Your argument is that the datasheet graph does not go beyond 1MHz. And you do that on a thread in which is obvious from the very first post that it goes beyond datasheet spec and performance gains.
I asked if you can provide some measurements that would settle this and would also make everyone wiser yet you seem to have an affinity of calling it false claims, false information etc.
So I am asking you again, are you sure that LM317 does not offer ANY kind of PSRR post 1MHz, to let's say 5MHz? And are you basing that on just the datasheet graph?
I asked if you can provide some measurements that would settle this and would also make everyone wiser yet you seem to have an affinity of calling it false claims, false information etc.
So I am asking you again, are you sure that LM317 does not offer ANY kind of PSRR post 1MHz, to let's say 5MHz? And are you basing that on just the datasheet graph?
You are once again putting words in my mouth. My argument has nothing to do with how far beyond 1MHz the datasheet graph goes. You should read the article you yourself quoted. It has the measurement and it also gives you an explanation of the behaviour.
And all your bickering does not add any credibility to your blatantly false claim "I know that LM317 is good up to around 5MHz."
There's nothing false in that statement as that is what I know from all the info I got. I don't really understand what you are aiming at. That is what I know, if it isn't so you could provide some measurements showing the opposite so I'll know better, that LM317 does not offer any kind of PSRR up to 5MHz. But there's nothing false in that statement. I didn't say the LM317 offers PSRR up to 5MHz as an absolute reality, as in that everybody knows and it's an absolute fact. There's a difference there that you'll continue to ignore. Saying that I know that means that from all the info I got that is what I know, which is very different from stating it as universal fact.
After I said that, your argument against it was that the datasheet does not go beyond 1MHz so it can't offer PSRR past 1MHz. When I asked you if by that line of reasoning the LT1085 does not offer PSRR past 100KHz, to say 500KHz you started accusing me of putting words in your mouth, which is kind of strange because that is your own line of reasoning.
So, your sole argument that the LM317 does not offer PSRR past 1MHz was because the datasheet graph for PSRR does not got beyond that point. Do you have other arguments than that? Because that argument implies that LT1085 does not offer PSRR past 100KHz.
Gentlemen, I see this thread derailing so please let's stop this.
When reading the document (nice document indeed) is is clear that LM317 is limited to 1MHz but the addition of a parallel cap to Cadj can make it work higher up in frequency. This is not datasheet wisdom but empirical and probably depending on brand too. I just examined LM317 data sheets and not a single one mentions LM317 capable of higher than 1 MHz.
The document in the link says it like this:"No misleading claim or inflated figures on PSRR beyond 1 MHz are shown because, beyond that, it’s no longer a measure of the native device characteristics." Well, that is clear enough I think.
When reading the document (nice document indeed) is is clear that LM317 is limited to 1MHz but the addition of a parallel cap to Cadj can make it work higher up in frequency. This is not datasheet wisdom but empirical and probably depending on brand too. I just examined LM317 data sheets and not a single one mentions LM317 capable of higher than 1 MHz.
The document in the link says it like this:"No misleading claim or inflated figures on PSRR beyond 1 MHz are shown because, beyond that, it’s no longer a measure of the native device characteristics." Well, that is clear enough I think.
Last edited:
I was personally hoping for someone to chime in with measurements settling this. I usually have more trust in Cunningham's Law than in people's good intentions and it usually pays out. Except when it doesn't and the interlocutor's intentions are hell bent on making the other side look bad instead of sharing knowledge. Which is usually wasteful of time and posts.
What are you steering at? Do you like conflict or what? Both the documents and the document in the link are clear. The device may be the best in the world but no guarantee is given that it functions above 1 MHz. Simple. Your new ones may measure till 100 MHz but the ones that are in drawers for 20 years may not. So 1 MHz, the official datasheet limit, should be followed.
Again: the document in the link says it like this:"No misleading claim or inflated figures on PSRR beyond 1 MHz are shown because, beyond that, it’s no longer a measure of the native device characteristics."
Again: the document in the link says it like this:"No misleading claim or inflated figures on PSRR beyond 1 MHz are shown because, beyond that, it’s no longer a measure of the native device characteristics."
