Hi,
Hey, I worked with those in the 80's. The actually at the time most precise and lowest drift solution we had at the time in the "100 Tons" class where classic mechnical balances with an attachment that rotated optical code disks. The microgram scales for the lab where all electronic digital of course.
I believe those capsules need a polarisation voltage, I would have to check how much. Difficult to do cleanly from 48V phantom power, well, there are 1MHz switchers these days, I would need to check if they can work in boost configuration...
For a preamp circuit, the common standard these days is a pair of PNP Emitter followers which use the phantom power resistors in the Mike Pre/Desk as loads, suitably biased, with their collectors supplying a zenner diode which in turn supplies a J-Fet that operates as Buffer and Concertina Phase Splitter. I believe this originated with Georg Neumann GMBH.
Scott Helmke's Alice is a good example:
As you can see, shooting for super low noise is pointless, actually some people tried 2SK170 and even 2SK147, they generally have lower output and more noise, due to their high capacitance, FET noise is much less important than low capacitance.
One can probably use a non-phantom power standard scheme with more current and other optimisations to lower noise, but the capsules capacitance is one of the main factors limiting noise.
Personally, I'd probably get Jonathan Billington to make me a transformer I can put straight into the source of a suitable cascoded J-Fet, this would minimise both noise and parts count. But such a transformer would probably NOT be cheap...
Ciao T
Hey, I worked with those in the 80's. The actually at the time most precise and lowest drift solution we had at the time in the "100 Tons" class where classic mechnical balances with an attachment that rotated optical code disks. The microgram scales for the lab where all electronic digital of course.
I believe those capsules need a polarisation voltage, I would have to check how much. Difficult to do cleanly from 48V phantom power, well, there are 1MHz switchers these days, I would need to check if they can work in boost configuration...
For a preamp circuit, the common standard these days is a pair of PNP Emitter followers which use the phantom power resistors in the Mike Pre/Desk as loads, suitably biased, with their collectors supplying a zenner diode which in turn supplies a J-Fet that operates as Buffer and Concertina Phase Splitter. I believe this originated with Georg Neumann GMBH.
Scott Helmke's Alice is a good example:
As you can see, shooting for super low noise is pointless, actually some people tried 2SK170 and even 2SK147, they generally have lower output and more noise, due to their high capacitance, FET noise is much less important than low capacitance.
One can probably use a non-phantom power standard scheme with more current and other optimisations to lower noise, but the capsules capacitance is one of the main factors limiting noise.
Personally, I'd probably get Jonathan Billington to make me a transformer I can put straight into the source of a suitable cascoded J-Fet, this would minimise both noise and parts count. But such a transformer would probably NOT be cheap...
Ciao T
Jorg Wuttke of Schoeps. An all FET version with SJ74's works too, gets rid of numerous problems and allows very nice film caps for the coupling. I think these capsules are designed for optimum tension at 60V. You can generate 60v from 3V without an oscillator (buy it off the shelf too). I don't want to give away too much now.
Scott,
First, the noise of the mike capsules falls with frequency, actually, to be precise, the noise of the load resistor gets lower as frequency rises because the capsules capacitance creates a lower impedance shunt across the resistor...
Second, hence, while the smaller size (and capacitance) capsule will have more noise at low frequencies, I suspect by 500 Hz or so things should be low enough to not loose a lot of sleep...
What I would worry about is if I can get a seamless enough crossover...
Heck, we probably could make a 2" capsule then...
Ciao T
First, the noise of the mike capsules falls with frequency, actually, to be precise, the noise of the load resistor gets lower as frequency rises because the capsules capacitance creates a lower impedance shunt across the resistor...
Second, hence, while the smaller size (and capacitance) capsule will have more noise at low frequencies, I suspect by 500 Hz or so things should be low enough to not loose a lot of sleep...
What I would worry about is if I can get a seamless enough crossover...
Heck, we probably could make a 2" capsule then...
Ciao T
Interchannel time delay is a more general phrase that covers ITD as the specific phrase for the ear as channel .
Frindle reported ~11µs and Holman ~10µs in an article.
As the topic is closely related to AB microphonic setups, there exists a plethora of research papers with (as usual
) diverging results.I havn´t seen so far a research paper in which delays as small as 10µs were used, but given the graphs it would map for an mimimum audible angle of 1 degree to a ITD of ~12µs, see:
http://www.uni-koeln.de/phil-fak/muwi/ag/tec/lok.pdf
Unfortunately written in german.
The Author researched the dependance of localization effects from loudspeaker quality.
Of course, in case folks didn't notice this is simply a rephrasing of the heisenberg uncertainty principle. Audio is tinkering with the laws of physics again. There is a Heisenberg limit for A/D's and D/A's and we are not even close yet.
Last edited:
Hi,
Really? I used 2SJ109. Works well.
Well, 60V is not too bad. I remember some that where quite a bit higher.
Ciao T
Really? I used 2SJ109. Works well.
Well, 60V is not too bad. I remember some that where quite a bit higher.
Ciao T
Definitely, the best reasoning would be the captured transient, rather than walking and laughing. Everyone is able to walk and laugh
Or just let's show something like this:
Pavel please explain this I can't read the caption. If these are 1 LSB steps on a 12bit converter, what's the problem? My take on this is that some take this as 4096 LSB steps on a 24bit converter or something like that. It is trivial to make a 12bit staircase +- full scale in a 24bit file and play it.
I don't see the relevance to our discussion - these are pipelined flash converters - not Delta-Sigma
looking at datasheet DNL, input noise spec I would conclude the hardware, test setup has a noise problem - the AD922x have ridiculous high input BW - can be used for direct IF down conversion by exploiting aliasing - DS example of down conversion from 5th Nyquist zone - so these ADC will respond to much higher frequency noise
looking at datasheet DNL, input noise spec I would conclude the hardware, test setup has a noise problem - the AD922x have ridiculous high input BW - can be used for direct IF down conversion by exploiting aliasing - DS example of down conversion from 5th Nyquist zone - so these ADC will respond to much higher frequency noise
Last edited:
First of all, a self-professed expert such as yourself must know that balanced has a +6 dB advantage over unbalanced. Factoring in the 3 dB noise only occurs when you combine the balanced signals into an unbalanced signal, but with Papa's SuSy, you can avoid combining the signals until the speaker. In any event, the worst case for balanced is a +3 dB signal to noise advantage. I have no idea why you would ignore the +6 dB advantage unless you're just trolling.
Second of all, the PCM1704 already has this 3 dB noise penalty by nature of using two DACs internally. All I ask is that they take each output to a separate pin rather than combine them internally. There's no way according to Kirchoff's laws that they can avoid the 3 dB noise penalty simply by combining internally.
Layout will never match two separate DACs as closely as the two DACs within one chip are matched.
Thanks, but I don't see anything in the data sheets which says there are two DACs. You do realize that it is possible to design a single DAC with differential outputs, don't you? Again, you claim to be an expert on weighted ladder DAC technologies ... are you familiar with the weighted current source ladder technique of DAC?
Why can't you just ignore one of the differential outputs? Other DACs that I've used tell you how to connect an unused differential output so that it does not destroy the accuracy of the remaining output. You'll be giving up the +6 dB advantage of balanced, but you don't believe in that anyway, so go for it!
Hey, we brought Thorsten to tears.
Venio,Vidio,Fleo would also be correct if I remember my Latin sounds a little better.
Last edited:
- Status
- Not open for further replies.