# Photo-diode for a high-frequency tachometer

I intend to build a tachometer, which is capable of detecting rotations up to 1kHz (60 000 RPM). From the Internet I know that this is possible to do with a magnet and a Hall detector. However, I would prefer a non-invasive solution with a light and a photo-detector, but from Internet it seems that ordinary low-cost photo-diodes are just too slow for such a task (700 RPM in this video https://www.youtube.com/watch?v=WG_4iChrvR4).

I assume that I would need a photo detecting device, which response time is at least an order of magnitude lower than the period of the oscillation, that is less than 100 micro-s. Are there any photo-detectors (photo-diodes) capable of that? If possible, please suggest a product.

• A cheap pair of an IR dectector and an IR led will do – Long Pham Apr 21 '18 at 18:26
• Lots of reflective sensors work in the 3mm range ballpark if you have slots or a a good black Sharpie stripe on rotating metal. Sensitivity and response time both are controlled load R. Try 10k with a comparator with x % hysteresis. Then a stator bracket or hotglue or better if it gets hot like PU. – Tony Stewart EE75 Apr 21 '18 at 18:27
• Tony, I think you need to create a reference page for the large number of acronyms you use - some of which are rather unusual and, I suspect, often way over the head of the OP in many cases. What's PU? Polyurethane? – Transistor Apr 21 '18 at 18:59
• @LongPham What do you mean by "IR detector"? Video which I referred in my question claims that 700 RPM is limit for a chip IR photodiode. – Pygmalion Apr 21 '18 at 19:08
• "ordinary low-cost photo-diodes are just too slow" I had a quick look at a major component site and mid-range photo diodes of ~45ns (that is >20MHz) are \$0.17. I assume you are using a fan. How many blades has a car fan? Op-amps go up to GHz so they are not the limiting factor. In electronics 1KHz nowadays is.....so............. slow. Even 1MHz is not that much of an issue anymore. – Oldfart Apr 21 '18 at 19:18

Never, ever, assume that a Youtube video presenter knows what he or she is talking about, unless he or she presents some credentials. From the video, it appears that the circuit being shown looks like

simulate this circuit – Schematic created using CircuitLab

Granted, backing out connections from a video is chancy, but that's what it looks like to me.

With this setup, it's the R1/C1 combination which limits the bandwidth, not the photodiode. I would suggest that the photodiode under discussion is probably good well into the MHz regime (since it's in a 5 mm package, which implies a small die area and low capacitance).

Assuming, of course, that you used a decent conditioning circuit and don't look at the wrong component as the culprit.

Here, for instance is a 50-cent IR diode in the same package. If you'll note the data sheet, the rise and fall times are under 10 nsec, which implies operation to something like 100 MHz. Granted, stray capacitance would make this a bit difficult to achieve, especially if you make the mistake of trying to do it on a breadboard (like the video), but that's the fault of the breadboard.

TLDR: The video is wrong.

• If I understand you right, ordinary cheap photodiode will do. I assumed that the Youtube creator is reliable, as he is connected to adafruit, a well know component seller. – Pygmalion Apr 22 '18 at 11:08
• I bought cheap IR diode / IR photodiode combination from China and it worked like a charm for frequencies up to 400 Hz (24 000 RPM) and it could easily go up to 1kHz. – Pygmalion May 12 '18 at 20:11

The video setup of jumper wires and opto-mechanics were "mickey-mouse" for any precision sensing at high speed.

However in order to get more speed with a fixed diode capacitance and lower Load R the signal amplitude of the expected 1~10uA for this application.

I picked a PD for you that is cheap and narrow angle adn radial THT suitable for reducing the aperture easily to block stray light.

https://www.digikey.ca/product-detail/en/on-semiconductor/QSD2030F/QSD2030F-ND/1049944

## Specs:

Photo Diode (PD) 5mm radial plastic , daylight filter.
BW: 40 deg
$I_L$ = 25 uA @ 0.5mW/cm² @950nm
C = 15 pF, if R=50Ω $~~~t_r=t_f$= 5ns typ.

• Assume: 1% or >= 3.6° of circumference pulse ( transmissive or reflective )
• also : 1kHz max but a 1% stripe response ~ 100kHz
• thus choose $t_r$ = 0.35/100kHz of 1ms or t=5us
• thus choose max gain R for a Trans-Impedance Amp config (TIA)
• using T=5ns=RC * k , k = 5ns/(15pF * 50Ω)= 11 approx. ( at rated current)
• thus for t=5us , R=50kΩ max. and wwith 1uA of light current you a get 50mV pulse which can be compared to create a logic level at 50% using a stable Vref.

I trust you can make a suitable reflective aperture for the PD to block stray light or slot interruptor.

here is one good IR LED . Use heatshrink tubing over the lens to reduce spraying stray light. The chip will be <1.5mm so the aperture can be 1~3mm depending how careful you can align this. The goal is to get at least a stable signal for entire rotation +/-10% adn then >95% reduction with the "Index" mark to get 10:1 SNR minimum, Then the index can be sliced at a fixed threshold with a comparator or saturated TIA Op Amp as a logic level. Use a high GBW RRIO type Op Amp.

Consider this Dual RRIO 200MHz GBW OA TLV3542

A divider or dual aperture /|\ holder is ideal with the angles chosen to given < 10% surface reflection variation with rotation yet as close as possible.

Good luck.