Hey everyone,
maybe I'm old school but I'm still building my regulated supplies using LM317 and LM337. Now that I'm doing another project, I thought it might be time to revisit the PSU design - maybe hoping to find something that generates less internal heat. I'm typically drawing around 300mA from each rail.
I see some voltage regulators considered "low drop out", does that do anything for efficiency? Or is the only way to get more efficient than a linear regulator to just go with a switching mode regulator at this point?
Maybe it's worth a separate thread, but I'm also wondering what the current opinions are on switching mode regulators anyways. I've always been against them, but the more I think about it.. wouldn't it be better to have the ripple frequency way above human hearing instead of 120Hz smack dab in the middle of the frequencies we care about?
maybe I'm old school but I'm still building my regulated supplies using LM317 and LM337. Now that I'm doing another project, I thought it might be time to revisit the PSU design - maybe hoping to find something that generates less internal heat. I'm typically drawing around 300mA from each rail.
I see some voltage regulators considered "low drop out", does that do anything for efficiency? Or is the only way to get more efficient than a linear regulator to just go with a switching mode regulator at this point?
Maybe it's worth a separate thread, but I'm also wondering what the current opinions are on switching mode regulators anyways. I've always been against them, but the more I think about it.. wouldn't it be better to have the ripple frequency way above human hearing instead of 120Hz smack dab in the middle of the frequencies we care about?
Both a low dropout regulator type, and a smaller DC voltage drop across the regulator, are needed to increase
the efficiency of a linear regulator circuit. You still would need to take the AC line voltage variation into account.
Many in recent years have successfully used SMPS in audio, even in phono preamps, with good results.
Insight | Sutherland Engineering
the efficiency of a linear regulator circuit. You still would need to take the AC line voltage variation into account.
Many in recent years have successfully used SMPS in audio, even in phono preamps, with good results.
Insight | Sutherland Engineering
Last edited:
Thank you Rayma, so I wouldn't be reducing heat unless I actually also dropped the voltage going into the LDO regulators?
Exactly, it's just P = V x I for the loss in the linear regulator, where V = Vinput - V output.
If things are carefully worked out, the V across the regulator can be made lower than usual
when using a low dropout type of regulator, so that can help. Or the input voltage can instead
be pre-regulated, so line voltage variation would be of no consequence, and the low dropout type
would then let you cut the voltage drop across the regulator (and the power loss) to a minimum.
If things are carefully worked out, the V across the regulator can be made lower than usual
when using a low dropout type of regulator, so that can help. Or the input voltage can instead
be pre-regulated, so line voltage variation would be of no consequence, and the low dropout type
would then let you cut the voltage drop across the regulator (and the power loss) to a minimum.
Last edited:
The only way to beat a linear regulator is with a SMPS.
Lineaar regulators just burn excess power.
SMPS converts the power instead.
Jan's "Silent Switcher" is a great example. In using Sam Groner's Ultra low noise amplifier, I find no difference whether using his switcher or batteries.
If you don't want to go the SMPS route, Linear Tech (now ADI), LT1963 is much preferred to the LM317.
In 40 years in electronics I had never used a LM317. Always used lm78xx series.
Just recently I designed a circuit which ran off 3v3 but I wanted it to interface a to a 5v logic IC. 3v3 was borderline for the Vh threshold.
So I picked up an LM317 and set it to 3v6 (max for driving IC) and it worked a treat.
Just recently I designed a circuit which ran off 3v3 but I wanted it to interface a to a 5v logic IC. 3v3 was borderline for the Vh threshold.
So I picked up an LM317 and set it to 3v6 (max for driving IC) and it worked a treat.
Other than Jan´s regulator, there are integrated packages from Tracopower that transform 5V to +-15V or the like.
For your current needs these would be a more expensive than DIY (TEL 10-4823 for example) but it´s an option,
if you don´t mind the heavy integration.
They are pretty clean noisewise too. We use them at work a lot.
Alternatives are Murata, Recom or Vicor (the latter if you want to spend lots of money;-))
For your current needs these would be a more expensive than DIY (TEL 10-4823 for example) but it´s an option,
if you don´t mind the heavy integration.
They are pretty clean noisewise too. We use them at work a lot.
Alternatives are Murata, Recom or Vicor (the latter if you want to spend lots of money;-))
Thanks for all the responses so far, I have a lot of research to do. I'm tempted to try a few of the encapsulated switch mode supplies now. I like the wide voltage input, which makes it very easy. My current PSU uses a 16VAC 1000mA power adapter and I could only find one (jameco).
What is preferred about the LT1963? Is there an equivalent part # for a negative rail?
What is preferred about the LT1963? Is there an equivalent part # for a negative rail?
One of the respondents to the Linear Audio article comparing power supply regulators remarked that the error amplifier of some created distortion, lack of gain and bandwidth -- he suggested putting a signal onto the power supply rail and looking to see if the error amplifier created harmonics. LT1963 came out reasonably well on this measure, not as good as the Jung, Sulzer or Sjostrom, but good for an OEM regulator which you can eat right out of the can.
You can also use Kelvin voltage sensing for the LT1963 -- it's shown in the datasheet.
