Triac Efficiency...

On Wed, 07 Jun 2023 10:33:01 -0700, boB <boB@K7IQ.com> wrote:

On Wed, 07 Jun 2023 07:09:09 -0700, John Larkin
jlarkin@highlandSNIPMEtechnology.com> wrote:

On Tue, 06 Jun 2023 20:36:01 -0700, boB <boB@K7IQ.com> wrote:

On Mon, 05 Jun 2023 19:33:17 +0100, Cursitor Doom <cd@notformail.com
wrote:

On Mon, 05 Jun 2023 11:04:03 -0700, John Larkin
jlarkin@highlandSNIPMEtechnology.com> wrote:

On Mon, 05 Jun 2023 11:01:33 -0700, John Larkin
jlarkin@highlandSNIPMEtechnology.com> wrote:

On Mon, 05 Jun 2023 18:54:16 +0100, Cursitor Doom <cd@notformail.com
wrote:

On Mon, 05 Jun 2023 18:50:38 +0100, Cursitor Doom <cd@notformail.com
wrote:

On Mon, 05 Jun 2023 09:21:32 -0700, John Larkin
jlarkin@highlandSNIPMEtechnology.com> wrote:

On Mon, 5 Jun 2023 12:10:05 -0400, Phil Hobbs
pcdhSpamMeSenseless@electrooptical.net> wrote:

On 2023-06-04 23:05, John Larkin wrote:
On Mon, 05 Jun 2023 00:03:20 +0100, Cursitor Doom <cd@notformail.com
wrote:

On Sun, 4 Jun 2023 15:45:44 -0700 (PDT), Ricky
gnuarm.deletethisbit@gmail.com> wrote:

On Sunday, June 4, 2023 at 6:13:39?PM UTC-4, John Larkin wrote:
On Sun, 04 Jun 2023 22:34:00 +0100, Cursitor Doom <c...@notformail.com
wrote:
Greetings, gentlemen,

As we all know, to get maximum efficiency out of a MOSFET, it needs to
be driven hard and fast and spend as little time between fully on and
fully off as possible. Does that same principle extend to Triacs? And
if so, how can that best be implemented when one has to use AC to
drive the gate?

CD.
Once a triac fires, the gate drive doesn\'t matter any more.

But they do need a hard, fast drive to fire properly, long enough to
let the drain current rise above the latch limit.

AC gate drive will be slow-rise, which could just possibly damage the
triac if the load current is high.

That should be on the data sheets.

I believe that is why a diac is typically used in the control circuit.

I thought diacs were sometimes used because triacs don\'t fire
symetrically, but I could be wrong.
So, as regards AC drive, is a higher frequency drive signal a safer
bet? Also, the load is AC too, so the thing will turn off at every
zero-crossing point, which makes life more complicated. :-(

With an AC load, an AC gate drive is probably OK. The triac will turn
on before the voltage gets too high, so the spreading damage mechanism
won\'t happen much.

The bad case is when there\'s a lot of voltage across the device with a
lot of instantaneous current behind it, and it gets a slow, wimpy
trigger.

Post your proposed circuit and we can discuss it.

I still call the terminals anode, gate, cathode. I can never remember
what\'s MT1 and MT2. Top and bottom would be OK too.


The triac is one of those parts where folks get confused by fossilized
analogies.

A triac is a bidirectional thyristor, but it does not behave like two
SCRs in antiparallel. SCR triggering only works in one quadrant, for a
start, so two antiparallel SCRs would be a two-quadrant device, whereas
triacs trigger in all four quadrants.

The triggering also isn\'t that symmetrical, because the structures
aren\'t--the second and fourth quadrant triggering mechanisms are fairly
different, and the four-layer stacks aren\'t well separated, as they
would be in two SCRs.

Triacs are also very slow even compared with SCRs, which are pretty poky
devices by modern standards.

Someplace I have a paper that goes into the details of real triac
operation--iirc it\'s a good read, but I\'m not laying my hands on it at
the moment.

(I haven\'t actually used a triac since I was a teenager, so this is just
out of general interest.)

Cheers

Phil Hobbs

I haven\'t designed in a triac for maybe 15 years now. I used a couple
of triacs to soft-start a giant power supply that had a 60 Hz power
transformer. Switch in a giant resistor on the primary side, wait a
second, then short it out. That was hard on resistors but prevented
the occasional 1000 amp startup surge. And allowed a little C&K power
switch on the front panel.

The customer insisted that we not use a switching power supply!

Real relays are usually cheaper than triacs, easier to drive, and
don\'t need heat sinks.

Yes, but you can\'t fire them on and off anything like as fast and they
don\'t have the same longevity AFAIK.

Plus a relay capable of switching 16A is going to be a hog, take up
loads of space compared to a triac and cost a hell of a lot more, I
would imagine.

It\'s easy to look up. Digikey has 15 amp relays starting around 60
cents.

The triac will need a heat sink, which will be big and cost more than
the triac.

Relays don\'t need isolated gate drivers, TVS/MOV protection, or
snubbers either.


Well, you got me there!


I have seen and used snubber across relays. To extinguish DC arcs
across the contacts. Sometimes you have to increase contact life.

Do you know why a snubber extinguishes a DC arc?

Yes. Of course... An R-C-D snubber.

The C has been discharged by the R when the relay opens.
Then, the C, an electrolytic typically, (large-ish C) charges up
instead of the contacts taking the arc. The C is a short across the
contacts when the contacts open.

Then, the R discharges the Cap so it is ready for the next opening.

So the snubber prevents the arc but doesn\'t extinguish one. The R-C-D
will be polarity sensitive. And if the contacts cycle before the
large-ish C discharges, the snubber won\'t work.

What sort of RC values are good?
 
On Tuesday, June 6, 2023 at 2:43:22 PM UTC-4, John Larkin wrote:
On Tue, 6 Jun 2023 10:43:19 -0700 (PDT), Lasse Langwadt Christensen
lang...@fonz.dk> wrote:

tirsdag den 6. juni 2023 kl. 05.03.52 UTC+2 skrev John Larkin:

There\'s a shop in Fremont that will anodize infinite heat sinks.

it takes forever and the bill is infinite? ;)
Oh, don\'t get technical. I hate it when people get technical.

Actually, you don\'t often need an infinite heat sink. The thermal
spreading resistance eventually dominates.

https://www.dropbox.com/scl/fi/zf1a1lds726f2y2682v7z/Infinite_Sheet.jpg?raw=1

Yeah, that\'s how we get net global warming despite our radiative sink for heat (outer space) having
no end anywhere in sight, and even (in its distant parts) admirably low temperature.
Our polluted atmosphere has lessened the coupling for outgoing heat.
 
