Octal Buffer or Latch with Schmitt Trigger inputs

[John Anderson]

Hi,

I just reviewed the forum topics searching for this but I think it has not been asked yet. If it did, please let me know. Thank you!

I am designing a PCB and need to interface some analog sensors signal (a lot of them!) to a digital reading bus. The input signal can has some noise, so it is pre-filtered through a basic RC filter to debounce it.

Then I need to input this pre-filtered signal into an octal buffer (or permanent input enabled latch) with 3-state output (i.e. 74HC244/74HC573...) in order to read the desired sensors. The obvious question is that I need to add some hysteresis to these inputs in order to avoid false detection.

I googled around and found that there is the 74HC7541 IC, which seems exactly what I find, but unfortunately it is not available in any of the stores here (Barcelona) (it seems a 2016-newer one?). So I was wondering if exists another option analogue to this, or maybe an alternative solution (as the number of input signals is huge, cost and room are importants, so the number of components should the minimum).

Thank you so much in advance!

[edavid]

Is it a 5V design?  Can you find a 74LS244 or 74LS245?

[John Anderson]

Hi David,

Yes, its a 5V design, and yes the 74LS244 are still available, but at a very expensive cost (~x4). I wonder it is probably related with the fact that the LS family is obsolete nowadays, so it will be hard to find replacement parts in the future, isn't it?

However, after this octal buffer the outputs will be connected to a 74HC573 octal latch. In case I finally could find and use LS244 parts, is there any problem interfacing the 2 different families (LS + HC)?

Thank you again!

[edavid]

Yes, its a 5V design, and yes the 74LS244 are still available, but at a very expensive cost (~x4). I wonder it is probably related with the fact that the LS family is obsolete nowadays, so it will be hard to find replacement parts in the future, isn't it?

OK, perhaps you could use 74LS541 now, and switch to the 74HC7541 when you can get some.

However, after this octal buffer the outputs will be connected to a 74HC573 octal latch. In case I finally could find and use LS244 parts, is there any problem interfacing the 2 different families (LS + HC)?

You would have to use 74HCT573.

[John Anderson]

OK, I just simulated what you proposed but I think this is not working for me.

The reason: due to the presence of the RC pre-filter, I need that the buffer IC doesn't supply nor sink current out of the input pins, because the small capacitor has to be recharged through the resistor to obtain the proper filtering timming.

Using 74HC244 this is OK, as the Datasheet states that the Input Leakage Current (I hope this is the correct parameter to review) is max. 1uA, while with the 74LS244 this can be up to 2 mA, so with the LS, the capacitor is recharged through the IC instead of the R. And with the 74LS541 it seems it will happen the same, as the figures are equal to the 244.

I attach the simplified schematic.

Any other idea? Thank you again!

[edavid]

Can you get CD40106/74C14/74HC14 hex Schmitt triggers?

[David Hess]

The 74x244 does not have schmitt trigger inputs.  I did not even know there were octal and higher schmitt trigger buffers until you mentioned it.

A pair of 74HC132s can be used to make 8 schmitt trigger inverters or like edavid says, a 74HC14 is 6 schmitt trigger inverters.

[Zero999]

What about adding resistors to the 74HC244 to give it Schmitt trigger inputs?

[Alex Eisenhut]

What kind of analog sensor needs debouncing?

[Zero999]

What about adding resistors to the 74HC244 to give it Schmitt trigger inputs?

Just to clarify.

The hysteresis voltage will be equal to Vdd*Rin/Rf and will be centred around half the supply voltage.

Then the gate is in its high impedance state, the output impedance will be no less than Rin+Rf.

The resistor values will need to be reduced if it needs to work at high frequencies.

[edavid]

The 74x244 does not have schmitt trigger inputs.  I did not even know there were octal and higher schmitt trigger buffers until you mentioned it.

It depends on the x.  The bipolar versions do have Schmitt triggers.

http://www.ti.com/lit/ds/symlink/sn74ls244.pdf

[David Hess]

The 74x244 does not have schmitt trigger inputs.  I did not even know there were octal and higher schmitt trigger buffers until you mentioned it.

