What's new
What's new

Best way to go from Bridgeport to Low End CNC in 2024

I am not an advocate of retrofitting commodity machines. To be blunt I think it's a waste of time to retrofit a knee mill or a two axis lathe etc.

However.

I have used LinuxCNC in the past for some projects and was quite impressed by it and what can be achieved with it.

If I were doing this I'd go LinuxCNC without even considering other options.With mesa hardware in particular, it is really trivial to integrate it with pretty much any kind of existing servo hardware and encoder feedback that is already on the machine.

I have spent a good amount of time comparing all these options in the past, and while my info might be outdated, I have never seen any compelling reason to choose Centroid, and especially not Mach, over LinuxCNC.
 
As I understand things back EMF generated when driving a servo motor is a known potential problem for low voltage low power integrated circuit drivers. Spinning a small servo intended to operate on 3 volts or so can easily produce 9 volts or more of back EMF which the IC won't like. Texas Instruments make things like the TVS diodes for that sort of thing and issue application notes for motor drives covering the general issue.

Bit different for the typical 100 volt (nominal) industrial rated servos found in machine tools where rapid deceleration can produce significant back EMF spikes which the control has to deal with as a matter of course. Far greater than anything generated by the very low speeds you get from turning the handles.

According to Motion Dynamics https://www.motiondynamics.com.au/back-emf-and-why-its-a-nightmare-for-speed-controllers..html high current motors can generate very high back EMF spikes during fast deceleration. The worked example suggests 10,000 volts is possible from a 100 amp motor. More of a worry for the electric vehicle folks methinks.

Clive
 
I am not an advocate of retrofitting commodity machines. To be blunt I think it's a waste of time to retrofit a knee mill or a two axis lathe etc.

However.

I have used LinuxCNC in the past for some projects and was quite impressed by it and what can be achieved with it.

If I were doing this I'd go LinuxCNC without even considering other options.With mesa hardware in particular, it is really trivial to integrate it with pretty much any kind of existing servo hardware and encoder feedback that is already on the machine.

I have spent a good amount of time comparing all these options in the past, and while my info might be outdated, I have never seen any compelling reason to choose Centroid, and especially not Mach, over LinuxCNC.

Why do you think so? Ive done a Bridgeport Interact and a Hardinge CHNC 4 lathe retrofits. The Hardinge has basically gone back into production with its new owner ....
 
Why do you think so? Ive done a Bridgeport Interact and a Hardinge CHNC 4 lathe retrofits. The Hardinge has basically gone back into production with its new owner ....
Simply because the cost/benefit almost always favours replacing the machine with something newer, better, faster that doesn't need to be retrofitted.

This is especially true if one is aiming for a retrofit with true industrial quality components and function, and to have everything documented properly for if the machine is sold to a new owner.

This does not mean that I am against retrofitting, and I have done a few, but it's a process best reserved for higher value machines IMHO.

Not trying to ruffle anyone's feathers, just my pragmatic view of things.
 
Simply because the cost/benefit almost always favours replacing the machine with something newer, better, faster that doesn't need to be retrofitted.

This is especially true if one is aiming for a retrofit with true industrial quality components and function, and to have everything documented properly for if the machine is sold to a new owner.

This does not mean that I am against retrofitting, and I have done a few, but it's a process best reserved for higher value machines IMHO.

Not trying to ruffle anyone's feathers, just my pragmatic view of things.
Yep, all good. Just keen to understand.

I'm retrofitting a Triac with 6 Change ATC as we speak. Centroid Acorn, Panasonic 750W for Axis and 1.5KW for spindle. Hardware is probably the better part of £3K. Thoughts? (Not my machine BTW).
 
I posted the question over on the centroid forum and the following is the response I received. It looks like cranking by hand is ok. https://centroidcncforum.com/viewtopic.php?t=10215

I am fairly certain that, with power released, the Allin1DC essentially disconnects the motor, so there is no energy to absorb.

But, even if it were in a dynamic braking mode (and you would know that when you tried to turn the cranks) the drive should still be able to dissipate anything you can generate by hand.
 
