There is nothing wrong with using LT3045 and LT3094 for op-amps. They are top notch regulators.
They do regulate all right. Don't sound too good with opamps, according to many people who have tried them and then tried other options (such as good old shunt regulators, Jung-Didden still being a pretty good one, but hardly the only good one). Film caps are sometimes used, as a slightly different approach. The issue people seem to have with LT304x for opamps doesn't seem to have anything to do with noise. Perhaps output impedance vs frequency, that's what some people seem to think has to do with what they don't like. Be nice if somebody would take some measurements one of these days to settle the source of complaints.
Last edited:
Measurements of LT3042 output impedance had been done.. below 1mohm up to some kHz. Depends on the actual board. Can be extended like less than 1mohm up to 10kHz. Less than 10-20mohm even at 100kHz.
Measurements here:
LT3042 OUTPUT Impedance（2/2）: アナログ回路のおもちゃ箱
Ciao, george
Measurements here:
LT3042 OUTPUT Impedance（2/2）: アナログ回路のおもちゃ箱
Ciao, george
I don't put any stock in anyone's subjective impressions. The fact is that LT304x is most likely the very best IC series regulator on the market and outperforms 99% of discrete regulator implementations I have seen. Many people are not competent enough to use these parts correctly, to be honest.
Looks pretty good, at least until one compares it with an old favorite of shunt regulator users: A Jung regulator with AD797 is reputedly still better for audio opamps, and the output impedance numbers if of any value at all would seem to give some weight to the claims. Don't know at what point the numbers become ridiculously low relative to audible effects. Don't really want to get back into the never ending audibility debate either.
Maybe I can leave off this way: A number of people have found LT304x regulator designs that they have tried for audio not to be their liking. Why exactly I don't know. I do know that film caps seem to be audibly to liking of some people (same for some shunt regulators), and I doubt the effects can all be attributed to imagination. Understood that others may disagree. I will stop here and let others have last say.
Mark, I have a suspicion that the linear audio graphs, made by Jack, were done with the Jung super regulators.
Which are available classically with ad825 / ad797.
They are not shunt. And not LDO. And not small...
'Small' like in the frase: serve the target circuit from few mm away.. some mm more and the 10uohm output impedance is gone. But an 3042 can be placed there, and provide real 100uohm range performance.
Apart from this, I'm listening to what your people say and curious.
Ciao, g
Which are available classically with ad825 / ad797.
They are not shunt. And not LDO. And not small...
'Small' like in the frase: serve the target circuit from few mm away.. some mm more and the 10uohm output impedance is gone. But an 3042 can be placed there, and provide real 100uohm range performance.
Apart from this, I'm listening to what your people say and curious.
Ciao, g
Thanks
So, nothing wrong going with : Trafo -> CRCRC (quasi-DC)-> LT3042? Or a pre-reg is really needed ? On opamps? AVDD? DVDD?
I have a compact design in mind...
Right, we probably won't agree about the subjective impressions of a handful of people.
My only point is that I certainly would not be steering people away from it. There are megabuck A+ rated products using LM317 or worse regulators for op-amp supplies. Some even use RC filters on the power pins, so I don't think you can even generalize when it comes to supply impedance. It's really reaching to find a flaw with their performance, I mean, I would have to look more closely at that test setup because it may be limiting the regulator's performance in that test.
Thanks
So, nothing wrong going with : Trafo -> CRCRC (quasi-DC)-> LT3042? Or a pre-reg is really needed ? On opamps? AVDD? DVDD?
I have a compact design in mind...
You probably don't need the R's in the supply, honestly, but if you aren't driving a heavy load it will be fine. You probably don't need a pre-reg. If you look at the evaluation board, you can get an idea of what the minimum you need is to get the datasheet performance.
Last edited:
That't right now that think about it some more. The error amp opamp is located downstream of the series regulator.
