Signal Processing Via USB Port

Not when the netbook is probably 1000x the power consumption and 1000x the size of a circuit that will do a better job, and (depending upon application) maybe 100x more expensive.

I can't even think why you would have the idea that using a large, powerhungry and slow general-purpose digital computer would be *less* wasteful than an efficient well-designed circuit.

That's if the netbook can do it at all: many signal processing operations must be done much faster or with much lower latency than you can possibly get with USB and a netbook CPU.

- Tim

Is a circuit faster or cheaper than that entire assembly?

A netbook is a circuit. That's why the people with brains invented Post Disney Fiber Optics, External non Rotating Mini Computer Harddisks, Laser Disk Libraries, Blue Ray, All-In-One Printers, HDTV, Home Broadband, GPS, Digital Terrain Mapping, Data Fusion, Weather Satellites, Pv Cell Energy Arrays, Vacuum Tubeless Microcomputers, mp3, mpeg, Cyber Batteries, On-Line Publishing, Flat Screen Software Debuggers, non C++ Pointers, PGP, Holograms, UAVs, AAVs, Distibuted Proceesing Software, Biodiesel, neo Wind Energy, Post Wattian Gas Turbine Engines, Microwve Cooliing, Thermo-Electric Cooling, Self-Replicating Machines. and Self-Assembling Robots.

Rather than Quantum Machanics idiots or GM.

Like most everything else in life, the answer is "it depends". If you just need AC-coupled signals in the audio range, and you don't need small or low-power, then a USB sound card, notebook/netbook, and software may be just the ticket.

If you need something beyond this, consider that unless you already have software that works with your chosen A/D to do what you want, developing the software might be far more involved than building an analog circuit.

Best regards,

Bob Masta DAQARTA v4.51 Data AcQuisition And Real-Time Analysis

Scope, Spectrum, Spectrogram, Sound Level Meter FREE Signal Generator Science with your sound card!

What OS does the netbook run? How many seconds at a time does it randomly go away for?

If the USB ADC is the input, where's the output?

John

So why is Cahill here in a physics group to demonstrate his ignorance of electronics?

Bob, quit trying to pound some sense into these idiots. They are having a wonderful discussion based upon zero real information. They are all speculating wildly with absolutely nothing to base any of it on. They are probably cosmologists in their day jobs...

Has that been a problem recently? Maybe cooling is a bigger concern.

I'm only interested in < 1 Hz signals anyway.

The screen.

One problem is having a consumer item in the work place. Maybe a fake decal to make the netbook look like an industrial part.

But I'm still open minded to analog circuits. Are convolutions, i. e., multiplying signals in the frequency domain, possible with op amps?

Bret Cahill

What is your application? How many units do you need. What is the frequency range that you want to operate in? Do you need the system to give near real-time answers, or can you do post-processing? Without knowing what your application is, we can only guess. You can do signal processing using analog components, but again, that depends on your application. I would go analog when the frequency band extended into the RF range, otherwise I'd stick to DSP processors.

A more appropriate group might be comp.dsp for your question.

Kind Regards, Jaco

Dividing two signals. Without noise both signals should have the same shape with a 0 degree phase angle. Guessing wildly the noise in that frequency range might be 10 - 30% of the signal. I'll have a better idea in a couple weeks.

Maybe a dozen initially. I first need to know that I can eliminate most of the noise in several cycles. That was my interest in matched filtering.

A similar application might eventually be in the hundreds of thousands.

Less than 1 Hz.

Output should be every 5 - 6 cycles initially.

The larger market would allow a minute or so. Even a lock in would have enough time.

Thanks.

Bret Cahill

He does it as widely as possible.

John

Well, if all you want to do is crunch ADC data and display it, there are tons of signal-processing programs around. Mathematica, Matlab, LabView, stuff like that. And lots of specialized virtual oscilloscope and spectrum analyzer/FFT programs for working with data from sound cards and USB/PCI a/d converters. All that's old stuff by now.

There's even some public-domain vector impedance analyzer software that uses a sound card.

Nobody I know has a problem with using a laptop computer or a netbook in an engineering environment. Quite the contrary; we use tons of them.

Opamps don't multiply. Analog multipliers exist but aren't practical for array stuff like FFTs and convolutions. What is common is to digitize a signal and dump it into an FPGA to do FFTs, delays, convolutions, filters, whatever, in the digital domain. A $50 Xilinx FPGA can out-compute the fastest Pentium by a factor in the hundreds, because it is capable of brutal parallelism. The FPGA runs continuously, so can be used for true streaming, realtime processing, unlike the PC that halts and stutters.

Analog circuits can do some specialized filtering and equalization functions that are equivalent to some convolutions. When they can, they are zillions of times faster than any digital method.

John

I kinda liked this one

I suggest you put it on 7.865Hz Benj, will make you feel better :o)

______

I think he might be looking for this.

He does seem to want to demonstrate his ignorance to a wide audience.

>

I've been out at sea so long dozens of movies have been made about it.

OK, now were talking.

Kind of a sheltered environment, don't you think?

OK now were talking.

A few seconds and everything is A OK.

Bret Cahill

d h .

ke

mmm

Since he is talking about really low frequencies, there is another multiplication method that may be feasible, from back in the days when multiplying was tough and lots of clever methods to do it were bandied about. The idea is for one input to generate a stream of pulses whose duty cycle encodes the input voltage (offset to eliminate negative values). This PWM signal is then used to chop the signal to be multiplied, and the output is filtered to remove the PWM frequency.

Since the input frequencies are low, the PWM frequency could be high enough to be easily filtered without compromising the product or response time.

Best regards,

Bob Masta DAQARTA v4.51 Data AcQuisition And Real-Time Analysis

Scope, Spectrum, Spectrogram, Sound Level Meter FREE Signal Generator Science with your sound card!

The "carrier" frequency remains the same but the duration of each pulse increases linearly with the original signal's voltage.

Any ideas on how to use this for division? The reciprocal of the PWM signal needs to go to infinity when the original signal goes to zero.

Thanks again.

Bret Cahill