Thread: REM Detectors (again)

Only reason I mentioned it is Windows NT - XP make it extremely difficult to control the hardware directly. They have added functionality for easy access to the control lines on the rs232, but for some reason have not made it quite as easy for the parallel port.

I've sait it a hundred times: Damn the hardware abstraction layer!

Originally posted by Seeker
I've sait it a hundred times: Damn the hardware abstraction layer!

Damn windows in general. You can't write cool shell scripts like you can in Linux.

i got the program and the device working in linux under wine, before making any progress with win XP, so yeah, i think we can go for serial now. As i see it, if we want to use both serial and LPT eventually we can, because both cables go to the box and not to the face or hand, so you can add as many wires between computer or box... i was hoping the LPT wouldn't be such a drag to get working under XP... most of the I2C to computer devices out there are LPT based anyway, so i would have expected an easy solution... but i guess i was wrong.

anyway, for serial i found this: http://zebra.tky.hut.fi/~jap/Electronics/I...r/SerialI2C.pdf
the problem is that it needs 5V so we have to switch the 2 AAA batteries for a 9V or something and regulate that. Does the serial give anything out ? (the gsr AMP uses 2.5V)

hmmm...

http://www.logix4u.net/inpout32.htm

and

http://www.hytherion.com/beattidp/comput/pport.htm (or strait to the C source http://www.hytherion.com/beattidp/comput/pport/test.c )

i still have a weakness for the LPT port because of the voltage requirements, the ease with which i built the hardware and the fact there does seem to be some support for winXP (as easy as copying one DLL in your system directory)

anyway, check out the source code and see what you think and if it seems doable i'll ship you the LPT unit in about a week (maybe even less). if not i'll try to get a serial one going.

If the DLL actually works, then, yes, it's do-able

I need to test it, maybe later today. And get a device to shove into the parallel port to test

it'll be there soon... i'm trying to debug the GSR amp right now (i built it last night), it seems the op amp i got is a different one than the one listed, so when i attached the pins as explained it obviously didn't work. I got the schematic for the op amp I have but there's some weird things happening with the resistors (so i might have burnt the one i'm using or made some bad connections).

i'll try to get a complete unit done by the end of the week, but i also have some school work due beginning of next week so i can't promise anything. but i'll try

hehe, i just got a call from the TI representatives in Montreal, inquiring about the device I'm building and how many I'll potentially mass produce I knew ordering samples would have a catch to it

anyway... I'm almost done with one of the units and halfway done with another one. I have the ADS chip to connect, some wires, and the motor + leds... and that will be about it. I'll test the output with an osciloscope i guess, see if it gives anything out. I think the debugging software the LPT connector comes with also has a read function, I could probably use that as well.

I'm just afraid I might send you a unit and you'll have trouble with the programming not because of the port or anything, but because the unit doesn't work properly... which would suck. I'll just make sure it reads some changes when I breathe faster or something, I know i was using a voltage meter to measure my GSR and when you breathe faster it went up.

anyway, I'll attach 4 LEDS and a motor to the printer port you'll be testing on, but the wires will be short, so you'll have to make them longer when you decide to mount them... The GSR wires as well will be short, so you'll have to extend them. I figure the best way for the unit to work is to have both light and skin signaling and you can use either one or both.

Sounds good. Not suprising about TI
As for the device - it would be troublesome if the device is faulty when it gets to me
I could probably sort something out, but it would be a much better idea to avoid that

I can extend the wires, no prob. The light and skin signalling is a bonus for me at this stage so cool
First step is to see the data. But if I can start playing with the signalling, that's great too

I've decided to wait for your device and then put the work into programming
It's just such a pain in the arse for me to do anything with electronics with the limited knowledge I have

hmm... do you have access to a Win98 computer ? Testing the device out as soon as you get it would be pretty easy if you did. There are a few things you'd have to change though, to make sure the LPT port can communicate two ways:

- You go in Bios and change the port type from standard printer to ECP and the DMA you set to 3 (unless it's already done)
- In win 98 you go to the device manager, and change the resources to basic configuration 000

this would set us on the same page (i tested on a laptop with win 98 and on a 486 with win 98 and it worked in both cases). From there on we could figure out the rest. I tried this in XP and running the program in win 98 compatibility mode, but still no luck. It's all up to that little DLL driver now

Unfortunately I only have win2k and winxp
But don't worry about it, I'll find a way around the damn abstraction layer
Ring 0 drivers here we come

well... i'm not sure what i'm supposed to get or if what i'm getting is right, but i'll look into it.

i finally got the gsr amp hooked up to the lpt interface and in to the computer. Running the debug software, when i switch the sync on and i read the I2C with ACK, the hex value changes between 00 and FF... somewhat randomly... it only does when I send an ACK though, so I assume you have to send and recieve constantly to make some sense of what's coming through... i'm pretty confused and i know nothing about data transfer.

i'll try to get some help. but if this is what it's supposed to do, all i have left is adding the LEDs and the motor and it'll be on it's way pretty soon...

just about now i wish i'd know more about electronics...

