zener clamping

[thefamilyman]

Good evening gentlemen

A client has been having problems with over voltage on their DC motor controller circuit and wondering how's the best way to implement zener clamping.

a bit of an overview on their basic system setup:

high powered DC motor ~35kW
Motor controller nominal voltage of ~470V DC
power is supplied by a 60KVA diesel generator running at 330V AC at 60Hz, cable run of about 70m of 25mm2 copper.
AC voltage is then fed into a DC rectification box; inside is a 150A 3 phase bridge rectifier with 2* 350V 2200uF capacitors.

the problem they have is that when the DC motor is turned off via the motor controller the DC voltage shoots up to over 550VDC which is dangerously close to the absolute maximum voltage of the controller.
the AC side of the rectifier does not change by much between pre-load, on load and directly after load. The diesel generator only overshoots for about half a second while the AVR counters the load, this could cause the DC to charge the capacitors to a dangerously high voltage of over 550VDC.

Can a high power zener diode(s) can be used to clamp the sudden surge to say a safer voltage of 530VDC while the generator stabilizes?
If so, what precautions of design considerations should be taken into account?

I hope this makes sense.

Many thanks
Patrick
the client is wanting to   

[langwadt]

It's hard to absorb 35kW instantaneous power from a zener.
I would use an IGBT+zener as switch, then dump all excessive energy to a big resistor.

that would most certainly explode, you never want to run an IGBT in linear mode

[thefamilyman]

It's hard to absorb 35kW instantaneous power from a zener.
I would use an IGBT+zener as switch, then dump all excessive energy to a big resistor.

thanks for the reply, i was also thinking of some sort of transistor switching like this too.

just a question, the 550VDC would only be the power stored in the caps as they DC motor is off and no current drawn through the motor controller, thus the power needed to be dumped by the zener is not 35kW right?
during the motor operation the DC and AC voltage is stable and a safe operating value, only until the motor is instantaneously stopped the DC voltage suddenly shoots up (due to the generator AVR not reacting fast enough and perhaps the inductance in the long cable run) and is stored in the large capacitors.

[langwadt]

that would most certainly explode, you never want to run an IGBT in linear mode

What makes you think it operates in linear mode? By carefully choosing resistors, it will operate mostly in switching mode.

gate voltage follows the supply so it'll be linear, there is not thing to make it snap on and snap off

[dom0]

Honestly when I read about >xx kW DC motors I already know the whole concept is probably garbage. Doctoring on issues of such a thing is throwing money away ; remedy the root cause and be done.

[langwadt]

It's hard to absorb 35kW instantaneous power from a zener.
I would use an IGBT+zener as switch, then dump all excessive energy to a big resistor.

thanks for the reply, i was also thinking of some sort of transistor switching like this too.

just a question, the 550VDC would only be the power stored in the caps as they DC motor is off and no current drawn through the motor controller, thus the power needed to be dumped by the zener is not 35kW right?
during the motor operation the DC and AC voltage is stable and a safe operating value, only until the motor is instantaneously stopped the DC voltage suddenly shoots up (due to the generator AVR not reacting fast enough and perhaps the inductance in the long cable run) and is stored in the large capacitors.

if the generator is generating 35kW and you suddenly stop using that I think the clamp would have
handle 35kW at least initially as the generator ramps down.

what about the energy in the motor?

[uncle_bob]

the problem they have is that when the DC motor is turned off via the motor controller the DC voltage shoots up to over 550VDC which is dangerously close to the absolute maximum voltage of the controller.
the AC side of the rectifier does not change by much between pre-load, on load and directly after load. The diesel generator only overshoots for about half a second while the AVR counters the load, this could cause the DC to charge the capacitors to a dangerously high voltage of over 550VDC.

Can a high power zener diode(s) can be used to clamp the sudden surge to say a safer voltage of 530VDC while the generator stabilizes?
If so, what precautions of design considerations should be taken into account?

I hope this makes sense.

Many thanks
Patrick
the client is wanting to

Hi

The problem you are running into is commonly called a "load dump". A bit of searching on Google under that term may help give you some background on the nature of the problem and common limits concerning what is to be expected. Your situation is from a motor load, but it (if I understand correctly) is a load none the less.

If all of your data is correct and I understand correctly, after the load is taken off line, your 4,400 uf of caps have 550 volts on them. You want to get it back to a more normal voltage.

