Class A buffer using LM317 3 terminal reg's

This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
I just chucked this thing together just for fun to see what a power supply regulator ic would do if it was made to process audio instead of ripple and noise. (Not much different from some people's music. Anyhoo...) The LM317 has a ripple reduction of 80dB to 10kHz and then slopes gently downhill from there so what we have is basically a class A power amp on a chip. In the cct the top ic is a voltage follower and the bottom one is a constant current sink. My generator would only produce 15v p/p so that's all I tested it too but it should swing to within 2 volts of either rail. The whole thing is super linear. Frequency response is to 100kHz then the 100r and 560pF start to roll it off. Without these it wasn't stable producing a square wave. Standing current is set by the 2R2 resistor right at the bottom - it always has 1.2 volts across it so the current is according to that. I havent yet run a speaker or headphones off it (reactive load) but it will be interesting to see. Component values are not really optimised, e.g. the 10k voltage divider should really be that which gives a dc output voltage exactly half supply. On a square wave the rise and fall times are 1.1 uS with absolutely no overshoot. I used ST brand regs, I don't know how other people's would go. This has gotta be the most under-rated chip ever!



  • lm317 buffer.gif
    lm317 buffer.gif
    6.3 KB · Views: 3,879
truly stunning. next you can develop a fully symmetric version with 337. oops, maybe not, goes again the 'circlotron' spirit. how bout current sources with more 317s?
dunno why, but if something doesn't play music as regulator is it really likely to shine as an amp? did you listen to it?

Sacrelige! It's bad enough people want to put (gasp!) op-amps in amplifiers but now you dare to build one out of...(I can barely bring myself to say it)...voltage regulators!!!???? :yuck:

Ahem, please let us know how it sounds.

I can see it now, how NOS Fairchild regulators produce the cleanest sound.
Replies to everybody

Samuel, yes that would be interesting. Hope you can find and post it.

analog_sa, I wasn't thinking about circlotrons at that time, but yes you are certainly right :) heh heh

paulb, So now I have started a controversy hey? Well, the name of this site is diyAUDIO, not diyHIGH_ENDaudio, so it fits in with the scheme of things here. Will listen to it soon, and develop it further perhaps. yours and other's contributions to this thing most welcome. NOS Fairchild ;)

Tomo, that sneaky tubecad crowd! They beat me to it! Oh well. One of my next tricks is use a '317 as the top half of a cascode pair in conjunction wit a normal bipolar. Should work well because the ic input is a near perfect current sink. About 300k slope resistance.

Now if I use 2 '317s, one at the top and one at the bottom I will have about 160dB of open loop gain... That will be difficult to manage I think.

Music from regs? "There's music in a horse shoe, there's music in a nail, there's music in a tomcat if you'll only pull his tail."

Joined 2002
Paid Member
For a similar alternative, but simpler signal path:

consider the LM395 as the follower device; continue to use the 317 as the current source.

The LM395 has similar short circuit protection and thermal protection as the 317, but a much simpler cascaded PNP-NPN-NPN pseudo-triple-darlington as the transistor.

The downside, is that despite the slightly simpler circuit, the LM395 is more expensive than the LM317, most likely because it has nowhere near the volume of sales.

The result may be inferior in all respects; the LM395 has no op-amp to provide for error correction, so you get the nonlinearity of its transistors. But, it might not have any stability issues; and givenits protection might also be a good output stage for a bulletproof low power class A amp.

I'll have to try the 317 idea! I think I have a bunch of them lying around.


-- mirlo
I think I like where this is going! I have a bunch of '7805s laying about, and this may be an excuse to use them. For the front end, how about a cd4049? The whole thing could be tailored to run off of 48 volts so it could be powered from a telephone jack, and be useable during (some) power outages.
OK. Here's the rules ;-)

1/ Any amplifiers built must use parts not meant for audio usage, or at least as few as possible .

2/ The more unusual (or silly) the topology the better.

3/ Preferably use junk box parts, the lower the cost the better.

4/ Don't be afraid to think outside the box. e.g. 4049UB like Daniel, or how about a CRT for an amplifying stage? :whacko:

5/ Competitors can only vote for other people's stuff, not their own.

6/ A picture, even of a rat's nest would be nice.