On 07/06/2023 3:09 pm, John Larkin wrote:
On Tue, 06 Jun 2023 20:36:01 -0700, boB <boB@K7IQ.com> wrote:

On Mon, 05 Jun 2023 19:33:17 +0100, Cursitor Doom <cd@notformail.com
wrote:

On Mon, 05 Jun 2023 11:04:03 -0700, John Larkin
jlarkin@highlandSNIPMEtechnology.com> wrote:

On Mon, 05 Jun 2023 11:01:33 -0700, John Larkin
jlarkin@highlandSNIPMEtechnology.com> wrote:

On Mon, 05 Jun 2023 18:54:16 +0100, Cursitor Doom <cd@notformail.com
wrote:

On Mon, 05 Jun 2023 18:50:38 +0100, Cursitor Doom <cd@notformail.com
wrote:

On Mon, 05 Jun 2023 09:21:32 -0700, John Larkin
jlarkin@highlandSNIPMEtechnology.com> wrote:

On Mon, 5 Jun 2023 12:10:05 -0400, Phil Hobbs
pcdhSpamMeSenseless@electrooptical.net> wrote:

On 2023-06-04 23:05, John Larkin wrote:
On Mon, 05 Jun 2023 00:03:20 +0100, Cursitor Doom <cd@notformail.com
wrote:

On Sun, 4 Jun 2023 15:45:44 -0700 (PDT), Ricky
gnuarm.deletethisbit@gmail.com> wrote:

On Sunday, June 4, 2023 at 6:13:39?PM UTC-4, John Larkin wrote:
On Sun, 04 Jun 2023 22:34:00 +0100, Cursitor Doom <c...@notformail.com
wrote:
Greetings, gentlemen,

As we all know, to get maximum efficiency out of a MOSFET, it needs to
be driven hard and fast and spend as little time between fully on and
fully off as possible. Does that same principle extend to Triacs? And
if so, how can that best be implemented when one has to use AC to
drive the gate?

CD.
Once a triac fires, the gate drive doesn\'t matter any more.

But they do need a hard, fast drive to fire properly, long enough to
let the drain current rise above the latch limit.

AC gate drive will be slow-rise, which could just possibly damage the
triac if the load current is high.

That should be on the data sheets.

I believe that is why a diac is typically used in the control circuit.

I thought diacs were sometimes used because triacs don\'t fire
symetrically, but I could be wrong.
So, as regards AC drive, is a higher frequency drive signal a safer
bet? Also, the load is AC too, so the thing will turn off at every
zero-crossing point, which makes life more complicated. :-(

With an AC load, an AC gate drive is probably OK. The triac will turn
on before the voltage gets too high, so the spreading damage mechanism
won\'t happen much.

The bad case is when there\'s a lot of voltage across the device with a
lot of instantaneous current behind it, and it gets a slow, wimpy
trigger.

Post your proposed circuit and we can discuss it.

I still call the terminals anode, gate, cathode. I can never remember
what\'s MT1 and MT2. Top and bottom would be OK too.


The triac is one of those parts where folks get confused by fossilized
analogies.

A triac is a bidirectional thyristor, but it does not behave like two
SCRs in antiparallel. SCR triggering only works in one quadrant, for a
start, so two antiparallel SCRs would be a two-quadrant device, whereas
triacs trigger in all four quadrants.

The triggering also isn\'t that symmetrical, because the structures
aren\'t--the second and fourth quadrant triggering mechanisms are fairly
different, and the four-layer stacks aren\'t well separated, as they
would be in two SCRs.

Triacs are also very slow even compared with SCRs, which are pretty poky
devices by modern standards.

Someplace I have a paper that goes into the details of real triac
operation--iirc it\'s a good read, but I\'m not laying my hands on it at
the moment.

(I haven\'t actually used a triac since I was a teenager, so this is just
out of general interest.)

Cheers

Phil Hobbs

I haven\'t designed in a triac for maybe 15 years now. I used a couple
of triacs to soft-start a giant power supply that had a 60 Hz power
transformer. Switch in a giant resistor on the primary side, wait a
second, then short it out. That was hard on resistors but prevented
the occasional 1000 amp startup surge. And allowed a little C&K power
switch on the front panel.

The customer insisted that we not use a switching power supply!

Real relays are usually cheaper than triacs, easier to drive, and
don\'t need heat sinks.

Yes, but you can\'t fire them on and off anything like as fast and they
don\'t have the same longevity AFAIK.

Plus a relay capable of switching 16A is going to be a hog, take up
loads of space compared to a triac and cost a hell of a lot more, I
would imagine.

It\'s easy to look up. Digikey has 15 amp relays starting around 60
cents.

The triac will need a heat sink, which will be big and cost more than
the triac.

Relays don\'t need isolated gate drivers, TVS/MOV protection, or
snubbers either.


Well, you got me there!


I have seen and used snubber across relays. To extinguish DC arcs
across the contacts. Sometimes you have to increase contact life.

Do you know why a snubber extinguishes a DC arc?

What an interesting question! I have only ever used RC snubbers on
contacts switching AC or as contact wetters for very low level DC
contacts. In all these years it has never occurred to me that a snubber
could extinguish a DC arc. The fact you allude to it happening makes me
wonder if something devilish to do with arc\'s negative resistance is
involved?

piglet
 
On Wed, 07 Jun 2023 10:57:48 -0700, John Larkin
<jlarkin@highlandSNIPMEtechnology.com> wrote:

On Wed, 07 Jun 2023 10:33:01 -0700, boB <boB@K7IQ.com> wrote:

On Wed, 07 Jun 2023 07:09:09 -0700, John Larkin
jlarkin@highlandSNIPMEtechnology.com> wrote:

On Tue, 06 Jun 2023 20:36:01 -0700, boB <boB@K7IQ.com> wrote:

On Mon, 05 Jun 2023 19:33:17 +0100, Cursitor Doom <cd@notformail.com
wrote:

On Mon, 05 Jun 2023 11:04:03 -0700, John Larkin
jlarkin@highlandSNIPMEtechnology.com> wrote:

On Mon, 05 Jun 2023 11:01:33 -0700, John Larkin
jlarkin@highlandSNIPMEtechnology.com> wrote:

On Mon, 05 Jun 2023 18:54:16 +0100, Cursitor Doom <cd@notformail.com
wrote:

On Mon, 05 Jun 2023 18:50:38 +0100, Cursitor Doom <cd@notformail.com
wrote:

On Mon, 05 Jun 2023 09:21:32 -0700, John Larkin
jlarkin@highlandSNIPMEtechnology.com> wrote:

On Mon, 5 Jun 2023 12:10:05 -0400, Phil Hobbs
pcdhSpamMeSenseless@electrooptical.net> wrote:

On 2023-06-04 23:05, John Larkin wrote:
On Mon, 05 Jun 2023 00:03:20 +0100, Cursitor Doom <cd@notformail.com
wrote:

On Sun, 4 Jun 2023 15:45:44 -0700 (PDT), Ricky
gnuarm.deletethisbit@gmail.com> wrote:

On Sunday, June 4, 2023 at 6:13:39?PM UTC-4, John Larkin wrote:
On Sun, 04 Jun 2023 22:34:00 +0100, Cursitor Doom <c...@notformail.com
wrote:
Greetings, gentlemen,

As we all know, to get maximum efficiency out of a MOSFET, it needs to
be driven hard and fast and spend as little time between fully on and
fully off as possible. Does that same principle extend to Triacs? And
if so, how can that best be implemented when one has to use AC to
drive the gate?

CD.
Once a triac fires, the gate drive doesn\'t matter any more.

But they do need a hard, fast drive to fire properly, long enough to
let the drain current rise above the latch limit.

AC gate drive will be slow-rise, which could just possibly damage the
triac if the load current is high.

That should be on the data sheets.

I believe that is why a diac is typically used in the control circuit.

I thought diacs were sometimes used because triacs don\'t fire
symetrically, but I could be wrong.
So, as regards AC drive, is a higher frequency drive signal a safer
bet? Also, the load is AC too, so the thing will turn off at every
zero-crossing point, which makes life more complicated. :-(

With an AC load, an AC gate drive is probably OK. The triac will turn
on before the voltage gets too high, so the spreading damage mechanism
won\'t happen much.

The bad case is when there\'s a lot of voltage across the device with a
lot of instantaneous current behind it, and it gets a slow, wimpy
trigger.

Post your proposed circuit and we can discuss it.

I still call the terminals anode, gate, cathode. I can never remember
what\'s MT1 and MT2. Top and bottom would be OK too.


