Not being able to understand simple LM317 lead acid battery charger

[Mosaic]

I don't see any schematic diagram attached to your post#20.

 

[Willen]

Sorry, see in the post #20 again for the diagram.

 

[MikeMl]

If V1 and C3 in that circuit represent a transformer, rectifier, and filter capacitor, then C3 is much, much to0 small to be a proper filter capacitor. Hence the regulator drops out because its input voltage sags below its Drop-Out Voltage (See Vdo on the data sheet).

To figure the correct value for the filter capacitor, do this simple calculation: (You have to make certain assumptions):

Assume a load current for your power supply is I=5A
Assume that the voltage in C3 can drop 3V (peak to peak Ripple) between current pulses from the rectifier so Δ=3V.
Assume you are using 50Hz AC and full-wave rectification, so time between pulses is t=10ms

∫I*t = q = C*Δ

So: C3=∫I*t/Δ = 5*0.01/3 = 0.017F = 16,667uF

So here is a sim to confirm the above. Pay attention to the peak currents required of the simulated transformer and rectifiers. The ripple in the sim is ~2V (not 3 as assumed) because the time that the filter capacitor has to supply the entire 5A load current is only ~7ms (not 10ms, as assumed) so the calculation yields a filter capacitor value which is a bit bigger than necessary.

 

[Mosaic]

Here's a fundamental explanation of why filter cap sizing is required as per max current demand. As a side note, this is why higher frequency switching power supplies can use smaller xformers & caps etc. Mike's airplane prob runs @ 400hz or so power gen. to reduce transformer and cap sizing requirements.
https://www.radio-electronics.com/i...er-filtering-smoothing-capacitor-circuits.php

 

[MikeMl]

...
I note@MikeMl has 'disliked' my post. It seems he believes his posts/circuits cannot be improved upon.

No, my posts can always be improved on. However, when an OP asks a specific question, in this case about how a current-limiter works, why did you feel it is necessary to inject a posting that had nothing to do with the question at hand? I dislike your disagreement with my critisism of your irrevelent posting.

 

[Mosaic]

You may think it is irrelevant, but your circuit would have led to further functional problems for the OP, thus I pointed that out. This is why.

If a member asks for advice and is shown a circuit that has thermal failure staring it in the face with no mention of such issues, further advice is required to prevent poor results if such a circuit is attempted.
Such is the benefit of a forum. Now the OP has more than one POV on that sample circuit and that is to his benefit.

 

[MikeMl]

You may think it is irrelevant, but your circuit would have led to further functional problems for the OP, thus I pointed that out.

But what you didn't notice is that the regulator used by Willen in his original post is a LT1083, which is rated at 8A and a whopping 60W dissipation when bolted to an appropriate heatsink.

 

[Willen]

Hi again,
I had seen some capacitor multiplier circuits where the base was regulated by zener and was used just 10uF capacitor and the transistor were passing supply from Collector to Emitter, smooth regulated output (same as with 4600uF). So it was called capacitor multiplier.

Same way cannot I use a small value capacitor in the ADJ pin or on its divider point to get more smooth supply with small filtering capacitor?

Another: I downloaded a LM317-TI model and simulated; but got heavy ripple output. Run this simulation (attached) and what is the wrong? (wrong model of wrong circuit?) maybe you have different model of LM317, no matter anyway.

 

[Mosaic]

But what you didn't notice is that the regulator used by Willen in his original post is a LT1083, which is rated at 8A and a whopping 60W dissipation when bolted to an appropriate heatsink.

Please note the TITLE of the thread. The OP is concerned with an LM317....I don't see u commenting on Tony's ref circuit to the LM317 as an issue...you are picking and choosing and its childish.

 

[MikeMl]

Hi again,
I had seen some capacitor multiplier circuits where the base was regulated by zener and was used just 10uF capacitor and the transistor were passing supply from Collector to Emitter, smooth regulated output (same as with 4600uF). So it was called capacitor multiplier.

