time operations and SOC boards that can handle the I/O. Considering my processor is 20MHz and a BeagleBone is 1 Ghz, I think it could handle it. Plus, there are LinuxCNC programs that have all the trajectory software done. What we need is an interface program that works with the Haas keypad, servo drives and I/O board.
The existing LinuxCNC programs are made to work with a PC, mouse and keyboard, all unsuitable to our application. Nobody wants to replace their controls with something that works completely different to what they are used to.
My project is to replace the computer stack with an SoC and FPGA board on a new board with all the existing connectors. Everything works the same plus upgrades such as rigid tapping and programmable coolant nozzle. And, of course, change the screen to an LCD.
A plug and machine upgrade.

Your plan sounds good but I am willing to bet that you will have more than 150 hours into getting this to work. At $100/hr.

You will be above the $15K Haas wants for their upgrade. Your approach is fine if you don't have the $15K but do have the 150hrs.

Do keep in mind, few people will consider buying a used machine with modified controls. Then there is the problem of liability with life safety issues.

One other mistake I see is that in most of these conversations, clock speed is being confused with machine instruction time.

Motorola's 68000 family which later became Coldfire had a slow clock speed but its instruction time was often quite good in that a lot of instructions were completed in only a few clock cycles and all of the registers were capable of 32bits. This gave the processor family a pretty respectable data throughput.

Intel in comparison used a design philosophy of utilizing lots of clock cycles for a single instruction. Nothing wrong with this approach and is a great marketing method in that our clock speed is 10X the competitor even though it takes us 10-20 clock cycles to do an instruction.

My point to this is that the simplistic approach probably will not yield the expected results after you get into it at the CPU instruction cycle level. Not saying it won't work but I seriously doubt that someone or a company can re-engineer the boards, get them to work and manufactured to the cost point that Haas is offering the upgrade board for at $15K. If you had a thousand customers willing to spend $5000 for an upgrade path, it would probably be a viable approach.
 
Several months ago I could have bought a vf2 without the control boards for $2000. It was cold and snowy and I had enough projects so I said not now. One of these days I will get one and do a Centroid retrofit. I will share all the plc programs and wiring instructions to anyone. The only problem it wont be a real Haas anymore. The operation and programming will be different.

I ran a centroid control, liked it well enough. I would not see the centroid retrofit being the problem as much as the retrofit itself. IE "oh... it's a retrofit so it's no good blah blah..."
 
$100/hour if I was making that much. I was making more than that when I worked in aerospace but now I am mostly prototyping parts for old British motorcycles. Not trying to be a hobby but the pay is about the same.
If my '91 dies I am not spending over $30k to $40k to upgrade it (it would need new servos, drives, spindle drive and motor plus encoder, monitor, keypad(?) and so forth). Nobody is going to pay more than scrap for a machine you cannot get parts for and nobody would pay for a 27 year old machine with upgrades what I would have to put into it.
So the point of this is to save me money down the line. I expect there are others in the same situation. If I were a company having to charge all my time to a number then yes, it would cost a lot.
Essentially all the code for parsing the .vnc files and coordination of the servos is done. What has to be done is making the interface work with the Haas control and look the same.
FPGAs can parallel process, unlike microprocessors (unless they are multi-core). They do not have to time slice the way a microprocessor does which is why the PC based CNC programs have to do a speed check on the hardware. Nobody running a CNC needs to check E-mails or play games. The FPGA board would handle the trajectory calculations. The processor only has to keep the input buffer full which is not a full time job.
There are many people debugging the LinuxCNC software. I doubt there are any problems there. It is only the interface control that is new.
By the way, the BeagleBone Arm Cortex-A8 processor is 64/128 bit single instruction cycle, two 64 bit ALUs and fully pipelined.
 
There are many people debugging the LinuxCNC software. I doubt there are any problems there. It is only the interface control that is new.
Put a K&T D control onto an fpga, make it a bolt-in kit and even *I* would buy a Haas :D


Okay, not gonna happen. But if you do LinuxCNC, could you please do a professional-grade interface instead of that stupid hobbyist crap ?
 
That is what I intend. The LinuxCNC variants all use a PC mouse and keyboard. Who wants to use a mouse to get to the Emergency Off button?
The machine has to work the same with the same keypad interface. When I asked for the code to rewrite the interface their response was "why? We already have this one". Obviously the reason is professional interfaces are what shops are used to.
G code is ancient but it would be tough to get everyone to agree on a modern update. A more conversational interface would be clearer. Computers used to be programmed by flipping switches on a panel. Just do not get SAE involved or they will require licenses on every machine.
 
That is what I intend. The LinuxCNC variants all use a PC mouse and keyboard. Who wants to use a mouse to get to the Emergency Off button?
The machine has to work the same with the same keypad interface. When I asked for the code to rewrite the interface their response was "why? We already have this one". Obviously the reason is professional interfaces are what shops are used to.
G code is ancient but it would be tough to get everyone to agree on a modern update. A more conversational interface would be clearer. Computers used to be programmed by flipping switches on a panel. Just do not get SAE involved or they will require licenses on every machine.