The triac is one of those parts where folks get confused by fossilized
analogies.

A triac is a bidirectional thyristor, but it does not behave like two
SCRs in antiparallel. SCR triggering only works in one quadrant, for a
start, so two antiparallel SCRs would be a two-quadrant device, whereas
triacs trigger in all four quadrants.

The triggering also isn\'t that symmetrical, because the structures
aren\'t--the second and fourth quadrant triggering mechanisms are fairly
different, and the four-layer stacks aren\'t well separated, as they
would be in two SCRs.

Triacs are also very slow even compared with SCRs, which are pretty poky
devices by modern standards.

Someplace I have a paper that goes into the details of real triac
operation--iirc it\'s a good read, but I\'m not laying my hands on it at
the moment.

(I haven\'t actually used a triac since I was a teenager, so this is just
out of general interest.)

Cheers

Phil Hobbs

I haven\'t designed in a triac for maybe 15 years now. I used a couple
of triacs to soft-start a giant power supply that had a 60 Hz power
transformer. Switch in a giant resistor on the primary side, wait a
second, then short it out. That was hard on resistors but prevented
the occasional 1000 amp startup surge. And allowed a little C&K power
switch on the front panel.

The customer insisted that we not use a switching power supply!

Real relays are usually cheaper than triacs, easier to drive, and
don\'t need heat sinks.

Yes, but you can\'t fire them on and off anything like as fast and they
don\'t have the same longevity AFAIK.

Plus a relay capable of switching 16A is going to be a hog, take up
loads of space compared to a triac and cost a hell of a lot more, I
would imagine.

It\'s easy to look up. Digikey has 15 amp relays starting around 60
cents.

The triac will need a heat sink, which will be big and cost more than
the triac.

Relays don\'t need isolated gate drivers, TVS/MOV protection, or
snubbers either.


Well, you got me there!


I have seen and used snubber across relays. To extinguish DC arcs
across the contacts. Sometimes you have to increase contact life.

Do you know why a snubber extinguishes a DC arc?

Yes. Of course... An R-C-D snubber.

The C has been discharged by the R when the relay opens.
Then, the C, an electrolytic typically, (large-ish C) charges up
instead of the contacts taking the arc. The C is a short across the
contacts when the contacts open.

Then, the R discharges the Cap so it is ready for the next opening.

So the snubber prevents the arc but doesn\'t extinguish one. The R-C-D
will be polarity sensitive. And if the contacts cycle before the
large-ish C discharges, the snubber won\'t work.

What sort of RC values are good?

All very well for switching DC currents, but how to prolong contact
life when the current is AC?
 
On 07/06/2023 10:50 pm, Cursitor Doom wrote:
On Wed, 07 Jun 2023 10:57:48 -0700, John Larkin
jlarkin@highlandSNIPMEtechnology.com> wrote:

On Wed, 07 Jun 2023 10:33:01 -0700, boB <boB@K7IQ.com> wrote:

On Wed, 07 Jun 2023 07:09:09 -0700, John Larkin
jlarkin@highlandSNIPMEtechnology.com> wrote:

On Tue, 06 Jun 2023 20:36:01 -0700, boB <boB@K7IQ.com> wrote:

On Mon, 05 Jun 2023 19:33:17 +0100, Cursitor Doom <cd@notformail.com
wrote:

On Mon, 05 Jun 2023 11:04:03 -0700, John Larkin
jlarkin@highlandSNIPMEtechnology.com> wrote:

On Mon, 05 Jun 2023 11:01:33 -0700, John Larkin
jlarkin@highlandSNIPMEtechnology.com> wrote:

On Mon, 05 Jun 2023 18:54:16 +0100, Cursitor Doom <cd@notformail.com
wrote:

On Mon, 05 Jun 2023 18:50:38 +0100, Cursitor Doom <cd@notformail.com
wrote:

On Mon, 05 Jun 2023 09:21:32 -0700, John Larkin
jlarkin@highlandSNIPMEtechnology.com> wrote:

On Mon, 5 Jun 2023 12:10:05 -0400, Phil Hobbs
pcdhSpamMeSenseless@electrooptical.net> wrote:

On 2023-06-04 23:05, John Larkin wrote:
On Mon, 05 Jun 2023 00:03:20 +0100, Cursitor Doom <cd@notformail.com
wrote:

On Sun, 4 Jun 2023 15:45:44 -0700 (PDT), Ricky
gnuarm.deletethisbit@gmail.com> wrote:

On Sunday, June 4, 2023 at 6:13:39?PM UTC-4, John Larkin wrote:
On Sun, 04 Jun 2023 22:34:00 +0100, Cursitor Doom <c...@notformail.com
wrote:
Greetings, gentlemen,

As we all know, to get maximum efficiency out of a MOSFET, it needs to
be driven hard and fast and spend as little time between fully on and
fully off as possible. Does that same principle extend to Triacs? And
if so, how can that best be implemented when one has to use AC to
drive the gate?

CD.
Once a triac fires, the gate drive doesn\'t matter any more.

But they do need a hard, fast drive to fire properly, long enough to
let the drain current rise above the latch limit.

AC gate drive will be slow-rise, which could just possibly damage the
triac if the load current is high.

That should be on the data sheets.

I believe that is why a diac is typically used in the control circuit.

I thought diacs were sometimes used because triacs don\'t fire
symetrically, but I could be wrong.
So, as regards AC drive, is a higher frequency drive signal a safer
bet? Also, the load is AC too, so the thing will turn off at every
zero-crossing point, which makes life more complicated. :-(

With an AC load, an AC gate drive is probably OK. The triac will turn
on before the voltage gets too high, so the spreading damage mechanism
won\'t happen much.

The bad case is when there\'s a lot of voltage across the device with a
lot of instantaneous current behind it, and it gets a slow, wimpy
trigger.

Post your proposed circuit and we can discuss it.

I still call the terminals anode, gate, cathode. I can never remember
what\'s MT1 and MT2. Top and bottom would be OK too.


The triac is one of those parts where folks get confused by fossilized
analogies.

A triac is a bidirectional thyristor, but it does not behave like two
SCRs in antiparallel. SCR triggering only works in one quadrant, for a
start, so two antiparallel SCRs would be a two-quadrant device, whereas
triacs trigger in all four quadrants.

The triggering also isn\'t that symmetrical, because the structures
aren\'t--the second and fourth quadrant triggering mechanisms are fairly
different, and the four-layer stacks aren\'t well separated, as they
would be in two SCRs.

Triacs are also very slow even compared with SCRs, which are pretty poky
devices by modern standards.

Someplace I have a paper that goes into the details of real triac
operation--iirc it\'s a good read, but I\'m not laying my hands on it at
the moment.

(I haven\'t actually used a triac since I was a teenager, so this is just
out of general interest.)

Cheers

Phil Hobbs

I haven\'t designed in a triac for maybe 15 years now. I used a couple
of triacs to soft-start a giant power supply that had a 60 Hz power
transformer. Switch in a giant resistor on the primary side, wait a
second, then short it out. That was hard on resistors but prevented
the occasional 1000 amp startup surge. And allowed a little C&K power
switch on the front panel.

The customer insisted that we not use a switching power supply!

Real relays are usually cheaper than triacs, easier to drive, and
don\'t need heat sinks.

Yes, but you can\'t fire them on and off anything like as fast and they
don\'t have the same longevity AFAIK.

Plus a relay capable of switching 16A is going to be a hog, take up
loads of space compared to a triac and cost a hell of a lot more, I
would imagine.

It\'s easy to look up. Digikey has 15 amp relays starting around 60
cents.

