35 MHz from a retro 74LS00 chip

I have an old 130-in-one kit with a 74LS00 quad NAND gate on it. What's the easiest thing to do with a NAND gate (besides using it for its intended purpose as a logic gate)? Build an oscillator! More specifically, a ring oscillator, formed by using the NAND gates as inverters. Three NANDs are wired in a ring. I expected to get 1 or 2 MHz from this configuration as the chips were late 80's low speed TTL... boy was I surprised when I fired up my oscilloscope... 35 MHz! From a '80s TTL chip. I wonder if I can go faster using only one stage. Might not be stable though.

I'm going to get a HP-IB compatible LaserJet soon so maybe I'll be able to print out things instead of struggling to get a digital camera to work on an oscilloscope screen (best mode was "Sports" probably due to the short exposure.)

Wow. I guess my 'board thread got you to pull out your old x-in-1 and have some fun?

Very impressive. How stable is the osc?

Have you tried any heating and cooling to see how much it changes?

Well, I had some wire... and an idea... and I couldn't resist. I had previously lost all the jumper wires that came with it but I now have 3x100m reels of wire (highly recommended: they were only £3 each.)

It's very stable but it starts at 33 MHz and goes up to about 35 MHz where it stays. I tried with one gate only but it won't oscillate at all. I'm going to try to add the "tuning capacitor" to the circuit to see if I can pull the frequency.

I know it's definitely the case for CMOS but I suspect it is for TTL as well, where tying the input and output together of an inverter is a stable operating point.

There's a well defined intersection point if you cascade the input/output characteristic, especially since there's a decently linear region. For an oscillation you need a well delayed trigger.

It's neat to see the frequency value though, I played around with ring oscillators a few years back without the ability to measure it... they're kind of finicky when you start adding capacitors. I did get one to get down to a Hz or so using some massive caps.

Ok, this is just bizarre. I have attached a tuning cap (presumably in the low pF range as it is designed to be wired up in an AM radio) and it's going FASTER! I'm able to "pull" it from around 39 MHz to 32 MHz, but in between it goes as low as 25 MHz (and becomes unstable.)

Interesting, where did you connect it?

At the input to gate #1 (both inputs tied together) and wired to ground. Putting a 100 ohm resistor in series seems to reduce the effect. I'm able to get it to put out an odd double-pulsing effect at a certain point on the dial: one big pulse, one small pulse and so on at about 12 MHz (~25 MHz if you count each pulse.)

I've noticed when it's going faster the Vp-p is much lower, around about 400mV, and the waveform is much more distorted.

Kind of a lame answer but I guess that the cap is having some effect with your large amounts of inductance there.

100 nH with 100 pF resonate at 50 MHz and that would be the order of magnitude of inductance you'd have there.
As it peaks and rings it could trigger the gates a bit earlier. The resistor would add to that idea I guess, it would damp things out.

A pic of the scope screen, please? For the double pulse and the higher speed action.

Your wish is granted.

Pic #1: dial at zero
Pic #2: dial midway: double pulse action
Pic #3: dial at end

To make pic #3 I had to stop the oscilloscope because the circuit was not stable and merely touching the dial would change the frequency.

What's the bandwidth of this scope anyway?
Does that "BW Lim" icon in pic#1 indicate you turned on a BW limiter?

100 MHz bandwidth. HP 54501A digitising 10 MS/a oscilloscope doing equivalent time sampling to reach an effective 10 GS/s sampling rate. If BW lim is highlighted (or as HP put it, "fullbright") it turns on a 20 MHz bandwidth limiter. It was disabled for all these measurements.

Well that would explain the look, you're only catching the 1st and 3rd harmonics.
You're probably also hitting a rise time limitation...

So the frequency is higher than 100 MHz?

The scope is more than capable of measuring past 100 MHz. I measured the FSB of a computer at 133 MHz with no issues, except the voltage was a little lower. HP made the same oscilloscope in 100 MHz, 200 MHz and 500 MHz versions I believe, so it would make sense to use the same frontend (i.e. trigger circuit and ADC) but change the analog frontend, right?

That's a good point on the risetime... perhaps I should try 10X probes.

What did that 133 MHz FSB look like?

If your 3dB is 100 MHz then your 3rd harmonic of 3x35 = 105 MHz is already down to 70% amplitude and your 5th = 175 MHz is down to 50%

I haven't dealt much with exceeding my scope but that peculiarly triangular looking sinusoid is really suspicious.

133 MHz FSB looked like a sine wave, but it was not perfect. I wouldn't expect it either, as it is coming from a PLL on the motherboard. I could clearly see when the computer started up it booted up on a 100 MHz FSB but within 5 seconds switched over to the faster 133 MHz FSB.

I tried 10X probes and the effect with the double bumps is much smaller, but it is there. I have attached a pic.

The triangular waveform is likely due to the chip's limitations. Remember we're working with 80's chips here, I would imagine the capacitance of the inputs is quite high. These are 74LSxx chips (low power schottky) not designed for high speed.

hmph.... Pretty interesting. I like what im seeing there.

Any noticeable heating in the chip?

Can you get a capacitance measurement on the vari-cap where it is set when the "double-bump" is happening?

It's possibly the chip and your wiring but you're most definitely running into scope limitations as well, to get a decent square wave you need to record many harmonics, you're barely catching 3.

There's a lot of guessing involved here because what you're showing is slower than the chip's slew rate according to this site, but at the same time it changes with load capacitance.
http://www.interfacebus.com/IC_Output_Slew_Rate.html

The fact that switching probes did something implies the probes either introduced or hid an effect.

The wiring you have is definitely awful, like I said you're going to have on the order of 100 nH's there, probably more.

Im thinking more like 250nH+



This is quite interesting.

retched said:

Im thinking more like 250nH+



This is quite interesting.

Click to expand...

I agree, that's why I said 'on the order of'