Do old 74-series logic NOT have Schmitt-Trigger inputs?

[jolshefsky]

I was working on a project that I think can be completed entirely with one quad NAND gate. I first wanted to just test that I could make a switchable-frequency ~1KHz oscillator, so I dug around and found I had a MC74HC14AN hex inverter with a 9630 date code—good enough for the oscillator.

I wired up an RC low-pass filter from one inverter's output-to-input: output-resistor-input; input-capacitor-ground. I also grounded the remaining inputs. No matter what filter I set up, I was getting an oscillation in the >10MHz range. The input was not swinging more than a few millivolts away from VCC/2, basically ignoring the RC time constant. This implied the device did not have Schmitt-triggered inputs.

1996 seems like a time when all logic gates would have Schmitt-triggered inputs ... am I wrong?

[MK14]

One thing that comes to mind is, did you put a decoupling capacitor on the power rails ?
Ideally, close to the actual ICs power pins.

[capt bullshot]

The HC14 definitely has a Schmitt-Trigger input, that is what it's supposed to. Others don't, even today they don't.
Anyway, if your oscillator doesn't work with the '14, something else must be wrong - bad decoupling or breadboard comes to my mind.

[spec]

Crossed posts 

Hi jolshefsky,

The 74HC14 has Schmidt trigger inputs on all six of its not gates. Few other logic chips have Schmidt inputs though. The 74HC132 quad two input nand gate is the other common logic chip to have Schmidt inputs.

Mk14 just beat me to it: you need a 100nf X7R ceramic capacitor between the supply pins.

Also all the other 5 gates need to have their inputs connected to either logic 0 or logic 1, or the whole chip may get into a state- that may be happening. The chip may even destroy itself.

Bear in mind that 74HC is very fast, so a good layout with short wires/ traces is essential.

Are you using the integrating oscillator circuit or the differentiating oscillator circuit? The former is well behaved and the latter is not. Your best bet is to post a schematic and a picture of your layout so we can have a look.

[eliocor]

Here it is: dual QUAD NAND SCHMITT TRIGGER port, the 7413...
https://www.unicornelectronics.com/ftp/Data%20Sheets/7413.pdf 
connect together all inputs of one gate and you'll get an inverter for your oscillator.

[tggzzz]

I was working on a project that I think can be completed entirely with one quad NAND gate. I first wanted to just test that I could make a switchable-frequency ~1KHz oscillator, so I dug around and found I had a MC74HC14AN hex inverter with a 9630 date code—good enough for the oscillator.

I wired up an RC low-pass filter from one inverter's output-to-input: output-resistor-input; input-capacitor-ground. I also grounded the remaining inputs. No matter what filter I set up, I was getting an oscillation in the >10MHz range. The input was not swinging more than a few millivolts away from VCC/2, basically ignoring the RC time constant. This implied the device did not have Schmitt-triggered inputs.

1996 seems like a time when all logic gates would have Schmitt-triggered inputs ... am I wrong?

Firstly HC is not TTL - it is CMOS rather than bipolar. TTL families are 7400, 74L00, 74LS00, 74S00, 74LS00, 74H00, 74ALS00, 74F00 and probably a few others I've forgotten about.

Secondly the CMOS variants typically come in two varieties: *C and *CT. The former's V_IH and V_IL are symmetric, as befits a CMOS logic. The *CT variants have non-symmetric V_IH and V_IL that are compatible with TTL output.

Thirdly, there are so many devices within one family and across different families that the only way to be sure whether there are schmitt trigger inputs is to RTFDS.

[SiliconWizard]

Showing us your exact schematic and pictures of your setup would probably help.
Yes you can easily design an RC oscillator with a 74HC14, up to a few tens of MHz depending on the power supply voltage.
But without actually seeing anything of what you did, this is just going to be endless guesswork.

[spec]

I was working on a project that I think can be completed entirely with one quad NAND gate. I first wanted to just test that I could make a switchable-frequency ~1KHz oscillator, so I dug around and found I had a MC74HC14AN hex inverter with a 9630 date code—good enough for the oscillator.

