Thread: Variable Lathe spindle speed control project for constant surface speed

From time to time on this forum a few threads have touched on the idea of controlling lathe speed to maintain a specified surface speed.
This thread will be a stub to pool ideas and perhaps develop a group designed project to make this possible.

The sytarting point is a variable speed spindle drive with external control.
VFDs running a 3-phase motor is one of the options. Others may be DC dries with PMW electronic control or even setro drives with amplifier control.

So far, the discussions have proposed to fit a multi-turn potentiometyer to the cross-feed screw and a linerar slide potentiometer moed by the cross slide mechanically.
Both options would provide a linear variation in resistance or - if used as a voltage divider - variable voltage in response to tool position in relation to the cutting radius.
I was quickly identified that this is not a direct relationship to desired spindle rpm, since other ariables are material cut, tool geometry, machine rigidity, turning operation (turning versus facing), and desired speed range for specific purposes like threading for example.

My last comment in another thread was to suggest using the positional data of a potentiometer as just one of the inputs to a 'black box' processor with additional operator inputs for measured starting radius, desired surface speed (maybe this could include a reference to material machined), allowances for tool geometry and chatter avoidance. In addition, the set motor to spindle drie ratio will need to be able to be set.
These variables would then be used to feed the actual motor controller - such as a VFD - with the signal required to set the spindle speed and vary it to maintain a desired surface speed as the turing operation changes the cutting radius.

I imagine CNC machines hae all these parameters already worked out and some input from our experienced CNC operators and designers/builders will be greatly appreciated.

Hi Joe,
I have been thinking about this for a while too, although with the machines i now own and use it is less of a concern to me. Whilst you could do it as an analog system i see some major flaws in it. I can't see a way to reach full speed for one, not without hitting the end of the pots travel. The other is that the speed to position ratio is not linear, it is exponential.
Far easier i think to program a PIC or Arduino with dual analog inputs, one for the cross slide position and one for the tweaking pot, maybe a "materials selection" switch and of coarse a manual overide switch. Then i think you will need a zeroing button to tell the unit where center is so it can calculate the correct speeds at whatever distance from the materials center.

Ew

Originally Posted by jhovel

It was quickly identified that this is not a direct relationship to desired spindle rpm, since other variables are material cut, tool geometry, machine rigidity, turning operation (turning versus facing), and desired speed range for specific purposes like threading for example.

It's worth pointing out that the surface speed is not exact - it has been worked out empirically as the speed that give optimum tool life under industrial settings, so while a target it is not necessarily to hit it exactly.
On the surfacing/ turning vs facing thing I can't see why it would make a difference. If you desire a particular speed at say diameter 50mm when facing, that should also be the speed when turning parallel (surfacing) at diameter 50mm. The thing is that ideally you would like it not to make a difference which direction the tool is moving provided the speed is right for the radius.

Michael

Hi Joe,

only just read this thread and prior to this I had never really given this subject any thought. So, as a question without notice I come up with the following suggestions:

Broadly, it would need 3 components. (1) A electro-mechanical device that measures tool movement wrt the spindle rotational axis. (2) A "black box" that would then interpret the signal from (1) and also allow user inputs for such things as you mentioned. i.e. threading, facing, tool type, material being cut et. etc. and (3) a method for varying the spindle speed. This would be controlled by the "black box" of electronics.

Part (1) could be as simple as a linear POT, or perhaps even a set of the cheap DRO scales. If using the dro scales, the black box would need to be able to interpret that protocol. I assume the black box would be made up of a PIC with user made firmware and allow inputs and perhaps an LCD screen. If the code becomes too involved then perhaps a simple interface and a laptop should be used. If the speed control is to be done with VFD, then the firmware would need to make allowance to input lathe gear selection so as to keep the VFD operating within manageable frequencies (say between 30 - 100hz depending on HP) Oh, also there would need to be information from spindle rpm fed back into the black box so as to keep the spindle speed correct regardless of cutting load and make allowances for motor slip.

The more I think about it, the more I think that several PIC's would be needed or indeed something with a little more computing power, depending on how "complete" you wish to make this and how many features you would want to include.

Well, that's my 2 cents worth anyway!

Cheers,

Simon

Ew,
as I wrote above, I agree that a processor is needed to integrate all the variables and the position and ratio data. I didn't know that the radius to surface speed is exponential. No doubt something a PIC or ATMEL processor can calculate on the run.
A zeroing button for the actual tool tip is a great idea. That could actually be any value, let's call it 'offset button' - or indeed hae both - since I can imagine it geing a nuiscance to try and set 0 once a workpiece has been possibly laboriously set up...

