How to select a relay for short, very high currents?

[Rich Grise]

(my first post to this group)

I have a magnetizer project that needs to discharge 3x4700uF at 300V
into a coil via an SCR, resulting in a current of around 2000A for about
5ms. The complication is that the direction of current through the coil
needs to be selectable, which implies the use of a DPDT style relay. ....
Can anyone suggest a type of relay that would be adequate for this
project?
Also, since there would never be any current flow during switching,
would contact maintenance be an issue?

Any help or suggestions will be greatly appreciated.

How much room do you have, and how accessible is the stuff? You
could go retro:

http://www.cshunt.com/pdf/kpg10.pdf

Cheers!
Rich

 

[Frank Bemelman]

Wrench:
http://www.mcli.dist.maricopa.edu/show/interact/images/wrench.gif

Wench:
http://www.totallycracker.clara.net/images/wench.jpg

So what is wrong with the wench?

 

[Yzordderex]

Many fine ideas here.

Does your CV and your integral of current over time (S(i)dt) agree?
Have you calculated or simulated what this pulse is going to look like
when it's trying to drive a coil? The coil inductance needs to be
small or you're not going to get the pulse you desire.

I like the 4 scr approach. So that's what you should do. Heheh. Turn
one on one end of coil and pulse the other one into udder end. Don't
need any mechanical stuff but you could use a switch to select which
gates to fire. If your really hooked on the mechanical thing take a
look at some starter relays - Ford might be good. You may want to
pull one apart to see what it looks like inside. I think anything
mechanical is going to give you probs.

I once designed a large crowbar for a 480vac motor drive. I cleared
some JJS or JJN 50amp fuses (the 600v ones) and discharged a 1000uf
cap bank charged to 680vdc. I was able to get some 7000amp pulses
using a 90amp scr. Most of the energy went into wiring R, but still a
bit into the scr.

You should be able to get by with smaller parts. No heatsink if
repetition rate is slow enough. Keep in mind you want to hit gate of
scr with a fast pulse of current so it turns on fast. Otherwise you
will just dump all energy into the scr and fail it.

Sounds like fun project. Good luck with it.

regards,
Bob

 

[Tim Shoppa]

(my first post to this group)

I have a magnetizer project that needs to discharge 3x4700uF at 300V
into a coil via an SCR, resulting in a current of around 2000A for about
5ms. The complication is that the direction of current through the coil
needs to be selectable, which implies the use of a DPDT style relay.

Have you completely ruled out non-contact switching? Thyristors?

So if that wire doesn't turn into a strip of smoke, surely I can get by
with a modest relay costing $50-$100?

A 100 Amp (continuous) thyristor is typically rated for 1000 or 1500 Amp peak
for 8.3 or 10ms, which sounds right up your alley. Cost is typically
around $100-$150. But if you really are thinking "DPDT relay" you're
going to need two or four.

Tim.

 

[Winfield Hill]

Fred Bloggs wrote...

YIKES!!!!! How do you keep the EMP from destroying everything in the lab

Well, that's nothing compared to the 100000A sub-us risetime pulses
Bert Hickman makes in his quarter shrinker (=> 1MA in the quarter)!
EMP city! http://205.243.100.155/frames/frameindex.html

 

[Rob Storey]

Thanks for going to the trouble of drawing that Fred, but 5 expensive
semiconductors and associated triggering is way too complex.
One semiconductor and a decent mechanical switch is all that's needed. I
just have to find that switch.

Rob Storey.

 

[Rob Storey]

You could go with a reversing contactor set like the Albright SW64

http://www.bridex-ets.com.au/albright_sw64_series_dc_contactors.htm

Full ratings

http://www.bridex-
ets.com.au/albright_miniature_contactor_performance_data.htm

I believe these are the smallest that they make. They are used a lot in
small electric vehicles and should be relatively inexpensive. I'd be
wary of using them to hold off large voltages but they might be worth a
look at for your application. I rather suspect they'd last longer than
your capacitor.

Thanks, that looks the sort of thing I need, subject to ratings.
I also put this question to them on Friday, waiting for their reply.

Rob Storey

 

[Rob Storey]

Many fine ideas here.

Does your CV and your integral of current over time (S(i)dt) agree?
Have you calculated or simulated what this pulse is going to look like
when it's trying to drive a coil? The coil inductance needs to be
small or you're not going to get the pulse you desire.

I'm just duplicating an existing manual contraption, so the basic design
of the system and coil is already done, and works. I'm just trying to
also make it able to be controlled by a PLC.

