Anyone tried these integrated DC servo motors?

I was thinking to replace my steppers with these servo motors instead of the common closed loop steppers. Has anyone used these to his laser?

A lot of people have tried servos and I donā€™t know about these, but they have timing issues of the servos being sluggish. Better description might be the machine lases before/after where it should.

Good luckā€¦ let us know if it works for youā€¦

:smile_cat:

Open loop is the normal cnc steppers, closed loop is with servos .
Iā€™m not sure if grbl can handle servos well but `a servo has a position feedback which allows it to remember where it was at any point .

Lasering does not really need servos which are more used in cnc applications where tools break and cause havoc.

I dont think lightburn is servo ready but the future is unknown .

Please report your progress

2 Likes

Iā€™ve not put them on a laser yet, but Iā€™ve worked with them.

Theyā€™re ā€œnewā€ in that itā€™s a BLDC motor with a very high resolution magnetic resolver on back, and a board running SimpleFOC.

But, what is it compared to a similar sized stepper?

Initially, the stepper appears to have significantly more torque, but not in a function way- the torque rolls off quickly, itā€™s down to about 20% by 1300 rpm. All laser motion controllers use ā€œtrapezoidal accelerationā€- basically, a fixed accel that doesnā€™t vary with speed. So when you put in your reduction ratio and set up the controller, you might set it up to go to only 800 rpm. At that point itā€™s only got half its holding torque, that sets your acceleration limit and you canā€™t make any use of the other half of its torque.

Steppers are almost all 200 steps/rev. Microstepping doesnā€™t actually work that well to increase resolution.

These new servos, you can program any steps/rev- like up to 16,384. Yes, you need to program them with an RS485 dongle.

Now, about 5 years ago or so ā€œclosed loopā€ steppers became the best. In addition to some performance improvements esp some better torque at higher rpms, they canā€™t ā€œlose stepsā€. Thereā€™s an encoder the driver reads and if forcefully stalled, it will remember how far off it is and try to get back there, although there is a correction limit which, if exceeded, could make the driver just go into an error state and stop moving, and express an ALARM pin which you can use to shut down the machine by, like, pull an interlock back to the OPEN state so the Ruida will stop.

Yes, you DO want these. Yes you do. But you may want to use them differently to get best results. What Iā€™d suggest:

  1. Direct-drive the axes! Do your calcs to match your axesā€™ linear speeds to the motorā€™s performance window by changing the pulley tooth count alone. In some setups, you may still want to couple it to another shaft so the pulley can sit between two fixed bearings so itā€™s not torquing the motorā€™s output shaft. In that case youā€™d still need to use a shaft coupler, but avoid a pulley reduction.
  2. Tie ALARM pins into interlocks
  3. Power with 24V.
  4. Be aware that when braking (it was going fast and is told to stop), it will generate back EMF that a power supply canā€™t just adsorb, it can overvolt. If thereā€™s other loads on the power supply, it might be consumed there. Otherwise you may need to add a simple circuit that will dump to a brake resistor. They do have onboard brakes built into the motors, but theyā€™re not very beefy- no room. The ihsv have stronger brakes than the xxAIMxx series, but still not a good idea to rely on.

Youā€™ve probably found the ihsv57 series (100W-180W). But check out the AIM series:
http://shop.smc-powers.com/index.php?route=product/search&search=aim

Secret knowledge: the 42AIMxx motors are the SAME as the larger 57AIMxx. Compare the specs. They just put the 42AIM innards into a larger case to work with existing 57 (NEMA23) motor faces. The 57AIM might dissipate heat better but I canā€™t back that up with facts.

1 Like

Your info is very dated. Closed-loop steppers took over >5yrs ago.

No problem at all, GRBL/Ruida donā€™t know the difference. Both closed-loop steppers and these new FOC servos just take the same step/dir signals. Most (but not all) of the closed-loop steppers used a separate driver. The FOC servos put the whole driver IN THE MOTOR. The motor has a plug for 24VDC and step/dir. Also can take RS485 or CANbus commands to simplify wiring. Ruida canā€™t put that out, and I havenā€™t heard if GRBL can yet, but that would actually be pretty easy to implement.

