power supply inductive load protector addon

[coppercone2]

So if you have a DC power supply with a maximum voltage, and you run a motor or solenoid on it that has back EMF, and you don't want to destroy the power supply (happened before to me from things that seemed innocent), you need to put some kind of protection on the power supply.

I was thinking about making a general 'addon' that is made to hook in between the screw terminals on the power supply, so it would be a piece of dielectric with 2 contacts and the component, so you can fit it on over the power supply to protect it, and use banana plugs ontop. It would be built like one of those earth-ground shorting strips.

My Sorensen power supply got destroyed by this before from a small solenoid.

Does anyone have a general idea of a fits all component that would generally work with most things? I am not sure what the best option is, I naturally go towards a diode but I wondered what peoples thoughts are on a good network or device to use for this purpose, maybe I did not think of something good.
RC in parallel with a diode (RCD)? just RC? some kind of varistor? Maybe all those together?

[T3sl4co1l]

Situational.

How are you overspeeding a motor, such that its EMF overvolts the power supply?  Well, obviously, don't do that...

It's possible that a solenoid coil or other inductive load generated EMI that caused failure, but the main load on the power supply is always positive and well defined in that case, aside from the nanosecond scale.  It's a common misconception that "back EMF" can damage supplies, but in fact the supply doesn't see anything wrong at all, just a current decreasing towards zero.

Tim

[PeteH]

Can use a diode to prevent backfeed and a TVS component on the device side to clamp and limit the peak voltage.... Continuously doing this though (regeneration) I would add a load resistor with a switch designed to absorb the regenerated power. (TVS would have lower average power capability)

[coppercone2]

Well I broke 2 supplies in 1 day with this solenoid, I think it must be responsible. I thought the first one was a fluke then the 2nd one happened and I got pissed off, it was a simple timer one shot circuit on a solenoid.

[floobydust]

One lab I worked at, where the engineers were working with large solenoids, had a room full of dead power supplies. Engineers just sneak them back to the supply shelf lol don't even label them as dead. I spent a day fixing them and ordering parts.

+ve voltage spike protection is usually a reverse-biased diode from the PSU output back to the raw DC at the filter caps. Like across C-E on a linear PSU pass transistor to prevent the  output voltage from being too high for backfeed. Aggressive regen can still kill it though.
-ve voltage spike protection is usually a reverse-biased diode across the PSU output, to handle inductive back-EMF.
If the PSU output is floating, it should have something to earth ground like 1MEG and 10nF so a back-EMF spike does not trash the common-mode insulation somewhere. If the solenoid has capacitance to its housing or the machine, it can have a back-EMF pulse wrt earth ground.

[coppercone2]

How do you feel about a varistor? Normally their bad for leakage, but with a power supply, does it matter? I have defiantly seen them fitted on schematic diagrams for motors I don't have.

Is it going to screw with sense ? The sorensen has sense, the lambda does not (in front). I suppose I should just try it out.

[Vovk_Z]

Does anyone have a general idea of a fits all component that would generally work with most things?

The answer depends on the commutation speed. The universal easy and cheap decision with a diode is for a low-speed start-stop mode.
And,  is a polarity the same always or it can be reversed?

[David Hess]

Just passively clamping the output voltage is not sufficient because the failure is caused by the voltage across the pass transistor reversing.  An anti-parallel diode across the pass transistor should prevent this by sinking the reverse current into the large input capacitors, but then if the input voltage rises too much, the power supply will fail anyway.  A voltage clamp or input crowbar would be needed then but this also causes issues.

One solution is to use a power supply which has a 2-quadrant output meaning that it can source and sink current.

[floobydust]

Working with solenoids, they are not just an inductor. They can have generator action due to the pintle or armature moving with their residual magnetism.
A mechanical system can kick the armature forward or backwards fast and the solenoid generates an unexpected impulse, of unexpected polarity.

You don't want HV (inductive) back-EMF stressing the solenoid's insulation either, and a clamp diode works but gives the slowest time for the B-field to collapse and a slugginsh solenoid action. Adding MOV/TVS can keep fast quenching times, as long as the energy seen is not too much for the part.
Whatever your solenoid current is, that is the peak current the clamp will see and should be rated above.

