LM317 good for constant current source?

[xxie24]

I used LM317 as constant current source in my Tori-amp, sound wise I am very happy of this amp.

But recently I read the LM317's datasheet from National, I found out the LM317 has quite bad ripple rejection rate from 10khz at only 50dB and lower for higher frequence.

Does that means LM317 could be a bad choice for constant current source? What is the effects of this ripple rejection on the stability of this amp?

Thanks

 

[robzy]

Please wait for someone more knowledgable to either confirm or deny what im saying here:

I would guess that it would be okay as it isnt actually in the signal path, all its doing is biasing the MOSFET and sinking current to V-.

(Also, doesnt ripple rejection refer to removing ripple from the input? Not about any ripple created on its own?)

Rob.

[edit: And bw, im planning on building a Tori - do you have any pics?]

 

[xxie24]

Robzy: I don't have a DC. Also my build quality is quite bad as it was my first build, after seeing so many beautiful diy amp here, I would feel bit shame if show my amp's picture. Sijosae has a good picture here, but seems to me he did not use LM317 as current source.

 

[ATAT]

Looks fine. Ripple rejection is fairly irrelavent if your 12 is clean, which it should be anyway. Sijosae even reccomended the use of LM317 as CCS for his Multi-hybrid amp.

 

[robzy]

xxie: My Tori is going to be my first ever p2p job ever (apart from tangent's cmoy which doesnt really count). So give me a week or two and then i'll show you messy!

Slightly back on topic - I have seen quite a few schematics relying on the LM317 as a CCS, so i'd guess that it would further support it?

(By the way - if possible - could you measure the current being pulled on the ground channel? I would love to know if i could stick an TLE in there instead of worrying about a dual supply)

Rob.

 

[Voodoochile]

Pics...pics...pics...

Sounds pretty cool.

 

[xxie24]

I think it a bit myself about my question: from the datasheet LM317 has 20dB ripple rejection around 100kHz, this is sort of like a wire connected from output to PSU's negative. Assuming my PSU has a 10ohm impedance, so at 100Khz the output is like shorted to ground with 10ohm resistor. Will it cause a fast OPamp oscillating?

I understand there is compensation capacitor in the Tori-amp, but from the designer it was mainly for the Mosfet's gate capacitance. So it would be nice to know whether LM317 will be another cause for the oscillation.

Robzy: I don't think TLE would be a nice here. The Tori-amp is designed for low-impedance headphone, so there would be rather large current in the ground. For batteries based amp (no class A)and high impedance headphone, TLE would be ok. I definetely think Tori-amp need a nice big dual-rail PSU to sound good.

 

[robzy]

Quote:

Originally Posted by xxie24 Robzy: I don't think TLE would be a nice here. The Tori-amp is designed for low-impedance headphone, so there would be rather large current in the ground. For batteries based amp (no class A)and high impedance headphone, TLE would be ok. I definetely think Tori-amp need a nice big dual-rail PSU to sound good.

Im not so sure about that. The PPAv2 uses a simple TLE for a virtual ground, and so does the m^3 (Which has a MOSFET output stage) - niether of them even have a buffer placed after it.

(Really sorry to thread hijack - perhaps the Tori should get a thread of its own?)

Rob.

 

[mono]

Quote:

Originally Posted by robzy Im not so sure about that. The PPAv2 uses a simple TLE for a virtual ground, and so does the m^3 (Which has a MOSFET output stage) - niether of them even have a buffer placed after it. Rob.

They only use TLE for a virtual ground reference, have the separate ground channel for handling return current from the load and handle very little themselves for the opamp power.

 

[ATAT]

You could of course do what the MINT did and put a large amount of rail capacitance after the TLE to source / sink current.

 

[Garbz]

There are arguements against a large amount of rail capacitance and how it effects sound quality but that is for a different thread.

Hey xxie24 how modular is the amp? The LM317 makes a very stable constant current source in my experience. I have used it before to bias amps without problem and use it quite frequently as a CSS for a TL431 based shunt regulator.

If the amp is modular enough that you can easily remove it and replace it with a dual mosfet CSS or something similar, I would be inclined to drag the entire mess into one of the labs at uni and start running it through the scopes. Even as undergrad they let me access the labs for non-uni related work providing they are manned.

 

[amb]

The ripple rejection spec is only slightly relevant if you're trying to achieve maximum PSRR. Even so, the LM317 here is much better in that regard than a single resistor solution. As an alternative, I suggest looking at the CCS found in the Borbely SE class A headphone amp (same design in both the JFET input and tube input input versions). It uses a BJT and MOSFET plus a few passives parts, so is a little more complex but IMO much more elegant.