Checking Tach

[NeverBlue]

How are you filtering other light sources that may be present?

[photofly]

In the physical domain (narrow angular response phototransistor) and the frequency domain (AC coupled amplifier stage and some high and low pass filtering)

[NeverBlue]

bandwidth?

[photofly]

of?

[iflyforpie]

I would imagine what your high and low pass filters let through. Can I check the RPM of my model airplane motor? Or the planet earth?

[photofly]

How fast does your model aircraft prop rotate?

The display goes from 1-9999. anything outside the range of about 200-9000rpm isn't displayed.

EDIT: from Wikipedia:

Under ideal conditions, the smallest .01 engines can turn a 3.5" (9 cm) propeller at speeds over 30,000 rpm, while the typical larger (.40-.60 cubic inch) engine will turn at 10–14,000 rpm.

So, no, you can't use it on your model.

And no, the 0.0006944 RPM of the earth won't show up. However, with a couple of bits of silver tape on the . it works quite nicely on a hand drill.

Which is a different question to how to filter out extraneous light sources, which on the basis that skylight is constant, is done by AC coupling the amplifier, 3dB point very low frequency. If you try to read a prop against a light source with its own 120Hz pulse rate, that won't work. The analogue low pass filtering prevents counting spikes (in whatever light source you're using to check) and so reading a too-high frequency. I put the 3dB point at 160Hz.

Out of interest, would you rather have a device that's programmed for two bladed props and you do the mental arithmetic for three or four blades? Or a device that measures blades per minutes and you do the arithmetic for two, three or four blades as appropriate? (Is anyone using single bladed propellers?)

[ahramin]

Personally, I'd prefer a blade count per minute and do the math but I can't see that it makes a difference.

[photofly]

Here's the first PCB version:

image2.JPG (87.34 KiB) Viewed 2165 times

image3.JPG (91.86 KiB) Viewed 2165 times

I'll let you know when I've tried it out on some props, with different sky conditions.

[NeverBlue]

Personally...I would use one that I can actually "calibrate" at many frequencies not look at only one (3600) that is out side of the RPM range you're actually measuring.
Secondly...a Tx , Rx version makes a lot more sense because there is less chance of eroneous light from other sources interfering....which is why I was asking about BW...are you only using Sunlight as your source? If so does your receiver only look for Sunlight?

...not to diminish at all what you've done photofly...very cool

[GyvAir]

Looks pretty slick! Where was the board made, that you have it in hand so quickly? I pictured it originating somewhere deep in Guangdong province and having to make its way via China Post.

If a 2-3-4 switch isn't an option (maybe for the 2.0 version?), I think having it display raw blade count would be simplest for anyone who will regularly see more than just 2 bladed props. I presume the majority of private owners would be happy with 2 blade programming though.

[photofly]

@NeverBlue: You don't get it. It doesn't need calibrating at different frequencies. If your wrist watch can tell you when it's 10am, and when it's noon, you don't need also to check that it can tell you the time when it's 10:30.

if you want to check the accuracy at any frequency you like, you can. You Tell me what accurate signal source you have to check the accuracy against, and I'll tell you how to feed that into the input so you can read the frequency. What do you have? an HP8657? HP5061A?

But bear in mind if your test source doesn't have at least a crystal oscillator that's stable and accurate to better than 50ppm, you should be using *this* device to check the accuracy of that.

[photofly]

Looks pretty slick! Where was the board made, that you have it in hand so quickly? I pictured it originating somewhere deep in Guangdong province and having to make its way via China Post.

DHL, but yes.

If a 2-3-4 switch isn't an option (maybe for the 2.0 version?), I think having it display raw blade count would be simplest for anyone who will regularly see more than just 2 bladed props. I presume the majority of private owners would be happy with 2 blade programming though.

No extra switches- this is about minimal components, and three position switches are $0.70.

[GyvAir]

Right. Forgot about the budgetary constraints in place!

Even with DHL, that seems like an impressively short lead time. I presume, given the non-urgent nature of the project, that you would have opted for most economical service levels.

Edit for spelling.

[photofly]

PCB manufacturing typically takes a week. There are lots of PCB companies all competing on price and time, so it's not hard to get good service. They were dispatched on Monday and arrived today.