Last edited:
I have not started any conflict, I'm just minding my own interests in this thread.
This thread is not really about datasheet applications so I don't understand the need to constantly mention the datasheet performance, since it's been proven with additional parts we can push the LM317 performance way past the datasheet. Seems that the dienoiser + LM317 offers better PSRR at audio frequency than the datasheet application of the cherished LT3042.
Yes, again, the LM317 by itself does not guarantee anything past the datasheet. That has been mentioned many times. But for experimental DIY stuff the datasheet specs are not all there is. Again, the LM317+dienoiser offers better PSRR (at audio frequencies) than the datasheet application of LT3042. Also better output impedance.
This thread is not really about datasheet applications so I don't understand the need to constantly mention the datasheet performance, since it's been proven with additional parts we can push the LM317 performance way past the datasheet. Seems that the dienoiser + LM317 offers better PSRR at audio frequency than the datasheet application of the cherished LT3042.
Yes, again, the LM317 by itself does not guarantee anything past the datasheet. That has been mentioned many times. But for experimental DIY stuff the datasheet specs are not all there is. Again, the LM317+dienoiser offers better PSRR (at audio frequencies) than the datasheet application of LT3042. Also better output impedance.
Also at the frequencies found in modern DACs ?
The way you are minding your own interests (really?) is aggressive towards others. You may not seem to notice this but still. It is clear that you will do everything to keep using the cherished LM317 in all scenarios and that is OK. No need to be aggressive and to mention Cunninghams law and the other under the belt techniques as you repeatedly do.
The way you are minding your own interests (really?) is aggressive towards others. You may not seem to notice this but still. It is clear that you will do everything to keep using the cherished LM317 in all scenarios and that is OK. No need to be aggressive and to mention Cunninghams law and the other under the belt techniques as you repeatedly do.
Last edited:
I have not measured that and simulations does not indicate it.
But simulation do indicate that LT3082 +dienoiser would have better PSRR than datasheet application of LT3042 to 5MHz. I won't be able to measure that sadly as I don't have the gear, I'm limited to audio only. But if the measurements match the simulation predictions I'd say it's a good indicative that it might offer better PSRR at higher frequencies as well.
But simulation do indicate that LT3082 +dienoiser would have better PSRR than datasheet application of LT3042 to 5MHz. I won't be able to measure that sadly as I don't have the gear, I'm limited to audio only. But if the measurements match the simulation predictions I'd say it's a good indicative that it might offer better PSRR at higher frequencies as well.
I'm the first to acknowledge my social skills are not top notch but you seem to have a mafia type approach to this. The other user gets aggressive by accusing me of spreading fake information than you came barging in under the pretext of cooling the discussion but one comment later you ask me if I like conflict "or what". You seem eager to pile on as well.
Yes my personal opinion is that LM317+extra parts is more than adequate in many applications. The output impedance of LM317+dienoiser is lower than that of datasheet app of LT3042 to about a bit over 700KHz, and that may matter for digital clocks. Might make the LM317+dienoiser a better option than LT3042.
This might be of interest, also mentiones that ferrite beads might worsen the output impedance of the regulators. Which is something I noticed while looking for ferrite beads for the digital part of a CS4398 DAC that I am working on right now. Lowest resistance I found for them was few mOhms which start to matter when also adding the tracks.
Optimizing Power for Clocks and other Sensitive Applications | Sierra Circuits
Optimizing Power for Clocks and other Sensitive Applications | Sierra Circuits
LM317N can be found in SOT223 package that make it suitable for digital work at 200mA max output. The additional dienoiser doesn't take that much space, and the whole thing is still easier to solder than the LT3042 package.
There's 6 dienoisers on this DAC board, +-12V for opamps, +5VACC/Vref, +5V analog for clock recovery, and two +3.3V digital for each WM8804 and CS4398. Being cheaper than LT3042 means I can easily use more.
Attachments
Last edited:
Again an under the belt remark to suggest I am mafia like. Please read your own posts and your own remarks in this thread. You are the one using fake information as bohrok pointed out.
