data recording with analog NTSC

Let's say 16-bit resolution, and a sampling rate of 100kHz. You could double both of those and it still wouldn't change the conclusion.

That's a storage requirement of 720 Mbytes/hour.

A 160 GByte hard drive is $80 and will store more than 222 hours of data.

Storing 222 hours of data on 6-hour videotapes will take about 40 tapes, which will cost more than that little hard drive (never mind the handling/loading/unloading issues or the cost of the mechanism) by a factor of several.

If the year were 1986 then what you're asking for would make sense to videotape (and indeed there were products on the market to do that). But today it doesn't make sense, it's just so much easier to dump it to a hard drive in digital form.

Analysis of data after-the-fact is way way faster from a hard drive too (by a factor of hundreds).

If you have legal requirements for archiving etc. these usually drive the solution rather than anything technical.

Tim.

A VHS video-tape recorder is an analog recorder. You'd need to scale your signal down to video levels, and add in frame and line synch signals - though the video recorder won't pay any attention to the line synch signals, except to extract the "black level" - the most negative possible signal excursion, nominally 0V for video - from the "back porch" on the line synch.

Around 1978 we used a standard consumer VHS recorder to record medical ultrasound scans, where the line period was about 300usec rather than

The video recorder was perfectly happy with the funny scan rate - in fact consumer video recorders happily recognise very crude approximations to the rather complicated frame signals specified in the video standards, which vastly simpliies the logic that you need to generate adequate frame signals.

---------- Bill Sloman, Nijmegen

Yes. Thats why I want to use a VHS recorder, rather than a digital system as another poster suggested. As I stated originally, I need an analog recorder.

Yeah, I figured this would be necessary..

Ok, now this is what I need to read about, and this response allows me to ask a few pointed questions:

a) Can the video signal be bipolar? Or must I add an appropriate DC offset?

b) The voltage range obviously needs to be scaled. What is the allowed range?

c) What are / where can I read about "frame" and "line synch" signals?

The video recorder couldn't care less about the colour burst - it just uses to the frame synchs to check that the drum the tape wraps around is spinning at the right speed.

------------------------ Bill Sloman, Nijmegen

What about dropout? It seems that could be major problem.

Alesis ADAT XT, modified for DC coupling. We use them for record and playback of laser show control signals with 30 Khz content.

Steve Roberts

That's a rather arbitrary decision considering that you've made no statements about resolution or noise. Or maybe it's only because you can make no statements about resolution or noise?

In the 60's and 70's your analog recorder would be an "instrumentation recorder". To get bandwidth down to DC you FM-modulate a carrier. To get 40kHz on your data you'd probably have to run the tape at 30IPS or faster. The instrumentation recorder from HP probably cost $10K (in 1960's dollars) and tape was another $10 a reel for a few minutes of data. Scaling and bias are done by knobs on the modulator (and hopefully matching knobs on the demodulator).

If you have no confidence in sampling your data and insist on using an NTSC VCR and analog recording then you're simply going to have to put up with horizontal sync/retrace at 15kHz and vertical sync at 60Hz and the random dropouts.

I suspect we could give you much more useful advice if you could parameterize your requirements and stop insisting that you know the solution. If I wanted vague requirements with mandated solutions that'll never work, I'd be at my day job instead of on Usenet!

Tim.

No, video signals are stored in a VCR using composite signals. A composite signal transmits luminance (Y) in the lower 4MHz and color information on a kind of chrom wheel at around 2MHz (Yr, Yc). A VHS VCR lowers those rates considerably and writes that to tape, using the sync signals to drive the drum with its heads. The Y signal can be recorded reliably between 0 and 0.7 V

Details are here:

The Y signal can be recorded reliably between 0 and 0.7 V.

Google.

You need to create at minimum horizontal and vertical sync signals and possibly a color burst. You have to fit all that into a tight timing scheme and you can't record any values during sync and retrace phases as well as the first and last few lines of a picture.

I should alos mention that, due to the helical scan recording of VHS, you can not slow down or speed up the tape transport beyon the usual SP and LP settings.

I'm confused, basically video recorders are analogue devices that record analogue, but they have digital gaps where nothing can be recorded, ie analogue with gaps in it. I cannot see an easy way without digitizing your signal and a buffering FIFO. the first 16 or so lines are more or less unusable( from memory), as the record head is not touching the tape, problem. The sony audio F1 system, 1970's digitised the analogue and spread it around the tape, for error correction purposes.

I dont think you have a hope in hell of directly recording 40khz on vhs without digitsing first. As others have pointed out, dropouts will be a very significant problem.

martin

Noise is exactly what I'm investigating, and in fact for my purposes I need it. I'm making a GPAC. Its storage will be magnetic tape (hence my desire to use a VHS -- because it is cheap. I'd consider other solutions, but it MUST be entirely analog).

Several systems of differential equations will be solved by this machine, and its qualitative results will be compared to the results obtained by a digital machine.

An all-analog system capable of solving differential equations, storing the results, and replaying those results as input to other equations is a requirement here.

I have the computation circuits nearly entirely built (I still need to wire up a few precision rectifiers). Its now time for me to consider the storage side.

