LM317/LM337 vs LM78xx/79xx regulators - any real difference?

Curious if the LM317/337 regulators are quieter or have any advantges over the LM78xx/79xx - aside from the fact that the LM317/337 are variable. Anyone does any tests or have any general info?

are you asking if LM317 devices are lower noise than LM78XX series parts? 

They both have the prefix LM generally..

And yes, the 317 types are lower noise however there are even more low noise types called "linear pass" regulators.  They are generally just a transistor pass element and a zener diode built into a single part.  The discrete parts used in the same configuration are called a zener-follower.

svart, you are right, i was asking if the 317 is lower noise.

is there a significant difference in noise and do you think it's something that is actually audible? im curious because i have a bunch of the ssl9k psu boards around...

is it something that might matter more in a preamp as opposed to another piece of outboard?

78xx are notoriously noisier than 317. The impact on the overall performance depends very much in the implementation, in particular if you let the noise current flow through ground circuits. This is significant on modular mixers. The noise current flows in the ground bus and produces longitudinal noise which is directly applied to the summing stage. Same problem with modular racks (900/500 type). But for a centralised PSU with serious central decoupling and only passive (RC) on-board decoupling, there should not be much difference.

The 78/79 series regulators were the first 3-terminal regulators. Newer generation parts like the 317 added new features and generally worked better, or else why do them?

Note: all 78/79 regulators are not created equal. I had to black-ball one vendor who refused to even spec noise. Many of these parts have gone through multiple die shrinks and process changes over the years to reduce cost.

As others have posted these are crude tools and results will depend on how you apply them.

While others (here) don't embrace my approach I have had good results from paralleling a 78/79 with a large electrolytic cap. Something like a 1000uF is a nice complement to the rising output impedance of the 78/79 reg with frequency due to circa 741 opamp technology inside. Of course you still need a smaller HF cap for very HF Z.

If the PS is important it isn't that hard to roll your own from a low noise opamp and pass transistor.

JR

LM317/337 regs are superior to their 78xx/79xx competition in almost every spec. Lower noise, better line rejection, and at least in the case of the LM317, significantly lower output impedance.

Walt Jung, in the series on his super-regulator, included tests of the various parameters of popular regulator chips. It might be on his webstie someplace.

Peace,
Paul

LM317/337 regs are superior to their 78xx/79xx competition in almost every spec.

Click to expand...

True. I'd like to point out though that this is mostly a result of the external bypass capacitor which reduces noise gain of the feedback loop to unity at audio frequencies. If this is omitted performance of the various parts will be similar as the internals of them ain't that different.

Samuel

The lower resistance of the feedback loop components may also account for a few dB less noise in the error amp.

The (Linear Technology) LT1086 (positive) and LT1033 (negative) are much better regulators and compatible with the LM317/337 series. LM78xx and 79xx are very inconsistent (too much manufacturers and unknow sources) and range from lousy to barely acceptable in terms of noise specs. I would never use these in any serious audio project. 

tubologic said:

The (Linear Technology) LT1086 (positive) and LT1033 (negative) are much better regulators

Click to expand...

Judged by what? Noise specs for the LT1086 and the 317 are equal, the LT part has ~6dB worse ripple rejection than the 317. The LT1086 has a low dropout voltage, but that's almost never relevant in audio applications. Output voltage accuracy is better (2% vs 4%), but that too has little impact. The LT1033 has a higher max current spec than the LM337, but you'll hit package dissipation limits before that gets relevant. For both the LT1086 and the LT1033 the max input-output voltage is lower than for the generic ones.

LT has nice parts, but in typical audio work these don't seem worth paying 3-5x the price of a more traditional LM317/337.

JDB.
[in my book, picking a single-sourced part over a multi-sourced industry standard had better have a damned good reason]

jdbakker said:

tubologic said:

The (Linear Technology) LT1086 (positive) and LT1033 (negative) are much better regulators

Click to expand...

Judged by what? Noise specs for the LT1086 and the 317 are equal, the LT part has ~6dB worse ripple rejection than the 317. The LT1086 has a low dropout voltage, but that's almost never relevant in audio applications. Output voltage accuracy is better (2% vs 4%), but that too has little impact. The LT1033 has a higher max current spec than the LM337, but you'll hit package dissipation limits before that gets relevant. For both the LT1086 and the LT1033 the max input-output voltage is lower than for the generic ones.