The triac will need a heat sink, which will be big and cost more than
the triac.

Relays don\'t need isolated gate drivers, TVS/MOV protection, or
snubbers either.


Well, you got me there!


I have seen and used snubber across relays. To extinguish DC arcs
across the contacts. Sometimes you have to increase contact life.

Do you know why a snubber extinguishes a DC arc?

Yes. Of course... An R-C-D snubber.

The C has been discharged by the R when the relay opens.
Then, the C, an electrolytic typically, (large-ish C) charges up
instead of the contacts taking the arc. The C is a short across the
contacts when the contacts open.

Then, the R discharges the Cap so it is ready for the next opening.

So the snubber prevents the arc but doesn\'t extinguish one. The R-C-D
will be polarity sensitive. And if the contacts cycle before the
large-ish C discharges, the snubber won\'t work.

What sort of RC values are good?

All very well for switching DC currents, but how to prolong contact
life when the current is AC?

This app note is one place...

<https://www.dropbox.com/s/4ktoqpj0pqb4zfv/Littelfuse_Arc_Suppression.pdf?raw=1>

The advice to watch the intensity of the arc is cute!

My starting point for general purpose AC powerline switching is RC
snubber of 100R and 0.1uF and adapt values up or down from there.

piglet
 
On Wed, 07 Jun 2023 22:50:31 +0100, Cursitor Doom <cd@notformail.com>
wrote:

On Wed, 07 Jun 2023 10:57:48 -0700, John Larkin
jlarkin@highlandSNIPMEtechnology.com> wrote:

On Wed, 07 Jun 2023 10:33:01 -0700, boB <boB@K7IQ.com> wrote:

On Wed, 07 Jun 2023 07:09:09 -0700, John Larkin
jlarkin@highlandSNIPMEtechnology.com> wrote:

On Tue, 06 Jun 2023 20:36:01 -0700, boB <boB@K7IQ.com> wrote:

On Mon, 05 Jun 2023 19:33:17 +0100, Cursitor Doom <cd@notformail.com
wrote:

On Mon, 05 Jun 2023 11:04:03 -0700, John Larkin
jlarkin@highlandSNIPMEtechnology.com> wrote:

On Mon, 05 Jun 2023 11:01:33 -0700, John Larkin
jlarkin@highlandSNIPMEtechnology.com> wrote:

On Mon, 05 Jun 2023 18:54:16 +0100, Cursitor Doom <cd@notformail.com
wrote:

On Mon, 05 Jun 2023 18:50:38 +0100, Cursitor Doom <cd@notformail.com
wrote:

On Mon, 05 Jun 2023 09:21:32 -0700, John Larkin
jlarkin@highlandSNIPMEtechnology.com> wrote:

On Mon, 5 Jun 2023 12:10:05 -0400, Phil Hobbs
pcdhSpamMeSenseless@electrooptical.net> wrote:

On 2023-06-04 23:05, John Larkin wrote:
On Mon, 05 Jun 2023 00:03:20 +0100, Cursitor Doom <cd@notformail.com
wrote:

On Sun, 4 Jun 2023 15:45:44 -0700 (PDT), Ricky
gnuarm.deletethisbit@gmail.com> wrote:

On Sunday, June 4, 2023 at 6:13:39?PM UTC-4, John Larkin wrote:
On Sun, 04 Jun 2023 22:34:00 +0100, Cursitor Doom <c...@notformail.com
wrote:
Greetings, gentlemen,

As we all know, to get maximum efficiency out of a MOSFET, it needs to
be driven hard and fast and spend as little time between fully on and
fully off as possible. Does that same principle extend to Triacs? And
if so, how can that best be implemented when one has to use AC to
drive the gate?

CD.
Once a triac fires, the gate drive doesn\'t matter any more.

But they do need a hard, fast drive to fire properly, long enough to
let the drain current rise above the latch limit.

AC gate drive will be slow-rise, which could just possibly damage the
triac if the load current is high.

That should be on the data sheets.

I believe that is why a diac is typically used in the control circuit.

I thought diacs were sometimes used because triacs don\'t fire
symetrically, but I could be wrong.
So, as regards AC drive, is a higher frequency drive signal a safer
bet? Also, the load is AC too, so the thing will turn off at every
zero-crossing point, which makes life more complicated. :-(

With an AC load, an AC gate drive is probably OK. The triac will turn
on before the voltage gets too high, so the spreading damage mechanism
won\'t happen much.

The bad case is when there\'s a lot of voltage across the device with a
lot of instantaneous current behind it, and it gets a slow, wimpy
trigger.

Post your proposed circuit and we can discuss it.

I still call the terminals anode, gate, cathode. I can never remember
what\'s MT1 and MT2. Top and bottom would be OK too.


The triac is one of those parts where folks get confused by fossilized
analogies.

A triac is a bidirectional thyristor, but it does not behave like two
SCRs in antiparallel. SCR triggering only works in one quadrant, for a
start, so two antiparallel SCRs would be a two-quadrant device, whereas
triacs trigger in all four quadrants.

The triggering also isn\'t that symmetrical, because the structures
aren\'t--the second and fourth quadrant triggering mechanisms are fairly
different, and the four-layer stacks aren\'t well separated, as they
would be in two SCRs.

Triacs are also very slow even compared with SCRs, which are pretty poky
devices by modern standards.

Someplace I have a paper that goes into the details of real triac
operation--iirc it\'s a good read, but I\'m not laying my hands on it at
the moment.

(I haven\'t actually used a triac since I was a teenager, so this is just
out of general interest.)

Cheers

Phil Hobbs

I haven\'t designed in a triac for maybe 15 years now. I used a couple
of triacs to soft-start a giant power supply that had a 60 Hz power
transformer. Switch in a giant resistor on the primary side, wait a
second, then short it out. That was hard on resistors but prevented
the occasional 1000 amp startup surge. And allowed a little C&K power
switch on the front panel.

The customer insisted that we not use a switching power supply!

Real relays are usually cheaper than triacs, easier to drive, and
don\'t need heat sinks.

Yes, but you can\'t fire them on and off anything like as fast and they
don\'t have the same longevity AFAIK.

Plus a relay capable of switching 16A is going to be a hog, take up
loads of space compared to a triac and cost a hell of a lot more, I
would imagine.

It\'s easy to look up. Digikey has 15 amp relays starting around 60
cents.

The triac will need a heat sink, which will be big and cost more than
the triac.

Relays don\'t need isolated gate drivers, TVS/MOV protection, or
snubbers either.


Well, you got me there!


I have seen and used snubber across relays. To extinguish DC arcs
across the contacts. Sometimes you have to increase contact life.

Do you know why a snubber extinguishes a DC arc?

Yes. Of course... An R-C-D snubber.

The C has been discharged by the R when the relay opens.
Then, the C, an electrolytic typically, (large-ish C) charges up
instead of the contacts taking the arc. The C is a short across the
contacts when the contacts open.

Then, the R discharges the Cap so it is ready for the next opening.

So the snubber prevents the arc but doesn\'t extinguish one. The R-C-D
will be polarity sensitive. And if the contacts cycle before the
large-ish C discharges, the snubber won\'t work.

What sort of RC values are good?

All very well for switching DC currents, but how to prolong contact
life when the current is AC?

AC arcs are usually self-extinguishing. That\'s why relay contacts have
different AC and DC current:voltage ratings, with the DC much lower.
 
