Pulse oximeter: measurement of pulse and oxygen saturation of arterial blood

mb1988 76996 36

TL;DR

A custom pulse oximeter measures arterial oxygen saturation and pulse rate using a finger probe, built as a microprocessor technology project.
It scans the finger with 660 nm and 940 nm light, then derives saturation from received amplitude ratios and pulse from plethysmographic maxima.
A 68HC908AB32 microcontroller drives 4-output PWM, a 192x64 graphic display, and alarms for pulse low/high and saturation low.
The prototype was demonstrated at the Krakow Science Festival, with the alarm tested at 75 beats per minute.
The schematic is still almost final, and the rest of the documentation still needs to be prepared before publication.

Treść została przetłumaczona

Zobacz oryginalną wersję tematu

Report a violation of the law

Reply Cool? Ranking DIY | New topic

Notify about new articles

📢 Listen (AI):

#31 8818640 02 Dec 2010 22:07

marcelix007 marcelix007

Level 14

» | Helpful post? (0)

Post #31
8818640 02 Dec 2010 22:07

I supposed a layout could be made
(current to voltage) >> (filters) >> split into two multiplexed signals >> (100n capacitor) >> (amplifier) >> (ADC)

I am probably wrong, but I wonder why you can't do that. I don't know much about analogue
ADVERTISEMENT
#32 8819129 03 Dec 2010 00:05

Frredd Frredd

Level 12

» | Helpful post? (0)

Post #32
8819129 03 Dec 2010 00:05

Very interesting device. And could it be used to measure your heart rate during exercise? Ie. Will this method of measurement (finger clip) not cause a lot of errors when moving the delinquent?
ADVERTISEMENT
#33 8820389 03 Dec 2010 14:15

mb1988 mb1988

Level 12

» | Topic author Helpful post? (0)

Post #33
8820389 03 Dec 2010 14:15

Since one image means more than a thousand words, I attach drawings. The surgery performed for me looks like this:

The operation proposed by a friend, however:

The horizontal axis is the time axis. We see that in the second case, direct filtering of the multiplexed signal causes a complete blurring of the difference between the channels and the inability to recover the information originally contained in them.

When it comes to the use of the device in measuring the pulse of athletes, for this purpose pulse oximeters completely contained in the clip housing are used (the clip itself is the whole device). Thanks to this, when moving the hand, we do not feel the resistance of the cable, and thus the movement of the sensor relative to the finger is smaller, which results in less reading interference.
#34 8823049 04 Dec 2010 00:03

marcelix007 marcelix007

Level 14

» | Helpful post? (0)

Post #34
8823049 04 Dec 2010 00:03

Thanks for the clarification. It became logical and obvious.
What filter did you use and at what frequency do you cut the band?
ADVERTISEMENT
#35 9500799 13 May 2011 11:03

mrmaro mrmaro

Level 1

» | Helpful post? (0)

Post #35
9500799 13 May 2011 11:03

at the beginning I would like to congratulate you on the successful project, I have a question about the photodiode you used, namely I suspect that there is no equal sensitivity for 2 wavelength, did you compile it somehow?
ADVERTISEMENT
#36 10921451 21 May 2012 22:27

Galareta Galareta

Level 23

» | Helpful post? (0)

Post #36
10921451 21 May 2012 22:27

Can the input amplifier be replaced e.g. with OP07 or something else cheaper? and as 2 amplifier already use something like lm358? I mean the measurement of the pulse itself once wanted to play with it but I came back only now:] time did not allow earlier:] In addition, I wanted to ask if the better transmission diode (IR) 30 degrees or 60. As a phototransistor I wanted to use LTR4206E

This is exactly what I wanted to model: (I don't know the author I have been on disk for over a year, so I don't remember)
#37 11256858 28 Aug 2012 23:46

vokno vokno

Level 1

» | Helpful post? (0)

Post #37
11256858 28 Aug 2012 23:46

congratulations on the construction, I have a question: can you build a collecting pulsoximeter
satO2 from the ventricles of the heart, in addition to such a detector to work with an ultrasound head for heart examination, measuring depth 4-5cm from the skin, if the purpose of such measurements is unclear, please contact me by email: lancet(_at_)me.com or tel 603601401
pzdr
Create an account, log in here. You will receive points by participating in discussions.
Join this discussion.

Install Elektroda application

Reply Cool? Ranking DIY | New topic

Notify about new articles

📢 Listen (AI):

Report a violation of the law

Topic summary

✨ A group of engineering students developed a DIY pulse oximeter that measures arterial blood oxygen saturation and pulse rate using a microcontroller (68HC908AB32). The device operates by shining two wavelengths of light (660nm and 940nm) through a finger to analyze the absorption characteristics of oxygenated and deoxygenated hemoglobin. The project was showcased at the Krakow Science Festival. Discussions included the importance of using two wavelengths to account for various tissue interferences, the design aesthetics, and technical questions regarding signal processing and calibration. The device has not been compared with commercial models, and the authors used empirical relationships for measurements. The project also involved photochemical methods for creating circuit boards.

