Heated bed noise

Velvia

@achrn Is there a solution I can use in RRF?

dc42

@Velvia if Prusa uses bang-bang mode and it doesn't cause artefacts on their machines, it shouldn't cause artefacts using Duet/RRF with the same bed and bed heater either.

alankilian

@dc42 said in Heated bed noise:

@Velvia if Prusa uses bang-bang mode and it doesn't cause artefacts on their machines, it shouldn't cause artefacts using Duet/RRF with the same bed and bed heater either.

I read the article and it was pretty interesting.

They investigated and decided:

The bed acted like a loudspraker. (Seems reasonable)
If they ran at a switching frequency of 62.5 KHz, no one head it. (Seems VERY reasonable)
But the MOSFET overheated. (Hmmmm.... Maybe the gate drive capability and gate capacitance was unmatched?)
So they ran the bed at 30 Hz (noisy) but on every on/off transition, they did a 16-step 62.5 KHZ transition from 0% to 100% pwn and then left it turned on. The opposite on the falling-edge.

So, VERY interesting way to handle the situation.

I'm not sure it's the RIGHT way to get it working, but it didn't require hardware changes, so that's probably a super compelling reason to do it this way.

Velvia

Using M307 H0 B1 still causes it to make the sound. Can I use Q parameter (PWM) at some really high value? Like over 20k so I don't hear it?

infiniteloop

@Velvia said in Heated bed noise:

Can I use Q parameter (PWM) at some really high value? Like over 20k so I don't hear it?

Did you follow the link @achrn gave you above? The Prusa guys explicitly stated that high frequencies were no option due to the heavy thermal load on the MOSFET. Same is true for the Duet. You better go "subsonic" instead. Try Q (PWM) values in the range of 5 to 15. The Prusa way, using "soft edges", doesn’t work with the Duet.

achrn

I don't have a solution - I have a printer that uses a Prusa Mk2.5 bed (so it's pcb-type bed with magnets and steel sheet as for the mk3, but it's at 12V) and I don't have a problem with sound from the heated bed. However, I did recall that I knew of Prusa acknowledging sound-from-bed problems, and that the fix was 'interesting', so I tracked down the blog.

As to why my bed doesn't make sound:

might be the 12V issue,
might be because my bed is insulated with cork sheets underneath (which might provide enough vibration damping to kill it),
might be because my bed carriage has non-standard printed bearing holders (likewise),
might just be because my hearing is appallingly bad (my max frequency is low enough that I fail the medical for railway trackside workers).

However, I have no way of knowing if any of them will help you (and I wouldn't advocate going deaf as a solution).

Although playing with the frequencies didn't work for Prusa, I'm not sure that it won't work for Duet - I don't think it necessarily follows that because Prusa's mosfet overheated the Duet will too.

If that doesn't work you could try insulating the bed - I think that was actually a beneficial mod to the printer anyway. If you do, you might want to recalibrate your temperature sensing - but it's easy to trap a exposed-junction wire thermocouple under the magnetic plate and do that.

Velvia

@infiniteloop I did. I've tried everything between Q2 and Q300. Nothing helped, just changed the tone of the noise. However the low "clicking" sound from Q10-20 is better than the singing/whining from 100+

Velvia

@achrn Thanks for the input. I guess I'll just have to live with the low clicking noise from a low PWM.

infiniteloop

@Velvia said in Heated bed noise:

I've tried everything between Q2 and Q300. Nothing helped

I’m sorry to hear that you still hear your bed heater. At any frequency below 10 Hz, however, the bed as a potential speaker membrane is inaudible to human beings. The clicking you describe thus must have another reason - something mechanical, I guess. I can imagine balls in a bearing, springs, cables, connectors, a metal sheet on top, the threaded rods of the Z axis, or other parts which might be induced by the magnetic field or by a tiny movement of the bed. Admittedly, the source of the clicking is very difficult to detect - either one has to localise it using acoustic sensors (= ears), or by mechanical means, i.e. touching, holding or pressing suspicious parts by hand (or a finger). That’s a lot of trial and error, success not guaranteed …

Velvia

@infiniteloop It's definitively the bed. It's from where the cables are soldered to the bed.

achrn

@infiniteloop said in Heated bed noise:

At any frequency below 10 Hz, however, the bed as a potential speaker membrane is inaudible to human beings.

You are assuming the response is constrained to match the exciting frequency.

I can't think of a reason why a (say) 1Hz square wave couldn't trigger a ringing at an audible frequency once each cycle.

T3P3Tony

@Velvia what current are you running the bed at, have you tried the high frequency?

mrehorstdmd

It's not hard to understand why it makes noise. You're switching a large current through the traces/wires in the bed. That sets up a magnetic field. That field interacts with the field created by the magnets that hold the bed plate down. Instant speaker.

At low PWM frequency or bang-bang mode, it will click audibly each time the bed heater switches on or off. At higher frequencies it will start to sing.

MOSFETs have very high off resistance and very low on resistance, but gate capacitance and drive current limit the speed of the transition between on and off. If you try to drive it at too high a frequency, the MOSFET experiences that transition for a greater % of the time and the MOSFET heats up. If you try to go too high in frequency, it may never switch all the way off or all the way on and you'll burn up the MOSFET.

If you want it to be silent, remove the magnets...

infiniteloop

@achrn

I can't think of a reason why a (say) 1Hz square wave couldn't trigger a ringing at an audible frequency once each cycle

As a former violin player, I know how harmonics work. The „clicking“ @Velvia describes tells another story. @mrehorstdmd may well be right with his thesis that, at low frequencies, the clicks originate from the interaction of the bed’s traces with the magnets holding down the bed plate. Personally, I won’t exclude other mechanical sources of the clicks, especially if @Velvia locates their origin in vicinity of the soldering ports. I don’t wanna repeat myself, but in order to produce clicks at very low frequencies, you need moving parts getting contact.

Velvia

@T3P3Tony It's a 24v bed. It says it's a 120w bed, so 5 amps? Have not tried higher frequencies than 300

raftaman

@achrnCan I use your files for my prusa Mk 2.5s? Would they work with a Super Pinda installed? thank