/diy/ - Do It Yourself

do you know anything about electronics?
you dont just connect the +- inputs like a battery or whatever.
if you connect them in different ways they operate in different modes, if you connect it like this way in the picture i'm sure the output jumps between + and - rail based on which input is greater (magnitude? not sure)

anyway just google non-inverting amplifier for a schematic, you will need two resistors or so. you can also do some frequency filtering by using a capacitor but thats slightly less basic theory.

in general all op-amps can be used in all 'modes', the only differences you will find are to do with gains and frequency ranges. (in practice i don't know if this is true)

good luck

>>543638
>lol. doesn't know what an instrumentation amp is.

>>543618
this is for the ina128. btw you'll need 3 contact points to get a waveform.

>>543638
that's obviously not an opamp, it doesn't even have the right number of inputs
>>543618
I don't really know anything about electrocardiographs, but that looks like a wheatstone bridge on the left. As the resistances change so will the voltage difference between the nodes you circled. The yellow bit just means that when the resistances are equal it will hover around 2v, and when they change one will go up and the other will go down. For what those actually hook up to you'll have to look at your instructions.

>>543618
OP I built an electrocardiograph in my second year of studying EE. It's not as simple as just hooking up a good amplifier to your electrodes. You need a lot of filtering or there will be too much static to see the ECG signal at all. This is the amp/filter my team ended up designing. I can upload the entire project report if you like.

>>543659
You wanted a gain of 10 from U2B but since you put R12 on the wrong side of R11, you get a gain of 1 instead and a 10k-ohm load on the output.

>>543660
>You wanted a gain of 10 from U2B
No we didn't. The output would be oversaturated with that much amplification in U2B.

>>543664
OK, then why the 10k resistor to ground on from the output of U2B when you DC couple the signal to a unity gain buffer, and why bother with R11 at all when all it does is make U2B another unity-gain buffer? You have 3 unity-gain buffers all in a row.

>>543643
>btw you'll need 3 contact points to get a waveform
Thanks, I'm currently looking into the "leg-drive" part of the reference circuitry provided by TI (pic related). I figured if I split the resistance (marked red) that sets the gain in two and run the split into another amp it should work.

>>543653
>hover around 2v
I figured as much. The thing is, the electrodes that connect to the human body are meant to be differential +/- so I figured if I impose them on the 2V points it should work. The pic related and pic in OP are from the same datasheet, but the former is a bit too abstracted so I have trouble understanding if I can presume that it will work if I connect it up using the wheatstone bridge.

>>543659
I've divided my project up into a lot of small steps. The first thing for me is to handle the analogue domain (of which the amp seemed like a good place to start). After I've amped the signal to a reasonable voltage level I'll handle the filtering (I don't know if this is the correct order of going about it, but this is the initial plan). I'm leaning towards a simple first order low-pass/high-pass combo with 0.5Hz and 30Hz cut-offs at the moment (if I'm a head of my plan then a higher order filter + some notching can be applied).

>>543664
Also, why are buffers U5 and U6 way up there on the schematic when the rest of the schematic is otherwise so neatly laid out? They look suspiciously like you threw them in after the fact because things weren't working the way you thought they were going to, and were too lazy to drag everything else around to make room for them.

>>543670
Idiot forgot to post pic :P

>>543659
I examined your schematic a little bit... the dual diodes (D1-D4) are there for ESD protection? Also, the RC combo before the amps is a filter, right? Is it better to filter before amplification?

>>543674

Yes, the diodes are there for ESD protection.
You don't have to filter before the very first amplifier stage, but usually you can't have lots of amplification before the first HP filter, as the input signal can have hundreds of millivolts of DC voltage.

>>543673

having a leg electrode is a pain in the ass coz it's a long-ass wire, and people have to pull down their socks.
a better idea is to have two electrodes, LEFT+GROUND on one hand, and RIGHT+GROUND on the other hand

>>543659

WTF is that? you have R1 =10K so the gain of the instrumentation amplifier is 0.1. Where the hell is the big gain necessary to amplify the tiny voltages?

