PT2399 pin 5 wierdness.

[anotherjim]

This chip is a constant source of mistrust and puzzlement for me.
I've learned to measure the VCO clock at pin5 as an easy way of finding actual delay time using the datasheet chart. Forget about using pin6 resistance on the chart. Eg, 1K gives 21MHz, chart says 17.

On a minimum breadboard circuit...
I happened to have my scope set to read out peak to peak volts as well as frequency.
It says, and looks like, 8v!
Pin 1 exactly 5volt.
Pins 3 & 4 exactly 0v.
All other pins around the 2.5V bias of pin 2.

I measure Si protection diodes in the chip between pin5 and power rails.
I even added Schottky BAT42 protection diodes to pin5.
Makes no difference. I can see that Pin 5 yanks the power rails a bit, presumably when the diodes conduct, but it don't limit the pin5 clock signal.

I wonder if anyone else has seen this behaviour? Wondering if it's a sign of 2399 variation or they all do it. This is a batch which do not like being forced down to chorus range without dying.

Talk of dying, I thought I'd killed the chip when it locked up. Power off until 5v gone would not restore it. In desperation - almost threw the chip - I grounded all the pins to discharge any residual. That fixed it.

[Kipper4]

So it can be reset....... Intresting thanks Jim.

[ElectricDruid]

I wonder if anyone else has seen this behaviour?

No, but I've got one on the breadboard currently, so I'll check it out and report back. Like you, I'm measuring the clock rate to find the delay time. The waveform gets heavily slew-rate limited at the top end of the range. It's ok for humans to read on a scope, but it long ago stopped being a valid digital logic signal!!

Tom

[anotherjim]

Tom, have your read the Belton Brick patent? It quotes required delay times, then goes on to list the delays for 3x PT2399 giving required pin6 resistance. They directly match the datasheet chart! I wonder how many Bricks actually match the invention in that respect?
Funnily enough, if you make delays equal Accutronics spring times, it sounds Sh!t - they have to be longer.

The Pin5 waveform is slightly differentiated, which I suppose is where the increased p-p comes from. I'm very surprised the diodes don't catch it though.
I did check my x10 probe compensation and it was DC coupled BTW.

[duck_arse]

aww, jim, you ruined my response. I've hadda peek at that pin too, on a 20MHz bandwidth cro. needless to say, it doesn't show a square wave. I wuz gonna suggest probe overshoot, or some such black-art-ness. ringing breadboard parasitics?

[anotherjim]

x1 probe shows more what I'd expect. Breadboard stray capacitance should integrate the wave - slewed edges - I see that with x1.

x10 probe is not suitable for fast edged signals.
I should have remembered that from a past life as a calibration 'droid. In my defence,  that was 30 years ago!
No need for x10 with logic - outputs are not loaded significantly by a x1.

[ElectricDruid]

On a minimum breadboard circuit...
I happened to have my scope set to read out peak to peak volts as well as frequency.
It says, and looks like, 8v!
Makes no difference. I can see that Pin 5 yanks the power rails a bit, presumably when the diodes conduct, but it don't limit the pin5 clock signal.

I wonder if anyone else has seen this behaviour? Wondering if it's a sign of 2399 variation or they all do it.

Ok, I've done some investigation with my PT2399.

I've got the same effect as you - considerable ringing on the leading edges of the 5V pulses, making the p-p level much higher. My scope reads 9.6V on the peaks! This is looking at the waveform with a x10 probe to reduce loading and the scope is a 100MHz digital OWON PDS7102. With the x1 probe, the waveform gets so slew-rate limited that the level drops markedly. That's what I was seeing before, and I'd incorrectly assumed it was the PT2399 doing it, but it's not.

The ringing output suggests that adding just a tiny touch of loading or even a basic RC LPF might clean the waveform up significantly. That might be my next experiment.

In practice though, since the ringing is all well above any logic threshold, it shouldn't cause any kind of false triggering of subsequent circuits. I'd expect the ripples to get blocked by protection diodes on the next chip. That's "hope" rather than "experience" though! Filtering might be safer.

HTH,
Tom


Edit: Also worth mentioning that I checked the probe compensation with the scope's internal square wave and if anything, the probes are slightly over-compensated (too little HF, not too much), so I'm fairly sure the ringing isn't from the probes.

[anotherjim]

Try different scope grounding. Night and day difference between using the probe ground clip versus a direct wire from the scopes ground post. Back in the calibration labs, we never used probe grounds (bloody nuisance, always breaking or catching on things). Instead used wire with croc clip from scopes chassis ground post. Probe clip adds some inductance in the wrong place I believe.

Direct chassis ground and x1 probe at 21Mhz gave me a wave more like a bent sine and 2V p-p! This might be closer to the truth.

I don't believe in magic. PT2399 definitely has standard CMOS protection diode clamps on it's pins. Pin5 cannot possibly be exceeding supply by more than 0.6v above pin1 or below pins 3/4. Even additional fast Schottky diodes would not change the scopes picture or measurements. Therefore I conclude that pin 5 output is not exceeding the rails.

[ElectricDruid]

Try different scope grounding. Night and day difference between using the probe ground clip versus a direct wire from the scopes ground post. Back in the calibration labs, we never used probe grounds (bloody nuisance, always breaking or catching on things). Instead used wire with croc clip from scopes chassis ground post. Probe clip adds some inductance in the wrong place I believe.

Ok, I'll give it a go next time I'm looking at the PT2399 circuit.

Thanks,
Tom