I once designed a large crowbar for a 480vac motor drive. I cleared
some JJS or JJN 50amp fuses (the 600v ones) and discharged a 1000uf
cap bank charged to 680vdc. I was able to get some 7000amp pulses
using a 90amp scr. Most of the energy went into wiring R, but still a
bit into the scr.

You should be able to get by with smaller parts. No heatsink if
repetition rate is slow enough. Keep in mind you want to hit gate of
scr with a fast pulse of current so it turns on fast. Otherwise you
will just dump all energy into the scr and fail it.

Yes, active drive of gate with 250ma at 3.5v should do it.
The discharge rate will be low, once or twice a minute at most. I'm
expecting a hockey puck SCR rated at 700A continuous, clamped between
small 1/4" copper plates will laugh at it. I can always add a heatsink
if not.

Rob Storey

 

[Fred Bloggs]

Thanks for going to the trouble of drawing that Fred, but 5 expensive
semiconductors and associated triggering is way too complex.
One semiconductor and a decent mechanical switch is all that's needed. I
just have to find that switch.

Rob Storey.

That's what you say, but then again you don't much. Those
semiconductors, as expensive as they are, will still be humming along
while you're replacing your 10th set of switches with welded contacts.
Then there is the issue of production throughput and its associated
costs,hmmm and maybe the liability for treating burn victim users of
that hazardous knife switch. So the cheap buy-in switch approach may
just be a rip off.

 

[Fred Bloggs]

I'm
expecting a hockey puck SCR rated at 700A continuous, clamped between
small 1/4" copper plates will laugh at it. I can always add a heatsink
if not.

Rob Storey

700A continuous???? Sheesh, you could buy that whole SCR bridge+ TRIAC
at 125A and 400V parts for the price of that one ridiculously
over-specified part! So you blew your budget on that one part and have
to resort to buying cheap switches to do the rest of the job.

 

[Rob Storey]

Thanks to all who replied to my question.

The suggestions for 4 SCRs are not on. This is just way too complex, and
will be too expensive.
There is no need for rapid reversal as in a servo drive, as this unit
will fire once or twice a minute at most, and reversal may be even less
frequent. Also, such deviation from the design of the existing unit
(that I need to duplicate) may not be acceptable to the client.

All I need is a bloody big relay! I just need to find out *how* big.

My gut feeling is that anything with contacts considerably larger than
the piddly wire used in the existing system should be adequate. Double
that again, and it's got to be enough (I hope!).

Rob Storey

 

[Rob Storey]

700A continuous???? Sheesh, you could buy that whole SCR bridge+ TRIAC

at 125A and 400V parts for the price of that one ridiculously
over-specified part! So you blew your budget on that one part and have
to resort to buying cheap switches to do the rest of the job.

No, the existing unit uses a 230A stud SCR, which Farnell & RS list at
around AU$160, but luckily had none in stock. I found the 700A device
elsewhere for $85. I saved on my budget.

 

[John Popelish]

Thanks for going to the trouble of drawing that Fred, but 5 expensive
semiconductors and associated triggering is way too complex.
One semiconductor and a decent mechanical switch is all that's needed. I
just have to find that switch.

Have you looked at Blue Line switches?
http://www.uksolenoid.com/switchgear/control_panel_switches/

 

[Fred Bloggs]

Thanks to all who replied to my question.

The suggestions for 4 SCRs are not on. This is just way too complex, and
will be too expensive.
There is no need for rapid reversal as in a servo drive, as this unit
will fire once or twice a minute at most, and reversal may be even less
frequent. Also, such deviation from the design of the existing unit
(that I need to duplicate) may not be acceptable to the client.

All I need is a bloody big relay! I just need to find out *how* big.

My gut feeling is that anything with contacts considerably larger than
the piddly wire used in the existing system should be adequate. Double
that again, and it's got to be enough (I hope!).

Rob Storey

Stop preaching to us- any sensible person would be in total disagreement
with every last thing you say...and like your "bloody big relay" will be
free.

 

[Rob Storey]

Stop preaching to us- any sensible person would be in total disagreement
with every last thing you say...and like your "bloody big relay" will be
free.

Preaching??!!

 

[Yzordderex]

Gonna have to agree with Mr. Bloggs here.

Did you read what I told you about the 90amp scr doing 7000amps? I
would imagine you can buy a little TO-220 scr or one of those cute
little stud mounted jobs that would do fine. Probably no heatsink. A
25amp device should be plenty. Your 350ma at 3.5v of gate drive might
be just the right amount to smoke a big hockey puck. The Key here is
to smack the gate with a nice big pulse of current in a hurry. A
little MLC discharge into gate would be a nice choice here.