[Rod]

So if you have a DC power supply with a maximum voltage, and you run a motor or solenoid on it that has back EMF, and you don't want to destroy the power supply (happened before to me from things that seemed innocent), you need to put some kind of protection on the power supply.  I was thinking about making a general 'addon' that is made to hook in between the screw terminals on the power supply...  My Sorensen power supply got destroyed by this before from a small solenoid.

Sorensen's blog (and some of its operating manuals) recommend both a parallel flyback diode and a series diode https://blog.powerandtest.com/blog/prevent-an-inductive-load-from-damaging-your-power-supply  This would be applicable to protect any unipolar power supply driving an inductive load, and will be stable as long as it is operating in constant voltage (not constant current) mode.

[coppercone2]

I am thinking two diodes soldered across a pinned banana to banana interconnect block would form a good lab protection network, if fitted with banana receptacles

for the series diode you would need to be a little tricky though

[coppercone2]

Am I right in understanding that there are two cases

1) arc stability case
if you have a inductor in series with a switch/electrode, opening the circuit just makes the inductor try to bridge the gap and stabilize the arc. The power supply sees nothing wrong, even if there is a HV discharge some where upstream?
2) turn off supply case
if you turn off a supply connected to a energized inductor it is the same as plugging in a battery

Does just disconnecting the inductor damage the supply (other then the arc trying to complete the circuit and damaging contacts)? because the current has no reason to reverse if the supply is still on right?

BUt then if you are using it and power goes out if you don't have #2 then bye bye supply

[Terry Bites]

So you want a wiget that tracks the set voltage and clamps anything above it for the duration of the over voltage?
I feel a TL431 on the way.
I dont know how you're killing psus. 

[coppercone2]

IDK either but it made me paranoid that a little solenoid can cause so many problems!

since then I Had made a bunch of banana jack D cell battery banks of different voltages to use for things like that! I am not going to be fucking dealing with those insane heat sink designs on these supplies again! hi 4th dimensional rubrix cube called

[johansen]

People say you cant unplug a stepper motor while the drive is powered up. They say you cant unplug a motor from a vfd during operation.

I dont understand it. The outputs are diode clamped to the rails.

In the case of old darlington transistor vfd's.. maybe the motor acts as a snubber, and unplugging it causes a high enough dv/dt to break something .

[T3sl4co1l]

Only way that a PSU would be damaged by powering down into a load, is if the load stores more energy than the output caps, and the PSU doesn't have clamp diodes at the output.

Normally the output rectifier, through the secondary winding, clamps at -1V or so.  Worst case I suppose more like 2-3V, and that would be for GaN sync rect.

It would have to be a very large inductor indeed for this to cause failure.  GaN most plausibly, but Si can take that for 100s of ms most likely, maybe even s.  (I've accidentally operated a prototype 12V 60A supply with SR disabled, i.e. body diodes passively rectifying.  Oh, so that's why the SR board was getting hotter faster than usual...)  And that's for SR; for passive rect (convection cooled), indefinitely.

Tim

[Doctorandus_P]

People say you cant unplug a stepper motor while the drive is powered up. They say you cant unplug a motor from a vfd during operation.

I dont understand it. The outputs are diode clamped to the rails.

I've also heard stories about motor drivers getting destroyed when a motor is disconnected while the system is powered, and I also do not fully understand why. This can be accidental, for example a bad screw connection or a breaking wire due to fatigue. Sure, the motor current has to go somewhere, and it will readily spark with a high voltage over an opened connector, but a high voltage over an (just) opened connector is in itself in a harmless place (it may damage the connector contacts a bit).

I guess that the main cause is the very high di/dt value, which induces secondary voltages in other locations.
For example, if the motor has long cables, and the cut is near the motor, the inductance of the long cables probably gets fed back into the motor driver.

Now that I've written this down, and read it back again, this sounds as a quite plausible reason.

Only way that a PSU would be damaged by powering down into a load, is if the load stores more energy than the output caps, and the PSU doesn't have clamp diodes at the output.

With a high di/dt value, the buffer electrolytic capacitors do not have enough time to react and absorb the energy, and the energy goes through whatever path has the least impedance. High energy transients are in a world of their own and they need special considerations to mitigate their effects. For example TVS diodes (there is a lot written about those) loose effectiveness if they are connected via a T-junction to the track they need to guard, because the track between the T-junction and the TVS has it's own inductance. The other side of the TVS also has to be connected directly and solidly to (usually) the GND plane.

