Author Topic: Schmitt disappointment  (Read 4394 times)

0 Members and 1 Guest are viewing this topic.

Offline PerranOakTopic starter

  • Frequent Contributor
  • **
  • Posts: 552
  • Country: gb
Schmitt disappointment
« on: October 01, 2018, 04:57:24 pm »
I have a Schmitt trigger (74HC14) with an input of: square wave, 1kHz, 5Vpp, 2.5V offset.

It triggers as expected but the overshoot is nearly 2V.

OK, I have it all in a breadboard with unshielded jumpers but 2V!

Is this normal?
You can release yourself but the only way to go is down!
RJD
 

Offline langwadt

  • Super Contributor
  • ***
  • Posts: 4857
  • Country: dk
Re: Schmitt disappointment
« Reply #1 on: October 01, 2018, 05:11:08 pm »
I have a Schmitt trigger (74HC14) with an input of: square wave, 1kHz, 5Vpp, 2.5V offset.

It triggers as expected but the overshoot is nearly 2V.

OK, I have it all in a breadboard with unshielded jumpers but 2V!

Is this normal?

how are you measuring it?
 

Offline schmitt trigger

  • Super Contributor
  • ***
  • Posts: 2431
  • Country: mx
Re: Schmitt disappointment
« Reply #2 on: October 01, 2018, 05:18:43 pm »
Without you providing any additional details, and without seeing your actual circuit or layout; I would say that the overshoot is caused by:
  • poor grounding and supply routing
  • lack of adequate capacitor decoupling
  • attaching the scope probe to an incorrect ground location

I mean, it is after all, only 1Khz.
To a logic circuit, any logic type, this is glacial speed.
 

Offline PerranOakTopic starter

  • Frequent Contributor
  • **
  • Posts: 552
  • Country: gb
Re: Schmitt disappointment
« Reply #3 on: October 01, 2018, 05:38:44 pm »
Thank you both.

I am looking at the input and output traces on a scope.

It is the most simple layout. One trigger has the sig gen input (scope channel 1) with a 1K resistor to ground on the output (scope channel 2). All other inputs are to Vdd. 5V battery pack as power with 100uF cap across it. 100nF cap across Vdd to ground - can't get is very close as Vdd and ground are at opposite ends of the chip and I keep shorting things if I try to connect the cap close to them!

It is a rats' nest of jumpers and the (10x) scope leads and sig gen leads all terminate in jumpers into the breadboard.

No matter what frequency (1kHz to 10MHz) the overshoot is remarkably stable!

Cheers.
You can release yourself but the only way to go is down!
RJD
 

Offline schmitt trigger

  • Super Contributor
  • ***
  • Posts: 2431
  • Country: mx
Re: Schmitt disappointment
« Reply #4 on: October 01, 2018, 05:40:25 pm »
Scope photos?
Breadboard photos?
 
The following users thanked this post: nugglix

Offline Mr. Scram

  • Super Contributor
  • ***
  • Posts: 9821
  • Country: 00
  • Display aficionado
Re: Schmitt disappointment
« Reply #5 on: October 01, 2018, 06:35:39 pm »
Did you try a different chip, preferably from another batch?
 

Online wraper

  • Supporter
  • ****
  • Posts: 17952
  • Country: lv
Re: Schmitt disappointment
« Reply #6 on: October 01, 2018, 06:41:29 pm »
Crap circuit, carp probing. If you wan't to look at sharp edges, you need good circuit with solid ground and no ground loop at probe of the scope. Like on the top picture or at least ground spring included with the probe.



 
The following users thanked this post: nugglix

Offline abraxa

  • Frequent Contributor
  • **
  • Posts: 377
  • Country: de
  • Sigrok associate
Re: Schmitt disappointment
« Reply #7 on: October 01, 2018, 06:57:15 pm »
Without you providing any additional details, and without seeing your actual circuit or layout; I would say that the overshoot is caused by:
  • poor grounding and supply routing
  • lack of adequate capacitor decoupling
  • attaching the scope probe to an incorrect ground location

I mean, it is after all, only 1Khz.
To a logic circuit, any logic type, this is glacial speed.

While I agree with your assessment of the situation at hand, I'd just like to point out that a squave wave frequency of 1 kHz can still have tons of high-frequency components if the rise and fall times of the edges are short. I'd say that this is the case here.
 
The following users thanked this post: wraper

Online wraper

  • Supporter
  • ****
  • Posts: 17952
  • Country: lv
Re: Schmitt disappointment
« Reply #8 on: October 01, 2018, 07:02:26 pm »
Without you providing any additional details, and without seeing your actual circuit or layout; I would say that the overshoot is caused by:
  • poor grounding and supply routing
  • lack of adequate capacitor decoupling
  • attaching the scope probe to an incorrect ground location

I mean, it is after all, only 1Khz.
To a logic circuit, any logic type, this is glacial speed.