[ahramin]

Feck me that's cool. I'll take half a dozen.

[photofly]

I'd better go and check it still works on an actual propellor....

[ahramin]

As long as you have it working for my annual next July .

[NeverBlue]

@NeverBlue: You don't get it. It doesn't need calibrating at different frequencies. If your wrist watch can tell you when it's 10am, and when it's noon, you don't need also to check that it can tell you the time when it's 10:30. 

...OK...SURE...you're calibrating with incandescent but using with sun.

A calibration requires a closed loop system with absolute know variables.

That is a fundamental rule....

...and the next time I do a 24 month altimeter re-cert I'll only check it at 10, 000 ft. That'll be good right???

I've been calibrating equipment pretty much my entire career...I think I get it.

...and I'm disappointed you'd say that Photofly...I was only asking questions...and you don't know me.

[Troubleshot]

@NeverBlue: You don't get it. It doesn't need calibrating at different frequencies. If your wrist watch can tell you when it's 10am, and when it's noon, you don't need also to check that it can tell you the time when it's 10:30. 

...OK...SURE...you're calibrating with incandescent but using with sun.

A calibration requires a closed loop system with absolute know variables.

That is a fundamental rule....

...and the next time I do a 24 month altimeter re-cert I'll only check it at 10, 000 ft. That'll be good right???

I've been calibrating equipment pretty much my entire career...I think I get it.

...and I'm disappointed you'd say that Photofly...I was only asking questions...and you don't know me.

The problem is we all know who you are on Avcanada and that is reason enough to speak to you like a 3 year old, absolutely nobody here to wants have any sort of interaction with you because it is literally like talking to a cement wall with a emoticon painted on it.

[photofly]

@NeverBlue: You don't get it. It doesn't need calibrating at different frequencies. If your wrist watch can tell you when it's 10am, and when it's noon, you don't need also to check that it can tell you the time when it's 10:30. 

...OK...SURE...you're calibrating with incandescent but using with sun.

A calibration requires a closed loop system with absolute know variables.

That is a fundamental rule....

...and the next time I do a 24 month altimeter re-cert I'll only check it at 10, 000 ft. That'll be good right???

I've been calibrating equipment pretty much my entire career...I think I get it.

...and I'm disappointed you'd say that Photofly...I was only asking questions...and you don't know me.

I think the bit that you don't get is how this device measures frequency.

There's a single-chip computer in it (a microcontroller) running from a 4.000MHz crystal, so the instruction rate is exactly 1.000MIPS. The computer times the number of clock cycles between a fixed number of light-to-dark transitions sensed by the phototransistor. The wavelength of the light doesn't affect things - the phototransistor is sensitive to any radiation between about 400nm and 1000nm.

Each blade passing causes a change (light to dark, then dark-to-light) - if you can time those transitions you can calculate the frequency. It's a very linear process - if the timing and calculation is correct for one frequency you can have some confidence that it's correct for all.

That's distinct from a mechanical device which has potential non-linearities, or (like an altimeter) isn't linear at all, so should be checked at a variety of points.

As I said, it's not particularly difficult to check this device at frequencies other than the 60.00Hz of the national electricity supply but because we're simply measuring time and doing arithmetic there's not really much point. If you can measure one amount of time accurately you can measure other amounts of time accurately (hence my wristwatch analogy). I guess there could be some programming error on my part that does the arithmetic wrong and gives the wrong answer at some particular frequency - but we're just dividing one number by another - something computers are good at getting right. And it would manifest as a systemic error affecting every unit. if you trust me to get it right once it's going to be correct for all. I can't think of an error that would differ from sample to sample. (I'm open to suggestions if you can, though.)

As far as doing a measurement on a prop, you'll need to use sky light, so that any "flicker" comes from the prop and not the light source. I imagine that's common to all "through the blades" optical tach checkers.

The point of calibrating against a fluorescent source is simply that it has strong 120.00Hz flicker; it has nothing to do with the spectrum of the light.

What stable frequency source do you have in your lab that you'd like to check this against?

A calibration requires a closed loop system with absolute know variables.

