Interesting, thanks. That's sort of like my resistor-in-the-supply trick, but relies on the FB loop instead of the CMR+PSR. I sort of like putting them in the supplies, because then they can be bypassed without making the amp vulnerable to big AC overloads, but that doesn't allow it to recover as fast as your approach. Your way would just make the amp clip, instead of losing its supply voltage and taking a millisecond to recover. Did you have separate capacitive feedback direct from the op amp output? Otherwise an unterminated cable could cause all sorts of nasty oscillations.
Cheers
Phil Hobbs
--
Dr Philip C D Hobbs
Principal
ElectroOptical Innovations
55 Orchard Rd
Briarcliff Manor NY 10510
845-480-2058
hobbs at electrooptical dot net
http://electrooptical.net
Thanks. My main worry is preventing the big dissipation, rather than supplying enough output current.
Cheers
Phil Hobbs
--
Dr Philip C D Hobbs
Principal
ElectroOptical Innovations
55 Orchard Rd
Briarcliff Manor NY 10510
845-480-2058
hobbs at electrooptical dot net
http://electrooptical.net
Actually there is (at least) one: the TCA0372 from ON-semi
--
Failure does not prove something is impossible, failure simply
indicates you are not using the right tools...
"If it doesn\'t fit, use a bigger hammer!"
--------------------------------------------------------------
Sorry, it wasn't clear which lines were connected and which not.
Cheers
Phil Hobbs
--
Dr Philip C D Hobbs
Principal
ElectroOptical Innovations
55 Orchard Rd
Briarcliff Manor NY 10510
845-480-2058
hobbs at electrooptical dot net
http://electrooptical.net
My preference is to put an attenuator on the output, so it gets a known load regardless of the external connections (but this wastes power and lowers drive, of course).
Lots of fuse/breakover device ideas, of course...
Another popular option is an 'overload' alarm; if your driver op amp saturates, the inputs become unbalanced; instead of just letting that happen, you can detect the imbalance with a window comparator, and set a monostable to light the 'overload' light (or sound the buzzer, or e-mail the incident to a log file, or whatever). I've seen it for input circuits more than for output loads, but the principle seems applicable here.
Positive indication of something awry is worth more to me than physical protection of the apparatus is.
Probably current limiting the rails to the op amp is the best approach if you don't want to hamper the op amp AC performance. A LED indicating failure would be useful.
The chip datasheets generally use the word "indefinite" for the allowed time a pin with protection can be shorted. A weasel word if there ever was one. What generally goes in the customer's favor is the wiring to devices that leave the chip is generally fat, and electromigration rules are written for minimum width wiring. So the designer puts in a limiting circuit, does the electromigration limit computation, and has margin you can drive a truck through.
Okay, since you all were good enough to provide suggestions, I did some sims. The net:
Jamie's simple series resistor inside the FB loop:
The diode trick suggested by Fred Bartoli:
Modified version of (1), using a cap in parallel with the series resistor, plus a series RC across the series terminator (mildly reminiscent of a scope probe):
As shown, with R1=220, C1=340p, R2=330, C2=195p, RT=51, U1= ADA4898-1, it has almost perfect edges regardless of the load, right up until it clips. Its three drawbacks are: (1) it has an irreducible -0.4% dip when the open-circuit cable reflection arrives; (2) it's sensitive to the exact component values (and obviously to the accuracy of the macromodel); and (3) it current-limits on each large signal excursion rather than waiting for a bit.
All of the above share the problem of being too eager to limit the current. Joerg's cute 40-cent thermal cutout is a good alternative, but is maybe a bit on the slow side--by the time a 60 C cutout trips, the temperature gradients on the board have got big enough to be a serious bother in my circuit, and will take tens of minutes to go away. The op amp has current limiting already, it's over-temperature and excess dissipation we're trying to avoid. We'd prefer to wait a millisecond or two before turning down the juice.
It turns out that the champ so far is what I had in the first place, and thought was too crude: 220 ohms in each supply lead, with a 10-uF ceramic bypass, plus a 50-ohm series terminator in the output and a 1k resistor on the + input to limit the current going into the protection diodes when the supply droops.
That makes really nice pulses--the pulse response is essentially unaltered from the unprotected case, so the pulse top artifacts are below 0.1% from an open circuit reflection, and the output limiting doesn't happen until the 10 uF cap droops by at least a few volts, so it doesn't cause problems with low duty cycle transients.
In the words of the estimable Edith Ann: "And that's the truth, *pththth*."
Cheers
Phil Hobbs
--
Dr Philip C D Hobbs
Principal
ElectroOptical Innovations
55 Orchard Rd
Briarcliff Manor NY 10510
845-480-2058
hobbs at electrooptical dot net
http://electrooptical.net
-- | James E.Thompson, P.E. | mens | | Analog Innovations, Inc. | et | | Analog/Mixed-Signal ASIC's and Discrete Systems | manus | | Phoenix, Arizona 85048 Skype: Contacts Only | | | Voice:(480)460-2350 Fax: Available upon request | Brass Rat | | E-mail Icon at
formatting link
| 1962 | I love to cook with wine Sometimes I even put it in the food
ElectronDepot website is not affiliated with any of the manufacturers or service providers discussed here.
All logos and trade names are the property of their respective owners.