On Wed, 7 Jun 2023 23:21:31 +0100, piglet <erichpwagner@hotmail.com>
wrote:

On 07/06/2023 10:50 pm, Cursitor Doom wrote:
On Wed, 07 Jun 2023 10:57:48 -0700, John Larkin
jlarkin@highlandSNIPMEtechnology.com> wrote:

On Wed, 07 Jun 2023 10:33:01 -0700, boB <boB@K7IQ.com> wrote:

On Wed, 07 Jun 2023 07:09:09 -0700, John Larkin
jlarkin@highlandSNIPMEtechnology.com> wrote:

On Tue, 06 Jun 2023 20:36:01 -0700, boB <boB@K7IQ.com> wrote:

On Mon, 05 Jun 2023 19:33:17 +0100, Cursitor Doom <cd@notformail.com
wrote:

On Mon, 05 Jun 2023 11:04:03 -0700, John Larkin
jlarkin@highlandSNIPMEtechnology.com> wrote:

On Mon, 05 Jun 2023 11:01:33 -0700, John Larkin
jlarkin@highlandSNIPMEtechnology.com> wrote:

On Mon, 05 Jun 2023 18:54:16 +0100, Cursitor Doom <cd@notformail.com
wrote:

On Mon, 05 Jun 2023 18:50:38 +0100, Cursitor Doom <cd@notformail.com
wrote:

On Mon, 05 Jun 2023 09:21:32 -0700, John Larkin
jlarkin@highlandSNIPMEtechnology.com> wrote:

On Mon, 5 Jun 2023 12:10:05 -0400, Phil Hobbs
pcdhSpamMeSenseless@electrooptical.net> wrote:

On 2023-06-04 23:05, John Larkin wrote:
On Mon, 05 Jun 2023 00:03:20 +0100, Cursitor Doom <cd@notformail.com
wrote:

On Sun, 4 Jun 2023 15:45:44 -0700 (PDT), Ricky
gnuarm.deletethisbit@gmail.com> wrote:

On Sunday, June 4, 2023 at 6:13:39?PM UTC-4, John Larkin wrote:
On Sun, 04 Jun 2023 22:34:00 +0100, Cursitor Doom <c...@notformail.com
wrote:
Greetings, gentlemen,

As we all know, to get maximum efficiency out of a MOSFET, it needs to
be driven hard and fast and spend as little time between fully on and
fully off as possible. Does that same principle extend to Triacs? And
if so, how can that best be implemented when one has to use AC to
drive the gate?

CD.
Once a triac fires, the gate drive doesn\'t matter any more.

But they do need a hard, fast drive to fire properly, long enough to
let the drain current rise above the latch limit.

AC gate drive will be slow-rise, which could just possibly damage the
triac if the load current is high.

That should be on the data sheets.

I believe that is why a diac is typically used in the control circuit.

I thought diacs were sometimes used because triacs don\'t fire
symetrically, but I could be wrong.
So, as regards AC drive, is a higher frequency drive signal a safer
bet? Also, the load is AC too, so the thing will turn off at every
zero-crossing point, which makes life more complicated. :-(

With an AC load, an AC gate drive is probably OK. The triac will turn
on before the voltage gets too high, so the spreading damage mechanism
won\'t happen much.

The bad case is when there\'s a lot of voltage across the device with a
lot of instantaneous current behind it, and it gets a slow, wimpy
trigger.

Post your proposed circuit and we can discuss it.

I still call the terminals anode, gate, cathode. I can never remember
what\'s MT1 and MT2. Top and bottom would be OK too.


The triac is one of those parts where folks get confused by fossilized
analogies.

A triac is a bidirectional thyristor, but it does not behave like two
SCRs in antiparallel. SCR triggering only works in one quadrant, for a
start, so two antiparallel SCRs would be a two-quadrant device, whereas
triacs trigger in all four quadrants.

The triggering also isn\'t that symmetrical, because the structures
aren\'t--the second and fourth quadrant triggering mechanisms are fairly
different, and the four-layer stacks aren\'t well separated, as they
would be in two SCRs.

Triacs are also very slow even compared with SCRs, which are pretty poky
devices by modern standards.

Someplace I have a paper that goes into the details of real triac
operation--iirc it\'s a good read, but I\'m not laying my hands on it at
the moment.

(I haven\'t actually used a triac since I was a teenager, so this is just
out of general interest.)

Cheers

Phil Hobbs

I haven\'t designed in a triac for maybe 15 years now. I used a couple
of triacs to soft-start a giant power supply that had a 60 Hz power
transformer. Switch in a giant resistor on the primary side, wait a
second, then short it out. That was hard on resistors but prevented
the occasional 1000 amp startup surge. And allowed a little C&K power
switch on the front panel.

The customer insisted that we not use a switching power supply!

Real relays are usually cheaper than triacs, easier to drive, and
don\'t need heat sinks.

Yes, but you can\'t fire them on and off anything like as fast and they
don\'t have the same longevity AFAIK.

Plus a relay capable of switching 16A is going to be a hog, take up
loads of space compared to a triac and cost a hell of a lot more, I
would imagine.

It\'s easy to look up. Digikey has 15 amp relays starting around 60
cents.

The triac will need a heat sink, which will be big and cost more than
the triac.

Relays don\'t need isolated gate drivers, TVS/MOV protection, or
snubbers either.


Well, you got me there!


I have seen and used snubber across relays. To extinguish DC arcs
across the contacts. Sometimes you have to increase contact life.

Do you know why a snubber extinguishes a DC arc?

Yes. Of course... An R-C-D snubber.

The C has been discharged by the R when the relay opens.
Then, the C, an electrolytic typically, (large-ish C) charges up
instead of the contacts taking the arc. The C is a short across the
contacts when the contacts open.

Then, the R discharges the Cap so it is ready for the next opening.

So the snubber prevents the arc but doesn\'t extinguish one. The R-C-D
will be polarity sensitive. And if the contacts cycle before the
large-ish C discharges, the snubber won\'t work.

What sort of RC values are good?

All very well for switching DC currents, but how to prolong contact
life when the current is AC?

This app note is one place...

https://www.dropbox.com/s/4ktoqpj0pqb4zfv/Littelfuse_Arc_Suppression.pdf?raw=1

The advice to watch the intensity of the arc is cute!

That doc is pretty bad.


My starting point for general purpose AC powerline switching is RC
snubber of 100R and 0.1uF and adapt values up or down from there.

piglet
 
On Wed, 7 Jun 2023 13:22:17 -0700 (PDT), whit3rd <whit3rd@gmail.com>
wrote:

On Tuesday, June 6, 2023 at 2:43:22?PM UTC-4, John Larkin wrote:
On Tue, 6 Jun 2023 10:43:19 -0700 (PDT), Lasse Langwadt Christensen
lang...@fonz.dk> wrote:

tirsdag den 6. juni 2023 kl. 05.03.52 UTC+2 skrev John Larkin:

There\'s a shop in Fremont that will anodize infinite heat sinks.

it takes forever and the bill is infinite? ;)
Oh, don\'t get technical. I hate it when people get technical.

Actually, you don\'t often need an infinite heat sink. The thermal
spreading resistance eventually dominates.

https://www.dropbox.com/scl/fi/zf1a1lds726f2y2682v7z/Infinite_Sheet.jpg?raw=1

Yeah, that\'s how we get net global warming despite our radiative sink for heat (outer space) having
no end anywhere in sight, and even (in its distant parts) admirably low temperature.
Our polluted atmosphere has lessened the coupling for outgoing heat.

Earth is not a triac and I\'ve seen recent studies that the atmosphere
temperature went up during the covid thing because there were less
particulates in the stratosphere to radiate heat into space.

https://climate.nasa.gov/news/556/particles-in-upper-atmosphere-slow-down-global-warming/

Volcanoes shoot junk into the air and cool the planet. \"Volcanic
winter.\"
 
John Larkin wrote:
--------------------------------
This app note is one place...

https://www.dropbox.com/s/4ktoqpj0pqb4zfv/Littelfuse_Arc_Suppression.pdf?raw=1

The advice to watch the intensity of the arc is cute!