I donā€™t know your actual age, but if this is your first ā€œnobody needs this new fangled stuff!ā€ compliantā€¦ congrats on becoming old. I think my first moment like that was complaining about fuel injection and ECUs vs carbs and distributor sparks.

Lasers can get a lot of benefit from running these servos. They can run notably faster when cutting small curves or small vectors, like following text fonts. They can raster at higer speeds with better accel.

2 Likes

You seem to be a guru in this field!!!
Thanks for the great info. So they are not so simple as it seems, connect the step dir, the power and make some configurations. They need programming knowledge to make them work from the serial port?

There are some good & valid points here.
Our cnc router has servos - I love them.
All our lasers have steppers.
I watched some videos of how fast certain servos (or maybe closed loop steppers/easy servos) could accelerate a device, and I figured thatā€™s be great in engraving huge jobs.

Haha.

After much testing on two mahcines we had at the time, sure I could have the head running at 700 mm/sec, but only at the expense of some loss of fine detail as it moves faster than the PSU could fire single dots.

I eventually established that about 350 mm/sec was most efficient for speed plus accuracy for our needs. Going faster also led to more space being needed to speed up and slow down at each end of a scan. If I set the acceleration higher, you risked lost steps and gained other inaccuracies.

For cutting, youā€™re only moving relatively slowly - and there is not the resistance on a head as there is on a CNC routerā€¦ BUT I found I needed to go more slowly than ideal in order to not have corners overshoot, and circles develop wobbles, and some straights begin with dampened sine waves.
So although the machine can move faster, accuracy with slowness became my ideal - accuracy was more important than speedy slop.


When we had a big flat-bed, 2-tube laser custom made in China 6 or so years ago, I asked them to fit those easyservos instead of steppers.

They did, then they tested the machine, and set a complex path, and had it running for 48 hours. They do a pulse at the beginning, and one at the end, and then restart it after 48 hours and expect the same starting point and end point.
They did not achieve that, so they pulled those out and put standard Leadshine steppers in, and tested it and all was fine after two 48 hour stints.
They could not explain the easyservo failure - They blamed translation, but I feel inexperience was part of it - maybe it had not been tuned properly?

I paid for them and the 2 easyservos were shipped out with the machine - and the drives came with them too.

I begain to wonder if I could use a speedy one for X, and a stepper for Y.
They told me no, both had to have the same specifications or natureā€¦
I am not totally convinced. I was going to build a smaller one using those ā€˜spareā€™ parts - as an experiment - but Iā€™ve been too busy. I have all the parts needed - tube, laser PSU, spare controller and 24v PSU, and everything else.

Accuracy in small letters and curves is important - maybe the servos can do that better than the steppers canā€¦?

So after all these information, i think the most affordable, easy to install and reliable solution for now is closed loop steppers like leadshineā€™s, that are used on Thunderlaser machines.

The FOC servo will outperform closed loop steppers in every category. As always, you want to match the configurations with the motorā€™s performance curve.

When the torque of a closed loop is 2nm, but the torque of same size DC servo is 0.6nm, is it comparable or need to be same nm so bigger and heavier motor?

Good luckā€¦ Love you hear how you end up with this implementation ā€¦

You are aware that @Dannym has an RF machine and you have a co2 from what I can tellā€¦?

:smile_cat:

These are ready to work servos with integrated driver and encoder? Connecting them to ruida directly?

Didnā€™t know he has RF but I think also with CO2 i have benefits using servo. For accuracy for example, not loosing steps etc

I was thinking of speedsā€¦ most of these machines will outrun the response time of the tube/lps anywayā€¦ Thatā€™s why many find a slower scan speed results in better detail, at least with glass tube.

Generally, rf lasers have a larger diameter output beam, which results in a smaller dot size and the beam has a higher quality or M2 value.

RF tubes are digital devices, a glass tube is an analog deviceā€¦


I checked out servos. Although I had training on them 3 plus decades ago there was a bit to learnā€¦ same basic idea, a few decades of technology differences.

I found them to be pretty expensive, when used instead of plain olā€™ steppers. It is nice they can possibly recover. Iā€™ve run all of my time with the supplied stepper motors, even doing my share of abusing themā€¦

Mine has run at 1650mm/s, but itā€™s purely academic, what it does do it make for quicker jobs and faster response to changes. Acceleration is currently around 50,000mm/s^2ā€¦ itā€™s on the edge here, but rarely have any issues.

