# LM723 Current Limiting Questions

Discussion in 'General Electronics Discussion' started by Gristle McThornbody, May 12, 2012.

1. ### Gristle McThornbody

31
0
Mar 14, 2012
I'd like to understand how current limiting works in a LM723 voltage regulator. I get that the LM723 current sense and current limit pins (2 & 3) see a voltage created by the very low restistance current limiting resistors and that when this voltage reaches a certain value, the IC shuts down or limits the power supply output.

How or where is this voltage parameter noted in the device specs? I believe that I read somewhere that 0.6 volts is the trigger point, but I can't find anything like this in the spec sheet..

What happens as this voltage reaches and exceeds 0.6 volts? Does the device reduce output voltage? If so, does it do this until the voltage drops back to 0.6 volts and stop there?

What happens during a current surge (output short circuit)?

TIA for any help or for any links to published explanations.

2. ### (*steve*)¡sǝpodᴉʇuɐ ǝɥʇ ɹɐǝɥdModerator

25,505
2,849
Jan 21, 2010
Gee, I think it was 1979 that I last used an LM723. And yeah, it was in the TO-5 metal can too.

Here is the datasheet.

If you take a look at the equivalent circuit (on page 2) you will see that the current limit and current sense pins are simply tied to the base and emitter of an NPN transistor which has its collector tied to the output of the error amplifier.

Simply speaking, when the current limit pin of the 723 is more than 0.6V more positive than the current sense pin, the output of the error amplifier is pulled low, reducing drive to the output series pass transistor.

This may be done in a number of ways. The datasheet shows the sense resistor placed in the output prior to the voltage sensing, but it can also be done in the ground return.

The simplest example is on page 2 just under the equivalent circuit.

Table 2 on page 7 gives tot he formula for calculating the sense resistor value. Vsense is probably something like 0.6V. but it may be specified somewhere. Note that it varies with temperature, the table on page 4 illustrates that.

3. ### Gristle McThornbody

31
0
Mar 14, 2012
Better Choice than the LM723 ?

Okay, I'll grant that the LM723 has been around for quite a while. Is there a similar, newer device that will do a better job for me This is a 22 volt, 10 amp, low ripple power supply. I've attached an article with schematic that I have built. It works reasonably well but even with 10,000 mf filtering, I'm still getting about 0.5 volts of ripple at 8 amps output.

Also, I've played with the current limiting section (thanks for the help) by using a high quality multi-turn pot and monitoring the voltage across the current sensing pins. Sure enough when it reaches about 0.6 volts, the output starts to drop rapidly. However, when I reduce the load, the output doesn't come back by itself. Is this normal?

I notice that at full load, the 0.1 ohm emitter resistors don't get particularly warm, but the two paralleled 0.47 ohm current limit resistors get hot as hell. Is this expected?

Finally, I'd like to test the "short circuit robustness" of this circuit, but I'm hesitant to just short the output. any suggestions?

Well, anyway, if there is a better LR and circuit that I should be using, I'd like to know about it. Or, if there are some improvements to this existing circuit, that'd be cool, too. Thanks.

File size:
49.4 KB
Views:
14,604
4. ### (*steve*)¡sǝpodᴉʇuɐ ǝɥʇ ɹɐǝɥdModerator

25,505
2,849
Jan 21, 2010
There are *many* newer chips. You might consider a switchmode pre-regulator followed by a linear regulator for the best thermal efficiency combined with reduced high frequency noise.

I suspect you mean 10,000uF (10mF)

Getting that much ripple suggests to me that the regulator is either *very poor* or the input voltage is sagging under load.

I would take a close look at the input voltage waveform.

If you have managed to create a foldback current limit then this can happen, but it shouldn't from what I see of your circuit.

Let's assume that full load is 8A through 4 transistors. So each of those 0.1 ohm resistors has 2A flowing through it for a total power dissipation of 0.4W per resistor.

Now, each of the 0.47 ohm resistors have 4A flowing through them, leading to a power dissipation of 7.52 watts each. So they are dissipating about 20 times the amount of power. No wonder they get hotter.

