Advice on low noise stepper motor driver

[Boscoe]

I'm currently using the DRV8825 in some prototype electronics that has a includes sensitive analogue imaging electronics. The noise level of the imaging system is in the 10s of μV and my stepper motor driver is injecting noise through, I think, the common PSU. Now, I did design this properly with a good LDO powering the imaging sensor, using ferrite beads and a good amount of decoupling however there's some noise coupling there.

I do have some ideas on how to improve this from the PSU architecture perspective however I'd like to explore the idea of a linear stepper motor driver driven by a high frequency PWM generated from the FPGA filtered to a sinusoid. This will drive a class AB amplifier stage or buffer for the stepper motor. The motor is a small NEMA 11 and operates at a reasonable RPM and under very light load so only consuming 1-2W.

Ideally, I want to get to around 2USD at 1k quantity for this circuit. I think the best route might just be some discrete transistors in a class B setup. Does anyone have any experience of this? Is it a fool's errand?

Thanks!

[moffy]

To solve a noise issue you need to identify the source, the stepper motor electronics/wiring/stepper motor and the means of coupling, here you are assuming the common power supply, which sounds reasonable. But you need to test the theory, can you run the sensitive circuitry from say a battery? and does that eliminate the noise? There are lots of modes for noise at such a low level to couple, inductive, grounding, resistive, magnetic and electrostatic, you want to be sure you are fixing the actual cause.

[MarkT]

I'd like to explore the idea of a linear stepper motor driver driven by a high frequency PWM generated from the FPGA filtered to a sinusoid. This will drive a class AB amplifier stage or buffer for the stepper motor. The motor is a small NEMA 11 and operates at a reasonable RPM and under very light load so only consuming 1-2W.

No need for the complexity of an FPGA, many small microcontrollers have flexible PWM generation and its easy to generate quadrature using DDS with table lookup.

I suspect you will need better screening, in particular a screened cable for the wires to the stepper may pay off.  Are the wires to the motor screened?  Are they twisted pairs for each winding?  If not fix that first.  Perhaps adding chokes on each of the 4 motor wires on the output of the DRV8825 will help?

[Martinn]

Maybe have a look at Trinamic drivers: https://www.analog.com/en/product-category/stepper-motor-driver.html
(WTF? Yet another company acquired by Analog)
They have acoustically silent motor drivers. While that is a completely different thing, check if it could help with your problem.

[David Hess]

Stepper motor drivers limit current by chopping the output to each phase.  If you add enough resistance to each phase with power resistors, or set the current limit high enough in the driver, then the chopping will stop.

[mawyatt]

Maybe have a look at Trinamic drivers: https://www.analog.com/en/product-category/stepper-motor-driver.html
(WTF? Yet another company acquired by Analog)
They have acoustically silent motor drivers. While that is a completely different thing, check if it could help with your problem.

Agree, we utilize Trinamic chips for our custom stepper drivers for camera/subject positioning down to micron levels. Trinamic CS wasn't good but the chips work well.

If one is interested in precise positioning, stay away from the TI DRV types, they have a horrible transfer function when micro stepping.

Edit: here's some Info and crude videos we did ~4 years back with the Trinamic Controllers.


https://www.photomacrography.net/forum/viewtopic.php?f=25&t=42140&p=265427&hilit=Trinamic#p265427

https://drive.google.com/file/d/1P-95Ic2P9tuIHU2EgoVpmc5Y9WFqGWoe/view?usp=sharing

https://drive.google.com/file/d/1Ld40wU80KtxJoZsmEUSpN_mol4mkPIoH/view?usp=sharing

https://drive.google.com/file/d/19wE-B-dMesEXNtEZ552h_VnpWOjA0n0B/view?usp=sharing

https://www.photomacrography.net/forum/viewtopic.php?f=25&t=41483&p=262271&hilit=Trinamic#p262271

We moving into nanometer positioning utilizing Piezo Devices with Custom Controllers if folks are interested.

https://www.photomacrography.net/forum/viewtopic.php?f=25&t=40510&hilit=Piezo+Stage

Best,

[Doctorandus_P]

I agree with moffy first most. You have to identify the source of the noise first before you start modifying the system. It could be anything, but likely it is a combination of different things that all contribute a bit.

A common cause is voltage drop over PCB traces, especially the GND return. Be careful about PCB layout, star grounding and which point you define as a reference for the "clean ground" (star) point.

Ferrite beads usually do not do much for frequencies below 10MHz, but of course it depends on the ferrite bead itself. Low drop voltage regulators may have a quite bad ripple suppression. It's possible you get better results with a simple "capacitor multiplier" with a single BJT.

And things like re-routing the wiring or twisting the stepper motor wires may also have an effect. If the stepper motor is mounted on metal, Capacitive coupling between the motor windings and the metal frame of the motor can induce an AC current though the frame, and create noise this way.

[mawyatt]

The above mentioned Trinamic Stepper Motor Controllers were all part of a custom imaging system leading to create GigaPixel chip images, and was quite involved.

Best,

[zapta]

> This will drive a class AB amplifier stage or buffer for the stepper motor.

If you keep your own driver stage, keep in mind that it needs to be a current source rather than a voltage source. Also, the efficiency will be very low because the coils voltage, with not in motion, is very low compared to the power rails voltages (to be able to deal with the EMF during motion). Also, the current is bidirectional so you will need either a split ground or two A/B stages, so it's not straight forward.

[Boscoe]

Thank you all for the replies. You’re right I need to isolate the source of the noise however I do think it’s shrewd to try and reduce its source, too. Moreover, it turns out the poor micro step performance of the DRV is hurting performance dramatically in my application. It seems a natural choice to go with the Trinamic solution. Looks ideal and the budget is there.