Improving waveform shape in discrete Pierce oscillator with Schmitt trigger output

Question

I am making a simple clock source using discrete Pierce oscillator with its output attached to high speed Schmitt trigger buffer to get square wave in the final output.

I got the circuit for the discrete Pierce oscillator from National Semiconductor app. note AN-400 page 3 figure 5.

I did not add capacitor c1 and c2 just for the sake of testing (I will add them later to adjust the output frequency close to 16 MHZ).

The shape of the waveform coming from the Schmitt trigger is not smooth as a square wave should be (see attached image) and I would like know if the shape of the waveform is normal or not and I would like improve its shape a little bit but I do not Know what should I do about it.

I used 10X probe without the long ground wire and directly probed the output with minimal distance to the output with the tip and the ground spring.

Schematic and the PCB (2 layers PCB)

What do you mean when you say you did not use a ground wire? You did not connect scope ground to the board, not with the ground lead or ground spring? — Justme, Aug 19 '23 at 07:32
I made an edit to clarify the point, I used the ground spring instead of the long ground wire. — Muhammad Nour, Aug 19 '23 at 07:35
That square wave shape may be further disrupted by the transmission line out from J1 to a load if it is not properly terminated. — glen_geek, Aug 19 '23 at 14:16
This is a 16 MHZ clock and will be connected to motor driver IC, it will be on board, as near by the IC, how do you suggest it should be terminated? — Muhammad Nour, Aug 19 '23 at 15:30
The part number on the schematic is [3225-25M-SR](https://www.tme.eu/gb/details/3225-25m-sr/smd-quartz-crystals/sr-passives/) which has a frequency of 25MHz. Does the schematic use the wrong part number, or does the 3225-25M-SR actually output a 16MHz clock? — Chester Gillon, Aug 19 '23 at 16:16
@ChesterGillon It is the same footprint so I used it just for the sake of testing — Muhammad Nour, Aug 19 '23 at 20:12

Justme · Accepted Answer · 2023-08-19T08:25:33.777

The LVC1G17 buffer output is extremely strong and it means it will have very high slew rate so edges are very fast.

It can drive more than 32 mA and has switching time less than 1ns into 15 pF load at best. Which means it has edges fast enough for a 500 MHz output.

There is also no other load than the scope, probe wire, and probe. So the load is very light, and complex, due to capacitances and lossy probe wire, in addition to the time of flight of the signal through the probe cable.

So what you see is actually what you would expect to see. It could be a probing artifact or it could be that the signal really rings.

It is normal to see ringing and over/undershoots for an extremely fast signal edge, measured with scope probe. The chip output is basically unloaded, with signal with fast edges taking time to travel via lossy probe cable to scope, and maybe reflecting back from scope.

This is why designers delibrately add a series resitor to output, to limit the drive ability and slow down signal edges to have only enough bandwidth that is necessary. Or use chips that are only as strong or fast as necessary for the task.

Also these fast transtions require a good bypassing of the digital chip with capacitors. The only bypass cap you have is a 100nF ceramic, whih may not be very effective any more at the frequencies the chip tries to pull current out of it.

The capacitor is also quite poorly connected to the chip. There is just a few millimeters between cap 5V pin and chip, but the route between cap GND pin and chip is quite long loop around other components, so fast return currents may flow near sensitive analog crystal circuitry.

It might be better if the capacitor terminals are connected to chip terminals with shorter wires, and it might be better if there are multiple capacitors, or just a single capacitor that is more suitable for higher frequencies. Perhaps a 1nF right at the chip pins and only then the 100nF will make a difference.

I really did not make a lot of effort in the design of the pcb layout , I just wanted to know if it will work or not, but never the less, added 1.5n cap on the top of the buffer IC and I am already seeing some improvement, also I added 39 ohm in series with the output and it made it also better. I will re-design the layout and give it more though and see what I get, thanks — Muhammad Nour, Aug 19 '23 at 08:47

Improving waveform shape in discrete Pierce oscillator with Schmitt trigger output

1 Answers1