The first thing to do is to get a heavier and heavier load until you observe the power supply going into current limiting.

Then, whilst carefully monitoring the temperature of the output transistors, increase the load and monitor both the output current and voltage. You should notice the output voltage fall and the current remain close to constant.

Your output transistors will start to dissipate seriously large amounts of heat, so stop if the case temperature exceeds some safe value.

On a power supply like this, a friend of mine had a thermistor attached to the pass transistor which pulled the current limit low if a preset temperature was exceeded.

Another issue with the 723 is that it is seriously limited in power dissipation and this generally limits the output current to a relatively small value. You may find that the 723's output current needs to be monitored to ensure that you don't exceed this at maximum current and minimum voltage -- during short circuit conditions this is the worst case.

Most modern regulators are more rugged in that they have some sort of over temperature and over current protection. The 723 comes from the age of steam electronics and I would suspect that its performance is probably not up to par compared to current designs (most of which are a *lot* simpler to use).

What you might choose to use depends a great deal on what you want from your power supply.

Check out this for an example of the process you might go through.

5. ### Gristle McThornbody

31
0
Mar 14, 2012
Thanks, Steve. I watched the video you recommended and found it useful (that guy's a kick in the butt!) but I'm not at a point where I am comfortable designing a PS from scratch (there's a real understatement). However, I am understanding tha concepts better as I go along.

At this point, I guess I'm looking for a finished circuit that I can build. I'm probably going to end up building a second power supply for bench work as well, so I'll spec this to cover both PS's. Here are my criteria:

- Variable voltage output (30 or 35 volts dc max) at 15 amps (I've got the front end hardware for this current load.)

- Variable current limiting from 2 amps (or lower if possible) to max 15 amps

- Good short circuit protection

- I've got a bank of 2N3055's already built from the previous LM723 version and I'd like to use them for this supply if possible. (I know the 317 is good to only about 1.5 amps.)

Overall, the LM317 circuits I've been able to look up seem remarkably simple. Is this because the 317 has many features built in? Or do I need more sophistocated circuitry that I haven't yet seen?

Anyway, if you or someone has a PS circuit available that can do these things, I'd appreciate you sharing it. Thanks.

P.S. I assume that it will use a LM317 or something newer. I looked but I don't see info out there on how, for instance, current limiting happens with this device. Will I use something like the two regulators in series like in the video with the first for current limiting and the second for voltage control?

Last edited: May 14, 2012
6. ### (*steve*)¡sǝpodᴉʇuɐ ǝɥʇ ɹɐǝɥdModerator

25,505
2,849
Jan 21, 2010
The LM317 is pretty much the basic design that many linear regulators follow (sometimes with a couple more pins).

You have an input, and output and a voltage sense pin. In the 317 the voltage sense cleverly doubles up as the ground pin, but there are other regulators that have separate ground pins and pins to shut the regulator down, etc.

A linear regulator capable of a 15A load and variable output from 5? to 30 volts will be large, heavy, and expensive. you're looking at 450 watts of energy dissipated as heat. You would probably need 8 2N3055's to do this, coupled to a really big heatsink, probably with forced air cooling.

For these types of power outputs, a switchmode regulator is far better.

Something like this or this. And you'll almost certainly be able to source these for *far* lower prices. Here is one example.

7. ### Gristle McThornbody

31
0
Mar 14, 2012
Well, I've already got the front end hardware plus the bank of 3055's so I'm going to go ahead and build it. I think I'd like to stay with a linear design. Are there any good circuit designs laying around?

8. ### (*steve*)¡sǝpodᴉʇuɐ ǝɥʇ ɹɐǝɥdModerator

25,505
2,849
Jan 21, 2010
If you've already started with the 723, there's no real reason to change. You do have to be careful about the dissipation of the device, but that's the major factor I've seen bite people (including myself)

The amount of drive required by the pass transistors should not be underestimated.

