But so what? Why are people so fixated on noise for dac circuits when distortion is more the problem in most cases? The battery fad appears to be something a subset of hobbyists have turned to primarily because they couldn't get LDOs to sound good. I would suggest we go back and start over this time trying to do a better job with power supplies. Then we can get to output stages and making simple dac circuits work their best. Complexity can come later, after the basics are well understood.
My dear salesman, "chemical noise", if exists, is always excess noise, on top of the Johnson noise. Since the NiCd ESR measurements fully accounts for the measured noise, as Johnson noise, it results that no "chemical reaction noise" (whatever that is, for batteries) is involved. To add insult to injury, any "chemical noise" due to temperature fluctuations, and fluctuation of the number of ions (that could be alleged for certain batteries), have always spectral densities that are flat at low frequency and decrease with 1/f^a with a~3/2 or larger. Such a behavior was not observed in the NIST experiment for NiCd cells, therefore No Soup For You.
As of the origin of a battery ESR, do your homework and find its origin by yourself.
Last edited:
The linked measurements show batteries are lower noise than LDOs.
Sure it's not just about noise (this is the visible tip of the iceberg,IMO) & why I say stability (less interaction between battery & device being powered) seems to me to be also in the mix (probably other factors at play too?).
Discrete component regulators such as Paul Hynes designs seem to come close to battery/supercap power so it's not impossible but LDOs seem less capable
Dear Mr personality disorder - if you don't know what chemical noise in batteries is you need to go back to battery university (you may also learn about ESR in batteries) but stop off in the psychiatry dept first - I'm sure they'll find you fascinating
Yeah, now that I recall better my experiment, I don't think I averaged for 24 hours or more, since I was still getting those nasty fluctuation of likely thermal origin, so no wonder I couldn't get lower. But my gain was much higher than the NIST, 80dB if I recall correctly.
Let me know what you get, now I think you are on the right track (from paralleling 20 opamps).
Those 10 ADA4898-2 pairs have the advantage that they don't oscillate somewhere with the right input Z, as did all JFET amplifiers I checked (own&foreign) with the standard architecture, ( CS input, optional cascode, op amp for loop gain, feedback into the source, which makes the CS input no more CS at all. It could just as well say it's a follower with a nasty load. The source follows the gate with a delay.).
For me as a metrology guy, negative input impedance is simply a NoNo. It invalidates anything good you could say on an amplifier. I have given up on the feed back loop. Now the gain is no longer described by two resistors. Since the gain of the input FET is now proportional to sqrt(source current), it is fed by a current mirror. There stays a gain error of 0.8 dB / 10°C in simulation. When I find the time, maybe I'll tackle that with NTCs or a FET oven; I have one that works fine in LTspice.
Since the current mirror feeds the source of the input transistor, there must be something that keeps the sources at GND ac-wise. When the gate is dc-grounded, the sources will be at a +few 100 mV. I was thinking that a few 4700uF/2V7 Oscons would do the job. They did, but their leakage current subtracted from the astral clean current source. Result was 1/f**3 noise rising from 500 Hz. > 10V ordinary electrolytics healed that. That's 4 of the 6 large caps you see on the empty board.
The op amp solution is not the only way here from the beginning. The version number of the FET amp is 10.1.
Now, you gona be "routed to /dev/nul". But do not worry, in fact, it's lot better.
when being near rabid animals, better to be on the other side of the fence.
Last edited:
Merrill,
Don't know if you noticed, but Syn08 was trying a little experiment for part of today, conducting himself in a more gentlemanly and professional manner. Please acknowledge the improvement and respond accordingly, if possible. I know the 'Mr. so and so' thing was a little drift backwards, but please lets try to keep it positive for as long as folks are willing.
The datasheet you linked had 10 milliOhms for 100F at 2.7V. I can get 50 milliOhms with 1500 uF at 15V. The esr is not that low. Tantalum caps are lower.