LT has nice parts, but in typical audio work these don't seem worth paying 3-5x the price of a more traditional LM317/337.

JDB.
[in my book, picking a single-sourced part over a multi-sourced industry standard had better have a damned good reason]

Click to expand...

... Judged by ACTUAL performances in a REAL working circuit. Manufacturer's datasheets are great but doesn't allways tell you all the story. How could you be sure that the spec's of your 50 cents unknow sourced LM317 will be really met ? Ripple rejection is not the main issue (it can allways be cured by ample filtering) but wideband noise (including beyond the audio spectrum) definitely IS. You should be very careful when comparing spec's from different manufacturers who use different test procedures and criterias. I didn't measured the noise spec's of a LM317 vs LT1086 but actually compared how they perform in a very critical audio circuit (AC701 heater supply for a tube Mic.) Believe me or not,the AUDIBLE and MEASURABLE differences were not marginal or subjective: the LT powered Mic had a 6 to 8 dB lower wideband hum & noise than the LM version. To get the same results with a generic LM317 a 10,000 µF capacitor was required at his output,which is not recommended and partially defeat the purpose of the regulator. Of course,by carefully selecting a generic LM317 you may find one amongst 20 which will work equally well than a LT1086 (yes, I tried) but this will completely invalidate your economic argument. The LM317 is a good and cheap general purpose regulator and probably good enough for most applications BUT critical high quality audio circuits desserves the best. And this is just what this thread is about (I think...),not mass market engineering economical considerations which ultimately lead to crap audio equipment we're plagued with.

However, considering other things that the performances, a real vantage of 317/337 vs. LT1086  and LT1033 regulators, is that is very less likely 317/337 will be discontinued than these LT regs.
A very important thing for a manufacturer is to ensure the customer the repairability of its uncheap gear after ten years. But , seem LT regs can be replaced with 317/337 in several cases.

tubologic said:

I didn't measured the noise spec's of a LM317 vs LT1086 but actually compared how they perform in a very critical audio circuit (AC701 heater supply for a tube Mic.) Believe me or not,the AUDIBLE and MEASURABLE differences were not marginal or subjective: the LT powered Mic had a 6 to 8 dB lower wideband hum & noise than the LM version. To get the same results with a generic LM317 a 10,000 µF capacitor was required at his output,which is not recommended and partially defeat the purpose of the regulator.

Click to expand...

Sorry if this is getting OT, but you really had a 6 dB higher noise floor with a LM317 regulated heater supply? I didn't expect the heater supply to have such a huge effect on the noise floor, but it would definitely be worth trying it. I always believed that the effect of the heater supply cleaning was negligible as long as it's 'relatively clean'?

Michael

Michael Tibes said:

tubologic said:

I didn't measured the noise spec's of a LM317 vs LT1086 but actually compared how they perform in a very critical audio circuit (AC701 heater supply for a tube Mic.) Believe me or not,the AUDIBLE and MEASURABLE differences were not marginal or subjective: the LT powered Mic had a 6 to 8 dB lower wideband hum & noise than the LM version. To get the same results with a generic LM317 a 10,000 µF capacitor was required at his output,which is not recommended and partially defeat the purpose of the regulator.

Click to expand...

Sorry if this is getting OT, but you really had a 6 dB higher noise floor with a LM317 regulated heater supply? I didn't expect the heater supply to have such a huge effect on the noise floor, but it would definitely be worth trying it. I always believed that the effect of the heater supply cleaning was negligible as long as it's 'relatively clean'?

Michael

Click to expand...

If it is a KM5x microphone, the bias voltage is directly derived from the heater voltage, so any noise in the heater supply is transferred to the output at full gain.

I ignorantly quoted this to a friend [ 317's being quieter ] to which he replied
in regards a 9 v for gtr pedals

Actually, I think the MC7809CT O/P noise voltage is 52uV.  The LM317 is .003% of Vout, 9x.003% is .00027 volts...270uV.  Plus the 7809 will use less components

???