LM317 Power Supply for Small Amp...

[mac]

Is it me, or a LM317 PS is not the best choice to feed a 1w to 3w smal amp?

I've been using my 9v and 18v LM317-based PS for decades to experiment with pedals, they are as quiet as a Carl Martin PS I also have.
But recently I noticed that when I used them to power up a small amp like a Deacy on steroids I get an annoying, tiny distortion more noticeable when playing two notes in the middle of the freatboard using single coils and the amp set to clean (like intermodulation in Class B or a hi-freq sawtooth mounted on the signal, can't find the right words).
I always blamed the breadboard, the planetary aligment, the Gov, etc.
I took for granted that the venerable LM317 is king...

Yesterday I was playing with a Class A BJT+OT at 200ma idle,18v LM317 PS.
After an hour of tweaking here and there to get rid off the distortion I stopped for a while. For no reason, after a few seconds my eyes fell on the PS... "what if? naaa, can't be!... it's been built with good filtering, heat sink, and never had noise issues with high gain pedals... but these pedals do not suck 200ma or more..."
I took an old 220v - 12v used for dichroic lamps, added a rectifier, fired up... the annoying distortion was gone.

So...
Both power supplies have a defective filtering, leaky caps, etc,
or under pressure the LM317 is a bit noisy.

Thoughts?

mac

[antonis]

There are 2 cases (mainly) for "noisy" LM317..

1. Too much power dissipation [(Vin - Vout) x I]..

2. Too low Vin..

[EBK]

2. Too low Vin..

Without knowing more, that's my guess. 


Otherwise....
If you haven't tried it already, you can increase ripple rejection of the regulator by adding a 10μF cap between the adjustment pin and ground.  It may help.

[antonis]

In case of proposed by Eric cap implementation, place a diode across R1 for cap safely discharge in case of short-circuited output..

[Rob Strand]

Well it could be many things:

Checkout,
https://pdf.datasheetcatalog.com/datasheet_pdf/national-semiconductor/LM317AS_to_LM317MP.pdf

Missing surround caps (see datasheet, first page)
-  >= 100nF on input supply *close* to device; otherwise risk of oscillation
-  >= 1uF on output; datasheet says for improved transisients, but maybe it could oscillate as well
-  the 10uF EBK mentioned

The LM317 current limiting is probably kicking in on the peaks.  The way current limiting works is it drops the output voltage to the point where the current is maintained with the load.  You can imagine if the whole power rail is dipping down (to say 50% voltage) while you are playing it's going to do some weird stuff.   

If you have a preamp on the same rail then the modulated supply voltage will go to the preamp.   In this case to need to supply the preamp via a separately filtered rail,  for example a resistor + cap, or a diode + resistor + cap.

A year or two a go a forum member built an amplifier with a regulator on the power supply and he got quite a few weird effects cause by the current limit.   He ended up splitting/filtering the preamp supply.  Maybe this on,
https://www.diystompboxes.com/smfforum/index.php?topic=126501.0

When you use a straight rectifier and filter there is still some dipping at the peaks but the drop dips are quite small due to the large filter caps.    No current limit at all other than the transformer's natural regulation

You can try putting larger caps on the output of the regulator but if you need crazy large ones then it's a sign the design with the LM317 is going off course.

If you want to use large output caps on the regulator you need to add yet another diode to the LM317, see fig 3 in the datasheet.

[mac]

Both PS are the "full version" with caps, diodes, heat sink.
There is a fixed 2200uf 50v cap soldered on the power jack that goes to the breadboard.

There are 2 cases (mainly) for "noisy" LM317..

1. Too much power dissipation [(Vin - Vout) x I]..

This amp draws 200ma idle, (25-18)v*0.2a=1.4w, more while playing.

2. Too low Vin..

The 18v is built around a 220VAC to 18vAVC, 25VDC. Go to 1 :-)

In this case to need to supply the preamp via a separately filtered rail,  for example a resistor + cap, or a diode + resistor + cap.

The preamp is separated from the power amp --> 1k + 330uf to gnd. I always do this.

I guess it's a combination of the way the chip works when running at half speed, as Rob and Antonis noted.

A year or two a go a forum member built an amplifier with a regulator on the power supply and he got quite a few weird effects cause by the current limit.   He ended up splitting/filtering the preamp supply.  Maybe this on,
https://www.diystompboxes.com/smfforum/index.php?topic=126501.0

Missed that. Nice build.

mac

[antonis]

The 18v is built around a 220VAC to 18vAVC, 25VDC.

I presume 18VAC secondary is rated at double the regulator pass current, at least..

[PRR]

> This amp draws 200ma idle, ..., more while playing.

My car uses one gallon per hour idle, more when racing.

What is "more"?  202mA? 2,000mA?

Yes, it is hard to measure the dynamic power of an audio amp. For small stage-amps, you might fit the amp with a BIG speaker from a 50 Watt amp, put the guitar and amplifier into a bass feedback howl, then read current. That's about the absolute max, and stage amps may go there every night, though most "clean" audio won't.

[Rob Strand]

This amp draws 200ma idle, (25-18)v*0.2a=1.4w, more while playing.

If you tweak the 39 ohm resistor (or the other resistor in the divider) you could probably drop the idle current without affecting the sound.  It sets the VBE voltage of the output transistors.  Not all transistors have the same VBE's, not even the same part number.  Technically the bias voltage should be temperature compensated so the idle current doesn't creep-up when the amp is pushed.

[mac]

What is "more"?  202mA? 2,000mA?

The power amp is Class A with a 4.7 ohm emitter resistor. DIY output tranny is at its best at about 150m to 200ma idle, ~18v. That's 0.75v to 1v across the 4.7 resistor.
I always measure the "instant voltage" while playing with my the cheap chinese DMM... playing clean hardly reaches 1.5v, 300ma... let's say 400ma as the DMM is slow.
That's below the 1.5A limit of the LM317... or not?
A better reading: I can touch the TIP41C even after a few minutes of full drive. My cat belly is warmer

If you tweak the 39 ohm resistor (or the other resistor in the divider) you could probably drop the idle current without affecting the sound.  It sets the VBE voltage of the output transistors.  Not all transistors have the same VBE's, not even the same part number.  Technically the bias voltage should be temperature compensated so the idle current doesn't creep-up when the amp is pushed.

If you mean the regulator divider I have a 317 to experiment with.

The LM317 current limiting is probably kicking in on the peaks.  The way current limiting works is it drops the output voltage to the point where the current is maintained with the load.  You can imagine if the whole power rail is dipping down (to say 50% voltage) while you are playing it's going to do some weird stuff. 

This forgotten post made me put the 317 in the suspect lineup

https://www.diystompboxes.com/smfforum/index.php?topic=53241.msg404047#msg404047

mac

[Rob Strand]

If you mean the regulator divider I have a 317 to experiment with.

Actually it's my misunderstanding.  I thought you built a Deacy amp but you are actually talking about a class A amp.

[mac]

Actually it's my misunderstanding.  I thought you built a Deacy amp but you are actually talking about a class A amp.

The 39r of the Mullard schematic, minutes after I replied to you it came to my mind 

I presume 18VAC secondary is rated at double the regulator pass current, at least..

Checked |/

mac

[Rob Strand]

The 39r of the Mullard schematic, minutes after I replied to you it came to my mind 

Yes, that one      It tunes the idle current of the two output transistors.