Small SCRs with 1mA or less in holding current?

[Joerg]

Joerg a écrit :


Diodes' SOT363 HBDM60V600W or MMDT5451 and 2 x 0402/0201 resistors ?

Yeah, if I have to ... thing is, most contract mfgs shy away from 0201
and then you have to use 0402.

 

[Spehro Pefhany]

Yeah, if I have to ... thing is, most contract mfgs shy away from 0201
and then you have to use 0402.

For this sort of thing I almost always use networks rather than
individual resistors. Considerably less space and less mounting cost
at very little additional parts cost.


Best regards,
Spehro Pefhany

 

[Joerg]

For this sort of thing I almost always use networks rather than
individual resistors. Considerably less space and less mounting cost
at very little additional parts cost.

That is a good idea. If I can find uses for the remaining resistors in
there as pull-ups or something.

Probably I'll build some other self-latching gizmo for this one, looks
like there really ain't no <1mA hold curent SCRs.

 

[[email protected]]

We can't use 0201s (yet) but 0402 is the norm. We will soon be forced into
0201s for caps, though.

For this sort of thing I almost always use networks rather than
individual resistors. Considerably less space and less mounting cost
at very little additional parts cost.

Less space? Lower production cost (when that's counted), sure. Routing can be
a problem, though.

 

[Joerg]

If it's "extinguished" does the rail stay on? This implies original
triggering was due to the "desired" dv/dt?

T'is the idea, intentional dv/dt. But since the up-slope steepness is
not guaranteed there will be an extra "regular" trigger. And the rail
does stay on.

Only shallow thinkers have "several mA wasting away all day long" ;-)

Since I won't that means I can't be a shallow thinker, I assume.

 

[Joerg]

Specs? I know that's a naughty word around here. Minimum dv/dt
required to trigger. Further definition of 'extra 'regular" trigger'?
"Un"trigger specs?

Can't rely on dv/dt of the rail coming up. It can be a milliseconds or,
if contacts get dirty, tens of msec. My goal is that it shall work up to
100msec or more. If I did a Schmitt design it wouldn't matter.

So the extra trigger uses a capacitor and R feeding current into the
trigger input for a second or more, to be on the safe side.

"Untrigger": A digital circuit shall be able to issue a pulse as short
as 100usec which needs to open the SCR. This would be done via a FET
that goes in parallel to the SCR with its DS path. Of course this must
happen after the trigger current has petered out to below the trigger
threshold.

Maybe I'll just build a poor man's flip-flop. ICs are pretty much out
because of the 60V requirement.

These are the little things that make engineering fun

 

[Jamie]

Actually in this case dv/dt is even desired

It's supposed to come on when a voltage rail comes on and then remain
on, unless a uC deliberately "extinguishes" it but briefly shunting a
FET across the SCR.

I just can't have several mA wasting away all day long.

Ok, to save on space, how about using a GTO ? ETO's would work but
I find them mostly in large applications.

At least with a GTO, if you keep the gate voltage in the on state,
it'll keep the device switched in at low currents. Unless you have a
pulse transformer in there?

In any case, this could remove your need for the shunting fet and free
up some space to maybe enhance the GTO operation.

Jamie

 

[Jamie]

That would deplete the battery. There are periods of time where the load
draws very little and I can't have the SCR path burn off power that
whole time.





V can be up to 60V and I should be <1mA.

why not use a sol single package type RS flip flip to drive an open
Drain higher voltage Nmos?
It almost appears to me that you are doing a design that does not
involve a uP ?

I've done such cheap designs using a 555 timer but only for its
trigger and threshold latches to control a device. The trigger snapped
it on and
threshold input turned it off. It also works well for a ripple detector
in circuits where rush currents or voltages need to be acted on.

Jamie

 

[Joerg]

why not use a sol single package type RS flip flip to drive an open
Drain higher voltage Nmos?
It almost appears to me that you are doing a design that does not
involve a uP ?

I've done such cheap designs using a 555 timer but only for its trigger
and threshold latches to control a device. The trigger snapped it on and
threshold input turned it off. It also works well for a ripple detector
in circuits where rush currents or voltages need to be acted on.