I wired up an RC low-pass filter from one inverter's output-to-input: output-resistor-input; input-capacitor-ground. I also grounded the remaining inputs. No matter what filter I set up, I was getting an oscillation in the >10MHz range. The input was not swinging more than a few millivolts away from VCC/2, basically ignoring the RC time constant. This implied the device did not have Schmitt-triggered inputs.

1996 seems like a time when all logic gates would have Schmitt-triggered inputs ... am I wrong?

Firstly HC is not TTL - it is CMOS rather than bipolar. TTL families are 7400, 74L00, 74LS00, 74S00, 74LS00, 74H00, 74ALS00, 74F00 and probably a few others I've forgotten about.

Secondly the CMOS variants typically come in two varieties: *C and *CT. The former's V_IH and V_IL are symmetric, as befits a CMOS logic. The *CT variants have non-symmetric V_IH and V_IL that are compatible with TTL output.

Thirdly, there are so many devices within one family and across different families that the only way to be sure whether there are schmitt trigger inputs is to RTFDS.

Not sure of the objective of this post. The OP has said that he has an MC74HC14AN (Motorola) and all 74HC14As, old and new, have Schmitt inputs.

[tggzzz]

I was working on a project that I think can be completed entirely with one quad NAND gate. I first wanted to just test that I could make a switchable-frequency ~1KHz oscillator, so I dug around and found I had a MC74HC14AN hex inverter with a 9630 date code—good enough for the oscillator.

I wired up an RC low-pass filter from one inverter's output-to-input: output-resistor-input; input-capacitor-ground. I also grounded the remaining inputs. No matter what filter I set up, I was getting an oscillation in the >10MHz range. The input was not swinging more than a few millivolts away from VCC/2, basically ignoring the RC time constant. This implied the device did not have Schmitt-triggered inputs.

1996 seems like a time when all logic gates would have Schmitt-triggered inputs ... am I wrong?

Firstly HC is not TTL - it is CMOS rather than bipolar. TTL families are 7400, 74L00, 74LS00, 74S00, 74LS00, 74H00, 74ALS00, 74F00 and probably a few others I've forgotten about.

Secondly the CMOS variants typically come in two varieties: *C and *CT. The former's V_IH and V_IL are symmetric, as befits a CMOS logic. The *CT variants have non-symmetric V_IH and V_IL that are compatible with TTL output.

Thirdly, there are so many devices within one family and across different families that the only way to be sure whether there are schmitt trigger inputs is to RTFDS.

Not sure of the objective of this post. The OP has said that he has an MC74HC14AN (Motorola) and all 74HC14As, old and new, have Schmitt inputs.

It clearly answers the question in the OP's last sentence, plus provides extra clarification as to why he is wrong.

[Bassman59]

I was working on a project that I think can be completed entirely with one quad NAND gate. I first wanted to just test that I could make a switchable-frequency ~1KHz oscillator, so I dug around and found I had a MC74HC14AN hex inverter with a 9630 date code—good enough for the oscillator.

I wired up an RC low-pass filter from one inverter's output-to-input: output-resistor-input; input-capacitor-ground. I also grounded the remaining inputs. No matter what filter I set up, I was getting an oscillation in the >10MHz range. The input was not swinging more than a few millivolts away from VCC/2, basically ignoring the RC time constant. This implied the device did not have Schmitt-triggered inputs.

1996 seems like a time when all logic gates would have Schmitt-triggered inputs ... am I wrong?

74-series logic devices, of any family, do not have Schmitt trigger inputs except where explicitly noted (74xx14 and 74xx17 come to mind, there are a few others).

[jolshefsky]

Showing us your exact schematic and pictures of your setup would probably help.

Schematic attached. As it turned out, the chip was bad ... who knew, with it sitting in my junk bin.   

I finally went and tested all the gates and it seems the gates on pins 1-6 are not working right but the pins 8-13 gates are, so I tried it on gate 13-12 instead and it worked fine. D'oh!