I also found that the ones I were eyeballing would not fit in the confines of where the X motor was, even though they were the same NEMA motor size ā€¦ The encoding cover made the case too long.

@Dannym advised they loose torque as they speed up, which as he advises would limit acceleration values to their lowest pointā€¦ slowing down a job even further.

You can wire it to your Ruida door or water protect circuit and the machine will stop ā€” for me too much effort and cost for what seems little gain.

Few people with hobby machines have themā€¦ less people that have opinions about them results in less input with issues.

Glad to hear from you when your project is completed and how it works out for youā€¦

Best of luck with this

:smile_cat:

2 Likes

was thinking of speedsā€¦ most of these machines will outrun the response time of the tube/lps anywayā€¦ Thatā€™s why many find a slower scan speed results in better detail, at least with glass tube.

In rastering, yes, an HVDC excited tube is limited by its response time. Some of the drivers cite 1ms response time, so giving it 1ms to ramp on, 1 ms to get an even burn, and 1ms to ramp off would be limited to 300hz and would still have 2/3rds of the dash being blurry edges. However, if you are just carving out block fills with a powerful laser, a higher raster speed is a very good thing even on HVDC excited. The sloping of the fillā€™s edges may not be noticeable.

Generally, rf lasers have a larger diameter output beam, which results in a smaller dot size and the beam has a higher quality or M2 value.

RF lasers do not have an inherently larger beam size. For a given wattage, itā€™s usually much smaller actually.

RF tubes are digital devices, a glass tube is an analog deviceā€¦

No. The time from when the Ruida PWM goes high and actual power output of the tube peaks is in the tens of microseconds range. In HVDC excited, itā€™s about 1ms. The RF should effectively produce at any duty, even <1%. The HVDC excited laser has no ā€œlow endā€, somewhere about 15%-20% duty the laserā€™s output just stops. Thereā€™s a window of like 1%-2% where the beam will just randomly turn on and off and wonā€™t functionally produce fractional power in this region.

I found them to be pretty expensive, when used instead of plain olā€™ steppers.
No. Check again. Note you donā€™t need to buy a separate motor driver. Donā€™t compare them by holding torque. Also note you may be able to direct-drive it which eliminates the pulley-and-belt redction

Mine has run at 1650mm/s, but itā€™s purely academic, what it does do it make for quicker jobs and faster response to changes. Acceleration is currently around 50,000mm/s^2ā€¦ itā€™s on the edge here, but rarely have any issues.
If youā€™re just cutting large rectanges, very little gain.

But vectoring out small details and text outlines are typically slowed WAY down below the commanded speed. And also note than the ā€œdead zoneā€ at the lower end of the HVDC excited tube means if you slow down too much and hit the tubeā€™s min power point, you will be forced to burn at higher watt-sec per mm because the tube output will die if you reduce power any further, and you canā€™t get the motors to drive the axis any faster. Higher accel can make jobs run faster and that can be almost like running on a higher wattage machine. Better than a larger machine, actually

I also found that the ones I were eyeballing would not fit in the confines of where the X motor was, even though they were the same NEMA motor size ā€¦ The encoding cover made the case too long.

That can be true. The motor has to put a magnetic encoder on the other end of the shaft, that takes up space. The FOC controller is usually located inside the motor and that adds a lot to the size. However, the comparison is a bit complicated apples vs oranges to decide what stepper is ā€œequivalentā€ to comare the lengths.

@Dannym advised they loose torque as they speed up, which as he advises would limit acceleration values to their lowest pointā€¦ slowing down a job even further.

STEPPERS lose torque rapidly as they speed up. You can enter a high max axis speed but low accel, OR high accel but low max speed. A servo generally allows you to use higher speeds AND accels.

You can wire it to your Ruida door or water protect circuit and the machine will stop ā€” for me too much effort and cost for what seems little gain.

If nothing ever goes wrong, then there is no gain. But itā€™s good to have a plan if a motor stalls or jams. This is esp true if you are tuning for speed and push for the edge of motor performance- which does diminish at higher temps, so a max tune at room temp may stall after 30 min or so of running.

