It's always exciting to read how a component manages to break in its owner over the course of a few 100 hours
Jan
I have no idea what you mean here.
Thermals for solid state ic and board levels design? There's always more to learn, but it's not exactly a immature field.
It's always exciting to read how a component manages to break in its owner over the course of a few 100 hours
Hee! So instead of spending $10,000 or whatever on a DAC, maybe he could save some money and just get a cat.
We always can learn from tube aficionados.
I’ll start working again on my omnis, so I had a tour on Stewart Hegeman’s designs.
I found this interview, IMO it’s relevant to the interests of the members
retro vintage modern hi-fi: Stewart Hegeman interviewed by Norman Eisenburg 1961
George
Maybe you didn't read his reasoning, efficient things are "better" intuitively for no other reason than that they are efficient. Then go on to infer this is a scientific fact. Take this quote and then compute the wall plug efficiency of some of those flea power SET amps. Or just take the quote as a statement of "scientific" principle.
Your other comments were not necessary we (me too) moved into optics a decade ago. I would love to see this forum go no monikers.
Last edited:
I did my homework and I have to admit I was wrong. While I’m still convinced that pre and post truncation dithering would be indistinguishable in a DBT test, the correct way to dither is before truncation. That is, dither the low 8 bit of the 24 bit original with noise having a TPDF probability distribution. Other distribution (including flat) can be considered, but TPDF is theoretically optimum (in the distortion reduction sense).
Now, while dithering the 16bit truncated result LSB with either the XOR of the 8bits of the pre dithering signal, which is indeed as stochastic as you can get, or even simpler with a binomial distribution (aka “toss a coin” signal) will indeed provide distortion reduction, this is not optimal. It is equivalent with the pre dithering with a flat distribution noise, which can easily proven as suboptimal.
Unfortunately post dithering the 16 bit LSB doesn’t offer any other options; we could of course consider dithering the last 2.5 LSB and then considering TPDF is possible; but then this will add to the 16bit noise floor, which is once again suboptimal compared to the pre truncation dithering.
So the optimal way is dithering before truncation. Once again, I’ll place my bets on a NULL result in a sorting DBT, though, so this is an academic issue only.
Last edited:
Waly: Appreciate follow up on the dither question. May I ask, do you mean it is theoretically an academic issue, or proven to be an academic issue only?
I have no interest in betting on anything, just not much of a gambler. However, if you would like to put up some 16-bit files made from a high quality hi-res source you think would be a fair test for people to try to distinguish, I would be interested to give it a try. I'm sure other people would too.
Just to be clear, I'm not claiming to know in advance which way the outcome might go, just think it would be interesting to run the experiment and see.
Edit: To be sure we are on the same page with respect to a test, the idea would be to see if anyone can hear some remnants from, say bits 17 and 18 buried in dither noise in a 16-bit file, or if nobody can hear anything down at such a low level. Would that be a fair statement of what we would be trying to find out?
Last edited:
It might be difficult or impossible, certainly. Never tried that one is all. Not too hard to hear a little difference between a hi-res 24-bit digital recording and a 16-bit file made therefrom, but that isn't the same as what we are taking about in this case. 24-bit verses 16-bit might be easier to hear as compared this dither before/after 16-bit only test. In this latter test, the higher resolution information in the dither-before file is buried in noise, so no surprise if were harder to discern.
Last edited:
Yes. Positive.
My main experience comparing LF356 to LF357 as line stage in phono preamp.
I used LF356 for phono stage (per compensation needed) and LF357 for line
stage (per square wave response)......My main customer (whose ears I have
come to trust) complained " hard edgy " ( or something oh so close )
On listening myself I understood his observation .
Changing to 356 for line also made him happy .
My main experience comparing LF356 to LF357 as line stage in phono preamp.
I used LF356 for phono stage (per compensation needed) and LF357 for line
stage (per square wave response)......My main customer (whose ears I have
come to trust) complained " hard edgy " ( or something oh so close )
On listening myself I understood his observation .
Changing to 356 for line also made him happy .
The comment after Mr. Austin's nails it. I wore out my welcome with Mr. Hansen a long time ago.
What i meant is most if not all EE's never worry about the temperature of an IC as long as it doesn't reach the maximum. Resistivity and conductivity being temperature dependent will hence be different between low and high amplitude / energy signals. I know I can make an IC super efficient by simply talking a can of air, turning it upside down and spraying with the freezing air that comes out. So I propose, as I have in the past, why not attach a TEC (thermal electric cooler) and set to some dew safe temp like 15 C and see if you do not get more linear operation. This IS high end, no?
- Status
- Not open for further replies.