Overcurrent protection on analogue outputs

On a sunny day (Thu, 06 Aug 2009 15:14:59 -0400) it happened Phil Hobbs wrote in :

Normally, at least in a cable drive situation, I put 50 Ohm in series with the output. That requires 2x Uout if termminated with 50 Ohm, but it guarantees safe operating and correct match to the cable.

So what it comes down to: What do you intent to use it for?

It would not be too difficult to sense Uout and Uin with a second amp, and in case too much difference (after gain correction) assume a short, and clamp the input drive, or trigger a supply shutdown.

Yes, the output is series-terminated as you suggest. I'm not too worried about brief overcurrent spikes, e.g. when somebody is a bit fumblefingered connecting a BNC cable. It's the long slow bake that I'm trying to avoid.

Cheers

Phil Hobbs

On a sunny day (Thu, 06 Aug 2009 15:42:47 -0400) it happened Phil Hobbs wrote in :

LM135 temp sensor glued on opamp, 8 pin PIC as ADC that shuts down the voltage regulator?

regulator?

Okay, there's one PIC suggestion. I'm trying to reduce the parts count and complexity, though...and I don't want the rest of the circuit to shut down, just that output amp. I could use an ADA4898-1, which has a shutdown pin, and something like an LP339 quad comparator...one section looking at each supply, a third for a latch, and a fourth for MCU control of the shutdown.

Still needs a bunch of resistors...but it would also allow overtemp sensing as you suggest, since there would be a fourth section available. Maybe I could run two thermistors in parallel, one for each output amp, and detect overtemp that way. Or use two PTCs in series.

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

On a sunny day (Thu, 06 Aug 2009 17:34:14 -0400) it happened Phil Hobbs wrote in :

voltage regulator?

I have used a 1N4148 diode as sensor and an opamp as comparator in the past. It needed a trimpot to set the cut-off temperature though. Very simple circuit... It measured the temp in a cabinet with about 500 HV MOSFETS :-)

They come with fixed trip points, 60C, 70C, 80C, take your pick. Open drain logic output, reduces it to one part. Or as many as you have places to sense. Here is an example for under 40c, Digikey carries them:

I'd just sprinkle them where needed, connect them all together and shut the whole thing down when one of them goes off and turn on one of them blinkenlights.

--
Regards, Joerg

http://www.analogconsultants.com/

"gmail" domain blocked because of excessive spam.
Use another domain or send PM.

s

bbs

t

nt

a .

mp,

Thermal shutdown is pretty common in LDOs. Hasn't anyone made an opamp with such a function?

In LDOs, the design gets a bit ugly since they want high temp operation. Usually the shutdown is over 150 deg C.

I could use an LM317, I suppose. ;) But I'd like to limit the dissipation to well below the level that would reach those sorts of junction temperatures. The 50 ohm output is for the user's convenience, so he can use BNC patch cords--I don't really want him to hang 50 ohms on it.

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

That's the ticket. A virtual Ommegang to Herr Prof. von Blinkenlight. ;)

Thanks

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

What opamp are you using? If your output swings +-10 unloaded, that's

100 mA into an external 50 ohms, which most opamps can't do anyhow.

Polyfuses are really terrible as regards guaranteed trip/hold limits.

John

What kind of signal ?

If it is AC coupled e.g. audio with a lower frequency limit at several Hertz, a PTC or even an incandescent light bulb in series with the output could be usable.

This is not very practical for DC coupled signals, since the output signal can be sitting at one extreme for a longer period of time, heating the element and hence increasing the resistance even in normal operation.

Paul

It's an ADA4898-1, which has a short circuit current limit of 150 mA. The bandwidth is 10 MHz at most, so I don't expect anyone to be paying much attention to terminating the cable. Anyway, even if they do, it'll usually work fine--most of the time the voltage there will be fairly low.

I just don't want to bake the op amp or the box if someone mistreats it mildly, e.g. trying to run 10V into 50 ohms and making the op amp dissipate a watt or two trying. Even if it doesn't fry the amp, the temperature gradients in the box will take minutes to recover, and this is kind of a precision application.

I think one of Joerg's 40 cent thermostats mounted near each op amp (on top of the same VSS thermal paddle that the op amp sits on) will do a good job--a 60C trip point there should be reasonably good protection, and they're quite accurate. Since they hardly take any supply current, I can build a voltage rail at VSS+5V easily, to match the thermostat's active-low output to the AD8498's shutdown pin. Good medicine.