1 Like

Yes, connect directly to the Ruida. As always, make sure you have configured both Ruida and motor for the same active edge.

1 Like

The RF laser is also CO2. Theyā€™re in aluminum or ceramic resonator bodies. I call the common glass tubes ā€œHVDC excitedā€ as thatā€™s most accurate, vs. ā€œRF excitedā€.

The higher accel and speed can yield major benefits on the HVDC excited machine too. High detail jobs can run faster

1 Like

What do you mean having the same active edge?
Also have you tried them? DO they have big advantage to knows steppers?

A motor drive can be set to move on posedge or negedge of STEP signal.

At first, it wouldnā€™t seem to matter. There must be an equal number of posedge and negedge in the long run. If the controller thinks itā€™s posedge but the drive moves on negedge, it seems like it would just step very slightly later.

But thereā€™s actual probs. The controller can only change the DIR signal far from the active STEP edge, otherwise the drive could be confused as to whether that is a forward or reverse STEP. So it often changes DIR on the inactive negedge of STEP. If the controller is set to posedge and drive is negedge, you can get weird behavior where the stepper seems to drift because every time it changes dir it can be confused about which direction the STEP is supposed to go.

Itā€™s a bit more confusing in getting the terminology right- we often use the motorā€™s opto input by tying STEP+ to a fixed +5VDC and let the controller pull down STEP-. The optoā€™s LED inside the drive will turn on, so you might want to call that the posedge, but the STEP- wire is doing a negedge on the oscilliscope vs ground but positive if itā€™s vs +5VDC.

Also have you tried them? DO they have big advantage to knows steppers?
Iā€™ve been using closed-loop steppers for years. The new FOC servos look great, I need to make a direct drive motor mount, add a 24V power supply and maybe braking resistance, and try it out.

Also, BIG FYI- these are part of the SimpleFOC revolution, and ANY BLDC can be a precision stepper! Add a small magnet on the unused side of the shaft, add a magnetic hall encoder board just past it, and a controller like the

ODESC 4.2

That has a STEP/DIR input.

BLDCs can be pretty torque-y, but they donā€™t advertise that number. If I have this right, I believe the equation is : max power*9.55 /max rpm.
max power is in watts
max power should have been reported at the max rpm, which should be the motorā€™s recommended max voltage * kv number. And a ā€œ6sā€ motor is intended for the voltage of 6x 3.7V lipo. This part can be a bit unclear, Iā€™ve seen a motor listed as ā€œ24V/36Vā€ and also ā€œ8sā€ (29.6V) and also ā€œmax voltage 41.5vā€. To calc torque correctly, I need the correct rpm the reported 1800W power was taken at, which would be voltage * kv. These voltages arenā€™t all that different but Iā€™d like to be more certain of the voltage thus the torque.

So, the N5065 140kv 8s motor- used in skateboards- is 1800W and should have a holding torque of 4.2nm(!!!) And would go to >4100rpm if it was asked to do it.

Thatā€™s massive overkill, but you donā€™t have to give it that much power. That motor is under 1/2 kg whereas a big nema 23 can be 2x that. You can potentially use the rotating back half as the output and saw the shaft short and put the magnet on that shortened shaft, mount the hall encoder PCB just past that, locate the driver apart from the motor, now the motor and encoder are ~70mm long 50mm dia.

And thereā€™s plenty of other motors around, Iā€™ve been looking for low KV ones for their better torque while avoiding any need for gearboxes or reduction belts. Theyā€™re less common, physical construction limitations mean a smaller motor is usually less torque, and theyā€™re used on small quadcopters which have small dia, high rpm props that donā€™t have a large torque load

1 Like

Picked this up from a number of people that use and sell themā€¦ I donā€™t have one, but one of the engineers stated itā€™s the main reason for itā€™s exceptional ability for detailā€¦ same lens ā€“ smaller spotā€¦ However Iā€™ll suppress this.

Must have misunderstood you hereā€¦ My limited training with these never really addressed the torque

Iā€™m standing there I can shut it offā€¦ if itā€™s lost is place ā€“ is the Ruida smart enough to continue the job with servos?

:smile_cat:

1 Like