Well, I've got a design now. Just a transistor-based flip-flop. Logic
chips are not so suitable for this function because of the 60V rail.

 

[Joerg]

Ok, to save on space, how about using a GTO ? ETO's would work but
I find them mostly in large applications.

At least with a GTO, if you keep the gate voltage in the on state,
it'll keep the device switched in at low currents. Unless you have a
pulse transformer in there?

In any case, this could remove your need for the shunting fet and free
up some space to maybe enhance the GTO operation.

I can't really keep the gate voltage in the on-state. But I've decided
on ye olde two-transistor flip-flop. Have to look for arrays now, to
make it small. But my wife also has some other jobs planned, in the
yard, involving a ladder and a bow saw and all that.

 

[Joerg]

You can make the holding current as small as you want by injecting a bias current into the gate.

But I ain't got the luxury of a continuous current

 

[Joerg]

As long as you want to keep the CONTEXT a big mystery, you're not gonna get a reasonable answer.

Not a mystery. I've mentioned several times in the thread that I cannot
waste power. Also, if you sink current into the trigger pin that is
dicey. While they do guarantee that the thing will trigger at a certain
current they do not guarantee that it won't at 50% or so.

Anyhow, I'll do it discrete flip-flop style now because there does not
seem to be any suitable device.

 

[Tim Williams]

Can you make one with "high voltage" JFETs? Maybe you can find some with
60Vds (most are 30). Minimum turn on threshold may be too low, don't think
you mentioned minimum supply requirement. Of course, the canonical
structure would use a PFET, but you could substitute a PNP so you only need
half the resistors, or something like that.

MOS is similarly an option if you can find one with really high Vgs(max).
Unlikely..

Tim

 

[Spehro Pefhany]

That is a good idea. If I can find uses for the remaining resistors in
there as pull-ups or something.

There are 1mm x 1mm 2-resistor networks. No leftovers.


Best regards,
Spehro Pefhany

 

[Joerg]

There are 1mm x 1mm 2-resistor networks. No leftovers.

Those would be nice. But I am afraid they won't want anything smaller
than 0402 in terms of pitch.

 

[Joerg]

Can you make one with "high voltage" JFETs? Maybe you can find some
with 60Vds (most are 30). Minimum turn on threshold may be too low,
don't think you mentioned minimum supply requirement. Of course, the
canonical structure would use a PFET, but you could substitute a PNP so
you only need half the resistors, or something like that.

MOS is similarly an option if you can find one with really high
Vgs(max). Unlikely..

Well, I used two transistors as flip-flops. Works.

The min voltage can be as low as 10V, max just under 60V. There are no
high voltage JFETs but lots of depletion mode MOSFETs. Only N-Channel
though.

 

[[email protected]]

Those would be nice. But I am afraid they won't want anything smaller
than 0402 in terms of pitch.

Aren't two resistor 0402 (metric 1005) arrays (metric 1010) 1x1mm?

 

[Fred Bartoli]

Tim Williams a écrit :

Can you make one with "high voltage" JFETs? Maybe you can find some
with 60Vds (most are 30). Minimum turn on threshold may be too low,
don't think you mentioned minimum supply requirement. Of course, the
canonical structure would use a PFET, but you could substitute a PNP so
you only need half the resistors, or something like that.

MOS is similarly an option if you can find one with really high
Vgs(max). Unlikely..

MOSFETs with a gate zener clamp will take care of the 60V...

 

[[email protected]]

Typical Joerg ;-)

What's with the Outhouse Excuse style punctuation?

Not Outhouse Excuse, Google Gropes. Idiots insist on using both.

 

[Joerg]

Aren't two resistor 0402 (metric 1005) arrays (metric 1010) 1x1mm?

Not quite

You need space in between and the landing area is supposed to be 1.3mm
long plus spacing to the next part. Using many resistors is also an
issue in some apps because they are only rated 50V. Tough this is not an
aircraft project so we may be ok.