linear shunt failure / General Science and Electronics / Forums

Forums

4hv.org :: Forums :: General Science and Electronics

« Previous topic | Next topic »

linear shunt failure

1 2 3 4 next

Move Thread

LAN_403

IamSmooth

Wed Aug 19 2009, 04:50AM

Registered Member #190 Joined: Fri Feb 17 2006, 12:00AM
Location:
Posts: 1567

Oh well. I tried. My wind turbine was putting out something close to 1kw during a storm. The loaded voltage was somewhere near 200v. Then the grid went out during the storm. The unloaded blades took off and the open voltage rapidly rised to my 260v trigger voltage for turning on my mosfet linear shunt. I must have let some voltage get out of hand because some of the mosfets froze closed, and some froze open. It is a good thing I have a furling mechanism that prevents the rotor from spinning too fast.

I am open to any suggestions for a better solution to the problem before I decide to buy the companies shunting system. I would really like to build it myself.

I need a constant voltage source where I can set the voltage I would like out. Anything that tries to raise the voltage above my set point will get the current shunted through a dump load of resistors. In this case, if the voltage is below 260vdc the shunt does nothing. Once the voltage tries to rise above 260vdc I will start shunting however much current is necessary to keep the voltage at this point.

I am not knowledge in this subject, so if you are willing to help I will need some good pointers to references so I can read up on this. A basic schematic would help. If this needs me to program a chip then I think I will just have to bag the idea and go with the companies.

Carbon_Rod

Wed Aug 19 2009, 07:19AM

Registered Member #65 Joined: Thu Feb 09 2006, 06:43AM
Location:
Posts: 1155

You may wish to consider a Crowbar circuit + MOV:
http://en.wikipedia.org/wiki/Crowbar_(circuit)
and a resettable motor rated circuit breaker that fails open.

Then have an actuator reset the breaker with a fiberglass pole when the rpm indicator drops to usable levels for more than 1h etc.

What happened to your shear pin/plate?

Dr. Slack

Wed Aug 19 2009, 07:46AM

Registered Member #72 Joined: Thu Feb 09 2006, 08:29AM
Location: UK St. Albans
Posts: 1659

Thoughts

a) complicated is bad, especially if you am not knowledgable on this subject
b) linear in semiconductors is bad - heat needs to be lost and the data sheet lies by omission and obfuscation
c) there's a lot of heat to be lost, so parallel elements is suggested, maybe use that parallelism to simplify other stuff if possible
d) your application is really not critical for voltage regulation or cleanliness, there is absolutely no need to build something quiet or exact, let's use that sloppiness to simplify other stuff if possible
e) in your first post on this subject waaaay back, you worried about the instability of turning a shunt on, then off. So what? I failed to see why that was bad then, given the problem as set. I still don't think it's a problem now.

If you were building a load for testing power supplies, or a shunt regulator for delivering a clean rail, then you would have a much more difficult problem.

Consider the following possible schemes.

A) Hard bang-bang switch mode

Put a DC bus capacitor on the input of the inverter if it doesn't already have one. Have a 10A 260V load (a UK 3kW kettle element would be easy to water cool for example) switched in by a FET to shunt that caapcitor whenever the voltage rises above 260v, and off when it falls below 250V (applying some hysteresis to avoid accidently finding a destructive linear region for the FET) (from your figures, 10A appears to be more than enough to load the generator, if not, then use more).

Pros - insignificant FET dissipation, simple control, easy dissipation of the load, never exceeds 260v
Features - Switches on and off using the RC product and the size of the hysteresis window to control the switching rate / duty cycle
Cons - I really can't think of any

Ensure against failure with a crowbar thyristor and current limiting resistor in series, also in shunt to the capacitor, designed to trigger at say 300v if the switching shunt fails.

B) Progressively switched loads

Have a number of parallel paths, each with a small load (0.5A per step?). Arrange the first FET to switch hard on at 260V, the next at 265v, the next at 270v etc.

Pros - insignificant FET dissipation, simple control, easy dissipation for the loads, inherent redundancy if all of the parallel paths are actually independant
Features - voltage gradually rises as the load increases
cons - more silicon than A

C) Other

There are so many other ways of doing it, including building a fully fledged switch mode smooth load. Do you want to solve the overspeed problem in a simple way, or do you want to use the problem as a vehicle to kick you into learning about switch mode stuff?

