Very low noise 7805?

[carlmart]

Originally posted by Reina
hi,
so its more a 'hint' then a precise measurement...


MC7805C (Korea) 1.2mV
L7805CV (SGS) 0.4mV
L7805CV (ST) 1.3mV
SD7805D 1.0mV
MC7805CT (M) 0.4mV
JRC7805A 2.2mV
JRC78M05A 2.3mV
L78S05CV (ST) 1.5mV

comparing:
LM317T (M) @5V 1.7mV
LM317T (ST)@5V 1.4mV
LT1086CT-5 (LT) 6.1mV
L7808CV (SGS) 3.1mV
L7812CV (ST) 4.6mV

references:
TL431C (ST) 0.9mV
LM336Z-5.0 2.3mV
LM336Z-2.5 (LT) 0.8mV
LM336 (ST) 0.6mV
ZPD 5.6V (zener diode) 0.0mV
ZPD 3.3V (zener diode) 0.0mV

Zeners are considered one of the noisiest parts around to make supplies, so maybe there's something wrong on the way you are doing these measurements.

Carlos

[Reina]

>Zeners are considered one of the noisiest parts

i found the results surprising, too. the current through the zeners was several mA. the values are correct. the noise amplifier was my hifi-amp and my ears says the same. zener noise is important at low currents in amplifier circuits, of course. it is much higher than the noise from a diode (transistor) but nothing in compare to voltage references, you see.

[Christer]

Originally posted by carlmart

Zeners are considered one of the noisiest parts around to make supplies, so maybe there's something wrong on the way you are doing these measurements.

All zeners are noisy, but som zeners are noisier than others.

I don't know if you remember, or even read, about the experiments I did about a year ago (http://www.diyaudio.com/forums/showt...008#post417008).
According to my measurements there is a very big difference in zeners depending on their voltage drop. The noisiest ones seem to be those having a voltage drop close to the shift from zener breakdown to avalanche breakdown. These are better avoided if noise is of any concern at all. However, those zener diodes that have a sufficiently low or high voltage have quite respectable noise values. Also, don't forget that ultimately the noise should be correlated with the voltage. The 12 V zeners I measured had about the same noise voltage as LEDs. Then bear in mind that it would take a string of LEDs to achieve a 12 V drop, so for high voltage drops it seems that zeners are to be preferred.

It should also be noted that for many of the devices I tested, the noise varies enormeously with the current, while for others, the current hardly matters.

Finally, I only tested components that basically work as shunt regulators (diodes, LEDs, zeners, TL431 etc.). If measuring series regulators like 78xx and 317, one would have to come up with some test setup that gives a fair comparison. Probably (but I never measured them) series regulators will have a noise that depends on the output current. That is, for a fair comparison between shunt and series regulators, we must also consider the load. I only studied the case of no load, that is, shunt regulators used as voltage references. A series regulator, however, typically desires some minimum load to function properly, so even if it is to be used only as a voltage reference, it might need som suitable load to achieve its best noise behaviour.

[jackinnj]

it's helpful to bandwidth limit the noise measurements -- for references it's typically in a range of 10 milliHz to 10 Hz, for supplies, 10Hz to 100kHz. You get more repeatable results if you isolate the DUT in its own shielded environment and take care to make sure that your cabling to the test instrument remains unmoved.

further, the noise measurement is integrated over time -- take a look at Texas Instruments product folder for the TL431.

just like paralleling JFETs -- when you parallel shunt references the noise drops -- but there is a point of optimality which I found to be 3 LM329's for the super-reg.

fwiw -- here's a picture of the noise on a regulator for a tube preamp -- I previously posted it in the Tube Threads -- it's a variant of the "WJ Pooge" which appeared in Audio Amateur in 1979 for low voltage applications and was scaled up by Walt for the Marantz 7C and Dynaco PAS in 1982 -- the noise appears to be about 5 mV or so 16ppm. The top line is the noise from a Heath IP-17 tube-type regulated supply.

[Guido Tent]

Originally posted by jackinnj
it's helpful to bandwidth limit the noise measurements -- for references it's typically in a range of 10 milliHz to 10 Hz, for supplies, 10Hz to 100kHz.

Hi

If we want 2 single noise figures, that is a good proposal. Not everyone has a spectrum analyser at his or her lab.