REM Detectors (again)

hmm... didn't think processing sound frequency would be that complicated... i think i'm still a bit confused so i say we wait a while till i get one built then we see what it produces exactly...

from what you posted, you're getting too many samples... technically, the amplitude of the sine wave is always the same, the frequency being the distance between peaks, right ? if you just take the interval between peaks, it gives you the value you need. the amplitude of the sine wave doesn't matter, it's going to be constant.

the program you showed me did a good job. it seemed to show both frequency and amplitude (the redder the value, the higher the amplitude?)... we just care about the blue stuff that was getting shown, maybe in our case it's going to be an orange or a yellow (same amplitude so we'll have no variation in colour i don't think) only the positionning of the noise along that line will change...

say this is a generalized view of what we'd be getting (i was playing a song here, so there's A LOT more info than we'd have):

http://www.thrice-blessed.com/gsrimage2.jpg

what you need to plot are the maximum values, the points and resulting line in red. and if you DO need to take all those samples along the sine wave (but i don't think you have to, we don't care about the amplitude) to find out the frequency, you don't need to store those.. you take samples for... wow, 1/100th of a second or not even (at 1000kHz you have 1000 of those waves in a second, it's crazy), then take the average value and wait for one or two seconds, and do that again. I realize taking all these values is MAD (especially if we get sounds at 3000kHz or 6000kHz)... i was hoping you can just take a few consecutive zero values (say 10 or 20) and average the time between them... you multiply that by 2 (between two zero values you get half a wave) and you get the frequency of the wave.

there MUST be some sort of easy way to do this, all those plugins for winamp and windows media player are driven by both frequency and amplitude, and they monitor both in real time without significant CPU usage...

i guess an analog to digital might be a good idea, we could give it another go eventually.

i could also make the amplitude change rather than the frequency (technically speaking, shoving voltage into a speaker will increase it's noise output) but that's less reliable from person to person and would need some fine tuning. Frequency gives us a wider range and more precision.

anyway, i think it's some sort of misunderstanding between us, so the best thing to do is just wait for the device to be built and then we can keep on going... there's no point in you doing extra work if in the end you might have to rework (large?) parts of that.

the little exe you posted before will give me a decent reading for now... once i get something on that, i'll do some screen captures and show them to you.

Yeah, alright, I think there was some misunderstanding :)
BTW, that bottom chart doesn't work the way you describe ;) ... that one shows frequencies in straight lines, not waves
Only the top green line one shows sine wave. The thing is that multiple frequencies can be produced at once on the chart.

Yeah, I agree we should perhaps wait to see what the device spits out, and take it from there

yeah, but that's what i was saying... it shows different frequencies at different amplitudes (since it's a song)... we don't care about the wave, we're could be getting anywhere between 1000 and 15000 kHz (or waves/second) so displaying these for any reason won't help much with anything. What we'd get i think would probably look like this:

http://www.thrice-blessed.com/gsrimage3.jpg

the brighter blue are the peaks of the lines... if you refer them back to the scale, you get the frequency value. I think what you meant is that the red line in my previous post won't get plotted by itself (everything underneath it gets filled in as well)... and that's fine, but we don't really need all those filler points tho, so we only need to graph the peaks... i generalized in the red one and cut a lot of the peaks out, but this image should make a bit more sense...

No ? I thought I had it, but I could be wrong.

As for the unit, there's nothing very complicated about it, so I'll try to get all the missing parts tomorow and build it.... i'd do it now, i don't really feel like going to school ;) but i have responsibilities it would seem ;)

That's not exactly what I meant, but I think we should just wait and see what it does :)

sounds like a plan :)

i actually went to school AND i bought the missing parts, so we're doing good. i'll start putting it together tonight, by tomorow night it should be done if no major debugging issues start happening.

ugh... first problem: component sizes: the trimmers i need to figure out what future values will be for some resistors are too big and that voltage to frequency chip is just... a killer.. it's so small i can barely solder to it and if i do, i have to make sure it won't go anywhere, cause if it does, the pins will most likely break. so i guess i'll build this one on separate boards and connect each part with wires. this will help with debugging.

once i have that i can build two more without trimmers on smaller boards. there's a big problem in the design, because i don't think the voltage to frequency converter we're using comes in a regular sized package, only the small surface kind... it's good, because the design can go REALLY small scale... but it's also bad, because to the average person building it, this might be a problem.