I actually started out using 220 ohm 2W resistors in each supply lead, with a 10 uF ceramic bypass, but that just seemed way too Neanderthal somehow--it helped but didn't entirely solve the dissipation problem.

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

At a mere 10 MHz, you could add another opamp follower, using an amp that current and maybe thermal limits. Much simpler. Or even replace the 4898 with something less fierce.

Or current-limit the supply rails, with wimpy voltage regulators or Supertex depletion fets or even - gross! - other opamps that do current limit.

There is a nasty tendency to not provide thermal protections and, apparently, deliberate current limiting in fast opamps. On the other hand, most fast opamps seem to survive in this situation in real life.

John

On a sunny day (Thu, 06 Aug 2009 17:47:12 -0700) it happened Joerg wrote in :

Ha, just for fun Joerg: read Microchip application note AN720:

Measuring Temperature Using the Watchdog Timer (WDT)

The *example* uses a PIC 12C509A, the 12F508 is 41 cents in volume:

and should work just as well. I am not saying 'use this', but I am saying that by having a couple of 12F PICs in the drawer, as I have, you have most things covered that you need everyday. Especially at low volume, where one needs to order from for example the states, pay shipping, and VAT taxes on import here too, this is many times (no kidding) cheaper, and especially much faster. Plus this one example send the data via RS232 serial protocol to a PC, and you can have the PIC do other things too, as hardly any code space is used for this temperature measurement application, neither are any external parts needed. ;-)

That makes 41 cents versus your 40 cents, my 1 cent worth :-)

Polyfuse datasheets (and PTC datasheets generally) are horrible.

The output stage is a follower, for all the above reasons.

To keep the overall bandwidth, I'd like the output stage to be at least

30 MHz closed loop BW. To preserve the dynamic range, I need something with noise less than 4 nV or so, and enough output drive to be able to handle several metres of unterminated coax with good output accuracy--say 50 ohms in series with or 1000 pF. I could use a nice beefy, current limited, thermally protected buffer like the LMH6321 inside the FB loop, with two feedback paths to stabilize the loop. They cost $3 each, which isn't too bad, and their input noise is 2.8 nV, so it won't get noisy outside the FB bandwidth. I'm not a big fan of composite amps, though--back when, I'd probably have used an OP50.

I'm using the ADA4898 elsewhere in the design too...apart from the minor wart of having 8 mA supply current, it's the cure for a lot of ills.

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

PICs in the drawer, as I have, you have most

states, pay

cheaper, and especially

Ok, but having to write code just for a thermal shutdown thingie seems a bit excessive. If there's other jobs for the PIC it's ok though.

can have the PIC do other things too,

neither are any external parts needed.

But I can get mine for 36 cents :-)

--
Regards, Joerg

http://www.analogconsultants.com/

"gmail" domain blocked because of excessive spam.
Use another domain or send PM.

Phil Hobbs a écrit :

If your BW requirements are not too high, and as you have a 50R output that one is simple enough:

.----------- --------. | \\ ^ | | --- | | | | | | | | .--------+ | | | |\\ | | | '-|-\\ | ___ | ___ | | >---+--|___|--+----- ------|___|---+-----|+/ | | | |/ | | | | | | | | | --- | '-----------/ ^--------'

You can divide the output drop in case you want more that 10-12mA current...

--
Thanks,
Fred.

We have had similar issues protecting OP-AMPS on the output, what we did was to insert a R between the OP-AMP output and (-) feed back loop. The final out was where the other end of the R and feed back loop R comes together. This would allow the op-amp to max out with out damage if the output would become shorted and also because of the (-) feed back after this inserted R, minor loads would be compensated.

Your design may not allow for such a tactic, but it has worked well for us in many of op-amp output over load protection.

Fred Bartoli a écrit :

Huh... this one is calling for trouble (stability), but this simpler one is OK:

BAV99

.--||--+-----. | | | | | | | | | .--------. | | | |\\ | | | '-|-\\ | ___ | ___ | | >---+--|___|--+----- ------|___|---+-----|+/ |/

Again you can tap the output resistor to have more than 10mA current.

--
Thanks,
Fred.