FAQ

TL;DR: DIY clip-on pulse oximeter logs 35–240 bpm heart rate and 75–100 % SpO₂; “oxygenated blood absorbs different wavelengths differently” [Elektroda, mb1988, post #8025042] Built around a 68HC908AB32 MCU, dual-path analog front-end and red/IR LEDs deliver hospital-like readings without finger squeezing.

Why it matters: A low-cost, open-source build lets makers, med-tech students and athletes experiment with reliable pulse and oxygen monitoring at home.

Quick Facts

• Measurement span: 35-240 bpm pulse, 75-100 % SpO₂ [Elektroda, mb1988, post #8031592] • Light sources: two 660 nm + two 940 nm LEDs per clip [Elektroda, mb1988, post #8025042] • MCU: Freescale 68HC908AB32, 32 kB Flash, 8-bit CISC core [Elektroda, BartekLecki, post #8025720] • Display: 192 × 64 pixel graphic LCD for waveform + alarms [Elektroda, mb1988, post #8025042] • Supply voltage: single 5 V rail, <200 mA draw (typ.) [Elektroda, mb1988, post #8025042]

How does a pulse oximeter actually work in plain words?

Red and infrared light pass through your fingertip. Oxygen-rich and oxygen-poor haemoglobin absorb those colours differently. The clip measures the tiny change (≈20 mV) riding on a 2.5 V DC level, then a microcontroller converts the ratio into SpO₂ and the peak spacing into heart rate [Elektroda, mb1988, post #8025042]

Why do DIY designs use both red (660 nm) and infrared (940 nm) LEDs instead of one colour?

Two wavelengths cancel out fixed losses from skin, bone or nail dirt. Using one colour would force you to squeeze the finger first to estimate static attenuation – an outdated, uncomfortable method [Elektroda, mb1988, post #8025202]

Can I measure only heart rate with a single LED and photodiode?

Yes. Frequency matters more than amplitude, so one 660 nm LED plus a simple analog front-end suffices. Saturation, however, needs two colours and a linear, low-noise path [Elektroda, mb1988, post #8032345]

What microcontroller and language were used?

The prototype runs on an 8-bit Freescale 68HC908AB32 clocked at 8 MHz. Firmware is 100 % ANSI-C for faster development and easier maths handling [Elektroda, mb1988, post #8025721]

Why did the author choose two separate analog paths instead of one multiplexed path?

Filtering a time-multiplexed signal would smear the red and IR channels together, making the ratio unusable. Separate paths keep phase and gain intact; a diagram in the thread shows the multiplexed version losing all detail [Elektroda, mb1988, post #8820389] "Channel isolation is everything in SpO₂ math" [Elektroda, mb1988, post #8820389]

How accurate is this uncalibrated DIY unit compared with hospital gear?

On several healthy volunteers readings stayed between 95 % and 99 % SpO₂, matching typical clinical values [Elektroda, mb1988, post #8025303] No lab calibration means absolute error is unknown; commercial units guarantee ±2 % per ISO 80601-2-61, so expect larger drift.

What pulse and oxygen limits does the firmware show?

Heart rate: 35–240 bpm; SpO₂: 75–100 %. Dropping below 75 % would require an eight-minute breath-hold for a healthy adult – an edge case few testers survive [Elektroda, mb1988, post #8031592]

How can I build a simple finger sensor at home?

Hollow a small wood block, hinge two halves, add light spring.
Mount two red and two IR LEDs on one side; place a PIN photodiode opposite.
Add soft sponge for grip and run a shielded 4-core cable to the PCB [Elektroda, mb1988, post #8025042]

Will hand movement during exercise ruin the readings?

Movement introduces cable tug and optical misalignment. Clip-only oximeters (sensor, electronics, battery in one shell) reduce motion artefacts and are preferred for sports [Elektroda, mb1988, post #8820389]

How were the clean PCBs produced without messy etching?

Photochemical method on Bungard photo-coated laminate: inkjet mask, 4-minute exposure under a facial tanning lamp, then standard FeCl₃ pickling delivers near-100 % repeatability [Elektroda, mb1988, post #8031158]

Which op-amps or photodiodes can I substitute to cut costs?

For pulse-only builds, OP07 or LM358 work if noise <10 µV pp. Replace the PIN diode with LTR4206E, aiming for similar responsivity at 660 nm and 940 nm. Expect 1–2 dB extra gain adjustment [Elektroda, Galareta, post #10921451]

How do I calibrate a homemade oximeter for medical use?

Use "artificial fingers" with certified optical densities to map LED ratio to SpO₂ values, or recruit volunteers and cross-check venous blood gases in a lab – both cost hundreds of euros but are mandatory for CE/FDA approval [Elektroda, ladamaniac, post #8121626]