It's always wildly important to remember the source of such posts. This is the same person that has a (IIRC) suburban house with sub-anechoic chamber noise levels.
Please pass the salt.
If you have Bob´s book (of course you have) the relevant figures are 24.3, 24.4 and 24.5.
Jan
The transformer feedback amp remains on top at .065nV/rt Hz using only a pair of SK170's at the input. NIST has a nice article on using an auto-correlating amplifier to get -200dB resolution to characterize battery noise.
If I get ambitious I might revive my idea of using a sound card one channel as main source into a twin-tee of precious R's and C's and the second channel as distortion subtractor, divided down by 1000 or so. Now that the little Analog Discovery dongle has a Python interface I could automate the process of measuring the residual and computing magnitude and phase of the signal needed to null the distortion.
BTW my day was made yesterday when by sheer coincidence my son was having a yard sale and we came by to pick up the grandchildren and he had my Hafler 220 that I "Pooged" back around 1980 on the curb for $20 and lost track of. It ended up in his garage.
Last edited:
For less than $100 I'll play. I do have a ribbon mic input transformer to try.
Demian good point about the input capacitors. Curious to know how much effect the multiplier will have. Victors oscillators are quite nice and I may quit with a set. But I am really curious as to how low one can go.
DNF anechoic chambers are at least 5 dB quieter and as much as 24 dB quieter. Normal small town noise levels are often 25 dB outdoors at night. Bird chirping runs about 30 dB at 30'. Really stands out above the noise level. A foot thick brick wall is limited by the doors and windows. Not hard at all to get to 15 dBa. Ever actually make any measurements?
Yes that should work. Who knows we might be able to eventually get a true 26 bit signal.
Vaccy,
I set the design goal a bit high for a reason. Anything much less isn't a challenge. Now if I get there....
lowest noise voltage references?
Curious if anyone has been keeping up with these, as they should set the lower limit to regulator noise.
I just built a 60V series regulator from mostly discretes. It will provide power to a preamp which has some PSR to be sure. For fun and because they are plentiful and cheap, I am generating a 25V reference from a series string of 10 TL431s, which are not particularly quiet. And I was concerned that there might be oscillations, but things seem to be stable. I take the 25V and R-C filter for the reference input of the error amp. There is a little preregulation for the 431s and two-section R-C filtering for the unregulated input.
I think I posted somewhere, maybe not in here, the topology of the error amp, which contributes to very high line rejection. The pass transistor is a PMOS part.
Curious if anyone has been keeping up with these, as they should set the lower limit to regulator noise.
I just built a 60V series regulator from mostly discretes. It will provide power to a preamp which has some PSR to be sure. For fun and because they are plentiful and cheap, I am generating a 25V reference from a series string of 10 TL431s, which are not particularly quiet. And I was concerned that there might be oscillations, but things seem to be stable. I take the 25V and R-C filter for the reference input of the error amp. There is a little preregulation for the 431s and two-section R-C filtering for the unregulated input.
I think I posted somewhere, maybe not in here, the topology of the error amp, which contributes to very high line rejection. The pass transistor is a PMOS part.
Sipex's equivalent part for the '431 has dramatically better guaranteed noise specifications. Check out the SPX2431, SPX432, SPX2432, and SPX385.
Whichever "shunt voltage reference" zener equivalent you select, you can just use the good old National Semiconductor trick of 8 identical references fed by 8 identical JFET current sources, with 8 identical 0.1% resistors to average them at a summing node. If the noise is uncorrelated ...
Whichever "shunt voltage reference" zener equivalent you select, you can just use the good old National Semiconductor trick of 8 identical references fed by 8 identical JFET current sources, with 8 identical 0.1% resistors to average them at a summing node. If the noise is uncorrelated ...
Here's the reference.
ftp://picovolt.com/Lepaisant_preamp-xfmrs_RSI.pdf
Even if I do believe, I would think an anechoic environment would not be a very good listening space. We had one, I certainly would not enjoy being in it for more than a little at a time.
Last edited:
Scott,
Can we really compare a very low background noise level to an anechoic chamber? I would think that is not really what Ed would have internal to his house once you start playing music, you would still have reflective surfaces rather than the dead non reflective properties of a chamber. Ed is only talking about the background noise and not the decay rate of the inside of his house.
Can we really compare a very low background noise level to an anechoic chamber? I would think that is not really what Ed would have internal to his house once you start playing music, you would still have reflective surfaces rather than the dead non reflective properties of a chamber. Ed is only talking about the background noise and not the decay rate of the inside of his house.
Ed, it's all for the sport. I just spent a couple of days pushing the distortion of a soundcard below 110 dB. Different league, same ballgame.
I missed the discussion about digital cameras / iPods, but you might be interested to know that dxo, the company whose camera test suite has become a de facto standard on many review sites, recently released their first camera.
It's an add-on for iPhone 6. Naturally it has good dxo test scores. Resution, colour accuracy, barrel, pincushion distortion etc
If it wasn't for the on-cost of an iPhone and being pretty much invested in android, I'd be seriously thinking about one of these. I think it works with iPod too but I like to keep down my edc, so it would have to replace my phone.
It's an add-on for iPhone 6. Naturally it has good dxo test scores. Resution, colour accuracy, barrel, pincushion distortion etc
If it wasn't for the on-cost of an iPhone and being pretty much invested in android, I'd be seriously thinking about one of these. I think it works with iPod too but I like to keep down my edc, so it would have to replace my phone.
Are you sure it works that way? I remember Ed describing extensive wall treatments.
Thanks, I will look at the Sipex parts.
Yes, the parallel averaging works as well as the series averaging, and the latter can require an impractically high voltage. In this case the ~25V was useful for the overall regulator needs. In both cases the individual reference noises are uncorrelated, and in the series case the net voltage goes up as N while the noise goes up as the square root of N, the same net enhancement.
What I have done for really low noise references is the labor-intensive use of selected and multiple-cascoded JFETs with the first part and a source resistor selected for ~0 tempco. The voltage across the source resistor can be the reference, or the whole chain used as a current source to develop a voltage across some additional resistor, which can then be readily filtered and buffered as required.
When the temperature stability was critical I put the arrangement and the buffer amplifier in a temp-controlled chamber.
Scott,
I think you have to remember that he Ed was talking about thick masonry walls and a few windows. Even with some room treatments that isn't going to come close to being the same as an anechoic chamber. So he might have really low background noise which would me he would have great dynamic range capabilities but that is all that I would expect. He may even treat the room to dampen some early reflections but he would never get a chamber type of environment, I don't think anyone would want to actually live in a space that was that dead, I think it would be rather annoying to even talk to someone who wasn't facing you directly.
I think you have to remember that he Ed was talking about thick masonry walls and a few windows. Even with some room treatments that isn't going to come close to being the same as an anechoic chamber. So he might have really low background noise which would me he would have great dynamic range capabilities but that is all that I would expect. He may even treat the room to dampen some early reflections but he would never get a chamber type of environment, I don't think anyone would want to actually live in a space that was that dead, I think it would be rather annoying to even talk to someone who wasn't facing you directly.
Even if you don't talk, it is impossible to afford this feeling of being deaf.
We were obliged to set a radio to generate noises (or music) during all the time we were obliged to stay in an anechoic chamber.
- Status
- Not open for further replies.