double-shielded optical couplers

[Winfield Hill]

Does anyong know of any double-shielded optical couplers with higher
maximum output-slew-rate specs than the common wimpy 15kV/us rating?

Background:
Optical couplers are rated for a maximum allowed common-mode voltage
slew rate, which if exceeded may cause the coupler to malfunction
(e.g. the output may change state, etc.). A typical best-rated part
from a manufacturer like Agilent or Toshiba is rated at S < 15kV/us,
which means one could use the coupler to drive a high-side switch
limited to slower switching than say t = V/S = 42ns for a 630V swing.

While 40ns may be fast enough (even too fast) for large motor drives
or lamp ballasts, etc., for many applications it's very slow. While
one can solve the problem with optical-fibre links, or transformers,
etc., it would be preferable to use a single-package optical coupler
with appropriate internal double shielding for a faster slew rating,
like 250 or 500kV/us. That would be more like it!

Why double shielding? Generally one's first concern is maintaining
proper operation of the coupler's output in a high slew-rate scene.
If the photo-diode detector is operating on say 5uA of photocurrent,
then i = C dV/dt tells us that the inter-electrode capacitance C had
better be well under 0.33fF for a 15kV/us rating. That's the purpose
of the first internal shield. But at high enough slew rates, the LED
also needs a shield. E.g., Agilent's HCPL-3020 is rated at 15kV/us.
http://literature.agilent.com/litweb/pdf/5988-8736EN.pdf and its
LED-to-output capacitance is 0.6pF. When this part experiences a
15kV/us output slew, the induced current into the LED is 9mA, which
approaches the standard 10mA LED drive current. Clearly without an
input shield the part cannot withstand a higher output slew rate,
and one wonders if even 9mA of parasitic current is acceptable.

That's why I'm looking for some double-shielded optical couplers,
hopefully with _much_ better specs than 15kV/us.

Thanks,
- Win

 

[Fred Bartoli]

Does anyong know of any double-shielded optical couplers with higher
maximum output-slew-rate specs than the common wimpy 15kV/us rating?

Background:
Optical couplers are rated for a maximum allowed common-mode voltage
slew rate, which if exceeded may cause the coupler to malfunction
(e.g. the output may change state, etc.). A typical best-rated part
from a manufacturer like Agilent or Toshiba is rated at S < 15kV/us,
which means one could use the coupler to drive a high-side switch
limited to slower switching than say t = V/S = 42ns for a 630V swing.

While 40ns may be fast enough (even too fast) for large motor drives
or lamp ballasts, etc., for many applications it's very slow. While
one can solve the problem with optical-fibre links, or transformers,
etc., it would be preferable to use a single-package optical coupler
with appropriate internal double shielding for a faster slew rating,
like 250 or 500kV/us. That would be more like it!

Why double shielding? Generally one's first concern is maintaining
proper operation of the coupler's output in a high slew-rate scene.
If the photo-diode detector is operating on say 5uA of photocurrent,
then i = C dV/dt tells us that the inter-electrode capacitance C had
better be well under 0.33fF for a 15kV/us rating. That's the purpose
of the first internal shield. But at high enough slew rates, the LED
also needs a shield. E.g., Agilent's HCPL-3020 is rated at 15kV/us.
http://literature.agilent.com/litweb/pdf/5988-8736EN.pdf and its
LED-to-output capacitance is 0.6pF. When this part experiences a
15kV/us output slew, the induced current into the LED is 9mA, which
approaches the standard 10mA LED drive current. Clearly without an
input shield the part cannot withstand a higher output slew rate,
and one wonders if even 9mA of parasitic current is acceptable.

That's why I'm looking for some double-shielded optical couplers,
hopefully with _much_ better specs than 15kV/us.

Thanks,
- Win

Win,

why not, if the app allow this, making the MOSFET's side blind to any change
once triggered for say 100 or 200ns, i.e. longer than the common mode
voltage transition time and long enough to let the optocoupler recover from
it's possibly bogus state ?

I guess it could be fairly easy.

Fred.

 

[Winfield Hill]

Bruce Raymond wrote...

