Author Topic: Bowling foul detector (Read 6894 times)

james_s · « **Reply #25 on:** May 19, 2018, 06:03:55 pm »

Are the IR emitters working? Some digital cameras can see IR to varying degrees, you can test that with a TV remote. I've seen IR LEDs fail on a number of occasions.

RyanG · « **Reply #26 on:** May 19, 2018, 07:07:16 pm »

I used a 2 pin IR receiver to light a LED on a breadboard. That worked, so I assume it is working. I also looked straight into the emitter, which has a lens on it that magnifies the diode. I could see some glowing in there, which would stop when I unplugged it. Between those two things I am fairly sure they are working.

james_s · « **Reply #27 on:** May 19, 2018, 10:54:24 pm »

Yeah it certainly sounds that way. Careful staring directly into those things, I don't know if there's any risk of eye damage but some of them are very bright despite the fact that it's far enough into IR that you perceive only a very dim glow.

RyanG · « **Reply #28 on:** May 19, 2018, 11:39:39 pm »

Yeah, I thought about how it might be bad for my eyes. I did it anyway. I tried powering an IR LED from the circuit through the ULN2003 in case the flashing has to be precisely in time with the MCU signal and nothing changed.

floobydust · « **Reply #29 on:** May 20, 2018, 02:58:22 am »

I would measure the voltages at the 4-pin flex cable going to the sensors. Power, GND, IR TX LED, IR RX and see what the RX voltage is doing.
It looks like a vanilla photo-diode amp, so I would expect an analog output voltage, but pulsed at the 1.6kHz unless you use a flashlight. A DMM should still average it.

I think the trimpots are for setting a sensor comparator signal to the MCU. You can note their position and try turn them either way.
But both sensors drifting might be another problem elsewhere in the circuit.

If this unit was hit by a meteorite (bowling ball), it might have knocked out something in the optics.

RyanG · « **Reply #30 on:** May 20, 2018, 03:23:36 am »

On the 4 pin connector I have the following.
Pin 1 - Low side of emitter going to ULN2003 50% duty cycle voltage ~6v and ~5v, this goes to a resistor then the ULN2003 so the voltage difference would just be the voltage drop of the LED
Pin 2 - VCC at ~6.7v
Pin 3 - GND
Pin 4 - constant ~450mv from one module and under 50mv on the other

floobydust · « **Reply #31 on:** May 20, 2018, 05:36:48 am »

From the pics I can trace a bit of the circuit. Hard to see the traces.
Quad op-amp U1 is has a gain stage with trimpot, so it amplifies the sensor's output, and then a transistor switch and output to the test point, and another op-amp stage and transistor switch to the MCU. Measuring the voltages on U1 will tell what's going on I think.

On the sensor PCB, their output is AC-coupled through capacitor C2. So, little DC expected. You could try measure ACV at the sensors' output for signs of life.
I'd expect <50mV DC on their pin 4 so the ~450mV measurement seems suspect.

Where is the 6.7V rail from, I thought 12V for the relays. U3 might be a 8V voltage regulator for U1, and U7 a 5V reg for the MCU. What are the part numbers on U7, U3?

The trimpots set the sensitivity (gain), so I'd give them a whirl.

RyanG · « **Reply #32 on:** May 20, 2018, 08:02:12 am »

Those measurements I gave were taken with a scope, so it was a flat DC value on the scope.

The 6.7v is from the LM3578. U7 is a 78L05 and U3 is a MC34064 undervoltage IC. I'm starting to think that random 6.7v rail is the problem. The relay coils are connected to the output of the LM3578. That might be enough to pull them, but I highly doubt that is the correct voltage. I put 12v straight onto the cathode of the schottky diode coming off of the LM3578 and I do get some output signal from the sensors then and at the test points. There isn't much of a voltage difference between high and low, but there is an approximately 2Khz square wave signal coming out. The voltage isn't the same between the 2 sensors but the frequency and phase of the signals are the same. I did figure out which pins on the quad opamp are for the inputs from the sensors when they detect a reflected beam. The test points go to ~3.5v when the correct pin is pulled high. When I stop pulling it high the 2khz wave comes back, which also triggers one of the relays. Even with that signal, reflecting the light back in or hooking up a separate LED to do the flashing doesn't work.

james_s · « **Reply #33 on:** May 20, 2018, 05:54:24 pm »

Yeah that doesn't sound right, can you tell where that 6.7V rail is coming from? Often relays have their coil voltage printed right on them. It's possible that the 6.7V is supposed to be 12V, it may be coming from a regulator that is expecting to be powered by the 24V secondary.