Yeah, I think I'm going to have to look at a circlotron setup.

Circlotron > " Any amplifiers built must use parts not meant for audio usage, or at least as few as possible" .

2" The more unusual (or silly) the topology the better."

I could not agree more While alot of Folks have a Bias against using IC's and some Using any Solid state device at all, I seen a T-shirt once, saying Use a Transistor and go to jail. This is just limiting your options I think The more Devices available the Better and Yes please do use stuff not For Audio as that often also implies Low cost Restricted bandwidth. So Yes using Regulators as a gain stage Sure why not if it sounds good if it dose not Nothing lost and alot learend. Long time ago i made a Phono Amp for the Old Ceramic Phono's using a Regulator from a Circuit found in the data sheet.

Using a CRT as a gain stage. I tried that in my youth using a B & W CRT and could not get it to work Right it would pass a signal but was greatly Distorted. Nontheless i thought the concept was Nice just because I have not seen it done at the time. Heck it should work all the Elaments are present, it was just that i never could and all the Experts i asked about it at the time thought it was funny with out offering a reason why. So ya Go ahead get creative.
The verdict

Because you lot are hassling me so much ;) here is what you want to know, 12 hours earlier than I said I would. First of all it doesn't sound like a voltage regulator, it sounds like an audio amplifier! I havent tried the normal fixed voltage types but they might be a bit different sounding because internally they are nowhere near as sophisticated. I couldn't get a lot of volume out of it because I just shoved the cd output straight into it and so there was only 0.5 to 1 volt peak going into a 4 ohm speaker. Anyway, it was loud enough to get a bit of an idea. Straight away you could tell it was class A, no two ways about that. It had the typical breezy effortlessness about it with no abrasiveness or grain that I could hear. Good detailed bass; tish, snick and click type percussion sounds were well defined. Vocals almost as good as my upcoming mosfet class A.

But there was something about it that was lacking. I don't know whether it was just the lack of volume - about the same as you would get from an ordinary table radio if you were listening to speech, but it seemed as if the little amplifier was not really paying attention to what it was supposed to be doing. It sounded quite pleasant and all, but there was something vague about the depth of the sound that was missing. I used a 2200uF cap to couple to the speaker so that wouldn't really have helped things I suppose. I am being hyper critical but in all fairness it sounds way, way, way better than you might have ever expected.

Tomo did suggest I build a complementary symmetry version using it's negative counterpart. Get a wife for it. It's worth thinking about. I reckon the best way to think of a '317 is not as a voltage reg but as an npn transistor who's emitter (output terminal) voltage is always 1.2v above the base (adj terminal) voltage instead of *about* 0.6 volts below. What's more there are no problems with temperature compensation for bias, it just stays put. Linearity is several orders of magnitude better as an emitter follower than an ordinary transistor. As a current source the slope resistance is about 300k, I expect that's way higher than a bipolar. Go '317!

It would be interesting to see what this circuit will do if the Neg. LM-337 was used to complement the 317. I wonder about that because the Neg reg still uses an NPN series pass transistor However with the Col and Emt. rev. Q16 for the 317 has the Emit as the Output and Q26 for the 337uses the Col as the output.

Circlotron> Like i sed before I like your efforts and I have no problems using IC's and that includes Opamps in Audio equipment. Alot of spendy Well regarded High End Equipment uses Opamps and all kinds of other IC's as well.
"...12 hours earlier than I said I would"
Oh you are good to us! :D:D:D
What a relief to read your findings. Looks like it works pretty well, although I suppose the output current is limited to 1.5A or something. Perhaps a natty headphone amp though. You mentioned some limitations to the sound quality - any ideas why? I presume you've checked all the other stuff like grounding. Do you know what the input Z is between ADJ and gnd - could there be some distortion here that might require an input buffer? I'm not familiar with the 317 schematic but if I get time I'll look it up.
Great innovative attitude!
Re: The verdict

Circlotron said:
I am being hyper critical but in all fairness it sounds way, way, way better than you might have ever expected.
Good enough for me to try. Probably for headphones, whaddya think, ppl? Or maybe a small office amp. I'll put it on the list of things to try someday.
Thanks for posting & trying this, Circlotron.
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.