Regarding LT304x, I can't say that any listening comparisons have been made with the regulators and loads carefully matched. It may be that film caps, or regulators often favored for good audio circuit SQ are just more forgiving with respect to implementation. That remains entirely possible.
There also may be questions about performance of regulators with or without the benefits of multilayer PCBs including the possible use of power planes in addition to the usual ground planes.
Although some effort to sort out technical and listening impressions of various regulators was undertaken by John Walton, Linear Audio | your tech audio resource ...it now appears that a couple of the most highly rated regulators are no longer available.
To me, it still looks like there is more work to do, although my guess would be that mulitlayer PCBs and careful layout can probably help a lot. Unfortunately, low volume production of such PCBs may be beyond the budget of many diy'ers.
Regarding my own plans, I have here waiting to test a TP Placid HD, and Jung regulator set made from the diyaudio store pcbs, and a few other power supplies I want to try out. One that I feel obligated to test at some point is a three transistor nazar regulator (a bit odd design).
Regarding past experiences, one of the units I tried film caps (& LM317 type regulators) on was Allo Katana dac (for the +-15v output stage power). Although they seem to have competent engineers who are pushed to excel, for their more exacting customers who are the most picky about SQ, only linear regulators and film caps so far have been able to get the cleanest subjective SQ out of the dac. Allo is responding by coming out with their own linear supply with impressive noise performance. They say output impedance should be low enough without needing film caps. We will see.
I know DAC-3 did a great job with SQ starting with a SMPS and using some LDO regulators. However, they also have a multilayer PCB, which conceivably may help significantly (along with overall good engineering). So, not question if it can be done, it clearly can be. Only not sure exactly what it takes to get to that level.
I do know that film caps and linear supplies can come very, very close to the same sound quality as DAC-3, and without a fancy PCB. However, film caps and linear supplies aren't cheap, space efficient, or low cost to ship, things that make them not suitable for commercial products that can't justify their use unless by producing SQ superior to DAC-3. But, for diy use, I think linear supplies, film caps, or popular good sounding regulators are all fine.
Until someone can show how diy'ers can get equal or better sound quality out of LT304x, that's probably they way things will stay.
Last edited:
Here are a pretty good set of measurements of LT3042. Use Chrome / translate.
PSRR:
LT3042 PSRR Cset(PMLCAP vs MLCC): アナログ回路のおもちゃ箱
LT3042 Noise Spectral Density nV/√Hzの測定: アナログ回路のおもちゃ箱
OP Z:
LT3042 OUTPUT Impedance（2/2）: アナログ回路のおもちゃ箱
As you can see these regs are at least as good as data sheet and better than many other discrete series pass or shunt regs. I believe LT3045 *may have even better performance.
My experience with using LT3045 for clock supplies and general digital PS work has been very good.
Are there better regulators? - of course there are, but for a 6 dollar chip or 20 dollar complete pcb, they are pretty hard to go past, especially when you factor in that the bypass arrangement is a big part of the equation.
T
Terry,
It appears some people find LT304x perfectly fine for digital sections of dacs, just not for AVCC and the output stage.
What about ADM7150? I see it used for all dac digital and for AVCC too. Why do you think people prefer it over LT304x?
EDIT: There is something interesting in LT3045 data sheet page 16:
Given the high PSRR and low noise performance attained
using a single 10µF ceramic output capacitor, larger values
of output capacitor only marginally improves the performance because the regulator bandwidth decreases with
increasing output capacitance — hence, there is little to
be gained by using larger than the minimum 10µF output
capacitor. Nonetheless, larger values of output capacitance
do decrease peak output deviations during a load transient.
So, more output caps is better for load-related-transient voltage regulation, at the expense of reduced regulation bandwidth during non-load-transient conditions? What if 'peak output deviation' matters for opamps, or for AVCC?
It appears some people find LT304x perfectly fine for digital sections of dacs, just not for AVCC and the output stage.
