Query of compensation for fuse voltage drop on LM317 regulator.

[rfengg]

Hi there,

I am using a LM317 to create a regulated supply of output 5V from the data sheet using a fixed resistor and trimmer.
The supply is protected by a fuse of 1A which has a resistance of 0.2ohms.
When the supply is loaded with 1A, the output voltage drops to 5V-0.2ohms*1A=4.8V due to the finite resistance of the fuse.
Is there a way to dynamically compensate for the voltage drop across the fuse so that the voltage stays at 5V irrespective of the load current?

thanks

[Ian.M]

 Sure, put the fuse upstream of the LM317.   

There is no good way to compensate for the voltage drop across the fuse  if its downstream of the LM317, as the LM317's reference isn't ground referenced, so you cant simply sense the output voltage after the fuse.  That's a general problem with 3 pin adjustable regulators.

[rfengg]

Thanks Ian, there is already a fuse at the input of the 317, but our design requirement stipulates that the output should be protected by a 1A fuse as well.

[Ian.M]

Stupid design requirement.  Where's the current to blow the output fuse going to come from?

However if that requirement is set in stone, throw out the LM317 and look at regulators with more pins and ground-referenced internal references, so you can connect the feedback divider after the fuse.   Some will even give you adjustable current limiting as well.

[langwadt]

Thanks Ian, there is already a fuse at the input of the 317, but our design requirement stipulates that the output should be protected by a 1A fuse as well.

the 317 is current limited and thermally protected so that fuse will never blow ....

[rfengg]

Thanks Ian, there is already a fuse at the input of the 317, but our design requirement stipulates that the output should be protected by a 1A fuse as well.

the 317 is current limited and thermally protected so that fuse will never blow ....

The LM317 from Onsemi that I have is rated for 1.5A
I use 1A fast blow fuses from Shurter at the output and they have blown in the past before the current limit and thermal protection kicks in.

[Wolfram]

1.5 times the nominal current for a fast fuse is in the range where it might blow in a while, or never. The Schurter FSF 5x20 series specifies minimum 60 minute survival at 1.5 * Inom, with no maximum. LM317s also have a wide range of actual short circuit currents. Whether an LM317 willl be able to clear the fuse at all depends on tolerances of the fuse, the LM317, ambient temperature and the relative heatsinking of the fuse and the LM317. Not ideal design practice, if you want the design to behave in a predictable way in fault conditions.

[richard.cs]

If you knew the fuse resistance with confidence, you could pick a regulator with a slight negative output resistance, often referred to as "cable drop compensation" or similar, that winds up the voltage as the load current increases. It is significantly less good than using a regulator that senses after the fuse, but would be an option if it's hard to connect after the fuse (it's down the end of a cable or something). In any case a LM317 won't do either without an ugly external circuit.

In any case the best answer is usually to negotiate-out obviously bogus requirements that don't add any value to the customer. The trick is to deliver what they wanted rather than what they asked for. 

[Siwastaja]

Remove the fuse. LM317 has current limiting and thermal limiting, so it's unlikely the fuse will ever blow, so it has no purpose. Even if you manage to get it just barely blow, it still has no purpose. Input fuse is OK and protects against things smoking or melting if the LM317 fails as short circuit. Because with linear regulator, I_in = I_out, input fuse offers exactly same protection as the output fuse, PLUS additional protection if the LM317 fails as short between Vin-ADJ.

[rfengg]

Thanks to all for your valuable suggestions......am going back to the guys who spec'd  the requirement for an output fuse based on all your comments; might be it's an oversight on the requirements document.
 

[Siwastaja]

It's a weird requirement, I don't think I have ever seen one.

If you have a high-current regulator and branch out tiny branches, you might want to protect those with small fuses (like 100mA in case of 1.5A linear regulator), but a 1A fuse on a 1.5A linear regulator makes no sense. Also when you have just one fuse, it makes no sense regardless of values, because it could then be on the input side (even if it's much smaller than the regulator's rated current, for example a 63mA input fuse on an LM317 is just fine if the load does not draw more than this).

[mariush]

If you have the feedback for the regulator AFTER the fuse, then the regulator should automatically adjust output voltage to have 5v AFTER the fuse.
But like others said, if you already have fuse on the input there's no point for fuse on output.

The input current will be output current plus a few mA (regulator's internal power consumption).

[mikeselectricstuff]

Thanks Ian, there is already a fuse at the input of the 317, but our design requirement stipulates that the output should be protected by a 1A fuse as well.

If the input is protected by a 1A fuse, then so is the output by that same fuse, as all the output current comes from the input and flows through the fuse. Design requirement met.

[macboy]

If you have the feedback for the regulator AFTER the fuse, then the regulator should automatically adjust output voltage to have 5v AFTER the fuse.
But like others said, if you already have fuse on the input there's no point for fuse on output.
...