Haha... well we're doing the best we can (And I think you're doing quite well so far )
About the data transfer, I'd imagine that with 2 pins that it has it's own unusual way of slicing up the data
I'll have to figure out how to read I2C values

I wouldn't worry about the apparent randomness, but it would be a good idea to see if we can read the useful, decoded data
This is actually a catch-22 because technically that's my job But on the other hand, changing the electronics on my side isn't such a great idea.

Um... well.. if you can find a driver and program that will pump out the data, then good - I'll look around for one too
Otherwise we'll just have to hope for the best and if there's a problem I'll have to fix it on my side

hmmm... i don't know if it has to do with me not building the device right or me not knowing how to use the debugging software right, but something's not doing the trick. To my knowledge, i double checked everything in both parts, the LPT connector responds fine to win 98, the GSR amp doesn't seem to have any anomalies in the hardware. I think i might be getting the address of the ADS1110A0 wrong, it's 1001000, which in hex is 48. But it still doesn't change the reading when i enter that. Maybe i'll give it a shot on another computer and if not i have to get someone to help me out....

as it turns out i think the ADS converter is burning my amplifier chip (LM741). We checked the amplifier part with an oscilloscope and realized the IC's output pin (6) wasn't giving anything out. I changed it, measured it to make sure it worked, which it did... and then I connected the digital converter again. Turned the switch on, measured it again and nothing.

I'll be trying to connect the ADS straight to the computer (skipping the amplifying part) and see if that works.

no luck, i'm getting the same random numbers i was getting before (and i wouldn't mind that, but the randomness doesn't even change when i connect the probe to the ground or if the probe floats in the air, so something's obviously not good there).

Originally posted by MeusOpusMagnus
but the randomness doesn't even change when i connect the probe to the ground or if the probe floats in the air, so something's obviously not good there.

Hmm.. yes. That's not good
At least you found the problem with the amplifier...

i'll figure something out... i mean, how hard is it to make a simple voltmeter and to get the signal into digital? I just need to find a different analog to digital converter that won't fry up my amplifier (at least i THINK that's what fries it up, it has a momentary 100mA probably when it initializes and the op amplifier only supports up to 40mA)...

I don't get it though, ANYTHING that's called GSR anything costs hundreds of dollars... you can buy a digital voltmeter that you can plug in to a computer for 70$... and you can buy a small voltmeter for something like 15$ from radioshack... switch it to low current and you're measuring yourself some GSR.

Maybe i can try to figure out how those plug in to the computer.

Shit! I'm so dumb!

I assumed an LPT connection and all that crap was good and easy to do, you know, change the analog signal to digital and shove it in the computer through the printer port...

we don't NEED to

i was browsing around for GSR units at "low" prices, hoping maybe i could possibly rip one off and steal some technology... For 160$, the GSR2 is one of the "cheaper" ones out there... which is way, WAY too much for me. BUT! it comes with computer software, which made me wonder, since it doesn't seem to have any serial or lpt ports.

Then i found this:

http://webideas.com/biofeedback/calmlink/

basically the GSR2 gives you sound biofeedback, based on the voltage the sound changes. They then take the sound and plug it straight into the computer through the line in jack. All you have to do is plot a graph showing sound frequency against time. No need for digital conversions.

I'll use the op amp i have for the GSR, keep away from the I2C analog-digital, shove the sound in through the line in. We can still use the LPT port for the goggles or motor or both, but these can be built separately...

Also, i think the open EEG project already has some sound card software for their EEG device, we could probably give those a shot, work with the existing code and just add the new features we may need. Might be easier and faster.

finally, some cheering up... i was pretty pissed when i realized all that work turned out to be a dead end. So we're back on track and no need for you to figure out the hex values and all that crap.

oh, and to find out what the real voltage values are for the sound frequency we get, i could just measure some of the values associated to a certain sound frequency, but this will happen when we start working on the software.

i don't know how this can translate into wireless though. Wireless headphones might have what we need but i don't know how those work really.

and we're on our way (again)

i just ordered two samples of the AD7740 which is a small voltage to frequency converter (basically does the job for us) coming at a very cheap price (under 1$ us). I just have to make sure I don't burn any of them (i could only order 2).

I could directly connect a headphone jack to the voltage we're getting right now and we'd basically get white noise at different volumes, but i have a feeling the volume approach is not very accurate.

I'll be getting it by Wednesday so I'll start working on it on thursday or friday.

Sounds good - similar to a suggestion my friend gave me. Never really thought about the line in option
If the range of volume is not accurate enough (ie. if it scales down our values), then we could determine the min and max values we are expected to receive from human skin. And crop the values within that range. That way the volume can be spread over a smaller set of values and provide a more accurate value in that range

with the voltage -> frequency converter we shouldn't have to worry about scaling and everything... basically the output will always be the same for the same input...

say your GSR measures 0.35V and the output is 1000Hz, we'll have a scale to go from. I think the maximum input for the part i ordered is 2,5 V (which i don't think anyone's skin could produce)... I don't remember what the maximum frequency is, but it's a pretty wide range (i think it starts at 50Hz and it goes into the KHZ quite a lot)...

So we should be good with the accuracy and scale.

The part is taking longer to arrive (estimated shipment is the 7th) so we'll have to wait a bit longer to see.

What the fuck is this all about?

WS

[quote]What the fuck is this all about?
Try reading it - it makes perfect sense