RyanG · « **Reply #34 on:** May 20, 2018, 06:40:43 pm »

The 6.7v comes from the LM3578 that is connected to the big electrolytic cap, which is fed from the bridge rectifier. It would make more sense that the caps to are 50v rated if it is expecting a 24v winding on the transformer. 24v makes the LM3578 get extremely hot, but it puts out 12v.

floobydust · « **Reply #35 on:** May 20, 2018, 07:37:34 pm »

I'd say it should be regulating at 12V as the main rail for the relays, opto sensors and 78L05 input.
LM3578 is not a stellar part for efficiency.
But there's very little load, it should not get hot with say 18VDC input ((12VAC*120%)-2V)*1.414

I would check for other hot parts like those tantalum slug capacitor C18 to see if there is something drawing excess current. Otherwise the SMPS may be unstable if C20 is open, or maybe the IC is damaged from the backwards transformer overvoltage.

Another method is to reverse-engineer the LM3578 feedback resistors (R17 and nearby) to see what the design voltage is.

Getting close to fixing it

RyanG · « **Reply #36 on:** May 20, 2018, 08:06:41 pm »

Hard to see most stuff around the LM3578 on top of the board because they spilled the epoxy all over that stuff when they used it to stick the transformer down. I checked to see if it is conductive, and it is not.

RyanG · « **Reply #37 on:** May 20, 2018, 09:16:36 pm »

Found this picture on the TI website. The red values are the values I found in the circuit, and from what I can see the circuit on the board is the same as the one in the picture.

The formula in the data sheet for R1 is:

R1 = (Vout -1) * R2

That makes Vout = 12v

With power applied I don't see any oscillation on pin 3, it is at the input voltage, and pin 1 reads as the input voltage

floobydust · « **Reply #38 on:** May 21, 2018, 11:44:28 pm »

The oscillator at pin 3 is 0.55Vpp not a huge signal. If the IC is dead or transistor switch shorted it would not step-down (buck) at all, just pass-through. Does it step down at all?

I wondered if the toroid inductor core is cracked, I would look closely. If the unit was bashed hard. It would give low inductance and might explain the regulator's behaviour.

RyanG · « **Reply #39 on:** May 21, 2018, 11:56:37 pm »

When I was testing the it with a higher input voltage(18v-20v) the LM3578 was really hot and then the LEDs started blinking. After that the TLC274 let out the smoke. I ordered a new of both and had to break the other ones to get them off the board because of the spilled epoxy. I burned the rest of the epoxy that I could easily get to with my soldering iron and scraped it off. So, now I'll wait until I get the new ones and I'll see what happens. Hopefully nothing else is wrong after the smoke.

RyanG · « **Reply #40 on:** May 29, 2018, 03:15:26 am »

I finally got the stuff in to try to fix this things. I soldered in the new transformer, opamp, and LM3578. I replaced the blown fuse. The voltage is regulated at 12v out of the LM3578. It turns on, but I still have no detection of the reflected IR.

NiHaoMike · « **Reply #41 on:** May 30, 2018, 05:51:30 am »

I always assumed those worked with strain gauges or weight switches under the floor. Wouldn't an IR beam be easy to step over for someone wanting to cheat?

RyanG · « **Reply #42 on:** May 30, 2018, 03:04:17 pm »

If you mean step over without breaking the beam, then I would like to see someone do it. About 4 inches on the far side of the beam is where the oil starts on a typical bowling lane. If you step on that while approaching to throw a ball you will fall.