IamSmooth

Wed Aug 19 2009, 12:57PM

Registered Member #190 Joined: Fri Feb 17 2006, 12:00AM
Location:
Posts: 1567

Thanks for the replies. The crow-bar would solve the problem of limiting the high side, but would require additional circuitry to turn the system back on. The goal is to limit the high side, but only by that much; one wants to allow the voltage to stay up to allow the inverter to draw power again once the grid is on. The crow bar will drop the voltage to zero and it will stay there.

Neil, I would like to use this as an opportunity to grow and learn something new. Method B seems interesting or the last method. I have a 4500W/18ohm dump load. I would like to build something that smoothly keeps the voltage down. How would I go about with a switch-mode device? Would this be the best solution?

EDIT:

I was thinking. If I had a comparator to check when my voltage exceeds my limit, and then used the on state of the comparator to open the FET, would this work? The comparator would oscillate, opening and closing the FET, keeping the voltage at one level.

Bored Chemist

Wed Aug 19 2009, 05:40PM

Registered Member #193 Joined: Fri Feb 17 2006, 07:04AM
Location: sheffield
Posts: 1022

I think you need an "amplified zenner" circuit (or rather, several to dissipate the power.)

For this application I don't think you need the R or C though you will need lots of zenners in series to make the "260V zenner" you want.
I don't know if the generator is AC or DC; for AC you will need a big bridge rectifier too.

Of course, I'm a chemist so I'm not sure how well this will work, but I think that you can get transistors that will dissipate 100 watts easilly enough and, even if you need 10 of them and some current sharing resistors, this is still not a very expensive circuit.
Is there anyone out there who knows what they are talking about who would care to comment?

IamSmooth

Wed Aug 19 2009, 07:14PM

Registered Member #190 Joined: Fri Feb 17 2006, 12:00AM
Location:
Posts: 1567

What about using this IGBT?

It has a rating of 500a/1200v and 2900W power dissipation.

If I used a comparator to turn a shunt on hard above 240v and off when the voltage falls below 240v should this not be able to handle 2000W for short periods of a few seconds?

What do those of you who are experienced with these devices think?

Steve Conner

Wed Aug 19 2009, 07:55PM

Registered Member #30 Joined: Fri Feb 03 2006, 10:52AM
Location: Glasgow, Scotland
Posts: 6706

If you look at the datasheet, like all IGBTs and 99% of MOSFETs I know of, it'll say "Not for linear use".

We've been through all of this with you before, why don't you go back and read those threads again. I even linked you to a schematic of an amplified zener shunt that worked for me, albeit at 24V.

IamSmooth

Wed Aug 19 2009, 07:58PM

Registered Member #190 Joined: Fri Feb 17 2006, 12:00AM
Location:
Posts: 1567

So even if I open it all the way instead of gradually turning it on it is still linear? I thought if the gate was fully open I use it to conduct the current as a single element.

Dr. Slack

Wed Aug 19 2009, 08:29PM

Registered Member #72 Joined: Thu Feb 09 2006, 08:29AM
Location: UK St. Albans
Posts: 1659

If you "open it all the way", then it's not linear. I think you mean "open up" as in throttle, as in Vgs >> Vth, but I'm not sure.

If you make it go through the linear region fast enough where Vgs ~ Vth, then you can read the pulse length curves across the SOA graph to see how long it can stay in the linear region dissipating lots before you have to get to the hard-on or the hard-off condition to stop the dissipation.

BTW, the gate being "fully open" is a rather odd and confusing way of describing what you are doing. It may help if you use the correct terminology, for several reasons. It enables us to communicate more quickly and clearly - having to use the right language helps you to understand what the part is doing - and being able to use the right language indicates to us that you are making progress.

When you say you want the switchmode load to load it smoothly, would you specify what you mean by smooth.

IamSmooth

Wed Aug 19 2009, 08:48PM

Registered Member #190 Joined: Fri Feb 17 2006, 12:00AM
Location:
Posts: 1567

I am getting my terminology mixed up. I want to use the IGBT, or MOSFET, in the non-linear region. Rather than my previous approach where I "throttled up", I want to hit the gate at the maximum value so it fully conducts. The voltage will rapidly fall within a couple of seconds. If this is not acceptable, I could run a 555 with the duty set to zero, and then set the duty to 50% (or whatever value is appropriate) if I need to conduct current. Would either of these approaches be reasonable?

1 2 3 4 next

Moderator(s): Chris Russell, Noelle, Alex, Tesladownunder, Dave Marshall, Dave Billington, Bjørn, Steve Conner, Wolfram, Kizmo, Mads Barnkob