i'll look for alternatives eventually.

slow but steady progress:
http://thrice-blessed.com/working1.jpg

i got two of the three chips down (the one left is the voltage to frequency, and i'm trying to get everything perfect before i plug that one in, i'm afraid of burning it and i only have 2) and now i have to make the probes. the voltage is splitting right (0.5 across the skin) so now i have to check the resistance of the skin... the black electrical tape with a small metallic thing on it is the beginning of a probe. i'm connecting a wire to the back of the metallic stuff (which goes under all of the tape) and wrapping it all with velcro. once that's done i have to see how much comes out of the operational amplifier and amplify that voltage accordingly (if i need to)...

i know i'm taking long to make this thing, but it's coming along.

(also, if anyone's trying to build it, i changed some of the parts... the LM7805 for instance (which is ridiculously big) i replaced with a NTE 977, which is a lot smaller and does the same thing)

Good stuff.

I've been chatting to Seeker, who had a whole lot of good ideas.
If you blow the voltage to frequency chip, don't dispair:
Essentially you don't need that voltage to frequency converter or an A/D converter chip.
Simply use a cheap LM555 timer or equivalent and generate a pulse from the voltage. Either the duty cycle or the pulse can be variable to indicate the value.
Then stick it straight into the CTS line of COM1
In addition, I can then use COM1 to activate the motors/LEDs, with eg. the RTS line.
I'm sure you're familiar with the 555 timer, but if not : http://www.uoguelph.ca/~antoon/gadgets/555/555.html
You'd just need to tweak the voltage before and/or after the 555
This seems simple, cheap and easy :)
We would need to run some tests to see what the highest/lowest frequency would be that I can calculate over COM1


Here's a few links that could help:
http://www.bobblick.com/techref/projects/a...555/a2d555.html
http://www.aggsoft.com/rs232-pinout-cable/...connections.htm
http://www.electronics-lab.com/projects/misc/013/

that sounds like a really good idea...

i'll just finish this one since it's almost done, see what it does.

then i'll see what i can do with a timer (or even just a variable oscillator using a watch crystal) to see if it simplifies the design. it would probably be a good choice actually, because you could shove the resulting output into a microcontroller of sorts.

I did a bit more work (finger probes are done) but nothing significant... I have to put this on hold for a few days, it's the last 3-4 weeks of school for me so it's a pretty hectic time. Sometimes it's ok, but every now and then it gets real busy, and it so happens that these few days it's been real busy.

I'll come back to this as soon as I can, sorry about the delay and the slow development.

is this device being made or has it stopped.. iwas realy looking forward to buying one if they worked well..

Well, while Meus is busy, his side is on hold.
I've been working on it in my free time though
Progress = I have a lie detector, and a 555 A/D converter and a serial connector. Nice velcro finger thingy and little black box.
I just need to try tweak them to all work together :P
Until I get that right, I can't say if it's all a deadend...

After all, I suck at electronics. That's why Meus was helping :D

I'll try push a little harder, since I now know theres a bit of expectation out there :)

I've been so out of it lately, i'm sorry about that...

I finished my session ok, crazy crazy schedule and I had to work on posters/marketing for this theatre play at the same time, and right after i jumped straight in two weeks of working 65 hours/week...

BUT... it's all (slowly) coming to an end and I'm starting to be able to breathe.

The lie detector is a good start... for what i had in mind we're using a lie detector too. The serial too might be a good idea, I'm just crap at programming, so I don't know how to approach the serial issue. It's basically what we had before I jumped on the whole sound/line in. I'll try and think about it a little harder. I still think I'll just end up buying a GSR device, and all we'd have to do is reverse engineer it and then come up with something "different"... I'll start saving up.

I didn't forget about it though, so I'll get back to this as soon as I can (by the end of this week I should be back to the soldering iron)...

Yeah, I haven't forgotten either.
I have a half complete lie detector and serial connector at home :/
I've just been distracted and lazy

I've been lurking for a while, watching this thread, and I'm very interested as well. I'm sure I'm not the only one.

