Concede. At times I kind of thought of it as 'disassociation"! 😀 😉
I'm pretty much a neophyte compared to most of the people posting here, but every once in a while they touch on something I have some understanding of. I'm trying to glean tidbits from the various discussions/arguments.
I'm pretty much a neophyte compared to most of the people posting here, but every once in a while they touch on something I have some understanding of. I'm trying to glean tidbits from the various discussions/arguments.
I would have to agree. To be honest, Scott, Nelson and Yourself have been at this since I was either in diapers, or not even thought of yet. I know that there have been various mentions throughout the thread, but are there any particular articles/books recommended for the theory behind low noise design. As a background I have had my associates in EET for close to 17 years, and I am slowly working on my B.S. so I still have a lot of the junior and senior year courses to take. I do have the advantage of working with people very experienced in semiconductor design, so I know where I can bounce topics and ideas off besides just here at diyaudio. Sometimes it's nice to talk with people face to face.
Peace,
Dave
Peace,
Dave
Try Motchenbacher and Fitchen for a start. Unfortunately good stuff is spread pretty thin. you won't find a noise book with as much meat as, for instance, one of Donald Fink's books on TV.
http://www.amazon.co.uk/Low-noise-Electronic-Design-C-Motchenbacher/dp/0471619507
http://www.amazon.co.uk/Low-noise-Electronic-Design-C-Motchenbacher/dp/0471619507
Thank you Scott. I will have to see if either UVM has this, or if they can get it through inter-library loan. If I find it is in my interest (mental level), going to have to buy it.
Peace,
Dave
Peace,
Dave
Dave, if you want to know everything that you need to know (an approximation) about low noise design, and you have a technical library available, then get this article from 1981.
'The Design of Low-Noise Amplifiers' by Yashay Netzer 'Proceedings of the IEEE' Vol 69 No.6, June 1981
code: 0018-9219/81/0600-0728$00.75 1981 IEEE
'The Design of Low-Noise Amplifiers' by Yashay Netzer 'Proceedings of the IEEE' Vol 69 No.6, June 1981
code: 0018-9219/81/0600-0728$00.75 1981 IEEE
scott wurcer said:
The most recent incarnation of the book is by Motchenbacher and Connelly. Motchenbacher and Fitchen is out of print. Connelly taught the course out of Motchenbacher and Fitchen at GA Tech in the late '70s and early '80s. Looks like he's a co-author now.
Mr. Curl and andy_c, Thank You both for your contributions. Seems I'm going to have some heavy reading.
Peace,
Dave
Peace,
Dave
Hi, JCX,
Preamp output and amp's input is single ended (not balanced). The power amp is inverting input with 2k2 input resistor. Is this making it more or less demanding for shielding?
I'm thinking of using shielded twisted pair. The twisted pair is for signal and ground return.
I'm still not clear about shielding. If it is low impedance for outside/external EMI, then it will absorb more of the analog data in the signal wire, because it is closer to the shield. Like 2 face blade, it will protect from outside EMI, but it will also absorb analog data.
What do you think if the shield is floating at +15V (like patent# 5307416 by Demian Martin (1audio?))
Preamp output and amp's input is single ended (not balanced). The power amp is inverting input with 2k2 input resistor. Is this making it more or less demanding for shielding?
I'm thinking of using shielded twisted pair. The twisted pair is for signal and ground return.
I'm still not clear about shielding. If it is low impedance for outside/external EMI, then it will absorb more of the analog data in the signal wire, because it is closer to the shield. Like 2 face blade, it will protect from outside EMI, but it will also absorb analog data.
What do you think if the shield is floating at +15V (like patent# 5307416 by Demian Martin (1audio?))
syn08 said:
No, the answer is simple. Resolution bandwidth (RBW) is 0.37Hz, that's all. 48kHz sampling and 128K FFT memory.
Theoretical background is:
SNR = 6.02N + 1.76dB + 10log(M/2)
N .. number of bits
M .. number of samples in the FFT record
Try Wurcer/Kester/Bryant's AD seminars.
In theory you can get 194 dB for 24bits and 128K of narrow band noise background. So we are far from 'ideal'. Of course the mains line spectral components are shown in proper values, and it depends on your skills of circuit design, PCB design and instrument interconnection what you get.
PMA said:
I know that much theory 😀. I am not debating the 24bit D/A conversion SNR, but the lack of any mains (and harmonics) radiation in your place and the lack of opamp (I/V or whatever) noise, hence the about 200nV output noise. And the dynamic range (best D/A today are around 117dB, yours seem to be better than 120dB).
In my place, 1 inch of unshielded wire (or signal PCB trace) creates an about 1uV of 60Hz signal.
Oh, now I see. It is an analog generator with switchable output divider, and Dispre 2 rev. V3 PCB module. So there is no D/A.
But again, let's not mix narrow band noise and noise over whole audio band.
http://web.telecom.cz/macura/pcb_v3.jpg
But again, let's not mix narrow band noise and noise over whole audio band.
http://web.telecom.cz/macura/pcb_v3.jpg
The graph shows noise density and not total noise over a certain
bandwidth. You can always double the frequency resolution and the
noise power from one bin will be distributed to two frequency bins.
If you do that often enough, you get impressive numbers on the
amplitude scale - but the total noise from, say 20 to 20000 Hz, is still
the same because you have to add the noise power from twice,
4- 8-... times the number of frequency bins.
Signal components such as clean carriers that are narrow enough to fit
into one frequency bin even after repeated frequency zoom will
stand out clearly.
BTW that's the reason why they modulate the processor clock for
PC CPUs: Smearing the amplitude over some 100 KHz makes it
harder to see for a narrowband spectrum analyzer during
emission test.
Gerhard
bandwidth. You can always double the frequency resolution and the
noise power from one bin will be distributed to two frequency bins.
If you do that often enough, you get impressive numbers on the
amplitude scale - but the total noise from, say 20 to 20000 Hz, is still
the same because you have to add the noise power from twice,
4- 8-... times the number of frequency bins.
Signal components such as clean carriers that are narrow enough to fit
into one frequency bin even after repeated frequency zoom will
stand out clearly.
BTW that's the reason why they modulate the processor clock for
PC CPUs: Smearing the amplitude over some 100 KHz makes it
harder to see for a narrowband spectrum analyzer during
emission test.
Gerhard
Yes. But it makes sense to use narrow RBW as well, e.g. for the reason to find mains level component levels or distortion spectral lines hidden under noise level in case of wide RBW or measurement over whole 20kHz band.
lumanauw said:
The patent is about linearizing the dielectric of the cable. Your concern, I think, is the effect of the capacitive loading of the shield. If you want to be obsessive, drive the shield with a buffer at signal level. It would effectively reduce the capacitance to nearly zero.
The lower impedance of the amp reduces its electrostatic sensitivity, but increases its electromagnetic sensitivity. Keep the wires away from magnetic fields like transformers.
1audio said:
And don't forget to take into account the grounding scheme. You want to avoid ground currents through the screen. So maube you should ground the screen only at one side, but it all depends on the actual layout and grounding scheme.
Jan Didden
- Status
- Not open for further replies.