Using CMOS inverters to build oscillators works best when using "unbuffered"
inverters. These can be identified by the letter "U" in the part number, as
in "74HCU04".

If you use a buffered part or a part that has Schmitt trigger inputs, then
the input-output transfer function becomes very steep (too much gain) and
the oscillation becomes hard to control. When this happens, the circuit
often oscillates, but at some frequency much higher than the crystal
frequency. I've made oscillators out of XOR gates, but if it doesn't come up
on the right frequency it can be hard to debug without a scope.

To determine if it's oscillating at all, I suggest measuring the DC current
consumption with the feedback removed (crystal out of circuit), and then
with the feedback closed (crystal in circuit). When the circuit is
oscillating, the current will be markedly higher. Once you know it's
oscillating, then you have to determine where it's oscillating. This might
be challenging if all you have is a receiver.

Joe
W3JDR


"PaoloC" wrote in message
...
[Slightly off-topic request]

Hi.
As of the other post of mine, another part of my weekend was spent
trying to get a XOR gate to oscillate with a XTAL.

I have a working 74HC14 oscillator:
"One gate of a 74HC14. 470ohm resistor from gate output to the parallel
of 1Mohm//XTAL. 10pF and 33pF (from the junkbox) capacitors to ground on
each side of the XTAL."

Since my project calls for a frequency doubler, which I want to
implement with a XOR gate, I wanted to use one XOR gate of a 74HC86 as
oscillator. Remaining gates would work as buffer, delay line, digital
mixer.

(I have Googled newsgroups and found an interesting discussion about XOR
frequency doublers, so I am aware of its limitations.)

So I moved the oscillator circuitry (R; C; R, XTAL; C) from the 74HC14
to the 74HC86, pulling the other input of the XOR gate to "1", so that I
would produce an inverter.

Nothing happens. I used a 10.0 MHz XTAL.

I understand that the 74HC14 has Schmitt trigger inputs and that a
74HC86 might have higher propagation delays. Still, is it possible to
use a XOR gate as an inverter and oscillator? Do I overlook something?

In the end I recovered the 74HC14 oscillator, but this adds one
component to the final circuit I have in mind that wastes energy, space
and is underutilized.

Looking forward to your always helpful replies!
Paolo IK1ZYW

"JLB" wrote in message
.. .

...With an oscillator built from inverting logic gates, you need two
gates so
there is no overall phase inversion through the gates.

There are two types of oscillators with gates. Series and parallel.
The former requires no inversion the latter does.

If you are using XOR
gates, I would recommend tying one of the input lines on each gate high

He said he was.

I think the biggest issue here is thr Schmidt input. It can't be biased
*in the active region*.

--
Steve N, K,9;d, c. i My email has no u's.

"JLB" wrote in message
.. .

...With an oscillator built from inverting logic gates, you need two
gates so
there is no overall phase inversion through the gates.

There are two types of oscillators with gates. Series and parallel.
The former requires no inversion the latter does.

If you are using XOR
gates, I would recommend tying one of the input lines on each gate high

He said he was.

I think the biggest issue here is thr Schmidt input. It can't be biased
*in the active region*.

--
Steve N, K,9;d, c. i My email has no u's.

On Mon, 19 Apr 2004 18:53:16 GMT, "W3JDR" wrote:

Using CMOS inverters to build oscillators works best when using "unbuffered"
inverters. These can be identified by the letter "U" in the part number, as
in "74HCU04".

If you use a buffered part or a part that has Schmitt trigger inputs, then
the input-output transfer function becomes very steep (too much gain) and
the oscillation becomes hard to control. When this happens, the circuit
often oscillates, but at some frequency much higher than the crystal
frequency. I've made oscillators out of XOR gates, but if it doesn't come up
on the right frequency it can be hard to debug without a scope.

To determine if it's oscillating at all, I suggest measuring the DC current
consumption with the feedback removed (crystal out of circuit), and then
with the feedback closed (crystal in circuit). When the circuit is
oscillating, the current will be markedly higher. Once you know it's
oscillating, then you have to determine where it's oscillating. This might
be challenging if all you have is a receiver.

Certainly would be. The other thing about using Schmidt type inputs is
the likelihood of causing longer-term damage to the crystal. They
'prefer' not to be driven to hard and the sharpness of the buffered
gates ain't good for 'em.