Originally Posted by Ueee

Hi Joe,
The other is that the speed to position ratio is not linear, it is exponential.
Ew

Hi Ueee,

Why would this be? Circumference is directly proportional to diameter (and radius). So therefore, tool position should be a linear relationship to the spindle speed. I would have thought that part was easy?

You just need to make sure you have a linear/linear output in whatever electro/mechanical transducer you choose.

Unless I have misunderstood?

Simon

Ok, you have a 2" bar to face with HSS. You want 100fpm. So you start at roughly 191rpm at the outside. By the time you are a 1/4" in you need to be going 254. 1/2" (you have 1"dia)its 382. Now you go in another 1/4, you have a 1/2" dia piece. You need 764rpm then in another 1/8" to get you to a 1/4" dia piece, you need 1528 rpm.
I don't think we need a graph to show whats going on....

Ew

Originally Posted by Ueee

Ok, you have a 2" bar to face with HSS. You want 100fpm. So you start at roughly 191rpm at the outside. By the time you are a 1/4" in you need to be going 254. 1/2" (you have 1"dia)its 382. Now you go in another 1/4, you have a 1/2" dia piece. You need 942rpm then in another 1/8" to get you to a 1/4" dia piece, you need 1528 rpm.
I don't think we need a graph to show whats going on....

Ew

Hi Ueee,

I must be doing something wrong:

circumference = pi x diameter (inches)

circumerence (in feet) = pi x diameter/12

Surface speed (fpm) = circumference (feet) x rpm

Transposing,

rpm = surface speed/circumference (feet), so for a surface speed of 100 fpm,

rpm = 100/circumference (feet)
= 100/ pi x diameter (feet), converting to inches we need to divide by 12
= 100/ (pi x diameter/12)

Which by all accounts is a linear relationship...


I have the following calculations based on 100 fpm

Diameter RPM

2" 190 (same as your figure)
1.75" 218
1.50" 254
1.25" 305
1.00" 381
0.75" 509
0.50" 763
0.25" 1527

Simon

Constant surface speed is a standard feature in CNC machines.

I do not want to put you off the idea. Just some things to think about when considering to implement constant surface speed on a manual benchtop home shop lathe (assuming a VFD spindle drive):

- you may not have the power required for it to work as expected. You may find a given depth of cut to work fine at the center of the workpiece, where spindle speed (and therefore motor speed) is highest. But towards the outer regions of say a 200mm diameter workpiece the motor speed will have to drop substantially to keep surface speed constant. VFD's are not constant power below base speed, they are at best constant torque, and even this only over a speed range of 1:4 or 1:6 depending on technology used. The spindle may suddenly stall when facing from the center outwards, as the motor power cannot keep up with the power required for the selected metal removal rate. Power is torque * speed. If the torque is constant, the motor only delivers half the HP at half the rpm. Of course, you could avoid this problem by always facing from the outside towards the center. But this wastes a lot of time, as you have to manually retract the cross slide after every pass. You can also avoid this problem by using a several times oversized VFD/motor, but that is expensive for home use only.

- on light home shop lathes I found that following industrial surface speed tables does not work well. Different tools, different or no coolant application, often unknown workpiece alloy, much less power available, much less machine rigidity......
I personally have the speed control knob strategically located on top of the headstock where my left hand rests, the finger on the speed pot. That way I can manually tweak the spindle speed on the fly. My brain has inputs from the eyes detecting surface finish and chip shape and rising smoke to correct speed and if necessary halt if chip gets tangled, my hand detects vibrations of the lathe and corrects speed accordingly- and if feeding the cross slide manually I get feedback how smooth the tool is cutting, my ears hear the lathe chattering so I adjust the speed to avoid chatter, my ears also hear if spindle speed is dropping indicatiog I reach the maximum motor power, my nose smells if things get burning hot...... It is a complex analog control loop. A large industrial CNC lathe does this automatically - it has vibration sensors to detect chatter, it constantly monitors motor power and servo motor torque, there are temperature sensors etc etc.

- just a multiturn pot on the cross slide may not do. You somehow need to limit the maximum speed when facing towards the center, or it will be infinite and possibly endanger spindle bearings and chuck. If you have a DRO on the lathe, you may find a way to access it's diameter information to control the VFD speed. There are also long linear potentiometers that may do the job, like used on audio mixing tables. But is that all really worth the effort just for constant surface speed? Would you maybe be better off CNC'ing your lathe for good??

Which is pretty much the same numbers.

Think of it like this. If the diameter halves the speed must double.

