Searching \ for '[OT] SCR failures' in subject line. ()
Make payments with PayPal - it's fast, free and secure! Help us get a faster server
FAQ page: www.piclist.com/techref/index.htm?key=scr+failures
Search entire site for: 'SCR failures'.

Exact match. Not showing close matches.
PICList Thread
'[OT] SCR failures'
1998\07\31@100120 by lilel

flavicon
face
* This message contains the file 'nt5esch.gif', which has been
* uuencoded. If you are using Pegasus Mail, then you can use
* the browser's eXtract function to lift the original contents
* out to a file, otherwise you will have to extract the message
* and uudecode it manually.


Attachment converted: wonderland:nt5esch.gif (GIFf/JVWR) (00012240)

1998\07\31@100123 by lilel

flavicon
face
Sorry for the OT subject, but where else can I find answers to such
difficult questions?

I'm having a peculiar problem with failures in SCR's.   The circuit I
am working with is, in general, an RC timer with a DIAC connected
from the capacitor to the gate of an SCR.  The whole thing is driven
by about 40 volts DC.  R is 1.5 megohm, C is 100 Uf, and the SCR is a
sensitive gate type.  The general idea is this is a timer that fires
the SCR after 70 to 80 seconds.


I've attached a GIF of the schematic - I can't
ever make heads or tails out of those ASCII schematics you guys use.

The problem is, the SCR's fail rapidly, the gate shorts to the
cathode.  At first I thought I'd shocked the SCR with ESD, so I used
my most careful antistatic technique installing the next SCR.  It
worked fine, twice.  I replaced it, that one worked better, three
times.  The next one only worked once.

My next theory was that I had exceeded DI/DT spec for the SCR.
Having little or no way to measure this or check it, I decided to
try another theory.  (Hey, if you lose your keys in the dark, try
looking under the streetlamp.  Even if your keys are not there, at
least you have enough light to see by!)

My next theory was that maybe the charge surging out of that that big
100 UF capacitor was frying the sensitive gate.  I put a 200 ohm
resistor in series with the Diac, and this version has gone 7 cycles
without failing - a record.

Is anyone familiar enough with SCR's to give me a clue as to why they
fail?   Is there a gate current spec that I should know about? How
the heck to you even measure or calculate gate current in such a
circuit?


-- Lawrence Lile

    "An Engineer is simply a machine for
     turning coffee into assembler code."

Download AutoCad blocks for electrical drafting at:
http://home1.gte.net/llile/index.htm

1998\07\31@123006 by Mike Keitz

picon face
On Fri, 31 Jul 1998 08:59:14 +0000 Lawrence Lile <spam_OUTlilelTakeThisOuTspamtoastmaster.com>
writes:

>I'm having a peculiar problem with failures in SCR's.   The circuit I
>am working with is, in general, an RC timer with a DIAC connected
>from the capacitor to the gate of an SCR.

>
>My next theory was that maybe the charge surging out of that that big
>100 UF capacitor was frying the sensitive gate.

This is almost certainly the problem.  WHen the diac fires, it will
saturate at maybe 1V or so, so the entire charge in the capacitor tries
to go into the gate.  There is so much drive that a sensitive gate SCR is
not necessary.  A standard-gate one should work as well, cost less, and
be more reliable.  In either case a series resistor is absolutely needed
to limit the current since the capacitor is so large.  A resistor from
gate to cathode on the order of 50 ohms is also commonly used.  It helps
prevent false firing from leakage currents as well as limits any negative
voltage transients at the gate.


I put a 200 ohm
>resistor in series with the Diac, and this version has gone 7 cycles
>without failing - a record.

Showing you're definitely on the right track.  However, many sensitive
gate SCR's have a peak gate current rating of only 200 mA, which this
circuit barely meets.  Keep the 200 ohm resistor, replace the SCR with a
standard gate one, add the shunt resistor from gate to cathode and it
should work well.

Note that good standard-gate SCRs often show a low resistance from gate
to cathode in an ohmmeter test.  They are not defective.