(Edit: for what it's worth, I just connected the gates to one another and tied the first to a power rail because it's quick to use those 1-pin jumpers on a breadboard.)

[capt bullshot]

breadboard

I don't think that half of the gates in the chip were bad and the others were OK. That would be a quite rare kind of failure mode. I rather believe that the power supply routing and the negative connection of the 1uF cap was different, depending on which gates you used. Some reasoning: These gates switch their outputs in maybe 1 ... 2ns from High to Low and vice versa. This causes a high current spike drawn through the power pins, this in turn causes a voltage drop along the supply wire. Depending on how long the wires are, and where exactly the 1uF cap is connected, this voltage drop causes enough feedback to the input of the gate the make it switch back, in turn causing another current spike, causing another voltage drop (same as before, but into the other direction), causing the gate to switch, ... You get the picture, I hope. This results in the 10Mhz oscillation of small amplitude. Keep your power wires short, use a decoupling cap close to the IC, connect the 1uF cap directly at the GND Pin if the IC - it should work better then.

IMO the chip would work in a proper environment using a decoupling cap shortly tied to the power pins. Most probably some parasitic inductance and/or coupling caused the unwanted oscillations.

[tggzzz]

breadboard

I don't think that half of the gates in the chip were bad and the others were OK. That would be a quite rare kind of failure mode. I rather believe that the power supply routing and the negative connection of the 1uF cap was different, depending on which gates you used. Some reasoning: These gates switch their outputs in maybe 1 ... 2ns from High to Low and vice versa. This causes a high current spike drawn through the power pins, this in turn causes a voltage drop along the supply wire. Depending on how long the wires are, and where exactly the 1uF cap is connected, this voltage drop causes enough feedback to the input of the gate the make it switch back, in turn causing another current spike, causing another voltage drop (same as before, but into the other direction), causing the gate to switch, ... You get the picture, I hope. This results in the 10Mhz oscillation of small amplitude. Keep your power wires short, use a decoupling cap close to the IC, connect the 1uF cap directly at the GND Pin if the IC - it should work better then.

IMO the chip would work in a proper environment using a decoupling cap shortly tied to the power pins. Most probably some parasitic inductance and/or coupling caused the unwanted oscillations.

Just so.

With solderless breadboards you spend more time debugging the breadboard than your circuit. For the reasons why, and cheap easy effective alternatives, see the links in https://entertaininghacks.wordpress.com/2020/07/22/prototyping-circuits-easy-cheap-fast-reliable-techniques/

[SiliconWizard]

True that breadboards can give you a few headaches.

That said, building a simple 1 kHz RC oscillator with a CMOS Schmitt inverter should not be that tricky to build on a breadboard.
I haven't seen any bypass cap in the OP's schematic, so adding one close to the IC's Vdd pin should do the trick (unless the IC is indeed toast).

Haven't used a solderless breadboard in ages (I exclusively use SMD or through-hole prototyping boards when needed), but I remember using those long ago while learning and I had certainly built much more complex digital circuits on those than a simple low-freq oscillator, so I'd be curious to see a picture of the actual setup.
And don't forget to add a bypass cap.

[tggzzz]

True that breadboards can give you a few headaches.

That said, building a simple 1 kHz RC oscillator with a CMOS Schmitt inverter should not be that tricky to build on a breadboard.
I haven't seen any bypass cap in the OP's schematic, so adding one close to the IC's Vdd pin should do the trick (unless the IC is indeed toast).

Haven't used a solderless breadboard in ages (I exclusively use SMD or through-hole prototyping boards when needed), but I remember using those long ago while learning and I had certainly built much more complex digital circuits on those than a simple low-freq oscillator, so I'd be curious to see a picture of the actual setup.
And don't forget to add a bypass cap.