And this is just on a PCB scale. Now scale it up to several meters of motor cable.

[T3sl4co1l]

Meh, other than direct faulting, I wouldn't think that much of a cause.  Driving unterminated cable stubs would be one thing, but, they're unterminated anyway -- at high frequencies, the motor isn't much resistance, mostly inductance, with some core loss thrown in; if the VFD output isn't filtered already (another point against this possibility) it's basically already driving such a load, so presumably is robust against it (if not necessarily efficient into it ).

Another could be, the input side just isn't designed for a sudden shut-down, versus a controlled stop.  Maybe there's a DC front-end that doesn't stop as fast as it should, or there's overshoot at the inverter when its current suddenly drops (or spikes, or goes unloaded i.e. hard switching).  (Hard switching is likely a stretch, as synthetic sine output may encounter hard switching in normal operation.)  Which could be like, too much inductance (too long cable?) between input and output stages, or something.  Or erroneous filter inductors (or not enough damping thereon).

Still pretty stretch-y, but I mean, who knows, if we're talking about the entire space of possible sale-able "VFD" products, from and to anywhere on the planet... it might be more of a miracle that the thing's able to spin a motor at all, let alone any kind of slightly different load condition.

Tim

[coppercone2]

Well I got 3x 400W resistors, I guess I will just up the power and hopefully the power supply is happy with the electrolytic brush sparking . Still better then god damn polishing/grinding dust.

Thought it should be evaluated exactly how to stabilize this, because I guess with right lab PSU, you can try to do micro-tig welding or whatever, with inductance added.

[jonpaul]

Bonjour a tous ....

Seen a lot of these issues, worked on transient surpressors in avainoics, auto and HV ....

1/ The solenoid or inductive load stores energy in the magnetic filed, according to J= 1.2 L I exp 2 with L in Hy, I in amps.

A solenoid will have a different L depending on the slug in/out.

Estimate L and measure or guess max I at turn off, you have the stored energy.

2/ With the ckt break, the energy does not disappear but a high voltage is across the inductor as it maintains a continous  current flow.

As the rise is fast the suproession MUST be at the coil !

Normal practice si a damper diode across the coil, or a transorb.

MOVs have a soft clamp and limited life.

3/ Fine analog PSU for the lab eg Lambda, Power Designs, HP have a foward diode to block revers currents in the reguat and a shunt reversed dioe at the output.

Thus reverse extenal power or transients short thru the shunt diode and the regualtor cannot have a reversed bias.

4/ An external add on is OK for the shunt doide but internal placem of the fwd diode is need to plce within the regulation  loop.

Excellent ref is the 1980s Hewlett Packard app/tech note on power supplies.

Just the ramblings of an old retied EE....

Bon Chance

Jon
The Internet Dinasour

[coppercone2]

what is this tech note?

[jonpaul]

Rebonjour CopperCone: Reflecting on the HP PS notes and more details RE inductive load snubbers and transient protection of DC power supplies:

1/ REF BOOK: The HP PS hdbk has 2 versions, AN-90A and B.
https://worldradiohistory.com/BOOKSHELF-ARH/Technology/Technology-General/HP-Power-Supply-Handbook-1970.pdf
 https://archive.org/details/DC_Power_Supply_Handbook_Agilent_Technologies_Application_Note_90B

2/Energy, Voltage:  The solenoid had max L when energized or slug is fully inside, Stored  Energy ~100 mJ..many J depends on size and L, peak current.
Actcts like a flyback or igniton coil: The stored energy rings as an RLC resonant and voltage can be many kV.

3/Simple Protection:    reversed diode in parallel to the coil clamps to the diode fwd V, and entends the solenoid release time, T= L/R.
An alternative is a simple R-C seties snubber that can reduct the rise time and peak voltage by dissipating the transient in the resistor.
~  R 10..100 Ohm, C 10..100 nF. NO MOV or TAZ needed!

4/ RE DC PS:  add (or replace) shunt and series protection diodes. Our Lambda and HP lab PSU have these, by design.  See figure.

Hope this is useful to you and others.

HAVE AN ABSOLUTELY FANTASTIC DAY!

Jon
The Internet Dinasour

PS: HP memory project has a fine library of app notes, tech noes :
https://www.hpmemoryproject.org/ressources/resrc_an_01.htm