>>543618

using a 300-ohm bridge is nuts. the signals you're measuring have a significant output impedance, in the 1000's of ohms. by shorting them with 300 ohms, you are attenuating them significantly.
you need a large input impedance, like the 1meg in the >>543659 schematic.

>>543673

having a leg electrode is a pain in the ass coz it's a long-ass wire, and people have to pull down their socks.
a better idea is to have two electrodes, LEFT+GROUND on one hand, and RIGHT+GROUND on the other hand

>>543618

using a 300-ohm bridge is nuts. the signals you're measuring have a significant output impedance, in the 1000's of ohms. by shorting them with 300 ohms, you are attenuating them significantly.
you need a large input impedance, like the 1meg in the >>543659 schematic.

>>543760
>>543771
Thanks guys for some additional points to consider.
>filtering
After a short googling session and some analysis I have decided to add an additional high pass filter before the amplifier to get rid of the mentioned DC component.
>input impedance
It seems to me that the filter components (the large resistor) will serve a dual purpose of filtering and providing higher input impedance.
Checked the datasheet which says that the input impedance of INA333 is in the order of GOhm-s so I guess I'm safe in terms of impedance matching (pic of datasheet section)?
>ESD
Also, ESD protection seems so natural that it's a must in this design. How many diodes would be optimal? I've seen many designs where diodes are connected to Vdd and GND (the rationale provided in one design was to zero out the opposite leakage). So at least two in the former configuration would be OK?
Also, is the additional cost for specialized suppressor diodes justified by a large raise in performance or can I go on the cheap and still get the system up?

we made thousands of similar circuits and they were never returned coz ESD destroyed the OP-AMP. ESD did kill a few chips, but those were CMOS chips that were used to turn on the unit when the electrodes were applied. in short, you dont need diodes unless the op-amps are especially sensitive.

Holy crap OP are you still around?

I hadn't thought there'd actually be another thread on DIY about a /diy/ ECG.

>>543971


If you're working with TI parts anywhere they have some really nice docuemntation on ECG design and ICs (of course) - it's a bit biased towards the high end/cutting edge but still a good read

at first glance though i'd suprrised about the discussion on ESD, the designs i've seen utilize often a series resistor of fairly large magnitude to the leads, to prevent fault conditions from deliving current beyond miniscule microamps or so into the patient. this would have a beneficiary secondary effect on ESD, but frankly I hadn't heard of it being a real concern in the first place

>>545267
Yeah, I went through a heap of application notes etc. There's a lot of useful tips on their site for all sorts of possible problem domains.

I finally chose Analog Devices AD8223ARZ as my in-amp. Went for single supply operation mode.
The signal is first high passed with a cut off frequency of 0.05Hz.
The in-amp output is then low passed at 33.86Hz and fed into an ATMEGA32U4.
The ADC on it (10bit resolution) is nothing too flashy, but I believe it will do the job.
Chose the chip because it comes with a bootloader and no FTDI circuitry needed. Easy!
It also has sufficient amount of memory to do some digital filtering etc.
Brought out the PORTB pins to use them as an interface for the LCD.
About the ESD: the in-amp is rated at 1.5kV human body model... I added zeners on the inputs to play it safe.

At the moment I'm revisiting my board design. Just found out that the manufacturer I'm getting my board from allows me to put vias under the IC... which is a tremendous help considering the shit routing I've done so far.

Pic related is the current state of the board (using Eagle at the moment).
The analog/digital and mixed signal parts are coloured differently.

>>545574
>manufacturer I'm getting my board from allows me to put vias under the IC

They all allow that. PCB houses aren't really interested in your via positioning as long as you don't put multiple holes in the same or nearly the same location.
A more valid concern is a via in an smd pad. That's not a problem to the PCB house, but if you pay someone else to populate your board, you might get complaints, as the via sucks solder away. It's not a problem if you hand-solder your boards.

>>545574
>tfw a guy in my class used wires instead of nets
Gotta love Eagle.

>>545574
Try polygon pours for ground, so much easier.

>>546429
Ignore this post, I'm a dumbass