So I made W2AEW's pulse generator and now its behaving strangely

[AndrejaKo]

After watching these two videos:


 I made the oscillator described there, except that I used 74HC14, since I couldn't easily find the AC version.

For those who can't be bothered to watch the videos, it's a hex Schmitt inverter with an oscillator made from one inverter, 6.8 kilohm resistor and  4.7 nanofarad capacitopr. Its output is connected to the five remaining inputs. Those inputs are connected to female BNC connector via 220 ohm resistors. Schematic is attached.

Anyway, to the problem: It worked fine for a while and I was getting 5 V square wave of around 44 kHz with a 2.3 ns rise time and a volt or so of undershoot and overshoot. I could nicely see the reflection from the far end of coaxial cables I experimented with. Then I decided to take a look at how the signal would look like if I connected it directly to scope using only a coaxial cable. After I made the connection turned on the oscillator, it worked for a few seconds and then I lost the wave. I also noticed that current consumption increased a bit at that time.

After testing it out for a few minutes, I noticed that I'm getting a sinusoidal signal of around 600 mVpp and frequency going from 60 MHz to 62 MHz. After some probing, it turns out that pin 5Y is reading 75 ohms to the ground. Now that I disconnected the output resistor from it, I'm getting around 65 MHz at 8 Vpp. The signal is still sinusoidal and it has slow rise time of around 5.2 ns. It's interesting that the capacitor which I've used for the original oscillator now has no effect on the behavior of the circuit. Even if I remove it, the oscillations remain. Also I can affect the period a little bit by touching the chip itself.


Obviously the chip is dead, but I really have no idea what could have killed it. I mean, the scope has input resistance  of 1 megohm  and for its capacitance, there are the resistors on the output pins which should have stopped any big currents. Also all of my previous projects with 74xx chips worked fine, so any pointers on how to debug this would be appreciated.

[kfitch42]

Obviously the chip is dead, but I really have no idea what could have killed it.

To me it sounds like the cap failed 'open'. As a result you have created a 'ring oscillator' ( http://en.wikipedia.org/wiki/Ring_oscillator ) which is why the frequency went so high.

Why did the cap fail? I'm a noob, so I don't know.

[AndrejaKo]

Actually, no. It seems that I forgot to mention it, but one of my first ideas was that the capacitor failed and I removed it and tested it outside of the circuit. Capacitance shows up fine on my meter. Also putting another capacitor in parallel with that one doesn't affect the circuit's frequency at all. The oscillation does stop for a short amount o time, but then restarts.


Also how would that explain the very low resistance I get on the 74HC14 pins? I have another IC which is unused and it shows much different results when measured.

I do agree that the oscillations do look like effect of ring oscillator right now to me.

[w2aew]

Is this built on a solderless breadboard?  If so, the parasitics of those are terrible and can lead to oscillation like you've seen.  If you have it properly built on a PC board, soldered, etc., then you may want to play around with the position and number and value of the supply decoupling cap(s).  The HC family doesn't have the drive capability of the AC family, but you should still be below the damage limit, so I don't know why you've seen a failure.  Oscillations can sometimes be tough to squelch.  You may want to try adding a little loss in the feedback path (ferrite), since you really only want low frequency feedback here.

[hlavac]

I don't like how the source termination is done.
It is exceeding the current limit of the supply pins for 74HC family (20mA per pin, but 50mA total for GND/Vcc, you are trying to draw 53mA total).

My suggestion would be to put a voltage divider on the output that lowers the voltage/current output and still presents 50ohm impendance to the coax.

For example 150R/75R:
150R || 75R = 50R,
voltage divider will be 150R : (75R || 50R) or 150R:30R or 0.166, which makes for 0.833V signal amplitude at 5V power supply,
drivers will see resistance of (150R + (75R||50R) = 150R+30R = 180R, current draw will be 27.7mA at 5V.
Reflected 0.833V will not be higher than 1.66V which is harmless.

[hlavac]

Here is a table of suitable values of voltage divider for driving 50 Ohm coax  (assuming 5V supply):

R1	R2	Impedance	Amplitude	Drive current
100	100	50	1.25V	37.5mA
150	75	50	0.833V	27.8mA
470	56	50.038	266mV	10.1mA
2.7k	51	50.055	46.3mV	1.8mA
110	91	49.801	1.13V	35.1mA
130	82	50.283	0.96V	31mA
270	62	50.422	0.46V	16.8mA
180	68	49.355	0.69V	23.9mA

[hlavac]

This is what i mean by voltage divider on output:

[AndrejaKo]

I'll give that a try! In my attempt to remove the 74HC14 from the PCB, I made a ton of shorts everywhere, so the original circuit is pretty much ruined. Fortunately, I left enough room on it for a few more attempts.