From: Wayne.Beckman@l-3com.com Sent: Monday, September 14, 2009 2:56 PM To: Massion, Gene Subject: RE: A favor Attachments: dB Experiments.xmcd Gene: If the ADC has 1 V peak range and can go both polarities, it will have a 2Vpp range which is – 3 dBVrms full scale. With -160 hydrophone that means you could handle signals up to 157 dB re uPa rms. That seems pretty high. If ADC noise is per Hz then I would add about 30 dB of gain to get SS3 noise 46 dB) equal to ADC noise, that would still allow signals up to 127 dB which is pretty loud. I guess it is hard to guess the acoustic level of these mystery events. They must be releasing a lot of energy, but it is not obvious how much would be in 10 Hz to 100 kHz band. It might be better to go to 40 dB of gain and the worst that would happen is that the system would saturate during your event. It would give the duration and if the event is much smaller at least you would see it. I attached a Mathcad 14 sheet that shows how you can use dB. Wayne From: Massion, Gene [mailto:magene@mbari.org] Sent: Monday, September 14, 2009 9:56 AM To: Beckman, Wayne @ Maripro Subject: RE: A favor Wayne Hope your deployment went well. Thanks for taking a look at this. I don’t actually know how the original noise plot I sent you was calculated but I was assuming the units were RMS. Here is a more recent one that I took that I know is a spectral density with units of dB(power) re 1V / Hz. It was done in Matlab using Welch’s method. The data sheet isn’t explicit about the max signal level. I have a question into TI trying to get more detail and I’ll be testing to see where the ADC input clips but for now I’m assuming the ADC has a 1 Vpeak max signal. I think that means the voltage reference for the ADC is a 1 VDC reference. As you suggest, I’ve been leaning toward sticking with the 10 dB of gain the ITC8201 has in the preamp and not adding any more at least until we get some real data back. Thanks again for your help. Gene Massion Monterey Bay Aquarium Research Institute 7700 Sandholdt Road Moss Landing, California 95039 831 7751922 (tel) 831 7751646 (fax) -------------------------------------------------------------------------------- From: Wayne.Beckman@l-3com.com [mailto:Wayne.Beckman@l-3com.com] Sent: Saturday, September 12, 2009 5:58 PM To: Massion, Gene Subject: RE: A favor Gene: I have been at sea for the last couple of weeks, so I just saw this today. Is the -85 dBV noise out of the ADC an rms or per root Hz value? Is the 1V ADC range Peak or rms? One thing you have to worry about is crest factor for noise. A 1 Vrms noise signal will have peaks 2.6 times that 1% of the time. Without lots of analysis, I would think you would be better with low gain and not worry about SS0 kind of signals at high frequency, I would think what you are looking for is much above ambient and SS0 is not really a common condition in Monterey anyway. I’ll take a closer look at this on Monday. Wayne From: Massion, Gene [mailto:magene@mbari.org] Sent: Tuesday, September 01, 2009 10:29 AM To: Beckman, Wayne @ Maripro Subject: A favor Wayne I hope things are going well down there. At least according to Pete Yinger, you guys are busy. I have a small acoustic project that I'm tearing out what little hair I have left trying to remember how to do things I used to be able to do. I'm hoping you can help by answering a few questions that I think should be pretty easy. Here is some background you may find interesting. I've been working with one of the MBARI scientists, Charlie Paull, on his project to understand sediment transport in the Monterey Canyon. The old theory was that these large submarine canyons were generally formed primarily by major cataclysmic events over geologically long time spans. We've put a variety of instrumentation in the canyon to see if we could capture any kind of event. Turns out that some kind of significant sediment transport event is happening much more frequently than anyone suspected. We know because every time we put something in the Canyon, it gets moved and buried. We started off with 2 nodes with some basic instrumentation, current meters and so on, on 2500 lbf stands with a low CG. We lost both within a couple of weeks. We never found one and the other we were lucky to find buried under almost 2m of sediment several 100s of meters away from where we put it. That was interesting. The next step was to put some "disposable" stands with an even lower CG and no instrumentation except an acoustic pinger. We went out on a regular basis and surveyed in these stands (like we used to survey in hydrophones) from our vessel and were able to track their movement over several months. Eventually most were lost. But they all moved lending more weight to the theory that significant events are happening quite often. Unfortunately, this approach doesn’t tell us anything about the nature of these events. What we are trying now is to put an acoustic receiver/data logger on a ledge on the side of the Canyon (where it should be safe) and an "event detector" stand with a pop up float that releases at a current greater than 5 knots and phones home with the time and date. We've built and deployed a couple these event detectors and they seem to work. The acoustic receiver will log continuously and the event detector will give us a time stamp to tell us where to look in the record for a sediment transport event. The theory is if we know about where to look in the acoustic record we will be able to see a significant change in the ambient spectrum and be able to at least understand the duration of the these events and answer questions like, is it one big event, many smaller ones, is there something different in the ambient spectrum before the event? If these events have some characteristic signature, like a high level, low frequency rumble, that stands out from the ambient, we may be able to actually detect events acoustically without the mechanical event detector. My questions are about the acoustic receiver. I'm hoping you can double check some of my assumptions and calculations. We have a couple of ITC8201 pre-amplified hydrophones (data sheet attached). They have roughly a -160dBv/uPa receive response over our assumed band of interest (10 Hz to 20 kHz). I'm trying to figure out how much gain I need to get the hydrophone signal into the range of our ADC, The 16 bit ADC max signal is 1 V and we get a pretty flat noise floor roughly -85 dB down (conveniently -85 dBV). Seems like I should put in enough gain so the Wenz sea state 0 generates about a -85 dBV signal out of the hydrophone and it's gain stages. I remember the Wenz curve as being confusing with respect to units and nothing has changed. For the sake of simplicity I'll make this question easy. Assume I'm only interested in a 1 kHz band centered at 10 kHz. If I interpret the Wenz curve correctly (not the Knudsen curve on the 8201 data sheet), the ambient noise spectrum level is about 28 dB re 1 uPa/Hz. Q1) If you look at a Wenz curve, do you see ~28 dB at 10 kHz and sea state 0? Shouldn't the units be dB^2/Hz? Q2) Assuming the units are dB^2/Hz and for the sake of simplicity call it flat in a 1 kHz band around 10 kHz, is the following calculation correct (sorry I don’t know how to do dB in MathCad)? Q3) For this over simplified example it looks like I need about 20 dB of gain to get the ambient noise about the same level as the noise floor of the ADC? Q4) Given that the real Wenz curve shows 10 Hz, light shipping at about 63 dB, I would only get about 40 dB of dynamic range at the low end with 20 dB of gain? Q5) In our situation, we expect the sediment transport events to be “loud”. I have no idea how loud but visualize many metric tons of sediment moving in one event. We are trying to do this on a minimal budget and I’m wondering if I can just hang an instrumentation amplifier on the already preamplified ITC 8201 with no “pre-whitening”. Alternatively, if the calculations above are correct maybe I don’t need any gain except what is already in the 8201 and give up some sensitivity in the high frequencies? Know that I’ve written this thing, I realize I may be asking for too much free advice. Feel free to let me know if that is the case. Cheers and thanks for any help you care to provide Gene