Yes that is what people do when you answer like this with personal remarks, hidden insinuations and such. Like often people who act like this don't realize that it is their own fault to make themselves look bad:
While you are also to agree that LM317 is not for new designs you design new stuff with it
I guess you should build that DAC with your favorite stuff and one with other regs to measure and find out what is best in real life situations.
Have fun with your favorite part, capo di tutti capi.
Yes that is what people do when you answer like this with personal remarks, hidden insinuations and such. Like often people who act like this don't realize that it is their own fault to make themselves look bad:
While you are also to agree that LM317 is not for new designs you design new stuff with it
I guess you should build that DAC with your favorite stuff and one with other regs to measure and find out what is best in real life situations.Have fun with your favorite part, capo di tutti capi.
Last edited:
I wouldn't use LM317 by itself for new designs. I mean I might depending on the investment target of-course. It's still pretty decent for the money.
This thread is not about LM317, it's about LM317+dienoiser which has better PSRR at audio than many more expensive regulators, and also has lower output impedance higher up in frequency than many new more expensive regulators. With the dienoiser it's totally adequate for new designs.
This thread is not about LM317, it's about LM317+dienoiser which has better PSRR at audio than many more expensive regulators, and also has lower output impedance higher up in frequency than many new more expensive regulators. With the dienoiser it's totally adequate for new designs.
While you are also to agree that LM317 is not for new designs you design new stuff with it
LM317 is not intrinsically "dirty" as a regulator. As long as the performance is there if used with the dienoiser I don't get what you are trying to say. If it has better PSRR than LT3042 at audio which is pretty important for ripple for example, and has better output impedance to 700KHz then it makes it suitable for analog and digital. Considering the price and ease of soldering you'd be hard pressed to convince anyone of improvement in resulting sound by using LT3042 instead of LM317 + dienoiser. There's situations where datasheet applications of other regulators might measure better than LM317+dienoiser, but it seems pretty close to make me personally doubt that choosing LT3042 over LM317 +dienoiser is going to make any difference in output sound.
The difference in PSRR vs LT3042 is not that big at higher frequencies, 30dB or so, but LM317+dienoiser seems to have lower output impedance at those higher frequencies. Side by side the end result might very well be identical as perception by most people. Considering the price and ease of solder it seems pretty easy choice.
Last edited:
Although I already said goodbye: of course I do use various regulators in real devices so I don't have to doubt, assume, use words as "seems", "might", "perception" and other assumptions, call 30 dB a small difference etc. Nor do I have to lay words in another person's mouth and try to change the discussion by stating I don't know what another person is trying to say. We already had the "racism" thing which was already way below the belt. No one said LM317 is an intrinsically "dirty" regulator except you in post #1856. These are political tricks of the lowest kind.
Price and ease of soldering are non arguments considering the vast amount of also difficult to solder SMD parts in 2020 designs. DIY is more expensive than ready built anyway so one better uses the right weapons for the battle. We already agreed that LM317 is a no no for new digital designs so end of discussion. Bye!
Price and ease of soldering are non arguments considering the vast amount of also difficult to solder SMD parts in 2020 designs. DIY is more expensive than ready built anyway so one better uses the right weapons for the battle. We already agreed that LM317 is a no no for new digital designs so end of discussion. Bye!
Last edited:
For PSRR the extra performance is mostly relevant if there is anything there to attenuate. If there isn't noise in those frequencies then the extra PSRR isn't that important. The ripple is always there tho.
Price and ease of soldering are very important on a diyaudio forum. I'm not sure why you can't accept the majority of people for which price matters. Seems like you are in a very exclusive club. DIY can be more expensive yes, and it can also be way cheaper if one's looking just for good sound and no extra features, fancy displays and cases.
Also you agreed by yourself that LM317 is a no-no. I said I might use LM317 depending on the application and target cost.
And LM317+dienoiser I absolutely recommend for new designs. First off it's there with LT3042 for ripple so I don't see why I'd pay more and struggle more to properly solder the LT3042 package. You can't even solder it using a soldering iron, you'd need a hot air station which is something many beginners don't have. You could try it from the underside with the soldering iron, to get through the power pad but that can be messy and ruin the expensive regulator. Where a SOT223 package is dead easy to solder.