Your AC is at 330V, so I'm guessing you full wave rectify that. If so 466 V would be a good guess at the normal no load DC voltage.

Net result is that the caps are about 90V above where they should be at no load. A magic circuit that dropped them 90V would take care of the problem. From Q=CV, you get 0.39 coulombs. That gets you 35 Joules of energy to be disposed of with a magic / perfect circuit. If you spent a second discharging things it would be 35W. If it pulled to ground (as it almost certainly must) you have about 200 joules to deal with.

Ok, but .. a second?? didn't we already fry the controller? Indeed you need to know how fast the transient that charges the capacitor is. It could (and likely is) a 150A pulse for a lot less than a second. The key issue is that the energy involved is the same if it is fast or slow. The peak watts change. The watts averaged out over a second do not change.

So ... an answer ... A resistor that will handle 150A (for a short while) at 460V. A 3 ohm part is about right for the job. You need a switch on the bottom end that turns full on quickly (FET or whatever) If it has a 1V drop, you will get a very brief 150W pulse of heat into it. If it is over in 50us a surprisingly small heatsink may work fine. Things like bond wires and absolute current limits are not quite as easy to get around.

Yes that's only one way to do it. It makes a boat load of unsupportable assumptions. Given my track record for math errors today, it would be *very* wise to check my numbers.

There are a few issues like getting the switch turned off well before you get down to 460V. To some degree that drives you to a larger resistor and less current (You need the voltage to go up in order to detect the event). Net result is that the stuff above gets a bit easier rather than a bit harder. Switching losses in the FET do count, you may need a bit bigger heatsink ...details do not always work out in your favor.

Bob

[langwadt]

gate voltage follows the supply so it'll be linear, there is not thing to make it snap on and snap off

A sensitive IGBT can fully turn on (Vce<0.1*I*Rext) at gate voltage as low as 10V, so over the entire clamping range (580V to 480V, my set points) only 10% of the region it operates in linear mode.

there is feedback, when the IGBT turns on the voltage drops, lowering the gate drive
so it'll settle at some linear spot 

[thefamilyman]

thanks everyone for the replies.

I'm not an electronic engineer, so the concepts i understand just not the finer details so that's why i'm asking you guys for expert help.

I'll get the voltage and current curves from the client this afternoon and post them to show a bit more information.

I came across an IGBT from IR that's raded for DC switching (marked as a relay replacement) AUIRGDC0250, would this help?

[langwadt]

there is feedback, when the IGBT turns on the voltage drops, lowering the gate drive
so it'll settle at some linear spot

That's why there is a second zener across ground and gate. It sets non linear capping voltage.

all the that zener will do is keep you from blowing up the gate assuming it has the right voltage

[T3sl4co1l]

Surely this is more fundamental: a control problem?  Why does the motor have to shut off suddenly?  Why would you even want it to? (This situation being a prime example of why you wouldn't!)

Among passive solutions, it might be reasonable to inductively ballast the generator, by using a saturable reactor into an LC resonant tank.  With the right choice of values, it can be made to resonate just below line frequency, at nominal voltage; and therefore will draw little current.  Then, when voltage swells, the inductance drops, pushing resonance up to line frequency, sucking extra amps from the generator.  (Unless the resonant circuit has a heavy load, and even heavier construction, the current draw will be largely reactive.  I don't know the behavior of diesel generators, to be able to tell if this would help or not.  Or if the line frequency is even stable enough for this to work.)

Or if such a network is used as a pass element, then you have a ferroresonant transformer, which tends to regulate the output voltage (albeit proportional to line frequency).  These are extremely bulky and hot, for this kind of power level... but just saying, there may be a passive solution.

As for active solutions, I agree, you don't want to use an IGBT here, not with such a simple circuit anyway.  And it wouldn't work to use just one, with hysteresis (so it's only on or off), because you're just trading one load step for another: it can't just stay on, it has to turn off eventually!  It may end up working anyway (hysteretic control will result in PWM operation), or it may make things worse (pulsating load upsetting the generator, regulator or governor?).

If the controller will deliver full power at a lower voltage, I'd suggest simply turning down the whole thing.  Adjust the generator regulator, or add a transformer.

