Interesting, but I need an optoisalted solution

 

Half a mA is quite low anyway. You will not gain any appreciable battery life looking for lower current. And as you have a 12V supply for the relay anyway, what is your power problem? I don't really understand and you do seem a bit reluctant to elaborate on the purpose to any detail.
Why not have the phone charging from the 12V relay supply so the current is no longer an issue. If isolation is a requirement, use a 12V to 5V isolated supply and operate the relay by sensing the supply input current.

 

It sounds to me like you have already found your solution in message #1 "I have tested 4N35, it has 0.3mA"

I think that removing the optoisolator in favor of a MOSFET would bring that 0.3mA down, but you ruled that out.

A 4N35 is not a darlington output, and several such optoisolators have been suggested with the idea that you could lower the current through the LED while have a CTR great enough to operated the relay coil. These would need to be tested I would think and, again, I am very surprised that you can get the 4N35 to work at 0.3 mA, but that is what you are reporting.

 

Here it is:
U2 - CD40106BT

 

Hi Danko,
I don't undestand what all the added complexity achieves. You will only have about 1.75 uA flowing through the LED in the opto coupler (Assuming the LED has a forward voltage of 1.5 volts). You would get exactly result with just the 2M resistor in series with the LED. Also I don't think the opto coupler would work with only 1.75 uA current through the LED. The output side only needs a logic level mosfet to drive the relay.

Les.

 

One way I can see to reduce the load on the USB 5 volts would be to drive a mosfet with that and the mosfet would drive the LED in the opto coupler. The current for the opto coupler would come from an isolated DC to DC converter fed from the car's 12 volts..

Les.

 

Hi Danko,
I don't undestand what all the added complexity achieves. You will only have about 1.75 uA flowing through the LED in the opto coupler (Assuming the LED has a forward voltage of 1.5 volts). You would get exactly result with just the 2M resistor in series with the LED. Also I don't think the opto coupler would work with only 1.75 uA current through the LED. The output side only needs a logic level mosfet to drive the relay.

Les.

Hi LesJones,
You are right about current through the LED, if we mean average value. But pulse LED current is 35 mA peak.
See diagrams below:

After optocoupler we have series very short pulses, and we need not logic level mosfet, but kind of smart filter, discriminator for converting that series to DC voltage.

 

After optocoupler we have series very short pulses

Why? The TS merely wants to switch on a relay.

 

Why? The TS merely wants to switch on a relay.

I hope, TS will demonstrate us how to do it. I can not do it better.

 

Hi Danko,
Now you explain it I can see how it works. Q1 and Q2 working like a unijunction transistor. I had assumed that the top of C1 was connected to the base of Q3 and it was just a filter capacitor. I see on the output side the fast pulse discharges C2 which take longer than the gap between pulses to charge. I think your circuit should do what the TS wants. I don't think the 30 or so mS delay should matter.

Les.

 

Hi Les,
Thanks a lot!
Since my English is not enough to write explanations, I prefer to draw schematics.

 

A problem with any circuit that attempts to generate a pulse to cross the optical barrier is that the input side, as I understand it, gets its power at the same time that the pulse should be generated, though I assume even hundreds of milliseconds of delay would be of no consequence. If this happens with plug-in of the USB cable, which is not too likely to produce a monotonic clean rise, extra circuitry is likely to be required to avoid unwanted multiple clocking. Then there is the issue of generating a pulse on sloppy loss of input power.

It took me about 30 seconds to find an optocoupler characterized at 100 µA.

 

It took me about 30 seconds to find an optocoupler characterized at 100 µA.

Terrific. In post #5, one was characterized at 70 µA. For the one that you found in ~30 s, what was the CTR at 100 µA? Was it enough to operate the relay? If so, then please don't keep it a secret, because that addresses the original question (irrespective of the additional and likely valid points that you raised).

 

CTR isn't an issue for a circuit like this unless absolute minimum parts count is critical.

The coupler I found had a CTR of about 0.1 or so at 100 µA, 10 µA or even 1 µA is gobs of current and very easily amplified in any number of simple ways. It could drive the base of single NPN which, if carefully selected, should easily give a gain of at least 50, so we're up to half a milliamp. Onward and upward with another stage, e.g. PNP emitter follower. The coupler output would drive a suitable N-channel FET with a gate pull-down resistor of several megs. It would be slow, but that just isn't an issue, either. It would probably still be an order of magnitude faster than the relay. If for some reason you needed fast rise and fall, a schmitt trigger, discrete or integrated circuit, could be used. The limitation in amplifying very low current (and 1 µA is not "very low") is distinguishing coupled current from coupler transistor leakage current, which shoudn't be a problem unless the temperature is high.

