How to improve regulation of a power supply

lm338 short circuit breaker

How can we impove the load regulation of a power supply? By adding bypass capacitors. How do we calculate their values :?:
I have attached my circuit. I had pretty good regulation, when i didn't conncted my adjustatable current circuit. The regulation is around 20% after adding that circuit. Why does this effected. I thought Vce<0.5V(Q1) and shouldn't cause any problems. I tried to connect the resister bridge(R1 &R2 ) after Q1. Still that doesn't solve the problem
Also in the datasheet, it shows that a protection diode needed to protect from bypass caps. Will that comes to action when the output is switched off

Can any one help me with my regulation problem plz :roll:

It is a strange design where you try to achieve two different functions (voltage regulation and current limit) using two connected in series stages.
It will just not work!!!

Try to connect external transistor to the voltage regulator and implement current limit there..



Or, have a look at the L200 voltage regulator.
**broken link removed**
(Fig 35 and Fig36)..

Load regulation is not improved by the addition of capacitors, only by high loop gain and good layout. If you need more info, please post.

Your load regulation suffers because you now have a transistor (turned on by a simple resistor) and a 0.25Ω resistor in series with your regulator output. The voltage is sensed ahead of these components, so all they do is add to the "internal resistance" of your regulator.
The load regulation is in fact a measure of how low this internal resistance of the power supply is, so you want to keep it low.

Both functions should be implemented using the same pass element, but when you switch from voltage to current control, one error amplifier should take over.

One way to modify your schematic is to remove R4, Q1 (short it out), connect the bottom of pot R2 to "Output Ve-" and the now freed collector of Q2 to the ADJ pin of the LM318. That should make it work better.
NOTE THAT THIS WILL NOT LIMIT THE SHORT-CIRCUIT CURRENT TO THE VALUE YOU INTEND. THAT WILL ONLY BE LIMITED INTERNALLY BY THE LM318.

Your resistor R4 (10K) seems to be high enough to allow enough bias current for the transistor.

The problem with regulation ariases from the fact that the second stage, current llimitter, has high internal resistance. Either you redisign it or find another method to implement it.

It is also a good practice to put a smal electrolitic cap at the common point of R1,R2 and Adj of the regulator. The negative of the capacitor must go to ground. This does not affect the regulation directly but it helps the operation of the regulator.

One of the solution for your problem can be to move all your current limiting circuit in front of LM338. Current limit function will remain but regulation will stay intact.

I can't use L200 unfortunately, as I need upto 3 amps of current
And If connect Q1 between Adjust and ground, still there is 1.2V appears at the output and I want to isolate the power supply completely from load.

When I checked my circuit, I found most of the drop is across Vce of Q1 (3.5V with 220R load). The drop across rest of the circuit is in millivolts.
Are there any power transisters that has very little drop across the Vce/Can I use any other switching devices.

Also the reason for choosing 10k is to limit the amount of current when the current limit ckt operated.

Does the regulator consume considerable amount of current to maintain regulation. If so this will effect the final output current after the regulator. Can we avoid this? I would like to give a try anyway

How do the manufactures of normal bench power supplies do both regulation and
adj. current limiting. Any more ideas

Thanks

Hi,

LM339 is a comparator with a open collector output, is that correct? Is that the chip that you are using in your current limiting loop?
If so, where's your pull up?
Have you considered the stability of your current limit loop?
I do not think your circuit is stable, and I thought the first thing you must do is to make sure it's stable!
I have a simple suggestion but you need to get some new parts.
First, you can throw everything after the LM338 away, including your crude 5V zener based regulator.
Use LTC4210 from Linear Technology at the output of LM338, and use the typical app circuit in the datasheet. You disable the circuit breaker if you want to by connecting TIMER pin to ground. I think LTC4210 is not that expensive. You choose a small sensing resistor to improve the load regulation. The sense resistor also determine your current limit, as your original circuit does.
Is this a student project? You can request samples from Linear and get it pretty quickly if u r resided in US/California.

Dance with me? It's my last dance....

Power supply adjust the current limit and output voltage by over-powering the node that will subsequently drive the output power transistor. You are not doing that...and you can't bcos you do not have access to this node, it's in LM338.

How do the manufactures of normal bench power supplies do both regulation and
adj. current limiting.

Click to expand...

Both functions are performed by the same serial transistor which is controlled by 2 error amplifiers: voltage and current.
Depending on situation one of them takes control over this transistor.