It depends on the x.  The bipolar versions do have Schmitt triggers.

http://www.ti.com/lit/ds/symlink/sn74ls244.pdf

Huh, I missed that for 30+ years.  I checked and the Fairchild and Hitachi parts have Schmitt inputs also.

[John Anderson]

Hi,

OK, so using the Hex-Inverters (74HC14) the problem is the room in the PCB. There will be 32 sensors connected to it (with all the individual circuitry and connectors), so it will represent 6 additional "big" ICs (I am using DIP tech) for which I have no room. Even worse with 74HC132.

The sensors are actually photoresistors (in fact 2N2222A and R4 in the schematic represents the photoresisor). I use that sensors to obtain a clean reading of presence detection, and wait for about 0,5s that no further detection is made before giving a no-presence state. So this is why I need debouncing of the input signal.

I tried to simulate what Hero999 suggested (thank you a lot, this is exactly what I was looking for). It worked great for a single "sensor/channel" using 1 order of magnitude higher values (10M/100M), but only if the 74HC244 is always output enabled. As soon as the output is in high impedance (which I need because there will be 3 more sensors connected to that output in order to cascade-address the reading to the 74HC573) the functionallity is wrong, because in low input state, the output is forced to low, which results in a short with the high of another output-enabled channel. I attached the live values of the simulation of a low input state (sensor ON).

Thank you again!

[kony]

Why not to use just multiplexed ADC and do the filtering later in SW? Much more flexible, trims much of the ballast in schematic as well.

[Zero999]

I tried to simulate what Hero999 suggested (thank you a lot, this is exactly what I was looking for). It worked great for a single "sensor/channel" using 1 order of magnitude higher values (10M/100M), but only if the 74HC244 is always output enabled. As soon as the output is in high impedance (which I need because there will be 3 more sensors connected to that output in order to cascade-address the reading to the 74HC573) the functionallity is wrong, because in low input state, the output is forced to low, which results in a short with the high of another output-enabled channel. I attached the live values of the simulation of a low input state (sensor ON).

I don't understand. When the gate is set to high impedance mode, it effectively disappears, leaving the sensor connected to whatever is on the output via Rin+Rf, which is 110M in your case and will be too high to matter. If the output is being forced low, then the gate isn't set to its high impedance state but low and the same will happen, irrespective of whether Rin & Rf are there or not.

[edavid]

OK, so using the Hex-Inverters (74HC14) the problem is the room in the PCB. There will be 32 sensors connected to it (with all the individual circuitry and connectors), so it will represent 6 additional "big" ICs (I am using DIP tech) for which I have no room. Even worse with 74HC132.

If you don't have room for a 14 pin DIP, how can you have room for 12 TH resistors to replace it?

[John Anderson]

OK, sorry I understand it now. I was a little bit obfuscated thinking on the fact that the Rf does not 'phisically isolates' the input side from the output one.

The output value is forced to whatever state of the IC which is output enabled. So even if there is a low state in the output side through the Rf resistor, thanks to its higher value this will be forced when one of the 244 ICs will be output enabled.

Thank you again for your time.

Certainly, at the end of the day, adding 2 resistors per sensor is pretty much as using the 74HC14. I will reconsider room space and see if it is worth the flexibility of using discrete resistors than a single IC. Thanks edavid again!

Best regards.

[Zero999]

Personally I would go SMT so the correct IC can be used in the first place.

Failing that, what about using some SMT resistors or those small ¹/₈W through hole resistors? The good thing about the 74HC244 is the inputs and outputs are on opposite sides of the DIL package so RF can go on the other side of the board.

There's also an advantage of using resistors to get hysteresis: it becomes predicable and other resistors can be added to move the switching threshold up or down.

[hooverphonique]

The hysteresis voltage will be equal to Vdd*Rin/Rf and will be centred around half the supply voltage.

Then the gate is in its high impedance state, the output impedance will be no less than Rin+Rf.

The resistor values will need to be reduced if it needs to work at high frequencies.

I suppose this "trick" would also work with a transparent latch while it is open (and the output is enabled), so you could have schmidt-trigger inputs on e.g. 74LVC573 ?

[Kleinstein]

There is the relatively new 74HCS series from TI - essentially like 74HC with schmidet trigger function also for other functions.  At least Mouser has 74HCS245 as octal buffers and maybe a few others.