Price and ease of soldering are very important on a diyaudio forum. I'm not sure why you can't accept the majority of people for which price matters. Seems like you are in a very exclusive club. DIY can be more expensive yes, and it can also be way cheaper if one's looking just for good sound and no extra features, fancy displays and cases.
Also you agreed by yourself that LM317 is a no-no. I said I might use LM317 depending on the application and target cost.
And LM317+dienoiser I absolutely recommend for new designs. First off it's there with LT3042 for ripple so I don't see why I'd pay more and struggle more to properly solder the LT3042 package. You can't even solder it using a soldering iron, you'd need a hot air station which is something many beginners don't have. You could try it from the underside with the soldering iron, to get through the power pad but that can be messy and ruin the expensive regulator. Where a SOT223 package is dead easy to solder.
The way I see it you didn't make a good case for defending the newer more expensive fancy regulators vs LM317+dienoiser. For analog there wasn't much to defend, and for digital you started loosing ground at the LM317+dienoiser's lower output impedance at those higher frequencies. I think the only argument was what, higher PSRR at higher frequencies? That is relevant if there's noise there, and that can be very easily fixed with an inductor before the regulator. Considering the fact that the only advantage is that higher PSRR at those higher frequencies, and the LT3042 is more expensive, much easier to ruin at soldering, and the size advantage isn't that much vs a sot223 LM317+dienoiser I don't see any practical and financial (combined) reasons to go for the LT3042.
If anything, you made a better case for people using LM317+dienoiser in new designs, digital and analog for audio. Maybe not every case, as I admit I don't know that it would always be better, but you didn't make a compelling case.
What you also need to consider as maybe this wasn't clear, this is a hobby for me, I have no formal training in this field. I just enjoy tinkering with it. So I might be wrong on things, would be great if people weren't so aggressive in correcting my nefarious mistakes.
Adding 100uH in front of the LM317+dienoiser solves the PSRR at higher frequencies issue, and it gives very good performance, and still cheaper overall. Yes it is an extra part that takes extra space, not much tho. And it isn't to get to LT3042 performance, it goes way past it. Totally worth it.
Now we add ADM7150 into the mix, to see how the stock datasheet application stands between LT3042 and LM317+dienoiser and LM317+dienoiser with L in front. We also can have a look at the comparison between output impedances.
These are all simulations of-course and are no replacement for actual measurements. But if HF PSRR isn't an issue with the extra L, and LM317+dienoiser seems to have lower output impedance to 700KHz or so, what exactly is there that would outshine the LM317+dienoiser in digital? (edit: maybe low dropout? that is a good advantage indeed) We're also talking about sot223 package LM317.
If anything, you made a better case for people using LM317+dienoiser in new designs, digital and analog for audio. Maybe not every case, as I admit I don't know that it would always be better, but you didn't make a compelling case.
What you also need to consider as maybe this wasn't clear, this is a hobby for me, I have no formal training in this field. I just enjoy tinkering with it. So I might be wrong on things, would be great if people weren't so aggressive in correcting my nefarious mistakes.
Adding 100uH in front of the LM317+dienoiser solves the PSRR at higher frequencies issue, and it gives very good performance, and still cheaper overall. Yes it is an extra part that takes extra space, not much tho. And it isn't to get to LT3042 performance, it goes way past it. Totally worth it.
Now we add ADM7150 into the mix, to see how the stock datasheet application stands between LT3042 and LM317+dienoiser and LM317+dienoiser with L in front. We also can have a look at the comparison between output impedances.
These are all simulations of-course and are no replacement for actual measurements. But if HF PSRR isn't an issue with the extra L, and LM317+dienoiser seems to have lower output impedance to 700KHz or so, what exactly is there that would outshine the LM317+dienoiser in digital? (edit: maybe low dropout? that is a good advantage indeed) We're also talking about sot223 package LM317.
Attachments
Last edited: