MK1 Steel router build log 1500x800x200

[Doddy]

So there is some type of voltage regulator on the inputs circuit.

Agreed - our results tally.

http://www.everlight.com/file/ProductFile/EL817.pdf

If the 817 is preceded with an italic "L", then its the Liteon LVT-817. The 1024 is just the date-code. These are all pretty much the same.

I tested my 6-36 vdc NPN proximity switches in 24vdc and 12vdc situation
Hooked up a 4.7 k output load on black vs blue wire.
this to test inline resistance of the proximity switch by voltage divider calculation.

It looks like i have +/- 9.4k inline resistance in the output circuit of the proximity switch.

http://www.ekt2.com/pdf/14_PROXIMITY_INDUCTIVE__8BX.pdf

You don't say if the sensor is On or Off. What's the actual voltage with 24V applied across the supply, measured with a 4k7 pull-up, in both On and Off states? (I'll measure mine in the morning anyway). I'd expect the output voltage to be sat at 24V "Off" and near-as-dammit 0V "On".

Do you feel i can there for not go past 12 vdc to protect the opto`s against :
24volts PSU - 9.3volts input pin voltage = 13.7 volt reverse ?

Maybe we are trying to solve a non problem, please look at line 3 and tell me what you think.

It becomes a non-problem provided that there are no pull-ups to the supply voltage to the sensor. Once you have the sensor input floating to 24V (which you'll only have with a pull-up, for example if you wanted a 24V control circuit) then you're breaking the spec on the data sheet, and will likely damage the optocoupler.

3. External power supply input: 12-24V. Equiped with anti-reverse connection function.

I simply read that that there's reverse-polarity protection on (or around) the LM317 regulator. There is nothing on the optos to protect them (this would be an easy mod to the board to support 24V signalling).

I do think this is a non-problem by avoiding 24V switching levels.

[Doddy]

Okay, a nice Sunday experiment.

With reference to https://www.renesas.com/en-eu/produc...tandard-p.html - which discusses the effect of reverse-biasing an opto-coupler (section 1.5).

I've hooked up a brand-new 4n25 opto-coupler, 1K between anode and +10V, cathode connected to either 0V (opto = on) or through a 10k resistor to 24V (opto = off, reverse biased, simulating the sensor behaviour). Collector to +10V, emitter through a 1k to ground. DVM across emitter and ground.

With the cathode connected to 0V (on), the voltage at the emitter raises from 0.0 to 9.06V.

Now, I connect the cathode to the 24V supply via the 10k resistor (simulating the sensor internal pull-up of 10k). Predictably, the emitter voltage drops to 0V as the opto-coupler turns off.

So, the real test - leave like that for 5 minutes, then disconnect the 24V supply and ground the cathode.

Result - emitter voltage rises to 9.06V

Conclusion - for a short-duration (5 minute) exposure to 24V (and a reverse bias of 14V) the performance of the opto isolator is not changed.

However, measuring the voltage across the 10k gives 0V (i.e. no voltage drop - no current flow) - the voltage hasn't reached the avalanche voltage of the opto's LED.

Since then, I've added in a second PSU giving me a range upto 60V, in place of the 24V supply. Cranking the voltage up it was evident that the avalanche region of the LED in the opto was around 50V. Removing the 50V showed an emitter voltage of 9.01V (a 0.05V reduction from previous). Replacing the 10K resistor with a 2.2k resistor (to increase the current flow during the reverse-biasing), and the emitter voltage dropped to 8.84V (further degeneration). Note, this effect does appear to be permanent, but not increasing (in the 5-10 minutes I waited).

So, what do I think? Reading random articles on the internet does suggest that reverse-biasing the LED will result in a deterioration of performance over a period of years. The article above suggests that even short duration reverse biasing can result in deterioration.

However, the avalanche voltage may be substantially higher than the datasheet (clearly you can expect the opto-isolator to tolerate a 6V reverse bias without damage - but the actual headroom that you have between that and the actual avalanche voltage is unknown, and likely variable between difference devices, age, temperature etc).