Not with a LM317 or similar regulators like LM1117/AMS1117. You will unfortunately get the opposite effect that you hope for, that is, the output will sag even more under load. Build it or simulate it, you will see.

[rfengg]

Already tried this by putting a fuse directly at  the output of the 317, before the connection to the 240ohm fixed resistor.......does not work. 

[TheMG]

It's a linear regulator, for all intents and purposes current in = current out. So putting your fuse at the input of the regulator accomplishes the same thing as putting it at the output, and entirely avoids the issue of voltage drop on the output.

[harerod]

Quote from: mariush on Yesterday at 20:23:01

If you have the feedback for the regulator AFTER the fuse, then the regulator should automatically adjust output voltage to have 5v AFTER the fuse.
But like others said, if you already have fuse on the input there's no point for fuse on output.
...

Not with a LM317 or similar regulators like LM1117/AMS1117. You will unfortunately get the opposite effect that you hope for, that is, the output will sag even more under load. Build it or simulate it, you will see.

Or try to understand the operating principle of those regulators' feedback loop: The control circuit will try to maintain a constant voltage difference (Vref) between the output and the adjust pin. The "measurement resistor" will become part of the upper leg of the voltage divider.

That feedback trick may work with regulators which regulate the voltage on their feedback input in reference to their "GND". When expanding the feedback loop, i.e. introducing additional transfer funtions, one may need to adjust the control loop's compensation.

[floobydust]

Best picture I can find of the circuit layout for LM317, LM350, LM1117 etc. linear regulators. RW is the wiring resistance. From On-Semi LM1117 datasheet.
Note the need for a single-point ground, and for years I had it wrong- thinking the voltage sense resistor R1 had to connect right at the output terminals. No, it best connects right to the IC's output pin terminal, which is counter-intuitive.

I remember the 1976 National Semiconductor datasheet diagram showing how to deal with line resistance and the LM317, but would have to dig for it.
The IC can't correct for (+) line resistance, output is referenced to the output pin directly. Modern datasheets show how to do it wrong, with the resistor connected after the line resistance.
edit: added LT1083 line resistance circuit, same as LM317, LM3338, LM350 etc.

[David Hess]

There is no good way to compensate for the voltage drop across the fuse  if its downstream of the LM317, as the LM317's reference isn't ground referenced, so you cant simply sense the output voltage after the fuse.  That's a general problem with 3 pin adjustable regulators.

The circuit shown below adds remote sense to an adjustable 3 pin regulator.  The Linear Technology datasheet for the LT317A has the same circuit but this one is clearer.

[langwadt]

There is no good way to compensate for the voltage drop across the fuse  if its downstream of the LM317, as the LM317's reference isn't ground referenced, so you cant simply sense the output voltage after the fuse.  That's a general problem with 3 pin adjustable regulators.

The circuit shown below adds remote sense to an adjustable 3 pin regulator.  The Linear Technology datasheet for the LT317A has the same circuit but this one is clearer.

must be really really set on using a 317 to add all that stuff instead of picking another regulator

[Ian.M]

Yes.  Adding an OPAMP in the feedback loop is not going to help stability, and its telling that one with external compensation is used in the above circuit.  Also the OPAMP in question needs good performance right up to its Vcc rail.  I'd bet you could get in a lot of trouble substituting a modern OPAMP in that circuit . . .

I still assert that for a three terminal regulator: "There is no good way to compensate for the voltage drop ..."

You'd have to have an odd but *very* good reason to prefer the above circuit over using a modern ground referenced regulator with Gnd and the feedback divider connected at the load to implement remote sensing,  with diode clamping of the sense lines to limit the load overvoltage if they get disconnected.

[David Hess]

must be really really set on using a 317 to add all that stuff instead of picking another regulator

It is likely still less expensive than a regulator which includes remote sense, and the 317 has multiple sources.  There have been a lot of integrated voltage regulators with separate feedback and sense pins since the 70s, and a vast majority are out of production.

Yes.  Adding an OPAMP in the feedback loop is not going to help stability, and its telling that one with external compensation is used in the above circuit.  Also the OPAMP in question needs good performance right up to its Vcc rail.  I'd bet you could get in a lot of trouble substituting a modern OPAMP in that circuit . . .

Check the circuit again; the regulator is not within the feedback loop of the operational amplifier.  There is no requirement other than unity gain stability; the operational amplifier is operating as a follower.  The same compensation scheme could be done with a buffer or couple of transistors, but a jelly bean operational amplifier is the least expensive way to get low offset voltage.

The virtue of this circuit is not requiring an external reference, but of course that could be done also.  A reference, single operational amplifier, and regulator could implement remote sense rather than voltage drop compensation, as shown below, but the above circuit will usually be close enough.