The counterpart is, I believe, the LT3015.
Disadvantage of the LT1963A -- max input voltage!
Attachments
TPS7A4701/TPS7A3001 are quite good. Current is limited so one must divide the circuits over more than 1 regulator per channel/rail which does not hurt. The even better LT3045/LT3094 can deliver 500 mA.
Input voltage is a concern but a little calculation and you are there. It is of no concern till about +/-12V output voltage and a 2 x 12V AC transformer which works OK with LDO regs. When overdimensioning transformers I manage to have a stable 15V output with certain transformers, a not too high load and still enough headroom. This is not textbook practice though and only possible in specific cases.
Switchers/SMSP are given as advice but remember that no silent switcher exists without an ultra low linear LDO behind it. Besides that it is a hell of a job to get other parameters under control that come with SMPS (if you use mains voltage). When fed with batteries or power packs it is different but using those is for the true electro-sadomasochists that wants to charge as many batteries in life as possible. SMPS are cheap, light and high efficiency but they radiate, pollute mains voltage and what not. As a bonus they like to fail when they think the time is right.
Input voltage is a concern but a little calculation and you are there. It is of no concern till about +/-12V output voltage and a 2 x 12V AC transformer which works OK with LDO regs. When overdimensioning transformers I manage to have a stable 15V output with certain transformers, a not too high load and still enough headroom. This is not textbook practice though and only possible in specific cases.
Switchers/SMSP are given as advice but remember that no silent switcher exists without an ultra low linear LDO behind it. Besides that it is a hell of a job to get other parameters under control that come with SMPS (if you use mains voltage). When fed with batteries or power packs it is different but using those is for the true electro-sadomasochists that wants to charge as many batteries in life as possible. SMPS are cheap, light and high efficiency but they radiate, pollute mains voltage and what not. As a bonus they like to fail when they think the time is right.
Last edited:
I second TPS7A4901/TPS7A3001 recommendation. I use them in all of my projects. I can also recommend using isolated DC-DC converters with post regulation using a pair of LDO regs. The DC-DC converter can be powered using a battery pack or a USB charger. This eliminates the mains earth connection and thus reduces the chance to create a ground loop. On top of that it also eliminates the need to work with the mains voltage directly which is one of the most dangerous parts of the diy audio.
Regards,
Oleg
Regards,
Oleg
A real DIYer would know how to work with mains voltages IMHO. It normally all starts with electricity and then electronics. Using the grid is most economical, most green and quite user friendly. Since every home has it already it always was and still is the basis. For some unknown reason it is believed that everything with rechargeable batteries (and everything electric 🙂) is green but it is on the contrary....
PE is required yes, for safety but don't underestimate the other benefit of shielding metal cases to PE. As almost every home has wireless, bluetooth, cell phones, LED lighting, SMPS etc. so shielding and netfilters are meanwhile a necessity. Ground loop issues can be solved by clever implementation and a grounding scheme.
PE is required yes, for safety but don't underestimate the other benefit of shielding metal cases to PE. As almost every home has wireless, bluetooth, cell phones, LED lighting, SMPS etc. so shielding and netfilters are meanwhile a necessity. Ground loop issues can be solved by clever implementation and a grounding scheme.
Last edited:
Talking of the latter mentioned regulators:
TI offers gerber files for a design that could fit in here:
Just search for "TIDUA78B"
TIDUA78B - Google Search
TI offers gerber files for a design that could fit in here:
Just search for "TIDUA78B"
TIDUA78B - Google Search
Nice find Jens! I've implemented similar PSU using the off-the-shelf DC-DC converter. One thing I've done different is that I've put the HF ( C(R)LC filter) before the regs while TI board has it after the regs thus adding the ferrite beads series resistance to the regs output impedance. I find TI solution sub-optimal but probably they know something which I don't 🙂
Jean-Paul,
These days SMPS based PSUs are everywhere, even in the USB interfaces many of us using to measure the performance of our audio projects. The Didden's AutoRanger has incredible performance while using his SilentSwitcher PSU. So I see no problem using a competently designed SMPS PSU in any audio gear.
As for the mains voltage and average diyer being well skilled to work with it I tend to disagree with you.
These days SMPS based PSUs are everywhere, even in the USB interfaces many of us using to measure the performance of our audio projects. The Didden's AutoRanger has incredible performance while using his SilentSwitcher PSU. So I see no problem using a competently designed SMPS PSU in any audio gear.
As for the mains voltage and average diyer being well skilled to work with it I tend to disagree with you.
Yes they are everywhere but that does not take away their negative points. They are chosen for any other property than quality. Simple. I think you mean the Silent Switcher but batteries are simply not green and for me therefor unacceptable even if they would be best for audio. If there is something positive about Covid we can tell that the environment improved because of less pollution. Let's keep it that way.
The old way of learning was : electricity, switches, lamps, sometimes a spark or a molten fuse and only then electronics but I see it has changed 🙂
The old way of learning was : electricity, switches, lamps, sometimes a spark or a molten fuse and only then electronics but I see it has changed 🙂
In Finnish language there is an old saying among electricians: "When the smoke has cleared re-check the wiring" 😀