_____________________________________________________________________
You don't need to buy Internet access to use free Internet e-mail.
Get completely free e-mail from Juno at http://www.juno.com
Or call Juno at (800) 654-JUNO [654-5866]

1998\07\31@143815 by lilel

flavicon
face
Mike Kietz penned the following words of wisdom:

> >
> >My next theory was that maybe the charge surging out of that that big
> >100 UF capacitor was frying the sensitive gate.
>
> This is almost certainly the problem.  WHen the diac fires, it will
> saturate at maybe 1V or so, so the entire charge in the capacitor
> tries to go into the gate.  There is so much drive that a sensitive
> gate SCR is not necessary.  A standard-gate one should work as well,
> cost less, and be more reliable.

eu contrare (pardon my french).  I'm using an SCR rated at only 0.8
amps, and they just don't make standard SCR's that small.  I'm also
worried about another phenomenon.  I've seen cases where the Diac
doesn't fire, becuase it has a minimum trigger current that is not
met in this high impedance circuit.  The DIAC simply bleeds off a few
microamps, without firing.  I chose a sensitive gate SCR in case that
happens, so it still has a chance of firing anyway.

These things only cost US$0.056, so I doubt that a standard one would
really be cheaper anyway.


> In either case a series resistor
> is absolutely needed to limit the current since the capacitor is so
> large.  A resistor from gate to cathode on the order of 50 ohms is
> also commonly used.  It helps prevent false firing from leakage
> currents as well as limits any negative voltage transients at the
> gate.

Aha! looks like you hit the nail on the head!

My test circuit with a resistor in series (this time it's 110 ohm,
for no particular reason) has gone several hundred cycles without
failure.  I think that excessive gate current is definitely the
problem.

> Showing you're definitely on the right track.  However, many
> sensitive gate SCR's have a peak gate current rating of only 200 mA,
> which this circuit barely meets.


Aha!   Igm is the spec I'm looking for - maximum gate current - it
says it is  1 amp.  So with 32 volts on my cap, 200 ohms would give a
maximum possible current of  .16 amps assuming the diac is a dead
short.   Should be plenty safe.

A small resistor from gate to cathode would also be a good idea.  It
should improve noise immunity somewhat as well.  I think I'll use
that too.  Say 200 ohms from gate to diac, 50 ohms gate to cathode.



Ill build this and give it a test.

-- Lawrence Lile

    "An Engineer is simply a machine for
     turning coffee into assembler code."

Download AutoCad blocks for electrical drafting at:
http://home1.gte.net/llile/index.htm


'[OT] SCR failures'
1998\08\03@100537 by lilel
flavicon
face
Russell wrote:

> May need a little more information.
> 1.  Where lines cross, which ones are joined and which (if any) are
> not?

All lines are joined where they cross.  I should put nice big dots
there to create a more readable schematic.  Thanks for the suggestion

. 2.  What is the intended load of
> the circuit?

Coil1 is a holddown coil for an appliance.  It has a DC resistance of
about 480 ohms.  I am unsure of it's inductance.  It is being used as
an electromagnet.  The circuit is intended to operate once per
toasting cycle in the appliance.  It probably does have an inductive
kick.


> Q1 appears  to be intended to place R1/R2 across the input voltage
> BUT what is the function of "COIL1". Is it a load in its own right
> or some sort of a snubber or what.

The coil is the load.  The SCR turns off the coil by crowbarring it
(crude but effective.)  A mechanical mechanism then turns off the
power by actuating a switch after the coil lets go.

> 1.    Excess gate dissipation as you have already suggested.
> What is the DIAC trigger voltage?

32 volts.

> I have grabbed some specs for a typical scr which is rated at
> 400volts at 2.5amp average on current.

This one is 0.8 amps, 400 volts, gate dissipation of 1.0 watts for 10
uSec,  average gate dissipation of 0.1 watts.

>
> The gate has a rated dissipation of 0.5 watt for 20milliseconds (one
> mains cycle) and 5 watts peak. IF the 100uF was at 40volts when it
> switched energy is 1/2 x C x V^2 = 80 mJ = 80 milliwatt seconds. .
> IF this was dissipated lineraly over 20ms the average power would be
> 0.080/0.020 = 4 watts. About 10 times too high. The 200 ohm resistor
> over this time would dissipate about 2 watts (all this is really
> rough) so for my SCR you would still be over spec.

