1.) Select an other op.amp
2.) add a buffer (bipolar, FET, buffer amplifier, etc.)
3.) if a load is asymetrical, add a pull-up or down R.
4.) Use higher power supply voltage
5.) Add heat sink
Generally the best way depends on what do you have and
what do you want i.e. specipications, requirements and means
including urgency, quantity, cost, maturity of design,
experience of the designer, -- level of confidence--
reliability, -- MTBF, -- etc. etc...
The best way to get help is to describe exactly what
do you need -- with data.
I need to excite wheatstone bridge transducers with 10 volts. Currentely I have a single transistor providing current amplification from a 10 volt reference. The problem is that a short circuit will destroy the transistor. My idea is to use an op amp and a resettable fuse (which are very thermally sensitive) on the output and which will then be fed back to the -input. The +input will have the 10 volt reference. The idea being that any fluctuations caused by the fuse will be compensated. I also don't want to spend $$$ on a high current op amp. I have tried a NPN,PNP combination as suggested by this web site.
The transducers are strain gauge and DC only voltages. Any frequencies will be from the Transducer mechanical vibration which are typically < 1000 Hz.
What I am trying to do is increase the output current to be able to drive more than one transducer (typical impedance is 350 ohms) and also provide short circuit protection.
It will be going in a product currently under production.
If you only require DC with overload protection, a voltage regulator should work.
National Semiconductor among a lot of other manufactures make these regulators.
See LM317 at National Seminconductor Corporation site.
I have a feeling that you are spending energy on the wrong problem.
If you excite the bridge with DC, why do you want to power your bridges with opamps? Why not just with a linear voltage regulator, that can provide all the protection you need? Because the output is ratiometric?
Of course it all depends on what is your readout system. If you want precision, you will want to read your voltage at the power ends of the bridge (six-wire technique) to compensate for the cable losses, and take these losses into account in your calculations, thus the absolute precision on the excitation voltage isn't that important.
I've done many strain gage interfaces, and never used an opamp to power the bridge.
Make sure that your 350-ohm bridge can stand 10 volts of excitation. At 10 volts the strain gage foil will heat, and may offset the precision that you are looking for. Temperature compensation foils are not for the self heating of the gages, but for the heat of the the whole thing where the bridge is glued. Unless you are in a very noisy environment, another way to solve your problem would be to use similar strain gages that are over 1k.
Felix
I am using a precision 10 volt reference that is not capable of supplying enough current on it's own. The precision is necessary because the interface is not six wire and most sensors we supply are 4 wire
Are there any 10 regulators that can provide short circuit protection, and a stable voltage through different temperatures?
What are you using to read the output of your strain gage? An A/D converter? If it has multiple inputs you can also use it to read the excitation voltage and use it in your ratiometric calculations. Then you do not really need a precision reference. What kind of precision should your system have?
The voltage losses in the bridge supply wires may be non-negligible if you run long lengths.
Most linear regulators have a built-in thermal shutdown, in case of overload. For temperature, what kind of drift can you stand?
Integrated voltge regulators can provide the current and limiting along with overheat protection. For precision, you can use an op-amp to power or change the reference voltage pin of the regulator, and provide voltage feedback to it. YOu might find that you do not need the extra op-amp. Check out the voltage regulators from a few companies. Maybe get some suggestions from an applications engineer concerning which part would be best. Each major vendor has an apps engineer that can save you some experimenting time.