Testing thus far has shown that my computation circuits are capable of reliable operation from DC to 40kHz, so I'd really like a storage system that is reliable over that range. I can definately trade bandwidth for dollars, though.

Instrumentation recorders are an option, for sure -- I have occasionally seen untested devices on ebay for a few tens of dollars. Tested/known-working items for much more.

I'd like to keep costs to a minimum, hence my desire to use a VCR.

Thats funny. Honestly though, does ANY of the above actually HELP you in answering my question? I doubt it. You don't really need to know my application, or why I need analog.

My question was how to make some interface circuits that will allow me to record an arbitrary analog wave onto the video track of a VCR. I appreciate the digital proselytizing -- but really, in this instance, I have to be a pagan.

Mike

At the same time, you should realize that an analog recorder like a VCR will reproduce amplitudes to an accuracy of a few percent at best (and this includes tweaking knobs to get it aligned to this point - the accuracy will be worse than 10% without tweaking.)

If you don't want 16-bit digital recording because it has 0.0015% quantization error/noise but are willing to put up with a few percent error on an analog system, that sets off all sorts of alarm bells in my head.

I do share your interest in analog computers etc. (see my posts from earlier this year about building an integrator for the Lorentz equations) but you have to actually specify resolution and noise requirements before you can choose a solution.

Tim.

Yeah, I am aware of this. I am hoping to get 5% replay accuracy.

Are you saying that a few percent is all I can get from _any_ magnetic analog recording technology, or are you restricting your comment to VCRs?

Well, thats exactly what I'm saying. If I were strictly concerned with the precision of my solutions, I would certainly go with a digital recording method. Then again, I'd probably be doing digital computation too.

What I'm interested in is solving a certain set of equations with purely analog technology. I will also find the solution digitally. Comparison of the two is what I'm ultimately interested in. If the analog recording media increases my error by 10, or even 20% .. then thats life.

My hypothesis is that there are equations for which digital solutions are hard to find, and prone to error. The quantitative accuracy of the digital solution will degrade AND its qualitative behavior will vary from the actual solution. OTOH, the analog solution -- while having substantially less quantitative precision, will qualitatively behave similarly (analogously) to the actual solution.

That is why I am looking for an analog recording solution, but from what I've read so far, it doesn't appear as if a modified VCR is going to do the trick.

I'm afraid my perspective differs. I'm more interested in price than the noise contribution. I'd really like a cheap analog storage device, and a VCR with some signal conditioning circuitry seemed like the best bet.

Mike

Do you realize that the VCR requires a signal in the video format, including sync pulses? You cannot record a continuous arbitrary analog signal on a VCR without packaging it with some H-sync and V-sync pulses. These will interrupt your analog signal, unless you have some way of buffering the signal, which almost certainly involves digital conversion. However the general-purpose data acquisition tape decks will do what you want because they were designed for exactly that purpose.

-Robert Scott Ypsilanti, Michigan

Since you seem hell-bent on using a VCR, are you aware of the Sony PCM-F1?

It was a digital audio to NTSC (for a VCR) before CD but used the same sample rate. It had the error correction, dropout correction and metering, the whole nine yards. Right now, for 4 more days, there is one on eBay for $35 (paid $1000 for my PCM701, the line powered version, in 1983). It might be possible to alter the analog section for DC coupling. I KNOW its possible to intercept the digital data stream because I did it in '83 to make a digital audio fader using a TRW hardware multiplier and EPROM log (for dB) lookup table to fade in 1/16 dB/step. Crude but it worked.

GG

"Mike" wrote....

You are asking for the imossible. No VCR records a continuous signal. They all require horizontal and vertical sync intervals which consume a significant time period.

If you really must have analog, you are likely stuck with FM recording on an industrial data recorder. You have specified an impossible set of requirements if you insist on any kind of video recording device.

There are tricks like recording pilot tones that are filtered out on playback to let you get better amplitude calibration. And of course FM lets you go down to DC. But to say that the absolute error is less than a few percent full scale across 0 to 40kHz is going to be hard even with an instrumentation recorder.

Oh, yeah, lots of stiff differential equations out there. But solving them with error-prone analog circuitry won't help you any.

For many stiff PDE's there are numerous tricks that let you solve the problem through relaxation (especially when you know the values at extremes).

But all this stuff is first-semester numerical analysis and applies whether your computer is analog or digital.

Get rid of your DC to 40kHz requirement and make it 100Hz to 3kHz and a plain old cassette recorder will do OK (assuming you can tolerate the phase shifts, dropouts, etc.)

Make it DC to a kHz and a FM-modulated carrier on plain old cassette tape ought to do great (well, until you start noticing the wow and flutter, but you already told us that you're impervious to these because your soul is pure and intentions are good.)

Tim.

The analog system will almost certainly be measuring its own limitations, rather than what you think you are recording.

The error rate of VHS is totally unacceptable for any reasonable data recording app.

Odds are overwhelming that your "comparison" will only show up the gross deficiencies of analog data recording.

And have nothing whatsoever to do with whatever it is you are trying to measure.

Also, the ---> standard