Voltagr Reugulator NOISE

[fmak]

For sometime, I have been following conventional widom and replacing 78 and 79 regulators with LT 1085/6, LT323 and LM320/340 types. How wrong!!

I wanted to use lower noise units for a dac ; and so made some measurements on 5V units with 50mA current and 10uf tantalum output cap with an ac volmeter capable of up to 1 Meg with 0.5 uV or better resolution.

Result?

JRC 7805 - 40 uV - to spec
LT1086 - 150 uV - to spec
LT323 - 50 uV
LM340 - 60 uV

Why is the LT range so touted??

Does anyone know who makes the M5F7805L range of reulators; I want to check on spec as I have a whole bunch of them in my dac?

Has anyone designed/sell a pin compatible replacement on a pcb that costs less than £40 with say 10 uV or less noise wideband?

[jean-paul]

Conventional wisdom is using 78xx regulators. As I found out too they are not as bad as people think they are. Some exceptions are 78Lxx ( 100 mA version, tends to oscillate easily ) and 78Sxx ( 2 Amp version, for some reason screws up sound when used as a replacement for 78xx ).

Could it be that the internals of the newer 78xx series differs from the old ones ? I can remember that I had to match 7818's because some gave 17.2 and some even 19 Volt. Newer 78xx regulators are very precise and noise is lower too. Negative 79xx series were famous for oscillation when used without cap at the output and without load. Nowadays they don't have a problem with that

[janneman]

Originally posted by fmak
For sometime, I have been following conventional widom and replacing 78 and 79 regulators with LT 1085/6, LT323 and LM320/340 types. How wrong!![snip]

That is why some of us regard so-called "conventional wisdom" with a healthy dose of mistrust. Especially when the conventional wisdom is confirmed by undocumented, unspecified listening tests.


Jan Didden

[jackinnj]

Originally posted by fmak
I wanted to use lower noise units for a dac ; and so made some measurements on 5V units with 50mA current and 10uf tantalum output cap with an ac volmeter capable of up to 1 Meg with 0.5 uV or better resolution.

but how do you know if you were measuring noise or ripple, noise superimposed on ripple ?

[fmak]

Originally posted by jackinnj

but how do you know if you were measuring noise or ripple, noise superimposed on ripple ?

-----------------------------------------------------
You can see it on the scope. Supply is regulated. Only the LT1086 showed ripple from the supply; other rejected as stated.

[Fred Dieckmann]

Originally posted by fmak

-----------------------------------------------------
You can see it on the scope. Supply is regulated. Only the LT1086 showed ripple from the supply; other rejected as stated.

Go back and make sure you don't have a bad part. Also, adjustable regulators allow you to bypass the adjustment terminal to lower the noise below fixed voltage regulators. I have found the LT 1086 to be a measurable and sonic improvement over fixed voltage three terminal regulators. I think you guys just what to save money with 78XX types. They are much cheaper. Why not build your own discreet transistor regulators?

[jam]

You should listen to Fred. The difference between a good discreet regulator and a chip are usually huge. Much more than the difference between chip types.

I have replaced chip regulators in some rather expensive equipment with discreet regulators and always have had excellent results.

Not meaning to advertise for Fred but he has posted an exellent design in the forum, you owe it to yourselves to try.

Regards,
Jam

[fmak]

Originally posted by Fred Dieckmann



Go back and make sure you don't have a bad part. Also, adjustable regulators allow you to bypass the adjustment terminal to lower the noise below fixed voltage regulators. I have found the LT 1086 to be a measurable and sonic improvement over fixed voltage three terminal regulators. I think you guys just what to save money with 78XX types. They are much cheaper. Why not build your own discreet transistor regulators?

----------------------------------------------
Don't need to. Whatever people hear, go to the LT1086 data sheet; the noise voltage is 0.003% of output voltage which is 150 uV for 5V out. For a 7805, the noise voltage is 40 uV. Now the output impedances and other factors may differ, but that is another issue.

What have you measured about the 1086 which is better?

I agree that discrete regulators are better. Twenty, thirty years ago, we were building Japanese designs with current sources etc. which sounded much better. The problem now is that these and other designs are TOO big for stuffing into existing digital systems which may need 9-12 regulators. The only design I know off is the Audiocom regulator which can be plugged into a 78x or 79x designed board but they want £40 for one which to me is too much. Hence my post.

[jean-paul]

Not meaning to advertise for Fred but he has posted an exellent design in the forum, you owe it to yourselves to try.

Hi Jam, are you refering to the schematic with the green leds and LM1086 ?

Regards,

Jean-Paul

[jam]

Jean-Paul,

This one.

Regards,
Jam