Having run this experiment I'm inclined to think that you're not likely to have a significant problem running the BoB with the inputs connected to a 24V-driven NPN sensor. If might be that, over time, the opto-couplers degrade, but I'm not convinced that this would result in failure. If it does, the optos are easily replaced (or the BoB, if these are still available after the failure).

BUT, I have decided that I'll be powering the BoB and the sensors with a common 12V supply - it's a personal choice but I will keep the reverse bias voltage to less than the maximum expressed on the datasheet.

[driftspin]

Okay, a nice Sunday experiment.

With reference to https://www.renesas.com/en-eu/produc...tandard-p.html - which discusses the effect of reverse-biasing an opto-coupler (section 1.5).

I've hooked up a brand-new 4n25 opto-coupler, 1K between anode and +10V, cathode connected to either 0V (opto = on) or through a 10k resistor to 24V (opto = off, reverse biased, simulating the sensor behaviour). Collector to +10V, emitter through a 1k to ground. DVM across emitter and ground.

With the cathode connected to 0V (on), the voltage at the emitter raises from 0.0 to 9.06V.

Now, I connect the cathode to the 24V supply via the 10k resistor (simulating the sensor internal pull-up of 10k). Predictably, the emitter voltage drops to 0V as the opto-coupler turns off.

So, the real test - leave like that for 5 minutes, then disconnect the 24V supply and ground the cathode.

Result - emitter voltage rises to 9.06V

Conclusion - for a short-duration (5 minute) exposure to 24V (and a reverse bias of 14V) the performance of the opto isolator is not changed.

However, measuring the voltage across the 10k gives 0V (i.e. no voltage drop - no current flow) - the voltage hasn't reached the avalanche voltage of the opto's LED.

Since then, I've added in a second PSU giving me a range upto 60V, in place of the 24V supply. Cranking the voltage up it was evident that the avalanche region of the LED in the opto was around 50V. Removing the 50V showed an emitter voltage of 9.01V (a 0.05V reduction from previous). Replacing the 10K resistor with a 2.2k resistor (to increase the current flow during the reverse-biasing), and the emitter voltage dropped to 8.84V (further degeneration). Note, this effect does appear to be permanent, but not increasing (in the 5-10 minutes I waited).

So, what do I think? Reading random articles on the internet does suggest that reverse-biasing the LED will result in a deterioration of performance over a period of years. The article above suggests that even short duration reverse biasing can result in deterioration.

However, the avalanche voltage may be substantially higher than the datasheet (clearly you can expect the opto-isolator to tolerate a 6V reverse bias without damage - but the actual headroom that you have between that and the actual avalanche voltage is unknown, and likely variable between difference devices, age, temperature etc).

Having run this experiment I'm inclined to think that you're not likely to have a significant problem running the BoB with the inputs connected to a 24V-driven NPN sensor. If might be that, over time, the opto-couplers degrade, but I'm not convinced that this would result in failure. If it does, the optos are easily replaced (or the BoB, if these are still available after the failure).

BUT, I have decided that I'll be powering the BoB and the sensors with a common 12V supply - it's a personal choice but I will keep the reverse bias voltage to less than the maximum expressed on the datasheet.

Ok ,thoroughly does not even closely describe your commitment in helping me out.
Thank you for that.

Ok so...i decided to do the following based on this thread and other threads.

Since there seems to be 9.4k ish in line in the proximity switch emitter, i put a 4.7k resisitor in series connected to ground.

When connected to a 24vdc psu, this will give me about 8 volts at the point of connection at the bob. in the "open "position.

When in the " closed " position the emitter goes to 0v.. hope this will work.


I will report back when all is operational.

Doddy thanks again for doing research on this subject.


Grtz. Bert




Verstuurd vanaf mijn SM-A320FL met Tapatalk

[Doddy]

Ok ,thoroughly does not even closely describe your commitment in helping me out.
Thank you for that.

Ok so...i decided to do the following based on this thread and other threads.

Since there seems to be 9.4k ish in line in the proximity switch emitter, i put a 4.7k resisitor in series connected to ground.

When connected to a 24vdc psu, this will give me about 8 volts at the point of connection at the bob. in the "open "position.

When in the " closed " position the emitter goes to 0v.. hope this will work.


I will report back when all is operational.

Doddy thanks again for doing research on this subject.