According to Motion Dynamics https://www.motiondynamics.com.au/back-emf-and-why-its-a-nightmare-for-speed-controllers..html high current motors can generate very high back EMF spikes during fast deceleration. The worked example suggests 10,000 volts is possible from a 100 amp motor. More of a worry for the electric vehicle folks methinks.
Any servo motor is effectively open circuit if it's not being actively driven. Either it's an IGBT or a transitor H-Bridge if it's older, either way there is no output from the motor when it's not being driven because there is no circuit made across the windings. The reason they need regenerative/power sink circuitry is because they are full four quadrant control, i.e. they are always being driven, even when they are decelerating and absorbing inertia from the physical machine.

Yep, all good. Just keen to understand.

I'm retrofitting a Triac with 6 Change ATC as we speak. Centroid Acorn, Panasonic 750W for Axis and 1.5KW for spindle. Hardware is probably the better part of £3K. Thoughts? (Not my machine BTW).

Being a niche training machine, there is, IMHO, more value in retrofitting something like that little Denford than there is in retroing something that is an honest-to-god commodity machine like an Interact.

The Triac exists in a gray area between junk toy machines and expensive industrial micro machines. It's not easily replaced with something else, and nobody has high expectations of productivity from it. On the other hand, the Interact will always be an old (very)slow machine, probably worn, that isn't going to be very useful to anyone who is serious about making parts, and is easily replaced with something better.
 
I largely agree. The Triac is a 220V, 'office capable' decent bit of kit and certainly not a toy. The owner of the first one I sold (who also owns this one) uses his for making bits for his day job. He doesnt need an amazing MRR, he runs his Triac in his lounge.

I'm not sure I agree with the Interact statement - suppose depends on what you pay for one. The last one I had I paid less than £200 for, a bit of a bargain and cost me low £thousands to modernise.
 
I largely agree. The Triac is a 220V, 'office capable' decent bit of kit and certainly not a toy. The owner of the first one I sold (who also owns this one) uses his for making bits for his day job. He doesnt need an amazing MRR, he runs his Triac in his lounge.

I'm not sure I agree with the Interact statement - suppose depends on what you pay for one. The last one I had I paid less than £200 for, a bit of a bargain and cost me low £thousands to modernise.

Whether or not you agree with the Interact thing I imagine largely depends on what you intend to do with it and what your expectations are. I do think that most people would be better served with a newer, faster machine, but understand that doesn't apply to everyone.

I do think, given that you appear to be quite into doing these retrofits and will likely do more of them, that you should look seriously at LinuxCNC. It is much more capable than Mach or Centroid, in that it is completely open and you can make it do practically anything on any kind of machine (short of dual/multiple path control), and can make the hardware side as industrial or as hobby as you like. Conventional wisdom is that it is a steeper learning curve, and that's probably true, but I think the dividends of learning it would pay off for someone like you.
 
Thanks. I am a fan of Linux, I do IT as the dayjob. However the learning curve is a struggle and time invested is a rare commodity. Perhaps Ill take a look at it again based on your recommendation. I am about to do a fit out on a 5 Axis knee mill that I have ..... might be a place to 'play' and see.
 
Thanks. I am a fan of Linux, I do IT as the dayjob. However the learning curve is a struggle and time invested is a rare commodity. Perhaps Ill take a look at it again based on your recommendation. I am about to do a fit out on a 5 Axis knee mill that I have ..... might be a place to 'play' and see.

For 5ax definitely, as it can do real kinematic RTCP for 5ax simultaneous, which to the best of my knowledge the other options can not.
 
For 5ax definitely, as it can do real kinematic RTCP for 5ax simultaneous, which to the best of my knowledge the other options can not.

Thanks. Ill need to google that.

Is there a preferred controller to use and if so, what mode are servos typically run in?
 
OK, so, this thread seems to claim Centroid can ....


But Ill need to ask to be sure.

The second post in that thread confirms that Centroid does not have RTCP.

It's not a deal breaker, but it's a much nicer solution that relying on the cam software to linearise everything.

Here's someone using RTCP in jog mode - these motions would be impossible to do on Centroid or Mach for example.


Thanks. Ill need to google that.

Is there a preferred controller to use and if so, what mode are servos typically run in?

I think not really - I have only used it with Mesa motion control hardware and velocity mode amplifiers and TTL quadrature encoders. That was a pretty seamless setup even when I did it ~12 years ago.
 








 
Back
Top