Since this is a "one-shot " application there is no average gate
power dissipation.  This thing can fire at a maximum rate of once per
minute, never more than 6 times in a row,  making the average
dissipation pretty low.  The peak dissipation is another thing,
however.

100 Uf at 32 volts:   1/2 x C x V^2 =  51 mJ = 51 milliwatt secs.
Quite a bit higher than 10 mW-sec.   With a 1K resistor,  and a 50
ohm from gate to cathode, I'm getting a peak gate energy
dissipation of roughly 12 milliwatt-secs.  Much better

> Why can't you make Rgate MUCH higher.


Good idea.  I talked this over with a Teccor engineer and he
recommended a 1k resistor in serires, as well as a 50 ohm resistor
from gate to cathode.  The 50 ohm resistor will prevent false
triggering from noise on a high impedance gate.




>
> A PIC would be a good timner here :-).
>
>


No doubt.  I looked seriously at a 12c508 and a 16c54B.
Unfortunately this circuit must be at the absolute minimum cost.  I
can implement it with an RC timer for $0.87 each in volume, with a
PIC for about $1.25.  (that's for a complete board!)

I'll probably use a PIC in the high end version, add a few features.
-- Lawrence Lile

    "An Engineer is simply a machine for
     turning coffee into assembler code."

Download AutoCad blocks for electrical drafting at:
http://home1.gte.net/llile/index.htm

1998\08\03@122239 by Mike Keitz

picon face
On Mon, 3 Aug 1998 09:03:25 +0000 Lawrence Lile <.....lilelKILLspamspam@spam@toastmaster.com>
writes:

>Since this is a "one-shot " application there is no average gate
>power dissipation.  This thing can fire at a maximum rate of once per
>minute, never more than 6 times in a row,  making the average
>dissipation pretty low.  The peak dissipation is another thing,
>however.
>
>100 Uf at 32 volts:   1/2 x C x V^2 =  51 mJ = 51 milliwatt secs.
>Quite a bit higher than 10 mW-sec.   With a 1K resistor,  and a 50
>ohm from gate to cathode, I'm getting a peak gate energy
>dissipation of roughly 12 milliwatt-secs.  Much better

Consider placing a diode from the capacitor to the SCR anode, so when the
SCR fires most of the energy in the capacitor will dump into the anode
instead of the gate circuit.  Also the capacitor will rapidly discharge
almost completely through the diode and coil when the power is off,
leading to a more consistent time should the user immediately start
another cycle.  To keep the diode reverse biased during the time cycle
though the anode voltage has to be always higher than the triggering
voltage.

Another possible probelm with adding a diode or other method to turn off
the drive right away after the SCR fires is that the inductive "kick"
from the coil could pull the SCR anode negative momentarily and turn it
SCR off prematurely.  A slow "phase control" SCR probably won't be able
to turn off in time but other types may if the coil's "Q" is decent and
the power supply is rather soft.

If more precise timing is needed look into the circuit with interlocking
PNP and NPN transistors or maybe even the old PUT or UJT circuits (if you
can still buy those).  With these the trigger voltage can be
approximately ratiometric with the supply voltage instead of constant, so
the time will vary less over supply voltage changes.  On the other hand,
in a toaster you would want a longer time at low supply voltage anyway.
Properly setting the ratio of trigger voltage to nominal supply voltage
may be able to compensate for the varying toasting time quite well.

_____________________________________________________________________
You don't need to buy Internet access to use free Internet e-mail.
Get completely free e-mail from Juno at http://www.juno.com
Or call Juno at (800) 654-JUNO [654-5866]

1998\08\03@134000 by Reginald Neale

flavicon
face
Mike Keitz said:

much clipped

>the time will vary less over supply voltage changes.  On the other hand,
>in a toaster you would want a longer time at low supply voltage anyway.
>Properly setting the ratio of trigger voltage to nominal supply voltage
>may be able to compensate for the varying toasting time quite well.
>
In fact you can make a fairly effective regulator this way, with a circuit
that makes the "on" time for each half cycle slightly shorter when the
voltage goes up, and vice versa. For something like a toaster, you can make
the heat output relatively independent of line voltage over quite a range.
It helps to design for a reduced operating voltage, like say 80 or 90 VAC.

Reg Neale

More... (looser matching)
- Last day of these posts
- In 1998 , 1999 only
- Today
- New search...