Grtz. Bert




Verstuurd vanaf mijn SM-A320FL met Tapatalk

In honesty, I'm doing this for me, as much as anything - I will need to start thinking of wiring pretty soon, and you're bringing up the questions that I need to answer for myself.

I think the NPN sensors the black wire is actually the collector, not emitter (I guess the emitter is connected to blue/0v)

Re your proposal. Think carefully of what you're trying to achieve. With the NPN providing a switch to ground (current sink), by placing a resistor from the output to ground will give you a low(ish) resistance to ground even when the NPN transistor is OFF - the LED will still illuminate (you'll get around 1.5mA through it - enough to illuminate it, probably enough for the current through the opto's transistor to actuate the input to the UCx00 - the input impedance of that is something like 50k to ground (and 4k7 to +5).

Try it, by all means, but don't be surprised if the input is permanently on.

The easiest solution is to simply tack a small signal diode (1n4148 is probably the most common) across the pin 1/2 of each opto - in the opposite sense to the opto's LED (i.e. cathode to pin 1, anode to pin 2). That way you cannot reverse bias either diode beyond the forward conduction voltage of the opposing diode - max of 0.6V by the 1n4148 across the opto's LED, or around 1.5V by the opto's LED across the 1n4148. Essentially, this will limit the voltage at the input to the BoB to the V-Ref provided by the BoB's onboard regulator (~10V) plus 0.6V.

Of course, this (and the idea that you proposed) both serve to reduce the switching voltage to something around 10V, which is why I still think that just using a 12V supply to the BoB and to the sensor is an easy alternative.

[driftspin]

Try it, by all means, but don't be surprised if the input is permanently on.

.

Ok i get it.

So best would.be to raise voltage just over 9.3 or what ever comes out the bob with nothing connected.

I really like your idea of the reversed diode.

This will act as an indicator for status at the same time.


Grtz Bert.

Verstuurd vanaf mijn SM-A320FL met Tapatalk

[driftspin]

In honesty, I'm doing this for me, as much as anything - I will need to start thinking of wiring pretty soon, and you're bringing up the questions that I need to answer for myself.

I think the NPN sensors the black wire is actually the collector, not emitter (I guess the emitter is connected to blue/0v)

Re your proposal. Think carefully of what you're trying to achieve. With the NPN providing a switch to ground (current sink), by placing a resistor from the output to ground will give you a low(ish) resistance to ground even when the NPN transistor is OFF - the LED will still illuminate (you'll get around 1.5mA through it - enough to illuminate it, probably enough for the current through the opto's transistor to actuate the input to the UCx00 - the input impedance of that is something like 50k to ground (and 4k7 to +5).

Try it, by all means, but don't be surprised if the input is permanently on.

The easiest solution is to simply tack a small signal diode (1n4148 is probably the most common) across the pin 1/2 of each opto - in the opposite sense to the opto's LED (i.e. cathode to pin 1, anode to pin 2). That way you cannot reverse bias either diode beyond the forward conduction voltage of the opposing diode - max of 0.6V by the 1n4148 across the opto's LED, or around 1.5V by the opto's LED across the 1n4148. Essentially, this will limit the voltage at the input to the BoB to the V-Ref provided by the BoB's onboard regulator (~10V) plus 0.6V.

Of course, this (and the idea that you proposed) both serve to reduce the switching voltage to something around 10V, which is why I still think that just using a 12V supply to the BoB and to the sensor is an easy alternative.

Hi doddy,



All in all i had to add 500 ohm to 4.7k to have the inputs switch reliably. I ended up at 8.65 volt. after switching the inputs would not drop..

Adding 500 raised it to 9.15 ish. which is ok
So you where right it needed to go near 9 and or slightly over.

0.5volts still illuminated the input opto's just enough to screw up somtimes

I still have a glitch in 1 of the inputs though..

Need to find out why.
maybe wiring issue or a slightly off 4.7k resistor.

ill try 1k instead off 500 first... maybe there is a switching effect or so.


Grtz Brt




Verstuurd vanaf mijn SM-A320FL met Tapatalk

[AndyGuid]

It looks good from down here Bert

[Nickhofen]

Nifty Job!