If you want to have good power supply with voltage and current regulation (>2A) you can not relay on off-the-shelf IC. I would start from scratches and probably look at the old 723 voltage regulator ..

A simple and nice design is here: https://www.electro-tech-online.com/viewtopic.php?t=176

Just to get the idea of current limit implementation.

Looks like the +5V used as Vcc and reference of the OPAMP input doesn't regulated enough

lastdance said:

LM339 is a comparator with a open collector output, is that correct?

Click to expand...

Yes LM339 is a comaprator with open colletor output, so the circuit shown in the first post will never work correctly. Basis of Q2 will just float when it's not pulled down by LM339. LM339 can never turn on the tranisistor because it's open collector.
You need to move R5 as shown in my schematic.
The top of the first picture is your original design and the bottom of the first picture is with R5 moved to work correctly with an open colletor output.

But actually it would be easier to remove both R5 and Q2 completely from the schematic and connect the open collector output of LM339 directly where the collector of q2 is now connected.
When you do this you also need to swap IN+ and IN- of LM339 because q2 is not there anymore to invert the output of LM339.
This is shown in the second picture.

Even if u correct the open collector, there's another issue. Stability.
LM339 is a comparator, it's going to be a bang bang kind of response. imagine u short the output with a small resistor, the sense voltage would immediately increase, and your comparator, being fast, high gain and powerful, would in no time shut off the pass gate. it would then just turn on/off until the short goes away.
My point is, If you wanted to do regulation, it's an opamp u should use. but even if an opamp is used, u would still have stability issue because of the inherent dominant pole. you have a low frequncy pole at the base of BJT, coupled with the opamp pole, I think it's a miracle to get it to work.
If there's an opamp where you can control the dominant pole by external component, then you may give it a shot. It's difficult to compensate system like this.
Now, even with all these solved, you are left with the problem why this post arised, the regulation issue. The 0.26 Ohm is gonna modulate your output voltage as your load varies, which of course can be mitigated if u use a smaller value, like <10mOhm.

Yes you are right, this will never work correctly, but just oscillate.

Why not just throw everything after LM338 away and the ad an additional LM338 and use this second LM338 as current limiter.
Using LM338 as current limiter is shown at the top of page 12 in the datasheet:
**broken link removed**

Iout= Vref / R1 <=> R1 = Vref / Iout.

If you for example want Iout to be 4 A then you can calculate R1 (Vref = 1.24 V):
R1 = 1.24V / 4A = 0.31 Ω

well, u got your current reg, but loses the voltage reg.

You can use two LM338 and have both current limitation and voltage regulation.
Just use one LM338 as current limiter and the other as voltage regulator.

It is a strange design where you try to achieve two different functions (voltage regulation and current limit) using two connected in series stages.
It will just not work!!!

Click to expand...

I posted that 3 days ago and it look it is still valid ..

Yes it will work, just connect the LM338 used as current limiter after the LM338 used as voltage regulator. I don't know why you think this doesn't work. The current limiter is connected to a stable voltage from the voltage regulator.

A better method is shown in this schematic though. This is take from the LM317 datasheet, but you should be able to replace LM317 with LM338:

0.31 ohm multiplied by the load current, let's say 3A, you got about 1V. your voltage output is supposed to stay regulated as much as possible, before hitting the 4A current limit. 1V is too much, it's gonna kill the load regulation.

IanP, this can be done. (series/series)
Check out
https://www.linear.com/pc/productDetail.do?navId=H0,C1,C1003,C1006,C1163,P2237
before jumping to conclusion, but if you think it's wrong, pls point out the mistake.

I added what I beleive is the simplest and cheapest soloution in my previous post.
This is fom the LM317 datasheet, but should work with LM338 instead.
The good thing about this circuit is that the current sensing resistor are place after the GND point, so it doesn't influence the voltage regulation.

lastdance said:

0.31 ohm multiplied by the load current, let's say 3A, you got about 1V. your voltage output is supposed to stay regulated as much as possible, before hitting the 4A current limit. 1V is too much, it's gonna kill the load regulation.

Click to expand...

Then how about just placing the current limiter before the the voltage regulator. I have seen this done before with two LM317, so I suppose it works.

yes, the last circuit should work, and I agree it's one of the easiest method.
But I think the output can't be regulated down to 2V, can it?
It's good work from u, ME..