The E-Stop issue alone is a show stopper in an environment that has workers. This is the underlying problem with all of these cheap approaches, the hardware is not designed to meet the life safety requirements of the work place.

There are specific design requirements and specified control behaviors that must be met to be considered "Safe". That means when the E-Stop is pushed all power to the output channels is removed, servo power is removed, and all energy storage sources are drained of energy. This all must be done with only hardwired circuitry for fail safe operation which means the E-Stop strings must be normally closed when the E-Stop is off.

Just mentioning an E-Stop activated by a mouse tells me that there is a very very long way to go before this could be considered workplace ready.
 
Just mentioning an E-Stop activated by a mouse tells me that there is a very very long way to go before this could be considered workplace ready.
Actually, the original project was sponsored by some gubmint department, they retroed a KT 200 and it was fairly rational. You don't want to be sloppy with a 25,000 lb horizontal.

Then the kindergarteners got into the act.

But magno_grail here just said the magic words, he agrees with you that a mouse-driven e-stop is insane. The code for driving axes is pretty well proven. It's mostly the interface that is now bullshit. It was probably nice when they started.

About safety, I have to shake my head. Have read some posts here with Big Brand Names that are totally wrong. And I got a blob of runny egg on my face once troubleshooting a Mazak. I was certain that what we were seeing was impossible, because no machine tool manufacturer would write interface code that was so shitty.

Nope. I was wrong. Mazak did. Garbage.
 
RANT TIME!!!!

The E-Stop issue alone is a show stopper in an environment that has workers. This is the underlying problem with all of these cheap approaches, .


Full stop, right there!!!
And this is EXACTLY the correct forum and EXACTLY the correct time to do that full stop.

Y'all may have your opinions, wishes, how-to-do this and that with a retrofit .....

I DO NOT F@CKING CARE TO DO IT ON A MACHINE AND BUILDER which has already established itself as a reliable piece of equipment to have in an industrial environment!

And the jest of what I am saying is this:

One SHOULD NOT want to retrofit a Haas control with anything, and the current people running the Haas software/firmware department SHOULD NOT want to make excuses for
idiotic mistakes that are inherent in the NGC!

I do not have a problem with the MTB charging me $15K for a retrofit of a full control package of a 10+ year old machine. The price is what it is and I have a choice whether to pay it or else.
At the same time I absolutely will not accept any excuses why the new control CANNOT BE MADE TO FUNCTION as the older one it is replacing!

If Haas is listening ( that is Gene Haas ), please review the morons that are running the software / firmware show.
The NGC has potential and I am willing to bet that it can be configured to be used as a retrofit for older machines, but the way it is running or handled by the current crop of programmers
and support personnel is an absolute garbage.
Every new iteration has bugs, and I get that.
But the attitude that is coming out of the new development team OTOH sucks.

Here is the deal: The old control worked flawlessly and reliably. What it could do, it did. What it could not do, it had told you it cannot do. Period, the end.
The NGC on the other hand is a clusterfuck of JAVA garbage written by 20 somethings who couldn't get into working for a "Famous Giant" software firm, so they've ended up
getting an interim job at Haas Automation.

Are you fucking kiddin' me?
Who can possible accept that a CNC control newly built in 2016 cannot keep track of 2 different values that are labelled the same, when it's 25+ year old brother had
no problem doing so?

(For the Haas software people: I am talking about G8x cycles with subroutine calls and your inability to keep tracking the fucking P variables! )

Also, why is the MPG need a second pulse when reversing to be noticed by the control?
Why is the freaking cabinet lighting part of the control logic and need 2 presses to either turn on or off the lights when it is in "sleep" mode?
E-Stop means E-stop! After that the display should still be able to be put to sleep! Right now with E-stop in the display stays on forever.
Ditto for the luber! When the E-stop is in, there should not be any lubrication applied. Ever. PERIOD!!!

RANT-OFF ( for now, going to bed.... )
 
Linuxcnc isn't dictating how the estop works - you as a machine integrator are. You make it as safe or as scary as you want. If you want only the gui button to act as the only estop - that is your prerogative. is it safe? Heck no!

We have converted a few machines now. The very first circuit we setup is the physical estop loop. The computer has no control over it (other than reseting it after all the mushrooms/overloads/overtravels/whatevers are fixed.)

The K&T we converted kills power to everything except the computer. (servo drives, spindle, hydraulic units and so on) The thing has 8 tons of cutting force - and I trust that when I hit the big red button - it is going to stop.

The other thing when doing your own conversion - you get to know the machine inside and out. You don't have to call the tech to fix it.

sam
 








 
Back
Top