link .... ( crap sound like the chemist now )
They could have asked , I could have provided my data ...
The multitone really separates the sheep from the goats. I'm using 30 1/3 octave tones at about 12db crest factor. Artifacts show up on even the best boards.
The Japanese have long known that women are better than men at this kind of demanding detail work. I saw a Japanese woman on a production line solder a row of pins on an LSI with a deft swipe of the soldering tip. It was amazing.
Scott, are you doing this at work, or at home with Rightmark or Arta?
I see, right now I was just using a Fostex FR2LE and doing a diff on the results (there are obvious differences). There is a problem in general that one needs a reference A/D that is as they say "blessed". We should set this up, it's amazing how much work it is to put this all together.
At work, raw data and Python. I realize there is a disconnect with the A/D but this is a problem everyone has.
Last edited:
somebody has to if my stocks are going to keep me afloat in retirement - must have some ADI in my S&P 500 Index fund
I was thinking 16 is nice number of parallel chips
another thought is 30 pretty good DACs - each just making 1 frequency of the multitone, sum the analog outs in an inverting amp
could even self cal to a degree - one DAC working at the full operating amplitude for 1 tone, the rest at -1/29 at the same frequency to cancel the fundamental, each with much lower distortion components due to the lower level - do a iterative periodic disturbance cancellation algorithm to calc a feedforward correction table for the 1 full amplitude DAC with low demand on the measuring ADC which only sees the diff error - repeat 30x and you have a "perfect" multitone to diff your measurement DUT output against in analog
I was thinking 16 is nice number of parallel chips
another thought is 30 pretty good DACs - each just making 1 frequency of the multitone, sum the analog outs in an inverting amp
could even self cal to a degree - one DAC working at the full operating amplitude for 1 tone, the rest at -1/29 at the same frequency to cancel the fundamental, each with much lower distortion components due to the lower level - do a iterative periodic disturbance cancellation algorithm to calc a feedforward correction table for the 1 full amplitude DAC with low demand on the measuring ADC which only sees the diff error - repeat 30x and you have a "perfect" multitone to diff your measurement DUT output against in analog
Last edited:
Sorry, took a little to track it down. A fantastic resource for information about sound in the music world is at https://www.speech.kth.se, you could spend days wandering through this lot. Which is where I found my first paper, "Sound levels and long-time spectra recorded within the symphony orchestra and risk criteria for hearing loss" by Jansson, etc - this states 124dB as maximum. Then there was a later work, "Sound Exposure of Symphony Orchestra Musicians" by Schmidt et al, this raises the bar to 132dB for numerous situations.link .... ( crap sound like the chemist now )
Sorry, tied up at the moment, try Google to get the links for these,
.
Last edited:
Actually Pavel was right. As you know I did omit some details and changed others. What is a bit interesting is that his distortion measurements are limited on an unequalized line amp.
When I was trying to explain some of the limits in theoretical circuitry and how to use this to advantage the noise level just got too high.
Not really. A gas discharge tube will stop a surge but it crowbars the line and won't quench on 50 Hz or 60 Hz. That can have pretty serious consequences. Circuit breaker would be the easiest recovery. You can blow fuses or often traces or wires. The UL spec for testing is 240V 500A service. That can make lots of smoke and char very quickly. The usual solution is an MOV. In the EU we use MOV's in series with gas tubes because the relationship of the ground and the power is not fixed in some markets, it floats. The MOV allows the gas tube to quench on normal power.
Also you need to use an X-cap across the line (you probably know this) or fuse the cap so it doesn't crowbar things when it shorts.
I can surge test one of the boxes if you would like. I wonder how a 3KA 8/20 uS surge would alter the sound of a Bybee module.
More details or perhaps a plot? Using that tone set will plant harmonics and IM products on existing tones. The tone sets on the attached spreadsheet work around that issue. The plot below is a quick loopback test of a stock EMU 1616M using a triband test.
Last edited:
If you are interested in the evolution of audio circuits, there is also John Linsley Hood with his two books.
No 'prophet' and no fanatic 'followers'.
George
Oh yes, thank you George, I actually forgot to mention them, it's been some time.
I used to be his neighbor, he lived in Taunton, Somerset, some 12 miles from where my school was. I actually shook hands with him (just a cordial handshake, with me and hundreds of others) on an audio show in Taunton, 1968 or 1969, not sure which. He was demonstrating some of his works, and the audio boom of the 70ies was already starting then, low key, but it was there.
I learnt a lot from that man's works and texts.
- Status
- Not open for further replies.