I don't think I agree. A calibration requires a standard to calibrate against, and a method of calibrating. I worked for a while in the electrical division of the National Physical Laboratory, the UK's national metrology lab, where research and industry laboratories would send their own electrical equipment to be calibrated and certificated against the national standards. So I have some clue about this.

[photofly]

It occurs to me that you might be thinking I adjusted something or "tweaked" something until the display shows 3600. I didn't - there's nothing to tweak or adjust. The clock runs at 4.000MHz (which is what the crystal is cut to oscillate at - it's printed on the can), and the timing and frequency measurement follow by simple arithmetic. If the display shows 3600 it's merely a confirmation that the oscillator is running at 4.000MHz.

[NeverBlue]

No...I know what a crystal is and I know you can't "tweak it"

What I am saying is that in order for the tach to measure correctly there can't be any other source of light coming from anywhere...period...no reflections...no hydro lights...no vehicle lights...no runway lights...no stars etc..

Many of those light sources operate at different at different "frequencies" 400 -800 THz or something like that...HID headlights operate at 400 Hz

So how...when you're doing a test on your own aircraft tach...can you ensure that none of these will cause erroneous or inaccurate readings??...

...You can't

?..and you didn't answer my altimeter question...do you really think that if you check it at 10, 000 ft. It will have the exact same error at 20, 000 ft. and -1000 ft?

Your calibration experience is suspect me thinks...

[cgzro]

These things have been around for 30 years for model aircraft. They work perfectly from several thousand RPM up to 30,000 RPM or so in daylight. You can easily tell when they are not working because they read 0, otherwise they track the throttle very precisely and you can easily tell when they are working properly.

[NeverBlue]

Perfectly?

I happen to have proof that they are not accurate in the environment in which they are normaly used. I've seen people using incandescent lights, fluorescent lights, automotive lights, even sunlight at different times in the day and they never get the same readings when they do it multiple times...never...I understand why.

Why? Because the can't control the environment they're doing it in. Which is what true calibration is.

The tach calibrator I use of course uses a very accurate optical sensor but it is in a closed box with it'own light source with no possibility of any other ambient light sources causing inaccuracies.

One more question Photofly...what level of light is considered zero ( not there anymore) by the sensor?

... sorry one more..in your previous post why did you use 2 times instead of 1 to determine the accuracy of your watch at 10:30?...hummmm???

.because you know better...

[photofly]

No...I know what a crystal is and I know you can't "tweak it"

Well, actually you can. You can put a trimmer capacitor in the oscillator circuit. But in this case there's no need - the off the shelf frequency accuracy is easily good enough.

What I am saying is that in order for the tach to measure correctly there can't be any other source of light coming from anywhere...period...no reflections...no hydro lights...no vehicle lights...no runway lights...no stars etc..

I understand you're saying that, but you're not correct. The phototransistor has a field of view of about 8° - it's not affected by any light source outside that angle. So as long as you don't have a bright flickering light directly behind the prop where you aim the sensor, you're ok. Best measure against the clouds or sky, where there aren't any runway lights or vehicle lights. Stars are too dim to show up against the daytime sky, and won't affect the measurement.

Many of those light sources operate at different at different "frequencies" 400 -800 THz or something like that...HID headlights operate at 400 Hz

I think you're mixing up the frequency of the light waves (500THz is 500,000 GHz) with the modulation (on-off flickering) of the lamp - 120Hz for a fluorescent, 245Hz for my LED desk lamp, and (you tell me) 400Hz for a headlamp. You'll get incorrect results if you have any of those light sources behind the prop. Best measure against the sky. There are no headlamps in the sky.

One more question Photofly...what level of light is considered zero ( not there anymore) by the sensor?

Any constant level of light is ignored by the circuit because it's AC coupled. It only detects time-varying light levels, which is what we're looking for as the prop repeatedly moves in front of the sensor and away again.

... sorry one more..in your previous post why did you use 2 times instead of 1 to determine the accuracy of your watch at 10:30?...hummmm???

I'm sorry, I don't understand this question. Ask again, if it's a serious point.

?..and you didn't answer my altimeter question...do you really think that if you check it at 10, 000 ft. It will have the exact same error at 20, 000 ft. and -1000 ft?

Let's talk about altimeter calibration in another thread. This one's about tachometers.