Retrofitting a 1986 Maho MH400E

More
17 May 2019 18:19 #134083 by Glemigobles
Here's what I came up with today, afer several merry hours. Note that it's still not the same as it was on the Philips, at least at the top rapid speeds.

MAX_ACCELERATION
X 90
Y 100
Z 65

These values allowed me to retain a following error of ca. 0,025mm at max speeds. They were tuned at a feedrate of 1000 to within ca. 0,015mm, using a move of 25mm. The latter error value represents the largest spike, which in my case happens close to the beginning; most of the way the error is below 0,01mm. Due to absolutely zero experience as a CNC service tech I have no idea if this is all good or bad.

I got very anal about the P and ferror values and didn't do anything with I or D. The machine still has perceptible speed up and slow down periods in rapid moves. I haven't really done anything with the G64 modes though, it was running default for tuning.

The drives sound more annoying in Linux than they did on the Philips. I assume this is the combined effect of wear with the faster rate the controller uses to communicate with the Indramat. You could probably replace the motors with modern ones with better parameters that would be silent, and maybe raise the rapid speed to 5m/min. There are two sources of noise in my drives: the hum of the DC motors, and the noise of the reductions, ways and ballscrews. The Z axis in particular has a length of noisy travel. Your Maho is probably much better in this regard due to its excellent mechanical condition.

Regarding your points, it may well be that the low limit of programmable feeds on the Philips controller was connected to the following error at higher speeds (among other factors, chiefly the controller itself). Rapids just need to get to point B quickly, and any overshoot is going to be corrected instantenously. Nobody would think of rapid moves in material. Then there's the issue of spindle speed, design purpose and available tooling, which meant that in the 1980s Maho engineers didn't care about higher speeds that much. Deckels were the more agile machines, but they didn't have the obscene torque of the Maho geared spindles.

Please Log in or Create an account to join the conversation.

More
17 May 2019 20:19 - 17 May 2019 21:07 #134087 by RotarySMP
If you think about it, a following error of 25µm lag in the linear direction of travel is next to nothing. Maybe we should try tuning for 200µm Ferrors at rapid speed and see what that means for the overall performance?

If there is someone reading here who has experience with the setting up this sort of thing, please jump in.

I and D are not used with velocity controlled drives. FF1 and FF2 have a significant effect on the initial accel and final decel behavior.

Deckel FP CNC's also have gearboxes. Do they not have a similar range of speeds to the MAHO?
Mark
Last edit: 17 May 2019 21:07 by RotarySMP.

Please Log in or Create an account to join the conversation.

More
17 May 2019 22:26 #134091 by J Green
Looking at a 1982 Maho sales catalogue

400 P and 600P ( non CNC but straight line or point to point or not 432 control )
Resolution = .005mm ,.0002inch ( may be non LS403 scales ? )
Rapid Traverse = X and Z axis 2.4 m/min , 94.5 inches/min - Y axis = 1.6m/min , 63 inches/min

MH500C using the 432 control
Resolution XYZ axis= .001mm ,.00004 inch ( may be just the LS403 scales and not the complete machine resolution )
Feed Range for all XYZ axis = 1 to3000mm/min, .04 - 118.1 inches/min
Rapid Traverse for all XYZ axis = 5 m/min , 197 inches/min

MH700C and MH900C using the 432 control
Resolution XYZ axis = .001mm ,.00004 inch - -B axis =.001 degree ( may be just the LS403 scales and not the complete machine resolution )
Feed Range for all XYZ axis = 1 to3000mm/min, .04 - 118.1 inches/min
Rapid Traverse for all XYZ axis = 6 m/min , 236 inches/min

Note : The this is from a March-1982 sales brochure , but in 1982 the Marketing Departments didn't drive the engineering as they do in today's world.

Mark Is this of any use to you? Noticed the use of a radius vers sharp tip on the engraver tool. Don't know if your cutter grinder has a radius attachment - yet .


Wish I had the electric power hooked up for my MAHOs !
Bob


p

Please Log in or Create an account to join the conversation.

More
18 May 2019 05:45 #134100 by RotarySMP
Hi Bob,

I do have the Clarkson Mk1 radius grinding attachment, but that doesn't really help as long as I don't have a high speed engraving spindle. That is why I made the drag engraver, which does not spin, it is just a scriber which get pushed into the work and dragged through it. It needs that sharp point.

The resolution of the E series Mahos is also 1µm (although mine has only 5µm resolution on Y at present due to the replacement EXE used), and has max rapids of 2500mm/min, but those specs say nothing about how much following error Maho allowed. The high resolution allows the controller to know real position, verses commanded position (= following error).

What we don't know is what following error at rapid speed is considered normal and tolerable for a tool room mill in this class.

Please Log in or Create an account to join the conversation.

