over voltage mill shutdown ......

[niall]

i often think about this ..

but never got a simple  circuit running the way i wanted .....especially if the dc voltage is over 80 v , i know you can buy off the shelf units but some off these seem very expensive ...

the wardrobe has supplied the following ......a cheap lm 393 comparator....a hi v linear regulator to deal with 80 v plus  (hopefully)....and bits and pieces .....not very hi tech but i,d like to know the weak points ....



enter mosfets ....


the board has a relay but its way to small to emergency dump the high mill dc into a load ...... the comparator does have a very sharp switch on/off point ....with about 2 v gap ...at least i think it does ...

so if i added 5 big bruiser  fets ( have a good few of those ) driven of the small relay mounted on the aluminum plate ( potential heatsink ) with the correct power rating in parallel , would this sound problematic  .... a bit of a dog ?

i,ve tried pwm fets but its a hard one .....if  the   " all hell breaks loose "   an on /off  fet group via a ic comparator might  ... erm  " just shut  the thing down "

i suppose it,s driving the gates via a relay that seems interesting  ....i like relays ...theres something reassuring about them

nod to irc......solid state relays .....more fun

[ghurd]

I'll have a crack at it.

Is this going to stay shut down?  Manual reset?

What I am thinking about is 80V, then loaded with a large load will drop more than the 2V, and it will oscillate.  Relay will chatter to death.

If it will latch, finding a load that matches will be a problem.  It probably will not shut down completely, more slow down.  Cube square vs linear voltage vs fixed resistance?
Slow DOWN is still much better than blown UP.   

If it is going to latch, maybe add a 555 and a second high power / low resistance load 5 seconds after the 1st load comes on?  It would get it closer to 'shorted' without doing it all at once.

Parallel fets can be a problem.  They need derated for some extra safety.  And use gate ballast resistors.

If possible, best to have a separate load for each fet to get around a few of the issues.  The fet tabs should be isolated from each other (depends on the type of fet).

Not sure about this... contact bounce.  If the relay is fast switching, then the fet will be running faster, more cycles than expected, more time ramping the gave voltage down, more fet heat, etc.
Probably not much of an issue.

Not sure about this either... SSRs and lightning.  The SSR will be out there on the power wires.  I could imagine a regular old relay may have a better survivability % than an SSR.

This is what happens when I wake up too early,
G-

[niall]

thanks Ghurd...

i,m trying to incorporate some of your points in that i can .....if it works it may not be capable to be a fully shut down system....the mill is 48v and batteryless....a pwm unit will be the main controll .... i dont think i can do separate loads on the fets , the lm393 does have one unused op amp but i,m not sure if the existing heating resistors can be re wired to suit .... it does sound better to have a more balanced option ...

having some kind of extra control over the relay definitely seems worth it ...another dig for relays in the wardrobe it was then , the two grey ones are 60 second delay on ones ....the green one is interesting ...its timer can be set to delay on or off from 1/10 sec to 10 days ...i like that one ...but both these types need to be powered off to reset ...the delay timer would work well with a final short on a mill as you say...that would be an interesting option

i do have some 555 timers ...but am very unfamiliar with them 

i have some fets on now ...have lots of these 



mmm...about fets ..not sure if their de-rated  enough...their spec is 40A at 100v each ....the dump load may present an initial surge of 40 A ...maybe more 

the relay coils and comparator run at a regulated 24v which is (seems) to be working ok ...the mill is 48v ..so i hope any voltage fluctuations will be kind of above it ....

the relays are fun ...with a few different types it seems you can get a lot of combinations of switching and latching ...i hope the wardrobe approves when it,s switched on ..tomorrow maybe

otherwise the whole lot will wind up back in it again ....