Tim

[thefamilyman]

Surely this is more fundamental: a control problem?  Why does the motor have to shut off suddenly?  Why would you even want it to? (This situation being a prime example of why you wouldn't!)

Among passive solutions, it might be reasonable to inductively ballast the generator, by using a saturable reactor into an LC resonant tank.  With the right choice of values, it can be made to resonate just below line frequency, at nominal voltage; and therefore will draw little current.  Then, when voltage swells, the inductance drops, pushing resonance up to line frequency, sucking extra amps from the generator.  (Unless the resonant circuit has a heavy load, and even heavier construction, the current draw will be largely reactive.  I don't know the behavior of diesel generators, to be able to tell if this would help or not.  Or if the line frequency is even stable enough for this to work.)

Or if such a network is used as a pass element, then you have a ferroresonant transformer, which tends to regulate the output voltage (albeit proportional to line frequency).  These are extremely bulky and hot, for this kind of power level... but just saying, there may be a passive solution.

As for active solutions, I agree, you don't want to use an IGBT here, not with such a simple circuit anyway.  And it wouldn't work to use just one, with hysteresis (so it's only on or off), because you're just trading one load step for another: it can't just stay on, it has to turn off eventually!  It may end up working anyway (hysteretic control will result in PWM operation), or it may make things worse (pulsating load upsetting the generator, regulator or governor?).

If the controller will deliver full power at a lower voltage, I'd suggest simply turning down the whole thing.  Adjust the generator regulator, or add a transformer.

Tim

the control of the motor is not up to me, the client has finely tuned the characteristics of the pump for their application.
the application of using a generator is for their test site.

Unfortunately they've put the pressure on us to develop something with in the next few days, these motor controllers are custom designed, built they cant afford to blow them up.
we're already changing the AVR in the generator from an analog to a fast reacting digital AVR to hopefully react faster but they still want something that can clamp down any high transient voltages.

I know it might be asking a bit much for you guys to help but we've all have had this experience when the client puts the pressure on.

[Jay_Diddy_B]

Hi group,

Where did the motor controller come from? and how was it meant to be used?

Typically the dc voltage will come rectifying 3-phase.

Common voltages are 400V and 480V line-to-line.

If you rectify these with a capacitor input filter you get 1.34 times the line-to-line voltage, under load. The capacitor voltage will peak charge to 1.41 x Vac (line-to-line) under no load.

For a 400V system, the no load rectified voltage is 400 x 1.41 = 564V d.c. (nominal). Add 10% for line variation and you are at 620V d.c.

For the 480V systems the corresponding numbers are 676V and 743V.

IGBTs normally come in 600V and 1200V devices, there are other voltages.

I would be using at least 800V worth of electrolytics in these applications, either 2x 400V or 4x 200V parts with ballast resistors.

The voltage rating of the motor drive doesn't seem right 

Regards,

Jay_Diddy_B

[thefamilyman]

Hi group,

Where did the motor controller come from? and how was it meant to be used?

Typically the dc voltage will come rectifying 3-phase.

Common voltages are 400V and 480V line-to-line.

If you rectify these with a capacitor input filter you get 1.34 times the line-to-line voltage, under load. The capacitor voltage will peak charge to 1.41 x Vac (line-to-line) under no load.

For a 400V system, the no load rectified voltage is 400 x 1.41 = 564V d.c. (nominal). Add 10% for line variation and you are at 620V d.c.

For the 480V systems the corresponding numbers are 676V and 743V.

IGBTs normally come in 600V and 1200V devices, there are other voltages.

I would be using at least 800V worth of electrolytics in these applications, either 2x 400V or 4x 200V parts with ballast resistors.

The voltage rating of the motor drive doesn't seem right 

Regards,

Jay_Diddy_B

Hi Jay_Diddy_B

It's all custom made, including the motor.
the motor controller is designed for around 470VDC, with an upper limit of 580VDC before damage.
the generators are custom wound and controlled to supply 3 phase AC to the 3 phase rectifier.
i didn't design the rectifier circuit but here's a pic of it

[T3sl4co1l]

Well, if this is just for testing... why don't you rent a lab supply?  A few kilobucks/month should be more than enough to cover things, and it'll get you through development.

Tim

[thefamilyman]

Well, if this is just for testing... why don't you rent a lab supply?  A few kilobucks/month should be more than enough to cover things, and it'll get you through development.