The frequency is, of course, of little importance. What matters is the rise/fall time. For measurements confirming basic fourier transform theory, see https://entertaininghacks.wordpress.com/2018/05/08/digital-signal-integrity-and-bandwidth-signals-risetime-is-important-period-is-irrelevant/

While the OP appears to have used ancient CMOS technology that would probably work on a solderless breadboard (with decoupling caps!), modern CMOS technology certainly wouldn't. For an example of 250ps risetimes, see https://www.eevblog.com/forum/testgear/show-us-your-square-wave/msg1902941/#msg1902941 That shows 5V/100ohm = 50mA switching in 0.25ns, i.e. 200MA/s (not mA/s). Plug that into v=L^di/_dt and the voltages are significant.

[SiliconWizard]

The frequency is, of course, of little importance. What matters is the rise/fall time.

The "frequency" of the digital signal (as in the inverse of the period), obviously not. But rising and falling edges are of course just high-frequency components in disguise, so depends on what you mean by frequency. From an analog standpoint, digital signals with fast rising and falling edges are high-frequency signals, as you pointed out.

Granted the fact it's supposed to be a 1 kHz oscillator doesn't in itself matter much here (which is the part you may have misinterpreted) compared to the rise and fall times using even a 74HC14, although this kind of RC oscillator obviously severly limits the rise and fall time of the input signal of the first inverter, and also of the output (in a much less severe way of course) due to the RC loading the output (even slightly). The next chained inverters would admittedly not suffer from this.
So in this particular case, I still think the fact it's on a breadboard doesn't explain the issue he described, except maybe the lack of bypass (which could cause even more problems on a breadboard due to the inductive nature of the wiring) but again we haven't seen the setup and we may see something more obvious than suspected if we did.

But yes, for any fast circuits those breadboards should be avoided at all costs. (I don't consider the 74HC series as "fast" ICs and have never had any particular issue with those back when I was using solderless breadboards.)

The fact the IC could have been partially fried is not completely weird IMO (even if it's uncommon) as the apparently "fried" gates were not at random but all located on the same side of the IC (pins 1-6 if I got it right), so the chip may have suffered from some ESD on this side due to the way it was stored? The OP could try and test those gates with DC signals at the inputs and a load at the output, just to see if they appear to be dead or not. That could add to the learning lesson.

[spec]

Some interesting discussions on this thread.

Can I just say that the integrating multivibrator, which the OP is using, is one of the most reliable and well behaved multivibrators going. It is ubiquitous. It uses the same principal as many successful  timers, including the 555. Unlike differentiating multivibrators, which can get into a third state (both transistors conducting), like the classic two-transistor collector/base coupled multivibrator, the integrating multivibrator does not have this problem. Also, in contrast to the differentiating multivibrator, it has no chance of exceeding the voltage rating of its active elements.

The frequency of oscillation is controlled by the time of charging the capacitor up to and down, via the feedback resistor, to the input Schmidt high and low threshold voltages respectively- nothing to do with fast edges. If you put a scope on the input you will see an approximately symmetrical saw tooth waveform. The formula for the frequency, and associated k factor graph, is given on page twelve of this datasheet: https://assets.nexperia.com/documents/data-sheet/74HC_HCT14.pdf

A 100nF X7R ceramic capacitor needs to be connected directly across the supply pins of the 74HC14 chip to prevent it from oscillating.

An aluminum electrolytic timing capacitor is not advisable.  A tantalum electrolytic or solid capacitor is far better.  A polypropylene capacitor would be the Rolls Royce choice.

About the failure mode of the OP's 74HC14 chip, I hate to say this, but far from being uncommon, it is very common. I think the reason is that the individual gates rely on common circuitry for bias etc. The gates are also fabricated on a common substrate. Therefor, when one gate fails it takes some of the others with it. I have seen this many times and not just with the 74HC14. The other thing is that the 74HC14 was often used to interface to a bus or other difficult load, which gave them a hard time. Another cause of multiple gate failure is uncontrolled parasitic oscillation, where the chip overheats and destroys itself due to the high frequency. This can happen when the inputs and outputs are left floating.

[SiliconWizard]

As a quick sample simulation showing the effect of parasitics series inductance on power supply lines and bypass capacitors, attached is an LTSpice file.
Obviously this is a simplified model but this shows the principles.

For instance, remove the bypass cap C2 and see what happens to the oscillator.