 

[niall]

you were on the money Ghurd about the questionable behavior of the relay with a load that it would be comfortable ( or not ) with ...

for fun i hooked up the circuit on a 1k mill.....just the mill ,circuit  ,big capacitor bank and a 2k dump load...wind fairly good for testing

the mill quickly ran up to the 60v trip point and the relay kicked in the fets ...but it was a relatively fast switch rate , maybe 2 times a second so the mill didnt actually shut down ...i added the timer relay to set a 2 second hold on , but this would,nt  behave properly ....sometimes it missed its timing period ....dont know why

then big mistake time ...

i thought the second relay might have been underpowered  a little ( wrong ) so i adjusted the 24v voltage level pot.....(turned the pot the wrong way )....the relay went into a quick fast chatter mode and a 80v transorb across the source/drains on the fets smoked ......actually reddened more like ...its interesting  watching voltage that  " doesent exist "  materialize in a form that you can taste  .... ....

dont know anything about transorbs but they are pretty impressive ...this ickle one burned , spitted and hissed but refused to pop and had a bit of a go at shorting out the excess voltage


after i took out the timer relay and rewired the main one to latch on when it tripped ...this was simpler and much better , now the comparator would trip the relay and the mill would rumble down to about half its cut in rpm , then the relay would loose its power and drop out , the cycle would restart again ...much much slower switching ....i left it cycling that way for the rest of the afternoon

the fets seemed to run nice and cool throughout ....and oblivious to the transorb incident ...which was nice 

 

[ghurd]

"not sure if their de-rated  enough...their spec is 40A at 100v each ....the dump load may present an initial surge of 40 A ...maybe more"

Yikes!  Never ever run a fet at anywhere near what they claim as the max cont current.

If you have say 4(?), identical, and hopefully from the same production batch, then can try paralleling them.
Not exactly what you are doing in this sketch, but the paralleling thing is the same concept,
G-

[niall]

hi Ghurd

yes thats basically the circuit...i have 5 in parallel with a 10R  resistor on each gate ....if their full rated at 40A  each (and if they share relatively equal) that would bring down the potential A to about 5A each ...i think/ish  ..

the mill runs at between 15/ 20A 56v fully furled ....this would be flat out ...rare enough , but can happen ...

i think i was wrong about something earlier , i thought the transorb blew because of voltage spikes...its more likely that when i was turning (fiddling) the voltage power supply pot for the relay (fet driver) i knocked it completely off in the end for a few seconds ....this would be plenty enough time for the unloaded mill to jump from 60v to 80v ......

i,m leaving the trip circuit powered off for the time being  .....one afternoons testing in " good " wind is,nt really enough....interesting though         

 

[ghurd]

What voltage is the relay coil?  (Never mind, I see it is 24V)

I am thinking this is a slow switching circuit, or you wouldn't try a relay.
So probably can drive the fets from whatever (LM393?) is driving the relay?
Maybe increase the feedback on the comparitor, say 2V to 4~5~6V hysteresis, to slow it down a little more.

Then, even if it is switching kind of fast, still nothing mechanical to wear out.

==

I do like that "latching until there isn't enough voltage to hold it" relay idea.  Very old school stuff that works.

Can add diodes, resistor and a big fat giant cap (in parallel with the coil) to hold the relay On a little bit longer (probably help to change the relay to something with a smaller coil current).  Not sure its a great idea, but I have done it before.

G-

[niall]

by luck more than anything else the 393 circuit could drive the power fets directly ....the op amp output drives a small p type fet which switches on the hi side ....so this v could go directly to the 5 power fet gates ..i think

honestly i dont know why i put it in like that , i think i was just playing with a p type fet at the time...it seems to work ok

including relays into the loop might bring in extra options for switching loads which run of different voltages or systems ...switching low power mains ac might be an option 

as theres no battery on the mill the voltage drops out the latched relay at about 20v...a simple second relay could also trip a 24v wall adaptor to keep the first one fully latched ...possible full shut down ....all be it a strange one

i think the feedback resistor is fairly big maybe 1 meg ohm ....so that could be brought down a lot , (i,ll change that)  and a much smaller relay(s) could be used....i have some of those but no sockets for them ....i have lots for the bigger ones though

when i tried the latching option on the the mill it was definitely reassuring  ...you could hear it click in and out and see the contacts working ...