While I agree with your assessment of the situation at hand, I'd just like to point out that a squave wave frequency of 1 kHz can still have tons of high-frequency components if the rise and fall times of the edges are short. I'd say that this is the case here.
Yep, frequency does not mean much in this case. 1 kHz square wave can also have 1 picosecond rise time.
 

Offline 2N3055

  • Super Contributor
  • ***
  • Posts: 7462
  • Country: hr
Re: Schmitt disappointment
« Reply #9 on: October 01, 2018, 08:10:50 pm »
74HC14 without lot of capacitance on it's output will easily have edges faster than 10 nsec. That is frequency spectrum up to 40 -50 MHz....
"Just hard work is not enough - it must be applied sensibly."
Dr. Richard W. Hamming
 

Offline sdouble

  • Frequent Contributor
  • **
  • Posts: 280
  • Country: fr
Re: Schmitt disappointment
« Reply #10 on: October 01, 2018, 09:27:04 pm »
you could get rise time of 5-6 ns... so 100 MHz harmonics  should be considered.
 
The following users thanked this post: 2N3055

Offline 2N3055

  • Super Contributor
  • ***
  • Posts: 7462
  • Country: hr
Re: Schmitt disappointment
« Reply #11 on: October 01, 2018, 10:19:18 pm »
you could get rise time of 5-6 ns... so 100 MHz harmonics  should be considered.
You are right..I was going by datasheet value of avg 7ns... I had some chips from some manufacturers going 4ns on rise time without much capacity on output...
Anyways, moral of the story is that edges define frequency spectrum contents, not repetition frequency... And can be scary fast even with plain HC chips and 1 KHz clock..
"Just hard work is not enough - it must be applied sensibly."
Dr. Richard W. Hamming
 

Offline piguy101

  • Regular Contributor
  • *
  • Posts: 51
  • Country: us
Re: Schmitt disappointment
« Reply #12 on: October 02, 2018, 12:37:46 am »
Could the overshoot be due to improper probe compensation? Most scopes have a 1 kHz output waveform on where you can adjust the trimmer capacitor on the side of the probe in order to show a perfect square wave for a good input.
 

Offline macboy

  • Super Contributor
  • ***
  • Posts: 2309
  • Country: ca
Re: Schmitt disappointment
« Reply #13 on: October 02, 2018, 02:46:32 pm »
Could the overshoot be due to improper probe compensation? Most scopes have a 1 kHz output waveform on where you can adjust the trimmer capacitor on the side of the probe in order to show a perfect square wave for a good input.
This is possible, but the effect would be lost much above 1 kHz. At 100 kHz, the amplitude would be simply appear to be too high (in the case of undercompensated).

More likely, the overshoot is caused by the fast edge exciting the LC circuit formed by the probe tip capacitance (~ 8 to 20 pF) and the several inch long ground lead (~ 200 nH self-inductance).  The resonance falls in the order of 100 MHz, maybe quite a bit lower for cheap probes with high capacitance and longer ground leads. This causes overshoot and/or ringing on the observed edge. The solution is a very short ground lead to push the resonance frequency up beyond the capability of the scope and probe. If your probes didn't come with ground springs you can make your own.  A homemade "Low-Z" probe can also be very effective if you have a 50 ohm scope input and the circuit can handle a 500 or 1000 ohm resistive load (a logic gate can).
 

Offline PerranOakTopic starter

  • Frequent Contributor
  • **
  • Posts: 552
  • Country: gb
Re: Schmitt disappointment
« Reply #14 on: October 03, 2018, 05:22:00 pm »
Thank you all very much.

It's the same on every chip.

The scope picture with the two traces shows the input (yellow) and the output (blue) overshoot and ringing.
note: the zero base is different for each to get the whole traces on.

The scope picture with one trace is when I completely disconnected the input probe and jumpers and put as short a ground connection as I could.
(I wondered what those springs were for but now I can't find them!)

The final picture is the rats' nest!  :-[

So, this is a common problem with fast switching things is it? Is it only an artifact of the scope rather than a "problem" with the circuit?
You can release yourself but the only way to go is down!
RJD
 

Offline Zero999

  • Super Contributor
  • ***
  • Posts: 20357
  • Country: gb
  • 0999
Re: Schmitt disappointment
« Reply #15 on: October 03, 2018, 06:12:46 pm »
It's the physical layout of the circuit which is the problem. The long leads have inductance, which resonates with the parasitic capacitances, generating relatively high voltages.

Build the circuit on a strip-board with as shorter connections as possible, the supply decoupling capacitor right next to the supply pins and use the proper probing technique described above and it will be much better.
 

Offline sdouble

  • Frequent Contributor
  • **
  • Posts: 280
  • Country: fr
Re: Schmitt disappointment
« Reply #16 on: October 03, 2018, 07:05:16 pm »
yep,
a breadboard like taht one is not suitable for relatively high frequency circuitry.
I can't see low ESR/ESL caps close to the chip. Are they any ?
 

