Maybe:
Bandwidth
Damping factor
It's not just the characteristics of the IC, but the whole circuit.
Particularly when the implentation differs from the datasheet suggestion significantly.
Personally I'd attach more weight to measuring an effect than just disliking the sound.
No, I'd not seen that before.
Essientally it seems to be measuring noise levels of 21uV (doesn't seem to say what RBW but the plots go down to 20Hz....) and saying this sounds 'roary' in his headphones.
From which I am tempted to conclude that his amp has a very poor PSRR.
Interesting how noisy NiCd cells are.
NiCd cells are not noisy at all.
< http://www.hoffmann-hochfrequenz.de/downloads/NoiseMeasurementsOnChemicalBatteries.pdf >
In the references I have mentioned a NIST publication,
they could measure even lower to -205 dBV.
That does not count, it has external feedback via the reference pin.
Unconditionally stable means only that a device is stable with any
combination of input and output impedance, as long as the
terminations are passive and do not provide external feedback.
The LM317 is well-behaved if it is used according to the data sheet.
For very large capacitors on the output it may require protection
diodes, so it cannot be powered from the output side when the
input voltage drops too fast.
The 'roary' sound is being monitored by listening directly to the voltage regulator output with an amplifier.
I have measured the 3042, the 3045's little brother, against the rest
of the crop. There is one clear winner.
< VoltageRegulators | Output noise of some voltage regulators.… | Flickr >
Jim Williams predates the LT3045 :-( and his D cells were better anyway.
The influence of X7R is vastly overestimated unless you live in a helicopter.
That you have only half the capacitance in real live with the even lesser
dielectrics is a different story.
There are more interesting measurements on the Flickr page (left&right arrows)
Last edited:
I have used LM317 and LD1084 with 20 000uF input and 10 000uF + 0R1 output with few problems and very low noise. I did fit the diodes as per makers recomendaton. I used a 2A fuse as a Pi filter with 2 x 10 000 uF for the input. The fuse made a small useful improvement and I wanted a fuse. I had the recomended caps close to the device. I also used 470 uF with the feedback circa 2K ( 240R fixed value ). I felt it did no harm and was slightly better than 10 uF. The 0R1 started at 10R. Apart from useful current and drop out voltage 317 and 1084 seem identical in the same circuit.
My feeeling is a PSU made this way is using the regulator like we would a small engine in a hybrid car. If you like my output 10 000 uF is the electric motors in the anaology ( a car might only need 20 BHP average ). Having said that I fully realise 1uF would already be doing plenty and the regulator works mostly between circa DC and 1 kHz. I have real doubts a shunt circuit would beat this. If you see the picture. The regulator is doing as little as possible and parts are cheap. Add them as batteries if you like to remove need for using LM337. All you need is separte windings, rectifiers and caps. I measured no obvious problem doing this. You add plus to minus as the 0 volt common rail.
LM317/LD1084/LM78 series are very forgiving.
My feeeling is a PSU made this way is using the regulator like we would a small engine in a hybrid car. If you like my output 10 000 uF is the electric motors in the anaology ( a car might only need 20 BHP average ). Having said that I fully realise 1uF would already be doing plenty and the regulator works mostly between circa DC and 1 kHz. I have real doubts a shunt circuit would beat this. If you see the picture. The regulator is doing as little as possible and parts are cheap. Add them as batteries if you like to remove need for using LM337. All you need is separte windings, rectifiers and caps. I measured no obvious problem doing this. You add plus to minus as the 0 volt common rail.
LM317/LD1084/LM78 series are very forgiving.
I measured the impedance peak of the LM317 (using Bode 100), without any load or output cap at ~33kHz -- the feedback gives it that little push into excitation 15k||300p.
Well at least on power supply bypassing. Coming from the land of Roederstein + Wima that certainly means a lot.
PS
Fidelity of public address systems in choppers is important in some circles / films.
I am still left wondering if ANYBODY can beat this regulator for $10:
LM317 / LM337 +/-1.5V~37V Adjustable Dual Voltage Regulator Power Board | eBay
LM317 / LM337 +/-1.5V~37V Adjustable Dual Voltage Regulator Power Board | eBay
In my world, unconditional stability is quickly made void by bad grounding.
Which makes it easy to talk at cross purposes when you look at an IC which has no ground pin.
There is much talk of microvolts of noise. We build amplifiers with milliohms of ground impedance, then pull current from them.
A regulated power supply is still just a power supply.
If you want it to be a broad-band voltage reference, that's really a different animal.
It looks like a very handy thing to have on the bench.
There are a few variations on ebay.
I'm going to buy a couple just to run some op amps from if nothing else.
Fidelity of public address systems in choppers is important in some circles / films.
Yes, If you want to play the Ride of the Valkyrs for Vietnamese country people..
I love the smell of solder in the morning!