Same way cannot I use a small value capacitor in the ADJ pin or on its divider point to get more smooth supply with small filtering capacitor?

No. The energy-storage filter-capacitor required to eliminate ripple has to be connected at the rectifier, otherwise the load current discharges it between the current pulses from the rectifier. It must hold up the input voltage to the regulator above the regulator's Drop-Out Voltage.

The capacitor multiplier circuit is a second stage of filtering; it does not obviate the need for the primary filter capacitor. Just as with the regulator, the capacitive multiplier also has a "drop out" voltage, so the primary filter capacitor still has to be there, and be big...

Oh, and somebody should chime in and say that the transistor used for the capacitance multiplier needs to be mounted on a heat sink, too

 

[Willen]

Please run the file 'SIM_current boost LM317.asc' (Attached above in). I am getting few hundreds ripple at output and it's amazing to me.

 

[MikeMl]

It is oscillating. You have too many, and the wrong value bypass capacitors. I went back an looked up the current-booster recommended by National (ne TI) on the TI LM317 data sheet. They use nothing on the input pin, 10uF on the adjust pin, and nothing to 1uF on the output pin. That stops the oscillation.

 

[Willen]

It is oscillating. You have too many, and the wrong value bypass capacitors. I went back an looked up the current-booster recommended by National (ne TI) on the TI LM317 data sheet. They use nothing on the input pin, 10uF on the adjust pin, and nothing to 1uF on the output pin. That stops the oscillation.

Bypass capacitors? Did you mean decouple capacitors? I thought that adding too many decouple capacitors makes the supply stable.

Please will you post that particular ciircuit (by National)?

 

[MikeMl]

bypass = decouple

Can you go to TI.com and download the LM317 datasheet?

 

[Willen]

Hi,
Here I simulated again using some tricks found on TI datasheet (example circuits). I got oscillation of different KHz (10KHz to 35KHz). Then I used an inductor at output then got solve.

Can anyone tell me why some simulation runs within a second and some simulation takes almost a minute for same circuit and same stop time? Please check this circuit without and with L1. Without L1, it took me approx 20 seconds to complete the simulation. And with, L1 it took just a second! AMAZING! why?

If you found the core thing about such runtime that then I wouldn't face LONG....runtime problem again and again.

 

[ericgibbs]

hi Willen,
Ref your PM.
I had to use my LM317 and TIP32 models for your circuit, it runs equally fast with L1 in or out.

Where did you get the LM and TIP models from.?

Eric

 

[Willen]

hi Willen,
Ref your PM.
I had to use my LM317 and TIP32 models for your circuit, it runs equally fast with L1 in or out.

Where did you get the LM and TIP models from.?

Eric

I had downloaded a huge pack of zip file (approx 1.5MB) somewhere around internet. After extract, it become approx 15MB.

Equal fast, how long? My PC took more than 20 seconds when used the inductor. And took just a second when I connected directly (without the inductor).

Do you also get long runtime in some circuit sometime?

 

[audioguru]

I do not have a model of an LM317 so I could not run the sim. I think the circuit is oscillating because it cannot drive the HUGE 1000uF output capacitor. The datasheet for the LM317 does not show the circuit with the PNP current-booster circuit but shows a similar circuit with a 47uF solid tantalum output capacitor. Fairchild's datasheet for the LM7805 shows the PNP booster with a 0.1uF probably ceramic output capacitor.

 

[audioguru]

I do not have a model of an LM317 so I could not run the sim. The sim has a DC input so it could not be mains ripple but must be oscillation. Maybe because it cannot drive the HUGE 1000uF output capacitor.
The similar current-boosting circuit in the latest LM317 datasheet shows a 47uF solid tantalum output capacitor and Fairchild's LM7805 datasheet with the PNP current-boosting transistor shows a 0.1uF output capacitor.

 

[ericgibbs]

Equal fast, how long? My PC took more than 20 seconds when used the inductor. And took just a second when I connected directly (without the inductor).

hi,
Its less than 2 seconds for both simulations.
E

Extract from this zip the LM317 sub and asy files