DIY servo motor with wide speed range

[Gymnast]

I bought a Pfaff 230 about a year ago. It is a somewhat heavily built household sewing machine and can be considered a bit like a little sister to the Pfaff 138. Where household machines have typically been sold with motors with a maximum shaft power of 40 W, the Pfaff 230 was sold in Denmark with a larger motor with a shaft power of 80 W, which could make them sew at a maximum speed of approx. 1900 stitches/minute. According to German sources, the machine can withstand speeds of up to 2500 stitches/minute. Today's typical servo motors for industrial sewing machines have an input power of 550 W and a maximum shaft power of around 350 W.

Since I don't have room for a fixed table with a regular servo motor, I decided to build a smaller motor drive from scratch as a little DIY project to see what could be achieved. But at the same time cheap components were used. The idea was to achieve a wide speed range, and the result was that the difference between the lowest possible speed and the highest speed became a factor of approx. 600. Lowest speed was 2-3 stitches/minute and highest was 1780 stitches/minute. Most available servo motors that you can buy have a minimum to maximum speed ratio of about 50.

Most users here will probably think that the speed ratio of 50 is good enough, but maybe they haven't tried anything else. The reference may be the old clutch motors, which were more difficult to control. Here, for example, a seller who describes the sewing machine's lowest speed as super low:

My assessment would be that the lowest speed here is far too high. At the same time I don't like how the machine jump-starts with no warning. This is a video with a demonstration of my DIY motor drive:

I have now tried using the machine for a few projects, and I think the low possible speed is an advantage. This means that you get full control of the needle position via the pedal. A good pedal is also necessary to be able to achieve this. At the same time, the sewing machine can also sew quite fast.

[kgg]

4 hours ago, Gymnast said:

I bought a Pfaff 230 about a year ago.

I viewed your video and would be interested in seeing the motor and pedal setup.

kgg

[Gymnast]

9 minutes ago, kgg said:

I viewed your video and would be interested in seeing the motor and pedal setup.

kgg

Hello kgg

My YouTube channel got some more videos with this sewing machine. 
This is a speed test with some remarks using the original motor: https://youtu.be/VLU_s1RGasA

This is more about the mechanics regarding the new DIY motor drive: https://youtu.be/Ig69cM_0w7Q 

The pedal is like you see and attached to an air pressure sensor. Some electronics and an Arduino microcontroller is involved as well, but I have not yet made any "nice" documentation of that yet. To my knowledge, the important thing about the pedal is, that you need to have a large range of pressure applied from minimum speed to maximum speed of the machine. I use a pressure of 4 Newton (0.9 pounds) for minimum speed and 40 Newton (9 pounds) for max speed. So you got a factor 10 in pressure needed that allows for precise and fast control by the foot. The possible length of compression of the pedal have much less importance. When you using air as I do, you will have have a pressure change due to temperature changes (ideal gas law). Therefore a zero point adjustment are needed by some manual setting or by some software (I use software here). Furthermore the software implement some unilinear function, so you can get more accurate low speed control. I use parts of a parabola function. 

You are welcome to ask for more specific details, and I shall try to provide them.

 

[friquant]

I like it!

I saw this video a few months ago on youtube, and had not figured out who you were on leatherworker.

On 10/17/2024 at 10:13 AM, Gymnast said:

Furthermore the software implement some unilinear function, so you can get more accurate low speed control. I use parts of a parabola function. 

Using a variable frequency drive (VFD) and an AC gearmotor I have been able to get super low speed like you demonstrate. It would be convenient if the VFD had an option to add some parabola effect (or "exponential" as the radio control pilots call it) at the slow end. As it is I get the best low-end resolution when the max speed is on the slow side. Here's a link to a video in that thread: https://leatherworker.net/forum/topic/131403-my-first-vfd-w-3-phase-ac-gearmotor/#findComment-783471

On 10/17/2024 at 5:39 AM, Gymnast said:

Most available servo motors that you can buy have a minimum to maximum speed ratio of about 50.

As you built this system (or from your general knowledge) did you learn why it is that commercially available servo motors start at such a high speed? I'm still looking for a satisfactory answer to that. Or a way to adapt a commercially available servo to start at whatever speed I choose..

[GerryR]

14 hours ago, friquant said:

 

As you built this system (or from your general knowledge) did you learn why it is that commercially available servo motors start at such a high speed? I'm still looking for a satisfactory answer to that. Or a way to adapt a commercially available servo to start at whatever speed I choose..

Cooling, you can run a servo at very low speed (its just a motor), but you have to current limit so you don't over-heat it, and when you current limit, you lose torque.  By requiring a minimum start speed, you assure that someone can't run at very low speeds for any length of time and destroy the motor.  At that minimum speed, the cooling design parameters are met.  (A stepper motor is designed to remain stopped with the current being sufficient to hold the designed load and not destroy itself, though many systems drop to a lower current when stopped, as you don't usually need the full current to hold the load when stopped.  Many DC servo systems do the same when the load is stopped by current limiting.  That is not necessary on a sewing machine and would be more expensive to implement.)

[nejcek74]

At extremely low speeds the power of the motor is extremely low, so there is no point to engineer a motor that would be extremely powerful to overcome that. It's better to use a different gear ratio. In the production operators are skilled, and speeds are high. With a combination of needle positioner and other automatic controls low speeds are even less needed.

Outside of repairing stuff and hobby users there is not much need for such motors, we should be lucky for that what we have :))))

[Gymnast]

On 8/22/2025 at 12:47 AM, friquant said:

As you built this system (or from your general knowledge) did you learn why it is that commercially available servo motors start at such a high speed? I'm still looking for a satisfactory answer to that. Or a way to adapt a commercially available servo to start at whatever speed I choose..

My answer is, that it is a bit more expensive, to make a motor drive with good low speed control, and there is not sufficient demand in the market for that. EFKA do make expensive motor drives for sewing machines with a wider speed range.

In more technical terms, you got an issue with measuring the speed of the motor. For my drive I use another DC motor to measure speed, at it provide a very fast response.  But it got brushes that will wear out in time. The normal way is to use digital encoders of some kind. For good low speed control, you need a speed loop response of within about 1 ms. If you have got an optical encoder of 100 lines (as I have seen on some the servo motors), the minimum speed will be 10 rounds/sec or 600 rpm. You can get more expensive encoders with 1000 lines - and then you get minimum speed of 60 rpm. But some of these encoders can then get problems at high speed due to a high optical frequency. There is more ways to measure speed, but in general it is not that easy, and it can become more expensive to do it.

Another issue is how you control the winding current. The cheep way is to use three digital hall sensors in the motor, and it becomes a somewhat raw way of control the winding currents at low speeds. It is the old way of doing it for BLDC motors. More modern principles use the term Field Oriented Control, FOC, and they use other more expensive ways to measure the position of the rotor of the motor, and they use current waveforms near sine waves.

But I think the pedal could be improved without extra costs, and it puzzles me why no manufacturer have made a change here. You need to focus on the force, that your foot provides on the pedal, because with your muscles you can control force much better than a position. For me the lowest speed is with a force of 4 Newton (0.9 lbF) on the pedal and 40 N (9 lbF) on pedal at full speed. For some pedals it is like 20 N for lowest speed and 25 N for highest speed, and to me it makes the machine much harder to control.

[dikman]

I think your first sentence summed it up. The few manufacturers that need low speed (for leatherwork machines) overcome the issue with large handwheels and speed reducers. For everyone else the motors are fine for their requirements so no need to make them any more complex/expensive than they are.