Quick LM317 Question: Bypass

[thehallofshields]

Okay; I'm a fool. I designed a pedalboard with an LM317 without taking account of the 1.25v drop.

The goal was to allow myself to run off of 9v or 12v input by using the 5k pot in the standard circuit, and now my 9.5v 1Spot Power Supply is down to 8.4v.



I've got a couple ideas to circumvent rewiring and re-drilling for a proper bypass switch, but I'd like to avoid blowing the caps or frying the chip, so I thought I'd ask first.

1.) If I connect the input and output (connect pin 2 and pin 3 to bypass the 317) what is the expected result?

2.) If I attempt to bypass by creating a separate power jack that connects directly to the 317's output, will I loose significant current and/or voltage through the 5k pot?

Thanks,

-Matt-

[R.G.]

As a proposed alternative, why not switch in the LM317 when the voltage exceeds 11V, and switch it out when it's below 11V?

How (I can hear you asking...  ) can I do that?

Well, it takes some thinking. First you need good, low-voltage drop power switches. For those, you can use either PNP transistors or P-channel MOSFETs. Either of these will happily switch large(ish) currents with little loss in the switch device.

Then you need some way to turn the switch into the LM317 on and the switch around the LM317 off when the voltage is above 11V, and vice versa. Hmmm. How about a couple of comparators? The LM358 dual opamp can be mis-used as a couple of comparators, and will automate the switching. It'll take some resistors and such to set it up, but it can be done. The LM358 withstands up to 32Vdc as a power supply, and can turn on/off a pair of P-MOSFETs directly. The PNPs might need an additional transistor to suck enough base current through them to keep them well saturated.

This circuit would sit ahead of the voltage regulator, and monitor the input voltage. Below 11V, it runs the incoming voltage straight in, and turns off the switch to the input of the LM317. Above 11V, it turns off the direct path, and turns on the one to the LM317 to regulate.

I would not short the input and output of the LM317 without some in-depth study. Might be OK, might not.

The current lost to R2 and the 5K pot is just ohm's law: I = V/R. V is the voltage that appears on top of R2, and R is R2 plus the value of the 5K pot at its current setting - which can be as low as zero, right?

[thehallofshields]

Oh jeez, I'm trying to fit this in a smallish space.

Here's what I'm really hoping will work to switch between a 9v and 12v supply.


In (1), I fear that the 317 chip will overheat.

In (2), I'm worried that too much voltage will be dropped across the 5k pot.

are these suspicions well founded?

[PRR]

I would not trust (myself) with a manual switch.

However, this is safe:



> I loose significant current and/or voltage through the 5k pot?

> the value of the 5K pot at its current setting - which can be as low as zero, right?

Zero only if the output setting is 1.25V.

The puzzle says "9V". So the pot should be set near.... umadaumada, well, let THoS compute it or measure it. (Or better: find the value and insert a FIXED resistor.... don't trust pots, especially since in this case _when_ the pot-wiper loses contact, the LM317 Out goes to FULL supply (minus 1.1V).

Brown Green Red.

[PRR]

> I'd like to avoid blowing the caps or frying the chip

If it needs to be bullet-proof, there's one or two diodes you *really* should have.

If there's any significant capacitance on LM317 Output (you can't be sure) and you short the Input (stuff happens.... spike-heel on the power feed), that output cap dumps *backward into the short *through* the LM317, backwards! The datasheet suggests "breadboards" but I suggest that unless the WHOLE system is inside one case in a calm environment, that diode could save a '317 and a good gig.

There's another diode for any capacitance on the ADJ pin, but your plan omits that cap.

[thehallofshields]

Excellent!

Thank you very much, this is exactly what I was looking for. I just couldn't figure out if the input or output needed to be cut. I still wonder how the chip would behave if it had gone the other way...

Anyway, all boxed up and done. My pedalboard can now switch between center- and center+, and has a slider switch on the side to switch between unregulated (317 bypass) for a 9v wallwart and regulated (5k external pot) for 12v supplies (like my car).

I also threw a digital voltage readout in there which I've never done before. I would say that I couldn't be happier. Except....

I'm pretty disappointing with the LM317 with a 9.5v supply it drops down to 8.4v and then goes down to 7.8v under load from a single analog pedal.

That's not really going to be a problem when running off a 12v supply, but my 1Spot stays at 9.5v under the full load of digital reverb 4 analogs and a ruby amp. I don't quite understand.

[PRR]

> disappointing with the LM317 with a 9.5v supply it drops down to 8.4v and then goes down to 7.8v

It has to have a "thingie" in-line to drop high voltages down. This thingie has to have high current gain. And when the LM317 was new, it had to be basic IC processing (which pretty much meant NPN for all high-current paths).

LM317 uses a Darlington: http://www.fairchildsemi.com/ds/LM/LM317.pdf

So even if the triangle-thing could pull all the way up to 9.5V, the Darlington's emitter (the LM317's output) could apparently only go to 9.5V-1.2V= 8.3V.

You got a bit better no-load. Probably only half the Darlington can do that job. It looks like "single pedal causes the full 1.2V drop in the Darlington, plus a half-Volt drop in the triangle thingie (the control circuit).

Think of a water-valve, or an air pressure regulator. You can put in 100psi and get out 40psi. However if you try to set 99psi, but it takes 10psi of drop to make the regulator gizmo do its thing, it will give 90psi not 99psi.

There are newer regulators with PNP power devices. They are marketed as "low drop-out" or "LDO". LP2951 can do 0.1V up to 10mA, 0.4V at 100mA. Downside, it only comes in SMP (teeny) package. There must be others.

> my 1Spot stays at 9.5v under the full load of digital reverb 4 analogs and a ruby amp. I don't quite understand.

The 1Spot basically starts from 100+ Volts. It probably has some initial step-down, so the key parts can be 30V rating... but there's a lot of headroom and it was designed to work with 9.5V out at quite large currents.