More
18 May 2019 10:04 #134148 by Glemigobles
I had this idea yesterday after reading your post Mark that I could cross-referrence the Philips machine constants I have with the relevant manual from the Heidenhain website. They bought the Philips CNC branch and still have the documentation available online. But unfortunately, I'm having trouble accessing the file list today, their website posts an error of service unavailable.

There are manuals there with instructions for installing the controller in a machine, and finding various values to be entered in the MC list when you're starting from scratch. Checking them would help a lot I assume, but the control logic of the Philips was probably different and we'd have to search for any paramters that could be relevant to a Linux installation.

Bob, I wish my MAHO was that fast. All the FP Deckels from the 1980s I've seen for sale had rapid traverse speeds around 6m/min. I always assumed that the difference was down to spindle design. Deckels also had faster spindles. I'm sure Maho also made mills with various designs, but both Mark and I ended up with high torque models, probably designed for cutting hard steels with medium sized tools. They're somewhere between a 50 taper machine and a more standard toolroom mill with direct spindle speed control.

It's actually a blessing to get these nowadays I think, because it's super easy and rather cheap to attach a HF spindle for fast light cuts, while retaining the torque for fast material removal with larger tools. And Linux allows you to configure them beautifully, which wouldn't be a cheap option on a new industrial controller.

BTW, when I had my LCD monitor on the electric cabinet next to my VFD, the display would go completely haywire when the VFD was on. I had initially made some tentative plans to install the unit in the cabinet, but seeing the electrical noise it generates and having very little faith in the shielding of my wiring and components in the cabinet, I've decided against it. The VFD will stay outside, and I'll just make some holes, or use the existing holes to get the cables inside to connect the thing to LinuxCNC. I might make some kind of stand for it, so that it doesn't sit on its cables anymore and is protected from stray chips and oil.

Please Log in or Create an account to join the conversation.

More
18 May 2019 10:54 - 18 May 2019 10:57 #134161 by RotarySMP
Thanks for posting your acceleration values. They are nearly twice as high as mine, so I'll take another shot at tuning and see if I can get close.

the control logic of the Philips was probably different


Regardless what the internal motion control algorithms of the Phillips 432 looked like, the external control loop was the +/-10V velocity command to the Indramat, and position feedback from the LS403 scales. Whatever performance the Phillips could command, LinuxCNC must be able to as well.
Mark
Last edit: 18 May 2019 10:57 by RotarySMP.

Please Log in or Create an account to join the conversation.

More
18 May 2019 10:58 #134164 by Glemigobles
If anybody can check from their computer whether their website is available, it's heidenhain.com. Then you have to click on documentation, then mauals and finally O.M.A. controls. I don't know why it won't work for me, it used to in the past.

I'm attaching a picture of my old machine constants printout which has the software version of the controller listed at the start. Checking these against the manual is my best idea for getting a glimpse into what the engineers that made these machines thought about tuning servo drives.
Attachments:

Please Log in or Create an account to join the conversation.

More
18 May 2019 11:08 - 18 May 2019 20:20 #134167 by RotarySMP
A tapmatic 70X came up locally for only €99, so I took a chance on it sight unseen





When it arrived I was pleasantly surprised that it had a J441 collet in it.



This is a Swiss made one, with a B18 taper in it. I'll need to buy or make a BT40 - B18 shank, and the wedges for pulling the #3 morse shank currently fitted. It seems to work in that it drives forward with axial load on the output, and then reverses to a higher RPM when you pull on it. I have nothing with a MT3 taper to test it, as the Boleys tailstock is MT#2.
Mark
Attachments:
Last edit: 18 May 2019 20:20 by RotarySMP.

Please Log in or Create an account to join the conversation.

More
18 May 2019 11:11 #134168 by RotarySMP
It comes up "This service is unavailable" for me as well.

Please Log in or Create an account to join the conversation.

More
18 May 2019 11:29 #134169 by Glemigobles
That kind of BT40 to B18 shank sounds like a job for This Old Tony.

It's a shame the service is currently down, let's hope it becomes available again soon. I downloaded documentation from there some months ago but I didn't get the right controller version and the constants are completely different between versions.

There's also Anders' smoregrava file with a different version of the controller than mine. But I think Heidenhain has them all, or at least they have a lot.

I'm also having a CNC service tech come over to look at the alignments on my machine next week (if he actually makes the time, that's not a given over here). He was interested to hear about my retrofit on the phone and he's experienced with industrial controllers so I can ask him what he thinks about my servo setup. Experts like to be critical and educational at the same time when they see the work of a noob, also I might pay him to do a better job and post the results here. I'm sure someone who's serviced dozens of machines would have a very good idea of how a tool room mill should be expected to work.

Please Log in or Create an account to join the conversation.

Moderators: piasdom
Time to create page: 0.314 seconds
Powered by Kunena Forum