 

Hi ebp,
My understanding of how Danko's circuit works it that the current pulse comes from rapidly discharging C1. When power is first applied C1 charges via R1 (And R8 which we can ignore at this point.) The emitter of Q2 is at the same potential as the top of C1 The base of Q2 is at 2/3 of the 5 volt input (3.33 volt) When the emitter voltage reaches 3.33 + Vbe of Q2 Q2 starts to conduct at this point Q2 and Q1 latch on switching Q3 on dischaging C1 via R8 through the LED. So there is no current pulse taken from the supply. NOTE the base of Q3 is NOT connected to the top of C1, (This is the mistake I made when I first looked at the schematic.)

Les.

 

I shouldn't participate in threads I'm only paying token attention to.

I started writing my comments at 32, remembered I had something "important" * to do, came back to other threads then back here and finished my comments before even looking at Danko's circuit. His circuit would work very well for the TS's requirement where the longish time delay is of no consequence. It does require rather a lot of parts, but nothing that isn't really inexpensive.

[EDIT - my apologies to Raymond, I originally referred to him as Richard] Raymond found a very suitable optocoupler at #7, but I didn't look at the data sheet previously. It is characterized all the way down to 10 µA of input current - at -40°C to boot! Even it it had a CTR of only 0.1 at 10 µA input it would do the simple direct job nicely with some added current gain. The actual CTR at that input current is way higher. 10 µA for a week is 1.7 mA·H. I find it hard to believe that would even rise out of the uncertainly floor for power consumption of a phone. One pass through a tunnel where the phone screamed in its loudest voice to the nearest towers "Do not forsake meeeee!" would probably eat more energy. I'd actually probably run at at least 20 µA to have some margin above the lowest characterized current.

Hewlett-Packard opto division published a lot or really good stuff on optocouplers many years ago. I would advise anyone interested in learning about optocouplers to scour the web for the HP stuff. Motorola also published quite a lot of useful ap notes on opto (and tons more on other things).
~~~
* Actually, I remembered I wanted to place an Amazon order before someone else scooped the clearance item I wanted. I scored! - whereupon the strange but weird Amazon price adjuster raised the price of the remaining 5 of what I had ordered by a factor of 10. I've seen factors of 2 to 4 frequently, but never 10 before. For amusement, I've been watching another item on clearance that went from a price I don't remember to six thousand, seven thou, 8, 9, to, last time I looked ten thousand four hundred and some dollars ... and seven cents. The item is worth about $20. If anyone should discover the item I'm talking about, please don't notify Amazon. I needz my amuzemunks.

 

Hewlett-Packard opto division published a lot or really good stuff on optocouplers many years ago. I would advise anyone interested in learning about optocouplers to scour the web for the HP stuff. Motorola also published quite a lot of useful ap notes on opto (and tons more on other things).

That would later be Avago where things were arranged nicely. The data should be on the the Broadcom site well hidden after broadcom purchased them.

 

Avago - Yes, I couldn't remember the name - one of the earlier of the New Stupid Names (but not nearly as dumb as Keysight). But as you say, Avago knew how to organize a web site. I haven't looked at Broadcom's but it seemed to be a company with the attitude "we're much to important to be bothered with the likes of you."

There was a very old HP opto ap handbook that faded to oblivion, probably because it also had a lot of stuff on 7-segment displays, but it had some good stuff that I never saw in later publications.

 

There was a very old HP opto ap handbook that faded to oblivion, probably because it also had a lot of stuff on 7-segment displays, but it had some good stuff that I never saw in later publications.

This?
https://www.amazon.com/Opto-electronics-Applications-Manual-Hewlett-Packard-Company/dp/0070286051

 

Raymond (and my apologies to you for misnaming you at #36) - The one I was thinking of is Optocouplers and Fiber Optics Applications Handbook, published in 1986. Mostly it is a collection of individual ap notes, but includes other material such as articles originally published in EDN. Coincidentally, the very first thing in the book is Application Note 1002 Consideration of CTR Variations in Optocoupler Circuit Design.

Component manufacturers' ap notes can be a tremendous source for a free, if time consuming, education in electronics, especially about the "hooks" that can snag you in getting from paper designs to functioning, reliable real circuits. As your personal knowledge grows, they may be less likely to bring new understanding, but sometimes you can get a single point that makes the reading worthwhile.
(I once did an active power factor front end for a 1 kW SMPS, using a quite new Unitrode (now TI) controller. These are miserable things to cope with at the best of times. By the time I'd finished, the Unitrode ap note for the part was at revision G or H. I've never seen that many rev's for any other ap note. Ψ)

I haven't done anything out of the ordinary with conventional optocouplers for a long time. I have used some of the linear optocouplers in some industrial designs (cathodic protection power supplies/supply controllers). They're a reasonably simple solution to some serious problems and can be worth considering as an alternative to some of the very nice but rather expensive isolation amplifiers on the market.