What about ADM7150? I see it used for all dac digital and for AVCC too. Why do you think people prefer it over LT304x?
EDIT: There is something interesting in LT3045 data sheet page 16:
Given the high PSRR and low noise performance attained
using a single 10µF ceramic output capacitor, larger values
of output capacitor only marginally improves the performance because the regulator bandwidth decreases with
increasing output capacitance — hence, there is little to
be gained by using larger than the minimum 10µF output
capacitor. Nonetheless, larger values of output capacitance
do decrease peak output deviations during a load transient.
So, more output caps is better for load-related-transient voltage regulation, at the expense of reduced regulation bandwidth during non-load-transient conditions? What if 'peak output deviation' matters for opamps, or for AVCC?
Last edited:
Who are these people? I don’t think unsubstantiated claims from a few people with unknown implementations are even worth mentioning.
ADM7xxx came out earlier and had a negative version much earlier, same with TPS7A. They’re popular for that reason. They are all good regulators and how the layout is done is of high importance to achieve the datasheet specifications.
As soon as I see a daughterboard, long wires, huge film caps and the like I stop paying attention.
Last edited:
Oh, I don't know. There are various. I'll have to take a closer look inside DAC-3 one of these days and see what they use.
Trident-SR
Terry, do you use LT304x for AVCC? Please answer if you do.
Last edited:
This is true for every regulator, it’s nothing unique to these parts. Those pins and op-amps are benign loads, generally speaking.
When you can show me a dac that is reputed to have top notch sound quality and uses LT304x for AVCC and or for +-15v output stage, I will be very interested in it.
They are very new parts, but they will appear in time. As I said before, there was no negative complement for LT304x for something like 1.5 years. This is why they got used in single supply applications - because there was no negative regulator and presumably a lot of designers find it weird to use LT304x for positive and TPS7A33 or whatever for negative. Had LT3094 been available from day one, I guarantee that many of the designs using the ADM7xxx regulators would be using the LT parts.
In the pics of the DAC2, Benchmark is using two small SOT regulators near the ESS DAC. Not sure if that's changed, can't find high enough resolution pic of DAC3 to read markings off. It's also possible there are other linear regulators elsewhere but I can't see them in the pic I have.
http://www.ti.com/lit/ds/symlink/tps730.pdf
Not particularly impressive compared to the LT304x. Still probably sufficient for the job.
I trust these measurements from LT and others much more than unsubstantiated claims from a VERY small group of people that may or may not have used this regulator incorrectly. It's not possible to achieve the best performance without using the remote sense pins correctly. Any 3-terminal board is compromised from the start. There IS a reason that Benchmark crams those tiny regulators right at the point-of-load.
IMO, you should soften your statement to avoid leading well-intentioned members away from a top notch part with hearsay. The wording of your posts makes it seem like this is proven fact when it is debatable at best. I am done with this topic, I just wanted to be sure that new members are not mislead.
Last edited:
I cannot recommend for people to use LT304x for AVCC or output stages at this time. Too many reports of bad experiences from various people. For digital they are likely fine. I think somebody needs to do some work to find out what is going on that makes people unhappy with the part's performance. People were expecting really good performance based on the numbers, but report preferring the sound of opamp regulators for AVCC, or possibly ADM7150 (haven't tried 7150 myself yet, but have one here for that purpose).
My usual policy is to only recommend what I know from personal experience works, since for really good SQ performance, listening tests of prospective components is one qualifying factor, but not the only one. Obviously, measured performance is important too.
For people that are already expert or very experienced, they can do whatever they want as far as I'm concerned. However, I want to the less experienced people to have predictable outcomes from their modding work. The work is hard enough for them without adding potentially complicated puzzles of sound quality issues to solve.
Might be fine for the digital part of the dac or of the nearby FPGA. Don't know about AVCC. A multilayer PCB with power and ground planes might change things too. That and a set of distributed X7R bypass caps can make power supply impedance at HF very low indeed.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.