Offline abraxa

  • Frequent Contributor
  • **
  • Posts: 377
  • Country: de
  • Sigrok associate
Re: Schmitt disappointment
« Reply #17 on: October 03, 2018, 09:27:23 pm »
So, this is a common problem with fast switching things is it? Is it only an artifact of the scope rather than a "problem" with the circuit?

Yes, all switching things - be it due to fast switching or a high signal frequency (which naturally requires short fall/rise times, i.e. fast switching).
The ringing is a very real property of your circuit, the scope's inadequate probing merely exaggerates the effect.
 

Offline Zero999

  • Super Contributor
  • ***
  • Posts: 20357
  • Country: gb
  • 0999
Re: Schmitt disappointment
« Reply #18 on: October 04, 2018, 09:13:53 am »
A simple fix would be to swap the 74HC14 with the HEF/CD40106B, which is much slower and won't produce as much ringing on the output.

The most optimal solution is to use as smaller package as possible, so the 74HC1G14 if only one Schmitt inverter is required. Build it on a small PCB with a ground plane and a small packaged capacitor near the power supply pins. I was going to suggest a low-Z probe, but that would load the circuit down too much, as the 74HC14 can't drive a 500R load. You could make your own probe with a 5k input impedance, but it will be a x100, rather than 10 probe i.e. the output voltage will be 1/100 of the input voltage, but modern 'scopes can be set to compensate for that.
« Last Edit: October 04, 2018, 09:33:06 am by Hero999 »
 

Offline schmitt trigger

  • Super Contributor
  • ***
  • Posts: 2431
  • Country: mx
Re: Schmitt disappointment
« Reply #19 on: October 04, 2018, 02:05:57 pm »
Without you providing any additional details, and without seeing your actual circuit or layout; I would say that the overshoot is caused by:
  • poor grounding and supply routing
  • lack of adequate capacitor decoupling
  • attaching the scope probe to an incorrect ground location

I mean, it is after all, only 1Khz.
To a logic circuit, any logic type, this is glacial speed.

While I agree with your assessment of the situation at hand, I'd just like to point out that a squave wave frequency of 1 kHz can still have tons of high-frequency components if the rise and fall times of the edges are short. I'd say that this is the case here.

Thanks for pointing this out.  :-+ The repetition rate has nothing to do with the rise/fall times.
But my basic suggestion still apply.
 

Offline PerranOakTopic starter

  • Frequent Contributor
  • **
  • Posts: 552
  • Country: gb
Re: Schmitt disappointment
« Reply #20 on: October 04, 2018, 04:22:43 pm »
Excellent, thank you all.

There is a 100nF cap near-ish to the supply pins but the pins are on the diagonal opposites of the chip - mental! What would be a low ESR? My one has ESR = 15Ohms but I have no way of measuring the inductance of it.

How do you know what the impedance of the co-ax is? Is it as simple as measuring it in the normal way?

Cheers.
You can release yourself but the only way to go is down!
RJD
 

Online wraper

  • Supporter
  • ****
  • Posts: 17952
  • Country: lv
Re: Schmitt disappointment
« Reply #21 on: October 04, 2018, 06:16:08 pm »
My one has ESR = 15Ohms but I have no way of measuring the inductance of it.
It's useless to measure ESR of ceramic capacitors. At 100 kHz you won't get any meaningful figure.
 

Offline Mr. Scram

  • Super Contributor
  • ***
  • Posts: 9821
  • Country: 00
  • Display aficionado
Re: Schmitt disappointment
« Reply #22 on: October 04, 2018, 06:31:40 pm »
Why do many chips have power and ground on opposite sides of the chip? Isn't that exactly what you don't want from a loop area point of view?
 

Offline Zero999

  • Super Contributor
  • ***
  • Posts: 20357
  • Country: gb
  • 0999
Re: Schmitt disappointment
« Reply #23 on: October 04, 2018, 08:14:09 pm »
Why do many chips have power and ground on opposite sides of the chip? Isn't that exactly what you don't want from a loop area point of view?
It's probably a relic from back when decoupling wasn't essential. Many of the older CMOS ICs, such as the CD40106 will work perfectly with little or no decoupling capacitors.
 

Offline StillTrying

  • Super Contributor
  • ***
  • Posts: 2850
  • Country: se
  • Country: Broken Britain
Re: Schmitt disappointment
« Reply #24 on: October 05, 2018, 09:19:03 am »
There is a 100nF cap near-ish to the supply pins but the pins are on the diagonal opposites of the chip - mental!

On a breadboard I use a 100n with leads long enough so that the cap is above the IC, the breadboard's power lines are too many connections and uHs away to supply power for ns edges.  :)
.  That took much longer than I thought it would.
 
The following users thanked this post: PerranOak


Share me

Digg  Facebook  SlashDot  Delicious  Technorati  Twitter  Google  Yahoo
Smf