Back to the thing that is broken. I am going to get an actual reflector made for this thing from work next time I go in.

floobydust · « **Reply #43 on:** May 30, 2018, 07:47:56 pm »

I think you have the power supply fixed now, decent 12V? The TLC274 smoked, its max. is 16V I guess 18V killed it.

I would go back to the voltages/waveforms at the sensor 4-pin connector and see if they survived. You could even inject IR remote 38kHz light and see if you get a response.

For an IR mirror, a CD works. You can use a flashlight or laser pointer first to actually see where it aims.
The sensor's lens, I keep thinking it has a focal point, not a straight (collimated) beam. But it shouldn't be so finicky unless bowlers are to the mm.

In the old days, I'd use an IR detection card which has phosphor that glows when IR hits it. There are cheap ones out there.

james_s · « **Reply #44 on:** May 30, 2018, 08:20:45 pm »

Quote from: NiHaoMike on May 30, 2018, 05:51:30 am

I always assumed those worked with strain gauges or weight switches under the floor. Wouldn't an IR beam be easy to step over for someone wanting to cheat?

What is gained by cheating at bowling? Kinda takes the fun out of the game, and certainly won't win you any friends. For any type of official tournament surely there would be someone keeping an eye on things in case anyone gets any ideas.

RyanG · « **Reply #45 on:** May 31, 2018, 02:56:26 am »

I tried bouncing the IR back with an actual reflector and it didn't do anything.
The 4 pin connectors are:
1- low side of IR emitter
2- 12v power
3- ground
4- 0v on one of them and a constant ~2v on the other

The main difference between the two is that the voltage across the IR receiver is about 8v on one side and less than 1v on the other. Neither side works though. Could it be the opamps on each of the IR modules are blown?

james_s · « **Reply #46 on:** May 31, 2018, 04:13:32 am »

Possibly, doesn't seem like there are a lot of other things you haven't checked. Have you traced out the IR sensor circuit? Often there's a pullup resistor.

floobydust · « **Reply #47 on:** May 31, 2018, 05:44:06 pm »

The sensor looks like a vanilla (single-supply) photo-diode amplifier. It should have ~~1/2Vcc (6V)~~ 1.1V across C4 from R5, R6.
I think the amplifier is AC-coupled between the two op-amp sections and at the output with C1 and C2 perhaps. C3 is for 12V power filtering.
You don't want the sensor output to include (DC) from room lighting and shadows etc. Only the HF carrier pulses. Unless D1 is not a clamp but a demodulator.

If you can measure some voltages and compare the two sensors- although both seem to be not working despite different voltages. That could also be due to the main board input affecting the sensors' output.
I tried to sketch a schematic but the pic is a bit blurry.

RyanG · « **Reply #48 on:** June 01, 2018, 01:51:26 am »

Here is a terribly laid out but accurate(I think) reverse engineering schematic of the IR module. I put the parts on the schematic as they were on the board for easier following of the layout while going through the process of following everything. I forgot to connect the V+ pins on the opamp symbols, but they are just connected to pin 2 of the P1 connector. Also, C2 value is unknown not 1uf. I will go back and make the schematic better later.

floobydust · « **Reply #49 on:** June 01, 2018, 04:34:28 am »

That looks good. The first op-amp is a photo-diode transimpedance (I to V) stage with the second op-amp I think as DC-restoration to keep the amplifier from saturating due to ambient light. It is a voltage-follower that integrates over a long time constant R4/C1 (~3.2 seconds) and very slowly keeps the (output) first op-amp biased at about 1.1V
R5/R6 at C4 sets 1.1V as the bias point with 12V power.
I would look for 1.1V at IC pin 5 and 6; pin 3 and 2 should be the same voltage and a bit higher at about 1.5V but I could be wrong here.

There's really the op-amp and photo-diode to have gone bad, unless cracked resistors from a bowling ball hit.

SMD marking code 5D probably a MMBD914 diode.


EEVblog Main Site	EEVblog on Youtube	EEVblog on Twitter	EEVblog on Facebook	EEVblog on Odysee

Author Topic: Bowling foul detector (Read 6894 times)

Share me