John Curl's Blowtorch preamplifier part II

Status
Not open for further replies.
Jump to Latest
#62,144
I am listening to the PONO with other headphones...... great bass now. I'll have to check the Z of those buds. Well, the 24/96 downloads sound great. Listening to Chesky's Ultimate headphone demonstration disc right now while typing. Listening on HiFiMan HD500's. The volume is adequate but only at 75-80% of max. I'll try it with my headphone amp, later. But it isnt sounding distorted at all at even full max. Nice for a portable and the thing is large .. with a much higher battery capacity than iPODs etc.

Very cool. All those 32 and 64 bit consoles and mixing/processing hardware/software mixed down to 24 bits is working just fine.

I think I'll keep it. 🙂 Going to try to find some 24/96 binaural recordings, now.

BTW -- yes, How are the JC2-copy tests and mods coming along? Can you easily make it better?


THx-RNMarsh
 
Last edited:
#62,146
Of the two outputs, just the one with the headphone logo.... common ground wire, I presume. Will try a higher Z headphone also. A Sennheiser HD800 which is closer to 300 Ohms.
Then see what the other output is good for..... and if I have a bal adapter. Personally, in my system(s) I havent found balanced needed nor does anything for me. Always have short cables and good grounding practices. But.....

-RNM
 
Last edited:
#62,148
And, john has the schematic... so, John, are they balanced in some good way?

But, for a quickie listen, as-is... it is quit good sound. Still, as the listeners on HeadFi also attest.... its a lot better than 16bit/44.1.
I'll eventually, plug it into my big rig and listen, also. Compare it to other DAC's. That lack of pre/post processing for spinning discs also makes a difference.

But, that is going too far for this forum.


THx-RNMarsh
 
Last edited:
#62,149
RNMarsh said:
Of the two outputs, just the one with the headphone logo.... common ground wire, I presume. Will try a higher Z headphone also. A Sennheiser HD800 which is closer to 300 Ohms.
Then see what the other output is good for..... and if I have a bal adapter. Personally, in my system(s) I havent found balanced needed nor does anything for me. Always have short cables and good grounding practices. But.....
Click to expand...

The outputs are bridged in balanced mode (provided you're using the cable adapter) so you can double the voltage swing into headphones. The Sabre DAC has differential current outputs, but when you're using it single ended, it's only one side that drives the output.

se
 
#62,151
As far as the JC-2 clone is concerned, we have made progress. Sometimes it can be made to oscillate, I am not quite sure why. It has some added dominant pole compensation 2 X 10pf, which usually works well enough, however an LT Spice simulation implies potential trouble between 1MHz and 10MHz.
The input fets appear to be fakes, just like several people have previously pointed out. They have too little Gm and too much Idss to match their part numbers. The outputs are wired too close together to completely avoid Xtalk.
It is possible to improve this product, but not without added expense. For a hobbyist, it is probably worth the added time and expense. The essential part of the circuit is uncompromised and even improved slightly from the original Levinson JC-2.
 
#62,152
john curl said:
As far as the JC-2 clone is concerned, we have made progress. Sometimes it can be made to oscillate, I am not quite sure why. It has some added dominant pole compensation 2 X 10pf, which usually works well enough, however an LT Spice simulation implies potential trouble between 1MHz and 10MHz.
Click to expand...

Hi John,

Be careful there - LTspice doesn't know about parasitic C, R, L in the physical circuit, so that will not be accounted for in the sim.
That is why perfectly stable sim circuits do sometimes insist on oscillating in a physical implementation. And vice versa.

Jan
 
Last edited:
#62,154
Jakob2 said:
If you try to do such tests with different listeners you will most probable realize that most of them have difficulities in a controlled test to detect even quite large differences.
It is a difficult task because listeners have to evaluate (and memorize) complex sound events and have to do so reliable/consistent in as many trials that are needed.
Click to expand...

! yet can detect these changes when outside of a test...even when changing components!
 
#62,155
gizmodo-wont-post-my-comment-so-im-posting-them-here
Michael Fremer on Thu, 2015-01-29 06:19
Look clearly you haven't followed this thread for long. So I'll repeat what most here already know: I was challenged some years ago to a double blind test by someone who claimed that all amplifiers that measure the same sound the same. My experience says otherwise so I took the challenge. It was done at an AES meeting. There were 5 amplifiers under blind conditions AND I GOT EVERY IDENTIFICATION CORRECT OKAY? Does that mean anything? Apparently not. Since the population of AES engineers produced a statistically insignificant result (they couldn't hear any differences) guess what? I WAS DECLUARED A "LUCKY COIN" AND MY RESULT WAS TOSSED.
Now it's your turn to tell me there weren't enough samples (etc.) BUT I DIDN'T PRODUCE THE F...KING TEST AND YOU CAN BET HAD I FAILED LIKE THE OTHERS IT WOULD HAVE BEEN USED AGAINST ME.

I know how these ax-grinders work.