On Mon, 19 Apr 2004 18:53:16 GMT, "W3JDR" wrote:

Using CMOS inverters to build oscillators works best when using "unbuffered"
inverters. These can be identified by the letter "U" in the part number, as
in "74HCU04".

If you use a buffered part or a part that has Schmitt trigger inputs, then
the input-output transfer function becomes very steep (too much gain) and
the oscillation becomes hard to control. When this happens, the circuit
often oscillates, but at some frequency much higher than the crystal
frequency. I've made oscillators out of XOR gates, but if it doesn't come up
on the right frequency it can be hard to debug without a scope.

To determine if it's oscillating at all, I suggest measuring the DC current
consumption with the feedback removed (crystal out of circuit), and then
with the feedback closed (crystal in circuit). When the circuit is
oscillating, the current will be markedly higher. Once you know it's
oscillating, then you have to determine where it's oscillating. This might
be challenging if all you have is a receiver.

Certainly would be. The other thing about using Schmidt type inputs is
the likelihood of causing longer-term damage to the crystal. They
'prefer' not to be driven to hard and the sharpness of the buffered
gates ain't good for 'em.

Hi,
thanks to all those who replied and discussed this topic, providing me
with a lot of good suggestions.

In the end I might use a BJT/FET oscillator instead, especially if it
works with the ex-CB XTAL of my other post. Troubleshooting with a
receiver and a DVM is fun, but time consuming and often demotivating.

I dream of an oscilloscope as much as I dream of an HF dipole up on the
roof. Hopefully both will come with time, as well as children, a better
wage, ... ;-)

73,
Paolo IK1ZYW

Hi,
thanks to all those who replied and discussed this topic, providing me
with a lot of good suggestions.

In the end I might use a BJT/FET oscillator instead, especially if it
works with the ex-CB XTAL of my other post. Troubleshooting with a
receiver and a DVM is fun, but time consuming and often demotivating.

I dream of an oscilloscope as much as I dream of an HF dipole up on the
roof. Hopefully both will come with time, as well as children, a better
wage, ... ;-)

73,
Paolo IK1ZYW

like Figure 7:

http://www.northcountryradio.com/PDFs/column007.pdf


--
Steve N, K,9;d, c. i My email has no u's.


"Steve Nosko" wrote in message
...

"JLB" wrote in message
.. .

...With an oscillator built from inverting logic gates, you need two
gates so
there is no overall phase inversion through the gates.

There are two types of oscillators with gates. Series and parallel.
The former requires no inversion the latter does.

If you are using XOR
gates, I would recommend tying one of the input lines on each gate high

He said he was.

I think the biggest issue here is thr Schmidt input. It can't be
biased
*in the active region*.

--
Steve N, K,9;d, c. i My email has no u's.

like Figure 7:

http://www.northcountryradio.com/PDFs/column007.pdf


--
Steve N, K,9;d, c. i My email has no u's.


"Steve Nosko" wrote in message
...

"JLB" wrote in message
.. .

...With an oscillator built from inverting logic gates, you need two
gates so
there is no overall phase inversion through the gates.

There are two types of oscillators with gates. Series and parallel.
The former requires no inversion the latter does.

If you are using XOR
gates, I would recommend tying one of the input lines on each gate high

He said he was.

I think the biggest issue here is thr Schmidt input. It can't be
biased
*in the active region*.

--
Steve N, K,9;d, c. i My email has no u's.

Sorry. Poseted to wrong thread... , Steve

"Steve Nosko" wrote in message
...
like Figure 7:

http://www.northcountryradio.com/PDFs/column007.pdf


--
Steve N, K,9;d, c. i My email has no u's.


"Steve Nosko" wrote in message
...

"JLB" wrote in message
.. .

...With an oscillator built from inverting logic gates, you need two
gates so
there is no overall phase inversion through the gates.

There are two types of oscillators with gates. Series and parallel.
The former requires no inversion the latter does.

If you are using XOR
gates, I would recommend tying one of the input lines on each gate
high

He said he was.

I think the biggest issue here is thr Schmidt input. It can't be
biased
*in the active region*.

--
Steve N, K,9;d, c. i My email has no u's.