Tim

no onsite mains power hence the generators, and 470VDC @ 80A lab supply would be a bit hard to come by.

[T3sl4co1l]

Well if you didn't ask, it's no one else's fault... 

I've heard of industrial scale DC lab supplies before, though I don't know of a source offhand.  I'm sure they're out there!  Such a supplier may also have a fully portable option (i.e., including the generator, or at least offering support for use with your existing generator).

Tim

[Jay_Diddy_B]

Hi,

If you want to try a quick fix, try increasing the number of capacitors in the rectifier circuit. 2x 2200 uF 350V (in series) is only 1100uF. I would try adding another 3 sets, for a 4 in parallel, 2 in series arrangement. This will give you 4400uF.

Regards,

Jay_Diddy_B

[bills]

I used to work for a newspaper and the printing press had 5-150 hp dc v/s motors. we had a module made by
crydom that would shunt the back emf from the motors as they decelerated.

Even those would release the magic smoke , but the drives would still be ok. 







 

[uncle_bob]

Hi,

If you want to try a quick fix, try increasing the number of capacitors in the rectifier circuit. 2x 2200 uF 350V (in series) is only 1100uF. I would try adding another 3 sets, for a 4 in parallel, 2 in series arrangement. This will give you 4400uF.

Regards,

Jay_Diddy_B

Hi

I missed the "2200's in series" part in my math earlier.

If you plug 1/4 the capacitance I used (I assumed parallel) then all the energies and power's scale by 1/4. The current is still the same.  That gets a snub circuit down to a pretty reasonable size.

That said, for a test environment and never to be used again .... put in a bigger capacitor stack.

Bob

[thefamilyman]

morning.

last night during some tests i adjusted the stability of the generator's AVR and governor which has helped a lot with the overshoot but still has the overshoot.

looking into the rectifier box more, the caps are actually big 22,000uF 350V in series and a 100uF 800V Polypropylene cap across output.

as promised i've got the voltage and current curves for one of the tests.


the blue line is the current on the motor controller, max draw is 79A
red line is DC voltage measured at the output of the rectifier box. Nominal voltage of 480VDC, minimum voltage of 419VDC and max of 527VDC.
each division is 0.5sec.
as you can see on the red line at the end of the motor shutdown the voltage peaks to over 500VDC, the client wants to clamp that to around 500-520VDC to protect the motor controller.

[SeanB]

Have a look for inverter brake chopper resistors and the associated control gear. They are designed to do just what you are doing, dump the extra energy that a motor inverter dumps on the supply capacitors stopping a motor very quickly. Normally used on inverters where you do not have a few so you can share the DC bus, or where you are sharing but the energy returned is more than the other drives can use. They are available in assorted sizes from industrial inverter suppliers.

[dom0]

Well if you didn't ask, it's no one else's fault... 

I've heard of industrial scale DC lab supplies before, though I don't know of a source offhand.  I'm sure they're out there!  Such a supplier may also have a fully portable option (i.e., including the generator, or at least offering support for use with your existing generator).

Up to 200 kW cont. DC output or so: http://www.heinzinger.de/en/produkte/hochleistung.php

[thefamilyman]

Well if you didn't ask, it's no one else's fault... 

I've heard of industrial scale DC lab supplies before, though I don't know of a source offhand.  I'm sure they're out there!  Such a supplier may also have a fully portable option (i.e., including the generator, or at least offering support for use with your existing generator).

Up to 200 kW cont. DC output or so: http://www.heinzinger.de/en/produkte/hochleistung.php

time, availability and cost prohibitive.

[Jay_Diddy_B]

hi,

I think the best way to address this is to ramp the pump controller down, a speed sufficiently slow that the AVR on the generator can track the change.

With the abrupt stop, the total time is around 20 seconds. If you take the last second and stretch it 2 seconds, ramping the current to zero, you will have a total time of 21 seconds, but you would probably get the overshoot under control. You would start the tapering process when the transfer is 95% complete.

You may not need to do this with the LiPo battery pack. This is a work around because the generator is not a battery.

Any other methods would take too long to engineer.

I was thinking about a Buck regulator or a disconnect switch, but this would have to handle 80A for 20 seconds, not trivial.

Regards,

Jay_Diddy_B