That doc is pretty bad.


** This article from my colleague Rod Eliott is a lot better.

https://sound-au.com/articles/arc-prevention.htm

I particularly like the \" Mosfet Arc Extinguisher \" suppression circuit in fig 4.2.



..... Phil
 
On Wed, 7 Jun 2023 21:18:25 -0700 (PDT), Phil Allison
<pallison49@gmail.com> wrote:

John Larkin wrote:
--------------------------------
This app note is one place...

https://www.dropbox.com/s/4ktoqpj0pqb4zfv/Littelfuse_Arc_Suppression.pdf?raw=1

The advice to watch the intensity of the arc is cute!

That doc is pretty bad.


** This article from my colleague Rod Eliott is a lot better.

https://sound-au.com/articles/arc-prevention.htm

I particularly like the \" Mosfet Arc Extinguisher \" suppression circuit in fig 4.2.



.... Phil

That paper is better than the Littlefuse thing, but it doesn\'t explain
how the passive RC (fig 3.1) can extinguish an arc. The suggested time
constants are short.

The mosfet things are complex and maybe fragile.
 
On Monday, June 5, 2023 at 10:23:43 PM UTC-4, Ricky wrote:
On Monday, June 5, 2023 at 3:12:28 PM UTC-4, Lasse Langwadt Christensen wrote:
mandag den 5. juni 2023 kl. 20.45.03 UTC+2 skrev Cursitor Doom:

Seriously, I\'m not seeing 15A relays on Digikey for less than around
15 dollars. Where do you get 60 cents from ??
https://www.digikey.dk/en/products/detail/amphenol-anytek/AHQSH112LM1F00G/16721973

60 cent if you buy 250, 90cent if you buy one

Interesting. It is rated to switch DC too! What will they think of next?

It takes SCR plus MOSFET to switch DC; for \'off\' function, you just
crowbar the SCR with the MOSFET; at under 0.5V, the SCR cannot
stay latched ON. That MOSFET only needs to shoulder the full load
current for a millisecond or so; turnon can be slow but turnoff should be FAST so as not
to burn it up.

Gate turnoff SCRs were useless, but a crowbar using a power MOS device isn\'t
infeasible.
 
On Thu, 8 Jun 2023 07:15:54 -0700 (PDT), whit3rd <whit3rd@gmail.com>
wrote:

On Monday, June 5, 2023 at 10:23:43?PM UTC-4, Ricky wrote:
On Monday, June 5, 2023 at 3:12:28?PM UTC-4, Lasse Langwadt Christensen wrote:
mandag den 5. juni 2023 kl. 20.45.03 UTC+2 skrev Cursitor Doom:

Seriously, I\'m not seeing 15A relays on Digikey for less than around
15 dollars. Where do you get 60 cents from ??
https://www.digikey.dk/en/products/detail/amphenol-anytek/AHQSH112LM1F00G/16721973

60 cent if you buy 250, 90cent if you buy one

Interesting. It is rated to switch DC too! What will they think of next?

It takes SCR plus MOSFET to switch DC; for \'off\' function, you just
crowbar the SCR with the MOSFET; at under 0.5V, the SCR cannot
stay latched ON. That MOSFET only needs to shoulder the full load
current for a millisecond or so; turnon can be slow but turnoff should be FAST so as not
to burn it up.

Not too fast, or the SCR will stay on.

Gate turnoff SCRs were useless, but a crowbar using a power MOS device isn\'t
infeasible.

Why use an SCR if the mosfet can handle the load with under 0.5 volts
drop? Mosfets can get to single-digit-milliohm ON resistances these
days.

SCRs have huge ON voltage drops so need big heat sinks.
 
torsdag den 8. juni 2023 kl. 17.48.49 UTC+2 skrev John Larkin:
On Thu, 8 Jun 2023 07:15:54 -0700 (PDT), whit3rd <whi...@gmail.com
wrote:

On Monday, June 5, 2023 at 10:23:43?PM UTC-4, Ricky wrote:
On Monday, June 5, 2023 at 3:12:28?PM UTC-4, Lasse Langwadt Christensen wrote:
mandag den 5. juni 2023 kl. 20.45.03 UTC+2 skrev Cursitor Doom:

Seriously, I\'m not seeing 15A relays on Digikey for less than around
15 dollars. Where do you get 60 cents from ??
https://www.digikey.dk/en/products/detail/amphenol-anytek/AHQSH112LM1F00G/16721973

60 cent if you buy 250, 90cent if you buy one

Interesting. It is rated to switch DC too! What will they think of next?

It takes SCR plus MOSFET to switch DC; for \'off\' function, you just
crowbar the SCR with the MOSFET; at under 0.5V, the SCR cannot
stay latched ON. That MOSFET only needs to shoulder the full load
current for a millisecond or so; turnon can be slow but turnoff should be FAST so as not
to burn it up.
Not too fast, or the SCR will stay on.

Gate turnoff SCRs were useless, but a crowbar using a power MOS device isn\'t
infeasible.
Why use an SCR if the mosfet can handle the load with under 0.5 volts
drop? Mosfets can get to single-digit-milliohm ON resistances these
days.

SCRs have huge ON voltage drops so need big heat sinks.

once you get mains voltages single-digit-milliohm ON resistances gets a bit expensive

but you gain the advantage that you can turn off quickly in case of overload, you don\'t have to suffer a full half cycle
 
On 08/06/2023 17:45, Lasse Langwadt Christensen wrote:
torsdag den 8. juni 2023 kl. 17.48.49 UTC+2 skrev John Larkin:
On Thu, 8 Jun 2023 07:15:54 -0700 (PDT), whit3rd <whi...@gmail.com
wrote:

On Monday, June 5, 2023 at 10:23:43?PM UTC-4, Ricky wrote:
On Monday, June 5, 2023 at 3:12:28?PM UTC-4, Lasse Langwadt Christensen wrote:
mandag den 5. juni 2023 kl. 20.45.03 UTC+2 skrev Cursitor Doom:

Seriously, I\'m not seeing 15A relays on Digikey for less than around
15 dollars. Where do you get 60 cents from ??
https://www.digikey.dk/en/products/detail/amphenol-anytek/AHQSH112LM1F00G/16721973

60 cent if you buy 250, 90cent if you buy one

Interesting. It is rated to switch DC too! What will they think of next?

It takes SCR plus MOSFET to switch DC; for \'off\' function, you just
crowbar the SCR with the MOSFET; at under 0.5V, the SCR cannot
stay latched ON. That MOSFET only needs to shoulder the full load
current for a millisecond or so; turnon can be slow but turnoff should be FAST so as not
to burn it up.
Not too fast, or the SCR will stay on.

Gate turnoff SCRs were useless, but a crowbar using a power MOS device isn\'t
infeasible.
Why use an SCR if the mosfet can handle the load with under 0.5 volts
drop? Mosfets can get to single-digit-milliohm ON resistances these
days.

