Multiple Analog Channel Recording

This is much bigger than a tinkering job. Then again it's not my job, money or time.

First thing you must do is identify how fast each channel must be measured and how much resolution you need in each of these channels. By Nyquist theory, to resolve a 5kHz signal you must sample at 10kHz or faster. If eight bit resolution is sufficient for all 480 channels then you will be generating at least 4.8 megabytes of raw data every second. In about three and a half minutes you will have a gigabyte of raw data. That's without including all of the obligatory ancillary data of time stamps, channel numbers, communication routing bits and error correction.

Now there are tricks and techniques that one can use to compress this data size in a lossless manner. However, one must know much more about what is being measured to make any recommendations or suggestions.

I would expect that to achieve what you've asked for will take a distributed intelligence in more than one meaning of that phrase. Multiple data acquisition systems will aquire the data into a networked collection of data steering systems that will then place your data into data storage.

To get a better idea of what this will take, I recommend you review how EPICS handles this.

I have a 2F_s input and output system now, but that is what uses the isolation transformers. I have the ability to stream the data in/out from this configuration, but no way to record that data. That is why I need an A/D solution. The system I have now also takes up 4 full size racks and is cabling intensive, and requires me to move 16 channel cables at a time in order to select different analog inputs. Those inputs go into a switchbox where I can select 1-16 of the incoming channels. My spec-an and oscope load the channels to the point that when I have all 16 selected I get about 5 dB of loss across the spectrum. If I could record the data for later analysis on an as-needed basis it would make it much easier to pick and choose the stream I want and reduce the need to monitor them live in large groups.

I know very little about electronics, but I've got some EE buddies who do a lot of tinkering with 'Basic Stamp' electronics. Maybe this would be a way to get started.

http://www.parallax.com/Store/Microcontrollers/BASICStampProgrammingKits/tabid/136/ProductID/313/List/0/Default.aspx?SortField=ProductName,ProductName

I'll check into that link. I first went to the arduino boards since that is what came up in Google first. There is a 5 channel acquisition board there, so I would have to cobble up my own system with many arduino modules.

Data acquisition is what it sounds like you want to do.

You haven't mentioned what type of analog data you are logging.....not sure where your audio patch panel comes into this.

Anywho, this one can be configured to do up to 500 channels.

What search criteria did you use for your unsuccessful Googling?

Lots of solutions out there if you can work out what exactly you want to do.

The analog data is raw audio up to 5kHz.

I have googled analog recorder, analog to digital, arduino analog to digital, analog to digital capture, multi channel audio interface, multi channel sound card, record multiple audio channels, single-ended acoustic analog capture, and single-ended analog capture...so far.

I have followed numerous links off of the pages that came up...but they all seem to end in either very complex programming or very expensive design packages. I was really hoping that there was something similar that may already exist.

I am currently using a Graphtek DMS1000 which can capture up to 40 analog inputs...but only has a hardcopy for storage. I would really like to get my data into a PC with huge storage capacity which I would also be able to use existing software to analyze or replay the data for external analysis.

You're needing and interface also thought of as an external sound card. This can be very expensive or not. Presonus units ganged together could achieve your goal but there are other products too. http://www.sweetwater.com/c683--FireWire_Audio_Interfaces

Yikes...that is a little bit over the top for what I want to do. I don't want to modify the signal in any way. I am analyzing the pure waveform.

It seems almost impossible to find a product at a reasonable price. That is why I was looking at the Arduino type products, but that seems to require a lot of custom programming. I thought this type of thing would have been done already somewhere.

I am still looking.

You really should not be surprised. The nominal size of a large symphony orchestra is 100 musicians. You wish to record nearly five times this number of individual musicians with one channel per musician. There is no market for what you want to do. This means a custom design by you or a hired design team. The one realm that does what you want regularly are large scale scientific experiments. This is why I earlier offered the open source scientific data acquisition programming of EPICS. (I'm not going to make a new link.) You still will have to do a lot of your own coding to utilize this very expandable data acquisition software. You will also have to purchase many A/D hardware to access your very large array of analog signals.

Your dream configuration of 480 channels simultaneously acquired at 10 kHz is done. The ATLAS detector at the Large Hadron Collider gathers much more data than you wish to do. AFAIK this experiment does use EPICS to acquire data.

I think that you need a reality check here.

I hit this fresh again today...regardless of the supportive audience, and discovered Ethersound and Cobranet. This appears to be a promising direction for my venture...all done with simple PC interfaces and sound cards. It's still expensive, but I am whittling it down (below $10,000).

It's amazing what one can come up with a fresh perspective and an open mind.

You are gonna need a lot of PCs to get your 480 channel count.

How many channels can each of these cards handle?

64 differential or 128 single ended for one manfacturer...luckily I have single ended outputs.

There is one card which can address up to 512 single ended inputs.

I'm getting there.

Be careful here. You may find that this board can only sample a small fraction of the 512 channels simultaneously. Thus to sample all channels you will have to access different channels with a slight time delay skew as you scan across all of the channels you wish acquire. This may not be a problem in your application but you should be aware of this effect.

Could you flick some URLs for this onto the table please?

Redfred's observation re "sampling/polling rate" is a valid consideration.

modifying the wave form isn't a default or necessary part of using an interface but it does effectively digitize the wave form and gets it into a computer where using audio software you can analyse thousands of pure wave form samples

google signal processing or DSP