Oops my 1/2 dia speed is wrong....should be 764. (i'm working on pi being only 3.14)

You can make it a bit easier with the formula RPM= FPM / (dia/4) or even easier FPM/dia x 4 or even 4xFPM/dia (easy on a calculator)

you get-
2 200
1.75 228
1.5 266
1.25 320
1 400 (twice 2")
.75 533
.5 800 (twice 1")
.25 1600 (twice .5")
.125 3200 etc

Originally Posted by Ueee

The other is that the speed to position ratio is not linear, it is exponential.

I think you might have meant reciprocal.. ( which is exponential with negative exponents.. )

I can't see this working mechanically, I think it needs a processor. One nice thing to do would be to figure a way to get the DRO reading into the processor. Since it's going to be displaying diameter it would be perfect. I think also you could stick with open loop control and just output a 0-10v signal that scales according to some preset surface feet per minute.... depends on gearbox ratios and vfd configuration of course.

Zero changes to the lathe, just a bit of tinkering inside the DRO and a some cabling to the VFD.

Ray

PS.. I think it might be possible to fit the controller inside the DRO..?

Joe, have a read through this link

http://www.franksworkshop.com.au/Wor...o/LatheDro.htm

It is the DRO and controller I have on my lathe but I am only using the DRO part as I haven't installed the VFD/Motor that came with it.

As you can see it incorporates an SFM function.

Originally Posted by Ueee

Which is pretty much the same numbers.

Think of it like this. If the diameter halves the speed must double.

That part I understand. You had me confused when you mentioned it was an exponential relationship. Perhaps you meant an inverse relationship?

Cheers,

Simon

Originally Posted by Big Shed

Joe, have a read through this link

http://www.franksworkshop.com.au/Wor...o/LatheDro.htm

It is the DRO and controller I have on my lathe but I am only using the DRO part as I haven't installed the VFD/Motor that came with it.

As you can see it incorporates an SFM function.

Thanks a lot Fred! I just spent an hour reading through his website. What a wealth of information!
I wrote to Frank and asked him for some more detail and inited him to join this discussion-gien that he has actually implemented what I'm imagining. He did it exactly as RayG suggested: made it part of his homebuilt DRO.
I'e asked him to share a bit more about his thinking when he developed the algorithm for the VFD control. I'd like to learn more about howm he integrated all the concerns cha_melbourne has considered.

Thanks for your detailed and considered contribution cha! All grist for the mill.

Originally Posted by jhovel

Thanks a lot Fred! I just spent an hour reading through his website. What a wealth of information!
I wrote to Frank and asked him for some more detail and inited him to join this discussion-gien that he has actually implemented what I'm imagining. He did it exactly as RayG suggested: made it part of his homebuilt DRO.
I'e asked him to share a bit more about his thinking when he developed the algorithm for the VFD control. I'd like to learn more about howm he integrated all the concerns cha_melbourne has considered.

Thanks for your detailed and considered contribution cha! All grist for the mill.

I'm glad you found my site useful. I've created 2 versions of the DRO/speed controller. One for the lathe, which Fred now has, and a simpler one for my mill.

Logically, there isn't anything too tricky about the set up.

Basically, you specify the SFM and the RPM is calculated from that, as you have already discussed. My lathe dro supported different ways of specifying the final RPM, you can read that on my web site.

The DRO talks to the VFD via an RS485 serial link. Speed is controlled by changing the frequency of the VFD - 50Hz is normal. I limit the frequency to 2 - 100Hz, although as CBA says above, low speeds are pretty useless for cutting, but sometimes it is useful to have the spindle turning slowly.

I also have an optical speed sensor on the spindle - the lathe had a metal disk with 20 cut outs. My mill has a black and white paper disk, the diameter of the top pulley (lots more slots). This was to determine the actual speed of the spindle.

The speed control on the mill is pretty simple - I always starts the VFD at the previous frequency, then checks the RPM of the spindle, and corrects based on that - it only does it once after about 3 seconds. This "correction" will take into account changing pulley ratios. It isn't perfect, but a side effect (bug) is if you press the Run button again, it will correct the speed again - there is no CSS on the mill.

I tried to make the lathe a little more advanced - it uses a PID loop to try and maintain the speed. This way, the PID loop automatically adjusts when the diameter of the work piece changes. Under light loads, it is there to simply set the speed, again automatically taking into account the pulleys. Under heavy loads, ie when the spindle starts to slow down, it will increase the VFD speed to try and compensate for the heavy load. This doesn't work very well - if the spindle starts to slow down then the load lightens suddenly, the spindle will spin very quickly until it slows down again via the PID loop. You need an intelligent operator to recognise the problem.

PID + VFD + RS485 doesn't work very well in this scenario, because of the low update rate (25Hz from memory, compared to 1kHz for real servos), but it seems adequate for the spindle. The PID control does make speed control easier, supporting soft start and soft stop, just by adjusting the target output speed each pid cycle.

Frank