You might think this is science fiction but in truth I think it’s just a matter of time before someone invents a devise that will do what I'm about to explain. Years ago when I was helping my brother build his house I purposely dropped a coin in a specific location date side up before the concrete of the basement floor was pored. I recorded the type of coin and the date and predicted that someday there will be a tool sensitive enough to detect and read the date on the coin. Well it would be similar to radar and if a device like that tracked and image of your head and being connected to a computer could easily sense the REM period. A few light sources not even connected to the head could flash from many different directions so that there would be no need to attach anything to the body. It would give the computer a heliographic image to work with. It would require some form or radar. Maybe a devise that is similar to the tool used to see stuff on the ocean floor.

ok, so i saved up and I'll buy a GSR 2 next week, for inspiration. I'll figure out how their design manages to calibrate itself and provide a good reading range and (since that will eliminate trimmer pots) we'll have a much more simplified design.

Sounds good. Unfortunately I'm snowed under at work, and haven't touched it again

hey placebo. I have an idea for detecting REM. it should be simmilar to the novadreamer. using an optical mouse you can sense the movement of your eyeball even without touching it. when the mouse senses movement it gives more power to the LED. making it brighter ect. if you were to take 1 optical mouse and create a simple electronic circut, making the mouse use an infared LED when under normal power, and a bright blue LED ( or red if you are a conformist) when the mouse is giving more power to the LED, the blue one will be turned on.


if you have an optical mouse at the moment you can try this, stick it up by your closed eye, and let it sit until the light gets dim. move your eye about and watch as the light gets brighter.

i just dont have a free optical mouse to use thats the problem

:|

This seems like a great project. If you need any help, I would be more than happy to assist. I'm pretty good at programming, but really crappy at designing electronics. And right now, after having my first couple of serious LDs, I got devotion running out of my ears. :-) (Actually to the point that I'm almost regretting I took computer science at the university and not psychology.)

Maybe you got tired of it since you haven't posted in a while. I hope it's temporary... If you need any help, please tell me.

Well perhaps I got tired of the electronics, since I suck at it... but really I'm just snowed under at work. So the little bit of leisure time I get, I simply can't motivate myself to do MORE technical work :P

Anyway, the work stress is beginning to die down. Give me a week or so, and I should be able to continue with this. I guess I'll have to irritate the hardware nerd at work :)

About the optical mouse idea, I know about it. The problems are as follows:
1) The light is DAMN bright, and will probably wake you up
2) The mouse light + sensor would have to be right up against your eyeball, it seems. And it'd have to be a good (sensitive) mouse. This would be uncomfortable, and seems to hinder my eyeball from moving.
3) I was trying to keep away from using a face mask. I'm not a fan of that idea.
4) I wanted to be able to get some statistical data to record your sleeping state through the night. I can't do that with an optical mouse trick.

I am new here. I have read most of the thread on REM Detectors. I am wondering if an EEG biofeedback machine like this one
http://mindpeak.com/index.asp?PageAction=V...WPROD&ProdID=24
which monitors Brain, Heart, Muscle information could be used to accurately determine REM sleep? According to what I have read it can, but have never used one. The one at the mindpeak link above also has optional glasses that blink with LED's just like a NovaDreamer.
http://mindpeak.com/index.asp?PageAction=V...WPROD&ProdID=17

However this is all very expensive, $300 for the glasses! :holyshit:

Perahps there is a cheap basic system that can monitor some bio response associated with REM (skin temp [somewhere on the body] pulse whatever...REM Is associated with more than just eye movements)....that could be used with those glasses. I don't know enough about biofeedback.

This is interesting news article about dream detection thru the fingertip...
]http://www.ats.org/news.php?id=66[/url]

For some reason the link in my previous post did not work right, but I think this one will. Interesting article about a newly developed (medical) device that reliably identifies the onset and duration of REM sleep by tracking changes in the blood flow through the finger.

http://www.ats.org/news.php?id=66

Thanks for that link. I really appreciate it.
Now if I could get my hands on it :P
$5,400!!

I've been thinking of a different approach to this. The mechanism I'm thinking of is kind of like a dead-mans switch. Here is the general theory:

You hold this switch in your hand, it must have a very light tension. When you fall asleep, your muscles relax, releasing the switch, It starts a timer for maybe 5 minutes. After the timer has expired, the lights start flashing.

You would use this device late in the night, after 5-6 hours of sleep.

The switch could be a capacitive device or resistive.