And yes, the scrs will be cheaper than a big relay. And the relay
will wear out while the scrs will keep on ticking.

Just my $.03

 

[Winfield Hill]

Yzordderex wrote...

Did you read what I told you about the 90amp scr doing 7000amps?
I would imagine you can buy a little TO-220 scr or one of those
cute little stud mounted jobs that would do fine. Probably no
heatsink. A 25amp device should be plenty. ...

Not for the parts I'm familiar with. Most would be hard pressed
to do 300A for short pulses, and a few can make 600 or 1000A, but
that's about it for the TO-220 crowd that I'm aware of. They all
have rather excessive voltage drops (>>10V) at higher currents.

 

[Rob Storey]

Gonna have to agree with Mr. Bloggs here.

Did you read what I told you about the 90amp scr doing 7000amps? I
would imagine you can buy a little TO-220 scr or one of those cute
little stud mounted jobs that would do fine. Probably no heatsink. A
25amp device should be plenty.

While searching for suitable SCR's, I didn't see any that would handle
pulses of 70-80 times their continuous rating!
20 times is more typical. For example, the existing unit uses a
ST230508POV rated at 230A continuous and 5970A non-repetivive. Sure,
it's overkill, and the hockey-puck (DCR604-14) even more so, but the
puck was cheaper. There's no way I would try a TO-220 for this job.

Your 350ma at 3.5v of gate drive might
be just the right amount to smoke a big hockey puck. The Key here is
to smack the gate with a nice big pulse of current in a hurry. A
little MLC discharge into gate would be a nice choice here.

You may be right there, 3.5v is a touch high, make that 2.8v. The DCR604
has gate ratings of Igt=150mA and Vgt=3v. I'll be pulsing a pnp
transistor to switch +24v to the gate via a 68R resistor with a string
of diodes to clamp the gate to 2.8v.

And yes, the scrs will be cheaper than a big relay. And the relay
will wear out while the scrs will keep on ticking.

The guys I'm doing this for were happy to spring for the relay
(contactors), so the price was very good!
We deferred to the advice of NHP's technical people, who specified two
32A four pole contacters (two poles paralleled on each) with a
mechanical interlock that allows only one to operate at a time. They
said the current was no problem, except that it could cause the contacts
to lift and arc due to the magnetic field around them. Their solution
was a latch that fits to the top of each contactor, locking it down when
it activates. A pulse to the latch later releases the contactor. More
complex than I'd like, but it's just four coil simple driver circuits
with some caps to fire pulses to the latch drivers.

Rob Storey

 

[Yzordderex]

Yes you are both correct - 25amp part will never do. Looks closer to
100A to get 5ms capability. Forced into module or larger stud
device$$.

Sounds like job well done anyway.

Point about gate drive I was making was charge a low Z cap up to 10v
or 15v
and dump through a couple of ohms into gate. External scr circuit
will of course control current risetime, BUT you gotta make sure you
get the gate on fast enough so current spreads nice across die. If
not turned on fast enough current crowds around gate fingers and you
smoke die. For my design with crowbar circuit and little external
inductance to limit rise time (say probably more than 1000A/uS)I have
no choice but to smack gate hard. For your design with slow rising
current, probably not so critical.

Don't think you need to clamp the gates with diodes. But hey, been a
good 20yrs since I played with the crowbar or any other scr.

Good luck Rob.

regards,
Bob

 

[Rob Storey]

Point about gate drive I was making was charge a low Z cap up to 10v
or 15v
and dump through a couple of ohms into gate. External scr circuit
will of course control current risetime, BUT you gotta make sure you
get the gate on fast enough so current spreads nice across die. If
not turned on fast enough current crowds around gate fingers and you
smoke die. For my design with crowbar circuit and little external
inductance to limit rise time (say probably more than 1000A/uS)I have
no choice but to smack gate hard. For your design with slow rising
current, probably not so critical.

Yes, hopefully so!
Other designs I've seen use a small SCR to dump a charged cap into the
gate.
With my cicrcuit, when the transistor turns on, there's 2.8v on the
gate, which is able to suck up to 300mA from it. The SCR has a max Igt
of 150mA, which leads me to think it won't need some large initial
current (like 2A or so, but maybe it does - something to look into for
future uses). In any case, as you say, with the inductive load in this
application, it should still be happy.

Good luck Rob.

Thanks!