I'm probably waaaay off base here. Could you use separate
optical transmitters and receivers connected by fiber optic
cables? Agilent has optical receivers that will operate without
a huge optical budget (however the monetary budget is
something else). (I know it's not an optocoupler.)

http://literature.agilent.com/litweb/pdf/5988-8402EN.pdf

I've used Agilent HFBR-xxxx (?? I don't remember) to drive
a Powerex IGBT module for a 270 VDC prototype motor driver.
I'm sure it also depends on how many you're going to make.

No, you're not off base at all, in fact I previously mentioned
optical-fibre links, as well as transformers, for workarounds.
Fred Bartoli's suggestion to start a short timer and ignore any
changes from the optocoupler during each high slew-rate event
may also work. While using any one of these schemes requires
considerably more space and complexity, I'm prepared to accept
this if a double-shielded optical coupler cannot be found.

I'm working on a 500V 2ns push-pull 50-ohm cable driver using
a flying n-type MOSFET for the high-side switch. That's where
the 250kV/ns slewing for the optocoupler arises. Although the
option of using a p-type MOSFET for the pullup may be feasible
for 500V pulses, the flying-switch scheme is appealing because
it can be extended to 1000 or 1200V using n-type FETs.

Thanks Bruce, Win


Thanks,
- Win

 

[Fred Bloggs]

I'm working on a 500V 2ns push-pull 50-ohm cable driver using
a flying n-type MOSFET for the high-side switch. That's where
the 250kV/us slewing for the optocoupler arises. Although the
option of using a p-type MOSFET for the pullup may be feasible
for 500V pulses, the flying-switch scheme is appealing because
it can be extended to 1000 or 1200V using n-type FETs.

Some kind of variation of :

Please view in a fixed-width font such as Courier.




O.C.
+------+ o

--------| | O|----~~~~~~---->

|I | U| o ======
|N | T|----~~~~~~---->

---+----| | |

| +------+ CM Choke
| |
+-------+
----
///

How exactly do you have the O.C. configured- if it's not proprietary...

A 26dB reduction gets you in spec .

 

[Winfield Hill]

Fred Bloggs wrote...

A 26dB reduction gets you in spec.

Yes, a small 10-to-250MHz common-mode choke with 500V
insulation should help, good idea. It could be easily
placed on the LED side.

Thanks,
- Win

 

[John Larkin]

Does anyong know of any double-shielded optical couplers with higher
maximum output-slew-rate specs than the common wimpy 15kV/us rating?

Background:
Optical couplers are rated for a maximum allowed common-mode voltage
slew rate, which if exceeded may cause the coupler to malfunction
(e.g. the output may change state, etc.). A typical best-rated part
from a manufacturer like Agilent or Toshiba is rated at S < 15kV/us,
which means one could use the coupler to drive a high-side switch
limited to slower switching than say t = V/S = 42ns for a 630V swing.

While 40ns may be fast enough (even too fast) for large motor drives
or lamp ballasts, etc., for many applications it's very slow. While
one can solve the problem with optical-fibre links, or transformers,
etc., it would be preferable to use a single-package optical coupler
with appropriate internal double shielding for a faster slew rating,
like 250 or 500kV/us. That would be more like it!

Why double shielding? Generally one's first concern is maintaining
proper operation of the coupler's output in a high slew-rate scene.
If the photo-diode detector is operating on say 5uA of photocurrent,
then i = C dV/dt tells us that the inter-electrode capacitance C had
better be well under 0.33fF for a 15kV/us rating. That's the purpose
of the first internal shield. But at high enough slew rates, the LED
also needs a shield. E.g., Agilent's HCPL-3020 is rated at 15kV/us.
http://literature.agilent.com/litweb/pdf/5988-8736EN.pdf and its
LED-to-output capacitance is 0.6pF. When this part experiences a
15kV/us output slew, the induced current into the LED is 9mA, which
approaches the standard 10mA LED drive current. Clearly without an
input shield the part cannot withstand a higher output slew rate,
and one wonders if even 9mA of parasitic current is acceptable.

That's why I'm looking for some double-shielded optical couplers,
hopefully with _much_ better specs than 15kV/us.

Thanks,
- Win

Win,

how about adding a common-mode inductor or two? Like those little
Mini-Circuits baluns?

John