SCRs have huge ON voltage drops so need big heat sinks.

once you get mains voltages single-digit-milliohm ON resistances gets a bit expensive

but you gain the advantage that you can turn off quickly in case of overload, you don\'t have to suffer a full half cycle

There are affordable SCR/Triacs around with high enough I2t ratings to
pass on a downstream overload and trip the circuit breaker or blow a
fuse without turning themselves into three terminal short circuits. Are
rugged MOSFETs able to do that as cost competitive yet?

piglet
 
On Thursday, June 8, 2023 at 11:48:49 AM UTC-4, John Larkin wrote:
On Thu, 8 Jun 2023 07:15:54 -0700 (PDT), whit3rd <whi...@gmail.com
wrote:

It takes SCR plus MOSFET to switch DC; for \'off\' function, you just
crowbar the SCR with the MOSFET; at under 0.5V, the SCR cannot
stay latched ON. That MOSFET only needs to shoulder the full load
current for a millisecond or so; turnon can be slow but turnoff should be FAST so as not
to burn it up.

Not too fast, or the SCR will stay on.

That would be \'not too prompt\'; what I intended by FAST, was
slew rate, not delay time. Applying high current to turn OFF implies that
the voltage across the SCR starts low (circa 1.5V).

Gate turnoff SCRs were useless, but a crowbar using a power MOS device isn\'t
infeasible.

Why use an SCR if the mosfet can handle the load with under 0.5 volts
drop? Mosfets can get to single-digit-milliohm ON resistances these
days.

You can get SCRs with multikiloamps ratings, continuous. That\'s because
conduction is vertical, from one face of the wafer to the other, not
horizontal. That simplifies heatsinking and electrical connection.
 
On Thu, 8 Jun 2023 09:45:50 -0700 (PDT), Lasse Langwadt Christensen
<langwadt@fonz.dk> wrote:

torsdag den 8. juni 2023 kl. 17.48.49 UTC+2 skrev John Larkin:
On Thu, 8 Jun 2023 07:15:54 -0700 (PDT), whit3rd <whi...@gmail.com
wrote:

On Monday, June 5, 2023 at 10:23:43?PM UTC-4, Ricky wrote:
On Monday, June 5, 2023 at 3:12:28?PM UTC-4, Lasse Langwadt Christensen wrote:
mandag den 5. juni 2023 kl. 20.45.03 UTC+2 skrev Cursitor Doom:

Seriously, I\'m not seeing 15A relays on Digikey for less than around
15 dollars. Where do you get 60 cents from ??
https://www.digikey.dk/en/products/detail/amphenol-anytek/AHQSH112LM1F00G/16721973

60 cent if you buy 250, 90cent if you buy one

Interesting. It is rated to switch DC too! What will they think of next?

It takes SCR plus MOSFET to switch DC; for \'off\' function, you just
crowbar the SCR with the MOSFET; at under 0.5V, the SCR cannot
stay latched ON. That MOSFET only needs to shoulder the full load
current for a millisecond or so; turnon can be slow but turnoff should be FAST so as not
to burn it up.
Not too fast, or the SCR will stay on.

Gate turnoff SCRs were useless, but a crowbar using a power MOS device isn\'t
infeasible.
Why use an SCR if the mosfet can handle the load with under 0.5 volts
drop? Mosfets can get to single-digit-milliohm ON resistances these
days.

SCRs have huge ON voltage drops so need big heat sinks.

once you get mains voltages single-digit-milliohm ON resistances gets a bit expensive

SCRs typically drop 1.5 volts or so at rated current. It\'s not hard to
beat that. Two or even three mosfets is a lot cheaper than an SCR and
a mosfet and associated drivers and heat sinks.

but you gain the advantage that you can turn off quickly in case of overload, you don\'t have to suffer a full half cycle

I\'m designing some dummy loads these days that include short-circuit
programmability. I\'ll use a dual isolated ADC per channel to digitize
voltage and current. An FPGA will model mosfet junction temperature
and shut things down if the fet is in any danger.
 
On Thu, 8 Jun 2023 18:48:40 +0100, piglet <erichpwagner@hotmail.com>
wrote:

On 08/06/2023 17:45, Lasse Langwadt Christensen wrote:
torsdag den 8. juni 2023 kl. 17.48.49 UTC+2 skrev John Larkin:
On Thu, 8 Jun 2023 07:15:54 -0700 (PDT), whit3rd <whi...@gmail.com
wrote:

On Monday, June 5, 2023 at 10:23:43?PM UTC-4, Ricky wrote:
On Monday, June 5, 2023 at 3:12:28?PM UTC-4, Lasse Langwadt Christensen wrote:
mandag den 5. juni 2023 kl. 20.45.03 UTC+2 skrev Cursitor Doom:

Seriously, I\'m not seeing 15A relays on Digikey for less than around
15 dollars. Where do you get 60 cents from ??
https://www.digikey.dk/en/products/detail/amphenol-anytek/AHQSH112LM1F00G/16721973

60 cent if you buy 250, 90cent if you buy one

Interesting. It is rated to switch DC too! What will they think of next?

It takes SCR plus MOSFET to switch DC; for \'off\' function, you just
crowbar the SCR with the MOSFET; at under 0.5V, the SCR cannot
stay latched ON. That MOSFET only needs to shoulder the full load
current for a millisecond or so; turnon can be slow but turnoff should be FAST so as not
to burn it up.
Not too fast, or the SCR will stay on.

Gate turnoff SCRs were useless, but a crowbar using a power MOS device isn\'t
infeasible.
Why use an SCR if the mosfet can handle the load with under 0.5 volts
drop? Mosfets can get to single-digit-milliohm ON resistances these
days.

SCRs have huge ON voltage drops so need big heat sinks.

once you get mains voltages single-digit-milliohm ON resistances gets a bit expensive

but you gain the advantage that you can turn off quickly in case of overload, you don\'t have to suffer a full half cycle


There are affordable SCR/Triacs around with high enough I2t ratings to
pass on a downstream overload and trip the circuit breaker or blow a
fuse without turning themselves into three terminal short circuits. Are
rugged MOSFETs able to do that as cost competitive yet?

piglet

A mosfet can be turned off to prevent meltdown. That way no tripped
breaker or fuse has to be manually dealt with.
 
On Thu, 8 Jun 2023 11:38:56 -0700 (PDT), whit3rd <whit3rd@gmail.com>
wrote:

On Thursday, June 8, 2023 at 11:48:49?AM UTC-4, John Larkin wrote:
On Thu, 8 Jun 2023 07:15:54 -0700 (PDT), whit3rd <whi...@gmail.com
wrote:

It takes SCR plus MOSFET to switch DC; for \'off\' function, you just
crowbar the SCR with the MOSFET; at under 0.5V, the SCR cannot
stay latched ON. That MOSFET only needs to shoulder the full load
current for a millisecond or so; turnon can be slow but turnoff should be FAST so as not
to burn it up.

Not too fast, or the SCR will stay on.

That would be \'not too prompt\'; what I intended by FAST, was
slew rate, not delay time.

That\'s what I meant too. High dv/dt will re-fire an SCR.


Applying high current to turn OFF implies that
the voltage across the SCR starts low (circa 1.5V).

Gate turnoff SCRs were useless, but a crowbar using a power MOS device isn\'t
infeasible.

Why use an SCR if the mosfet can handle the load with under 0.5 volts
drop? Mosfets can get to single-digit-milliohm ON resistances these
days.

You can get SCRs with multikiloamps ratings, continuous. That\'s because
conduction is vertical, from one face of the wafer to the other, not
horizontal. That simplifies heatsinking and electrical connection.

Lateral mosfets are rare oddities. Most use vertical conduction.

Sure, use gigantic SCRs for those megawatt cases. But would you add
multikiloamp mosfets to commute them off?
 
