I have a board with a 74HCT573 with 10K pull-ups on the inputs. This is connected to a sensor with an opto-isolator output (4N25) via 2m of cable. The resistor for the opto led is 10K on a 5V supply, so according to datasheet the CTR can be as low as 20% at that drive level. So the transistor output could be pulling as little as 80uA. I can see them asking me to increase my 10K pull-ups by a factor of >
6.25.
What is the limiting factor here? Surely you cannot keep reducing the current and expect the cct to work. Even the 500uA seems low to me.
The minimum current that can drive the CMOS as a pull up depends on the leakage current into the COS, and the off state leakage current of the opto. But it is possible that you cannot get very close to this limit, because there is a speed penalty to pay when the currents are small. The node capacitances still need to be charged up by the current. And that takes time.
Hi, Ian. You're right -- 500uA of LED current seems really small for an optoisolator. And the current transfer ratio for the optoisolator is specified at one level of drive current (typically If = 10mA). You'll also notice when looking at the datasheet:
formatting link
that current transfer ration (CTR) goes way down when If goes below a couple of mA (see Fig. 3 and Fig. 4, p.6 of the datasheet).
If I were looking at this, I'd suggest that the HCT573 can easily drive more than 1/2mA, and I'd look at bumping up that current by reducing the series resistor from the HCT573 to the opto input. The data sheet says it can drive 15 ea. LSTTL loads, so it can sink (or source) 6mA. Assuming you're sinking current to turn on the opto, have a 5V supply, and the opto LED takes up 1.7V when on, you should be able to use a 560 ohm series resistor quite easily.
formatting link
You're asking a really intelligent question. The answer is in the data sheets.
I wasnt clear in my original post. There are 2 boards involved here with 2m of cable inbetween. The 74HCT573 with 10k pull-ups is on my board (it is the one I have control over - but not my design) and the opto-isolator is on the other board (I have no control over this). The
500uA referred to is the current required to pull down my boards inputs, and I mentioned I thought this was on the low side. As the opto-isolator is being under-driven it cannot sink this current, so I can see them asking me to increase the pull-up value. This seems wrong to me, hence my question about limits.
If the opto cannot be expected to sink more than 80uA (you haven't disclosed the rates required and that is important to know, along with the details, I suppose, about what composes the 2m of cabling impedance), you need to be sure that your end can work correctly with that much capability. Your inputs require at least 500uA, I gather, mostly because of the existing 10k pull-ups on your end (5v/10k =
500uA, and ignoring any HCT input requirements.) That simply needs to change.
You could increase the pull-ups on your end and yes that may very well impact the supportable rates. If so, you need to find another way than just a simple modification of a pull-up to get the job done, I'd imagine. Modifying a simple passive isn't necessarily the right solution, if you are trapped by other requirements, and you should be prepared to find a route that is correct for the application and not simply limited to finding a resistor value you aren't comfortable with. In other words, perhaps consider using an active front end.
John's point about the node capacitance is important and this will include everything from the 4N25 transistor itself, though the cable, and onto your input circuit, whatever that is. The 4N25 will have to drive it all with what it has in place at rates that work for you. You may need to think closely about this.
I gather that the 10k pull-up on your side is the only explicit pull-up that the 4N25 output transistor sinks against and that you don't have access to the transistor base at the 4N25? My guess is that you aren't doing rates much in excess of 1kHz (and I'm wondering about meeting your HCT input thresholds as it is.) Increasing that resistor will probably demolish your rate, my guess. If you can only expect about 80uA, then you might consider using a PNP on your end before the HCT input, with appropriate resistors. I think you should be able to acheive the same rates on that current. Also, perhaps, consider a current steering arrangement using two BJTs.
Actually, now that I think about it, you probably don't need a BJT to develop the steered current source. Just your larger valued pull-up resistor (say, 56k?) So one cheap NPN BJT, your increased value pull-up, and a collector resistor for the NPN (6.8k-10k?) would be enough, I'd imagine.
You have lost me here, I understood the PNP method, but not this. Rates are not an issue by the way (apart from 500ns max rise/fall time into 74HCT573), you can consider it DC.
I am going to try to get the manufacturer of the sensor to put a reasonable drive current into the emitter of the opto-coupler!! Otherwise it is a mod to our board similar to your description - nice mod, all 8 inputs are used!
Probably my fault. I'm just a hobbyist and see things in my own way, using words that may not be the best to use.
What I meant is simply this:
In the above case, the pull-up is now 56k, which is about the current you are talking about, I think. Since the NPN base and opto collector node itself stays within 6/10ths of a volt of ground, the current through R2 above stays roughly the same. But the current either is sunk by the opto's collector or else by the NPN base-emitter. In other words, the current doesn't change much but is 'steered' one way or another.
Slow. That's good. That 56k or 68k pull-up on your side, if the rates are truly close to DC (under 100Hz or so) and you don't have much voltage pickup on the cable, might be okay too.
Or either the PNP being pulled active or else the NPN steering arrangement will be fine, too, I think. But I can see that you want the least mod possible. And under the circumstances, just increasing your pull-up to 56k seems fine.
Ah... design push-back!
Well, I'm not so sure you need to mod it that much, given your comment about the signal being close to DC. Probably, you _can_ get by with just a resistor change, though I've no idea what your 2m of cable winds though and may pick up along the way.
Hi, Ian. Of course, as you increase the pullup resistance, you have more susceptibility to noise. Your vendor has put you between a rock and a hard place by choosing the 1960s refugee 4N25. There are better, faster optocoupler transistors with much higher CTRs.
In addition to Mr. Kirwan's great advice, I'd like to add that you may wish to use a 74HC input instead of an HCT. The input will be much more tolerant of a not-quite-saturated 4N25 (Vl(max) = 0.3*Vcc or 1.5V instead of 0.8V). That might help a bit.
Also, expecting fast enough transition times out of the 4N25 (especially when turning off) might be a little much. You may want to have a schmitt trigger buffer at the input (like a 74HC14), then you can use a 10K pullup along with a small (100pF) cap to GND, which will help a lot with noise immunity. This shouldn't slow down your input more than a microsecond, which you imply in your conversation with Mr. Kirwan is probably OK. Your input could then look something like this (view in fixed font or M$ Notepad):
ElectronDepot website is not affiliated with any of the manufacturers or service providers discussed here.
All logos and trade names are the property of their respective owners.