The supercaps look good for some mobile apps but don't bring a lot except turn on stress that I can see. Still some measurement should be able to show something.
https://www.mouser.com/pdfdocs/ADI_LDO_General_Presentation_2018Apr1.pdf
I didn't have a problem with their measurements, I just don't know the conditions under which the Linear/AD measurements were done.
You said it was unregulated yourself. Otherwise it would have defeated the entire purpose of your reply, which was to suggest NOT using a regulator and directly driving it with the battery. I mean, come on, are you even trying?
mmerrill99 said:
IMHO, the noise thing is way overblown. Output stages are being designed and built with resistors that are way too small for good sound quality, all in the name of low noise. Trade offs are always there to be dealt with in engineering, and going way overboard on noise is not a balanced trade off at all.
I don't think that the noise of batteries comes directly from the low source resistance. One of my 3V7 Samsung Lithium Ion cells delivers easily 30 A, so don't forget your fuses or you might be sorry.
If you compute the source resistance from that: that is much less than needed to explain the noise.
If you compute the source resistance from that: that is much less than needed to explain the noise.
I treat people the way they treat me - nasty gets nasty back; civil gets civil back.
Not just "Dear Mr salesman" but also "Chemical noise, whatever that is" & "As of the origin of the battery ESR, do your homework and find its origin by yourself."
The guy was obviously a backroom techy & should never be allowed out in public with his zero social skills
When Syn08 is civil to me he will find me civil to him. Until then, it is what it is
You guessed right, directly to /dev/nul
Though I have to admit, I was never called a "rabid animal", not even in high school, that's a first
. Say, how old are you?We were talking the NIST paper, about NiCd cells.
350F & up have <5mohm
As I said the noise thing is just the visible indication of something deeper going one - settling time? transient response?
And yes noise on a reference is pretty important
No doubt there are other areas to be tweaked but low hanging fruit is always tempting
I think he just has a low tolerance for factually incorrect statements, baseless assertions, and the non-stop repetition of said annoyances.
mmerrill99 said:
Go ahead and throw 350F 5 mOhm cap across your supplies. Hope you don't have a fuse and you have a robust soft-start circuit.
Yes, for sure, ultra-low noise for audio is mostly about bragging rights (in a DAC or preamp output, anyway). What's the noise floor of your recording, or the average listening room?
mmerrill99 said:
Or maybe noise is just the visible indication of noise?
It's like you are trying to cure a disease when you don't know what it is or how the body works. It would help to slow down and develop an understanding first, would it not?
Last edited:
Your statement that supercaps have high ESR is a baseless assertion.
Your statement that voltage regs are lower noise than batteries/supercaps is a baseless assertion.
Do you want to develop some understanding first before you make these baseless assertions or are you just happy to make a fool of yourself in public - I know trying to be right is a big ego thing with you - pity you fail at it?
Your statement that voltage regs are lower noise than batteries/supercaps is a baseless assertion.
Do you want to develop some understanding first before you make these baseless assertions or are you just happy to make a fool of yourself in public - I know trying to be right is a big ego thing with you - pity you fail at it?
Last edited:
We can do very well with a solid state low noise reference when we need one. Lack of batteries is not a real problem for dacs, it just isn't. We do need attention in other areas that are getting neglected though, and those things will continue to get neglected by the dac battery guys at the rate things are going.
http://www.ti.com/lit/an/slyt151/slyt151.pdf
Do some reading, please. You just continue to embarrass yourself. This is in reply to an earlier question about the effects of output capacitance on linear regulator response.
You can take my statements out of context if you want, sure. Everyone else here knows the truth.
Last edited:
We all have our unique ways on this journey - which path we take depend son so many factors but one path doesn't necessarily exclude the other
What has that got to do with your baseless & patently incorrect assertions?
Refresh the page for edit. It was a direct response to an earlier question you posed. It explains the compromise between output capacitor size and transient response.
Really, the level of idiocy you put on display here is quite impressive.
- Status
- Not open for further replies.