Electronics > Projects, Designs, and Technical Stuff
Small and quite +/-8v power supply/regulator
CJay:
I did a bit of reading about regulators when building a power meter with an AD8307, what I found was quite surprising to me, the 78xx and 79xx devices stack up pretty well against 'low noise' regulators, are you absolutely certain it's the regulators causing the problem and if you are, are you absolutely certain it's not your implementation?
Zero999:
The LM7808 and LM7908 should be quiet enough. If it's noisy, then it's because it's a bad design or there's a faulty component.
Replacing the LM7808 and LM7909 power supply with something else will make no difference.
SiliconWizard:
--- Quote from: ogden on July 25, 2019, 12:05:13 am ---
--- Quote from: langwadt on July 24, 2019, 11:01:51 pm ---
--- Quote from: imo on July 24, 2019, 10:25:55 pm ---LM317/337 are cheap and quiet too (with a capacitor in the divider).
--- End quote ---
afaict from a quick look at datasheets LM317 is much higher noise than LM7808
--- End quote ---
Yes you are right. @imo: :wtf:
Well.. LM7808 is obsolete. Modern version is UA78M08, 40uV RMS noise. LM317 have whooping 240uV RMS. MC7808 (from ON semi) have 10uV RMS.
Always doublecheck info from public forum where anyone can convincingly say anything.
--- End quote ---
Well, imo's comment was quick and inaccurate, but as I said above, there's still a reason why the LM317/337 was a favorite in many audio applications compared to the 78xx line, so maybe that's what he was remembering.
Yes the typical RMS output noise of the LM317 is higher (although I think they don't state a max value in many datasheets, whereas the max noise of the original 78xx, 200µV, is close to the typical of the 317...), so imo's short statement was wrong. But again ripple rejection was typically 8dB better in the LM317. Ripple noise (often typically harmonics of mains frequency) would be much more annoying in many audio applications compared to a linear reg's own noise, which is more wideband. So the choice would depend on the specific application and of course on the PSRR of the powered circuitry.
I think the LM317 load regulation was also slightly better. So all this was only to be considered maybe 2 decades ago or more.
Either way, for this level of noise to matter, the analog circuit would have to have high gain, and poor PSRR. Otherwise it would not be noticeable.
But yes there are now dozens of alternative regulators for this. The UA78M08 has better specs than the original LM7808, and similar ripple rejection than the LM317, and it's cheap, so it's better overall. It can put out 500mA only, but that would be a good fit in the OP's case.
To get back to what the OP said, they would have to explain what "noisy" means. If it's a matter of ripple noise, then yes ripple rejection would matter, but if it's so noisy as to be noticeable, a few dB better ripple rejection would probably not make a lot of difference. Maybe they could just post the original schematic. I suspect the input (rectified) voltage was very poorly filtered, or something like that. Or maybe the layout is a disaster. Or probably both.
iMo:
What is the rms noise of the LM317/337 with the capacitor (say 100u) in the divider (as in my post)??
SiliconWizard:
--- Quote from: imo on July 26, 2019, 05:13:48 pm ---What is the rms noise of the LM317/337 with the capacitor (say 100u) in the divider (as in my post)??
--- End quote ---
From memory, and then taking a look at the datasheet again, it was mainly to improve ripple rejection (what I'm talking about since first post), whereas ogden was talking about the RMS output noise voltage figure, which I take it as the regulator's own noise, which is a separate source of noise from ripple. That's how I interpret those figures anyway. I haven't seen anything regarding the RMS output noise figure with this additional capacitor in the DS I have. Do not hesitate to show the info if you have it.
The optional cap between the ADJ pin and GND is indeed to improve ripple rejection as I stated. I'm not sure how it affects the regulator's own noise. I admit this figure is not necessarily trivial to understand, and I'd be interested in seeing what kind of setup is used to characterize it independently from noise coming from the input voltage (haven't seen the test setup in the DS I looked at.)
As I already said, ripple noise in a typical PS stage coming directly from rectified AC (so often horrendous ripple even with big caps) would often matter more than the typical linear reg's own noise, especially for audio stuff, because its own noise is more wideband and thus more spread out spectrally.
Now there is probably some dependance between the two sources of noise, so again that would be nice to see how exactly they are characterized (if anyone has the info).
And of course if I have missed anything, do not hesitate to point that out.
Navigation
[0] Message Index
[#] Next page
[*] Previous page
Go to full version