9. ### Gristle McThornbody

31
0
Mar 14, 2012
Recall that I have a drive transister (BD139) between the 723 and the bank of 3055's, so no heat dissipation issues.

I've doubled the transformers (see other thread) and that is working great, so I've got lots more front end power. However, I am still seeing the regulated voltage start to droop below my 22.0 volt set point at about only 8 or 9 amps. At this point, I've still got about 28 volts at the front end supply. Any idea why it's drooping? Regulation is perfect up to that point. I have monitored the current limiting voltage (across 2 and 3) and made sure it is well below 0.6 volts (the current limiting appears to work well).

As an aside, I put my scope on the regulated output and it shows very little ripple up to about 3 amps output and then goes nuts. At all higher loadings, I get a sort of spirograph looking thing that I can 't stabilize no matter what. My Fluke MM shows no AC ripple. I want to believe it, but I'm confused by the scope image.

BTW, I've got my hands on a couple of 317's and I'd like to try a circuit with variable output and variable current limiting (like the video suggested). Do you know of a finished circuit I could try?

10. ### (*steve*)¡sǝpodᴉʇuɐ ǝɥʇ ɹɐǝɥdModerator

25,505
2,849
Jan 21, 2010
The problem is probably that your regulator needs quite a bit of headrom to supply the base of the transistor inside the 723 that is now a further 2 diode drops above the output voltage.

I imagine it's beginning to fall out of regulation.

Compare the voltage on the collector of the main pass transistors with that on the output pin of the 723. You probably need a volt or two difference here (check the specs)

If the voltage difference falls too far (check on your scope) then this would explain the bad things happening. Make sure your circuit is NOT earthed if you do these differential measurements with a scope. If it is, measure the ground to collector, and ground to 723 output as 2 different traces and look at the difference.

Current limiting is hard with 317s

11. ### (*steve*)¡sǝpodᴉʇuɐ ǝɥʇ ɹɐǝɥdModerator

25,505
2,849
Jan 21, 2010
Just checked the output of the 723 cannot exceed Vcc-3V.

Given that you have 2 further diode drops, this means your input voltage must always be at least 4.2V higher than the output.

I'm not overly confident that you will never need more than 85mA drive to the BD139 either...

12. ### Gristle McThornbody

31
0
Mar 14, 2012
If the 317 can't handle current regulation very well and the 723 is a dinosaur , then is there some other regulator that can do a better job? What is used these days in commercial bench power supplies?

I'm pretty well set in terms of the front end (transformers, rectification and filtering) and the back end (a heat sunk array of 3055's). So I'm really just looking for the regulation and control section, which is the most complicated, but costs next to nothing compared to the other hardware. I can build up the front and back sections in a permanent way and then muck around with different regulation circuits until I find something that I like. Does that make sense? At least in the meantime, I'll have something that more or less works, and I'll have fun playing with and understanding these circuits.

And in all fairness to you and the others, this has become a bit of a moving target. I started out with a simple set of performance criteria and have now added a second power supply with more requirements to the mix.

And as always, thanks for all the help.

13. ### (*steve*)¡sǝpodᴉʇuɐ ǝɥʇ ɹɐǝɥdModerator

25,505
2,849
Jan 21, 2010
There are about a bazzilion different regulators. In all fairness, I pointed you to a video which described one person's path to find a regulator, and you have come up with both the 723 and the 317.

The problem with the 723 and your losing regulation is one that will affect all regulators to some extent. The fact that the 723 has a separate Vcc means that you are able to operate it from a slightly higher (and more stable) voltage source. It may be as simple as a diode and a capacitor.

With a linear supply, you are always going to be up against the issues of drooping input voltage from the transformer(s) and also from your filter capacitors. The ripple current into these capacitors is also going to be huge. And this fact is totally independent of your regulator.

This is one reason for a switching pre-regulator. You can start with a higher voltage DC source and largely eliminate the problems caused by what happens between the mains cycles.

You are painting yourself into a corner with your unregulated supply that no linear regulator is going to get you out of.