But get this: one of the amps was a VTL 300 vacuum tube amp. One was a Crown DC-300 solid state amp known as a genuine "EAR BLEEDER". AND THE AES CLOWNS couldn't hear the difference between those. EVEN YOU COULD...but maybe not under double blind conditions. These tests are tests of the listener's double blind taking abilities as much if not more than anything. They are ABUSES not uses of science and that result is good demonstration of that.

Why did the test producer use two amplifiers that both sound and measure differently? I don't know but it says to me: get STUPID RESULTS, perhaps the test is STUPID.

I have done double blind speaker testing at Harman and done very well.

I DON'T HAVE TO PROVE MY LISTENING ABILITIES TO ANYONE including you.
Click to expand...

Dan.
 
#62,156
marce said:
! yet can detect these changes when outside of a test...even when changing components!
Click to expand...

Yes but you have much more help in that situation! Your whole perceptive, sensory and biasing system is available to produce a position for you.

In the controlled test you're thrown back on your ears only, having only the air vibrations to compare. And the differences in air vibrations are not enough to 'hear' a difference.

Jan
 
#62,157
jan.didden said:
On another note - has anybody here looked at Meridian's new MQA system?

Their Master Quality Authenticated codec offers 192kHz/24 bit uncompressed audio in about the same space as a 44.1kHz/16 bit WAV file.

Is it genuine? Is it worthwhile?

Jan
Click to expand...

I would be skeptical of claims that possibly violate first principles, but isn't this only 2X or so over FLAC? I don't save enough music to bother with compression. Anyone honest would offer a simple downloadable example.

EDIT - I take that back it's not clear what the claims are. Looks like some kind of proprietary encode/decode.

MQA music can be downloaded or streamed in any lossless format. It’s also 100% backwards-compatible with existing players, revealing full quality with a hardware or software MQA decoder.
Click to expand...
 
Last edited:
#62,158
scott wurcer said:
The maximum achievable dynamic range for a digital audio system with uniform quantization is calculated as the ratio of the largest sine-wave rms to rms noise.

Dick - Sorry to say the rms noise in a 20-20kHz BW is an AVERAGE measurement. In any case the DNR of a 24 bit number dithered at 1 LSB TPDF is 141dB this is mathematically provable. All deviations from this are due to noise and non-ideality in the converter/amplifier chain.
Click to expand...

Doing the Dynamic Range calculation as the ratio of Full Scale Output (V) to LSB (V) expressed in dB, I get:
Bit Rate / DR_undithered / DR_dithered (+/- 0.5LSB)
8 / 48.1 / 45.1
14 / 84.3 / 81.3
16 / 96.3 / 93.3
20 / 120.4 / 117.4
24 / 144.5 / 141.5
32 / 192.7 / 189.7
64 / 385.3 / 382.3

Are there any measured (DAC out) DR figures vs bit rate to compare ?



scott wurcer said:
DNR again. These plots are for -80db full scale at 24bits and 32bit floating point. Remember this is just the DNR possibility of a particular numeric representation, a purely math problem. The second undithered, yes the numerical noise floor of 64bit double precision math is -300dB or so. The first is dithered at 1 LSB with a triangular probability distribution, what you actually can save as a 24bit file. I still see nothing that indicates more bits has any benefit at all.
Click to expand...

Scott,
What does green and what does blue represent?
What are the spikes on the 2nd screenshot?

George
 
Last edited:
#62,160
gpapag said:
Doing the Dynamic Range calculation as the ratio of Full Scale Output (V) to LSB (V) expressed in dB, I get:
Bit Rate / DR_undithered / DR_dithered (+/- 0.5LSB)



Scott,
What does green and what does blue represent?
What are the spikes on the 2nd screenshot?

George
Click to expand...

George that was a numerical simulation unfettered by the usual constraints of having to use power of 2 only FFT's and then requiring windowing and its own problems. Those plots were of a 1kHz tone at 96K and a 96k FFT. The one with slightly lower (blue?) in floor is a 32bit float the other is 24bit integer. If you think about it a sine wave is periodic and any code errors repeat at the same period. In that sense the distortion can only be at harmonics (the spikes). The noise floor is purely accumulated round off error in the FFT. As an aside I got a footnote somewhere from an uber geek numerical programmer for pointing out the accumulation of noise in FFT's (in 1998). The bins are exactly 1Hz so any integer frequency gives the same result. I added 1 LSB TPDF dither to show that the harmonics for both disappear into the noise floor which has been raised to ~-100dB @ -80dB FS.

If folks don't want to learn the theory behind dither, I can't help them. Dither randomizes the harmonic energy and they are gone on any time scale. If you are so inclined take that horrible looking -80dB sine wave (only a few codes) and add the dither. Then subtract a perfect -80dB sine from the result. It will then be indistinguishable from noise at any time scale. There is no "hiding" the distortion or forcing the brain to average it out.
 
Last edited:
Status
Not open for further replies.
Top Bottom