On Thursday, June 8, 2023 at 3:16:53 PM UTC-4, John Larkin wrote:
On Thu, 8 Jun 2023 11:38:56 -0700 (PDT), whit3rd <whi...@gmail.com
wrote:
On Thursday, June 8, 2023 at 11:48:49?AM UTC-4, John Larkin wrote:
On Thu, 8 Jun 2023 07:15:54 -0700 (PDT), whit3rd <whi...@gmail.com
wrote:

It takes SCR plus MOSFET to switch DC; for \'off\' function, you just
crowbar the SCR with the MOSFET; at under 0.5V, the SCR cannot
stay latched ON.

Why use an SCR if the mosfet can handle the load with under 0.5 volts
drop? Mosfets can get to single-digit-milliohm ON resistances these
days.

You can get SCRs with multikiloamps ratings, continuous. That\'s because
conduction is vertical, from one face of the wafer to the other, not
horizontal. That simplifies heatsinking and electrical connection.

Lateral mosfets are rare oddities. Most use vertical conduction.

Rare? There\'s a lot of \'em in a CPU near you.
The power devices with vertical drains still have source currents flowing laterally.

Sure, use gigantic SCRs for those megawatt cases. But would you add
multikiloamp mosfets to commute them off?

Maybe. Or, maybe just a capacitor-coupled SCR. Gate turnoff, while
possible, of an N amp device takes about N amps...
 
On Wed, 07 Jun 2023 22:50:31 +0100, Cursitor Doom <cd@notformail.com>
wrote:

On Wed, 07 Jun 2023 10:57:48 -0700, John Larkin
jlarkin@highlandSNIPMEtechnology.com> wrote:

On Wed, 07 Jun 2023 10:33:01 -0700, boB <boB@K7IQ.com> wrote:

On Wed, 07 Jun 2023 07:09:09 -0700, John Larkin
jlarkin@highlandSNIPMEtechnology.com> wrote:

On Tue, 06 Jun 2023 20:36:01 -0700, boB <boB@K7IQ.com> wrote:

On Mon, 05 Jun 2023 19:33:17 +0100, Cursitor Doom <cd@notformail.com
wrote:

On Mon, 05 Jun 2023 11:04:03 -0700, John Larkin
jlarkin@highlandSNIPMEtechnology.com> wrote:

On Mon, 05 Jun 2023 11:01:33 -0700, John Larkin
jlarkin@highlandSNIPMEtechnology.com> wrote:

On Mon, 05 Jun 2023 18:54:16 +0100, Cursitor Doom <cd@notformail.com
wrote:

On Mon, 05 Jun 2023 18:50:38 +0100, Cursitor Doom <cd@notformail.com
wrote:

On Mon, 05 Jun 2023 09:21:32 -0700, John Larkin
jlarkin@highlandSNIPMEtechnology.com> wrote:

On Mon, 5 Jun 2023 12:10:05 -0400, Phil Hobbs
pcdhSpamMeSenseless@electrooptical.net> wrote:

On 2023-06-04 23:05, John Larkin wrote:
On Mon, 05 Jun 2023 00:03:20 +0100, Cursitor Doom <cd@notformail.com
wrote:

On Sun, 4 Jun 2023 15:45:44 -0700 (PDT), Ricky
gnuarm.deletethisbit@gmail.com> wrote:

On Sunday, June 4, 2023 at 6:13:39?PM UTC-4, John Larkin wrote:
On Sun, 04 Jun 2023 22:34:00 +0100, Cursitor Doom <c...@notformail.com
wrote:
Greetings, gentlemen,

As we all know, to get maximum efficiency out of a MOSFET, it needs to
be driven hard and fast and spend as little time between fully on and
fully off as possible. Does that same principle extend to Triacs? And
if so, how can that best be implemented when one has to use AC to
drive the gate?

CD.
Once a triac fires, the gate drive doesn\'t matter any more.

But they do need a hard, fast drive to fire properly, long enough to
let the drain current rise above the latch limit.

AC gate drive will be slow-rise, which could just possibly damage the
triac if the load current is high.

That should be on the data sheets.

I believe that is why a diac is typically used in the control circuit.

I thought diacs were sometimes used because triacs don\'t fire
symetrically, but I could be wrong.
So, as regards AC drive, is a higher frequency drive signal a safer
bet? Also, the load is AC too, so the thing will turn off at every
zero-crossing point, which makes life more complicated. :-(

With an AC load, an AC gate drive is probably OK. The triac will turn
on before the voltage gets too high, so the spreading damage mechanism
won\'t happen much.

The bad case is when there\'s a lot of voltage across the device with a
lot of instantaneous current behind it, and it gets a slow, wimpy
trigger.

Post your proposed circuit and we can discuss it.

I still call the terminals anode, gate, cathode. I can never remember
what\'s MT1 and MT2. Top and bottom would be OK too.


The triac is one of those parts where folks get confused by fossilized
analogies.

A triac is a bidirectional thyristor, but it does not behave like two
SCRs in antiparallel. SCR triggering only works in one quadrant, for a
start, so two antiparallel SCRs would be a two-quadrant device, whereas
triacs trigger in all four quadrants.

The triggering also isn\'t that symmetrical, because the structures
aren\'t--the second and fourth quadrant triggering mechanisms are fairly
different, and the four-layer stacks aren\'t well separated, as they
would be in two SCRs.

Triacs are also very slow even compared with SCRs, which are pretty poky
devices by modern standards.

Someplace I have a paper that goes into the details of real triac
operation--iirc it\'s a good read, but I\'m not laying my hands on it at
the moment.

(I haven\'t actually used a triac since I was a teenager, so this is just
out of general interest.)

Cheers

Phil Hobbs

I haven\'t designed in a triac for maybe 15 years now. I used a couple
of triacs to soft-start a giant power supply that had a 60 Hz power
transformer. Switch in a giant resistor on the primary side, wait a
second, then short it out. That was hard on resistors but prevented
the occasional 1000 amp startup surge. And allowed a little C&K power
switch on the front panel.

The customer insisted that we not use a switching power supply!

Real relays are usually cheaper than triacs, easier to drive, and
don\'t need heat sinks.

Yes, but you can\'t fire them on and off anything like as fast and they
don\'t have the same longevity AFAIK.

Plus a relay capable of switching 16A is going to be a hog, take up
loads of space compared to a triac and cost a hell of a lot more, I
would imagine.

It\'s easy to look up. Digikey has 15 amp relays starting around 60
cents.

The triac will need a heat sink, which will be big and cost more than
the triac.

Relays don\'t need isolated gate drivers, TVS/MOV protection, or
snubbers either.


Well, you got me there!


I have seen and used snubber across relays. To extinguish DC arcs
across the contacts. Sometimes you have to increase contact life.

Do you know why a snubber extinguishes a DC arc?

Yes. Of course... An R-C-D snubber.

The C has been discharged by the R when the relay opens.
Then, the C, an electrolytic typically, (large-ish C) charges up
instead of the contacts taking the arc. The C is a short across the
contacts when the contacts open.

Then, the R discharges the Cap so it is ready for the next opening.

So the snubber prevents the arc but doesn\'t extinguish one. The R-C-D
will be polarity sensitive. And if the contacts cycle before the
large-ish C discharges, the snubber won\'t work.

What sort of RC values are good?

All very well for switching DC currents, but how to prolong contact
life when the current is AC?


The are only happens when the contacts open and that is when the
capacitor shorts out the voltage to keep the arc from even happening.
I\'m talking BIG continuous arcing if this is not snubbed.

It WILL work and it DOES work.

Of course it is polarity sensitive ! This was used on a solar PV
array disconnect for ground fault protection in the 1990s. It worked
great.

No AC. Just DC here.

There are other ways of stopping relay arcs for AC too but involves
triacs across the contacts.

boB
 

Welcome to EDABoard.com

Sponsor

Back
Top