CMOS Schmidt trigger oscillator

[Julez]

Hi everyone!
I am trying to build a CMOS Schmidt trigger oscillator, based on 74LVC14A, which is required to run from 3.3V and generate complementary (Q and notQ) outputs at a nominal 25kHz with a 50% duty-cycle. I only have two gates available for this purpose. I got to a design, which I have drawn and include as an attachment. It works well apart from when first powered-up from a slowly-rising supply voltage, it locks up with both outputs at high logic-1 level. This is a big problem as it needs to be able to reliably start with these circuit conditions. I then found that a standard two-transistor astable circuit can be modified with a "sure-start network" consisting of just two 1N4148 diodes and a 100nF capacitor to fix the identical problem that this classic circuit suffers from. I enclose a schematic for this circuit aswell. I am wondering if this "sure-start" modification can be applied to my CMOS oscillator to fix my problem. I think it can, but I just can't see the exact way! I am hoping someone here can help me see the light. Thanks for reading!

 

[Mikebits]

Can you take a picture of the circuit you built,the layout may have some affect on start up.

 

[Les Jones]

You could try the same principle as the transistor astable. Join the bottom of the two 18K resistors together and connect them to ground via a 100 nF capacitor. Also add a high value resistor (Say 1 M) from this point to the positive supply. Now connect the two diodes with their negative ends to this point and the positive end of the diodes one to each output. If the circuit is locked up with both outputs high then the 100 nF capacitor will charge via the high value resistor. this will result in the gates switching to give a low output. This is unlikely to happen at exactly the same time so the oscillator should start. When it is running there will always one diode pulling the top of the 100 nF capacitor to ground.

Les.

 

[alec_t]

That CMOS oscillator is based on a differentiator. Try the simpler standard one based on an integrator, like this:-

With only one inverter it can't lock up the same way.

 

[crutschow]

Your circuit emulates the transistor oscillator circuit and suffers from the same latchup problem.
That complex circuit is not needed with a Schmitt trigger.
Use the circuit Alec suggested.
It's much simpler and can't latch up.

 

[MikeMl]

If you dont want to use a Schmitt Inverter, you can make one, like is shown in this app note,
which also talks about why multivibrators sometimes dont start-up, and how to prevent it.

 

[Julez]

Firstly, thanks to everyone who has taken the time and trouble to read my query and post their thoughts and solutions.
Its great that so many people became involved to help!

Mikebits: Yes the layout is bad - it's built on stripboard so it's going to have drawbacks compared to a PCB.
Its only a prototype at the moment - the PCB will come later on. I tried to keep everything as short and direct as possible, but stripboard is unfortunately always a huge compromise in this regard.

alec_t / crutschow: I know the circuit is a little oddball compared to the standard relaxation oscillator circuit, but the reason for this is I need 50% duty-cycle.
It must be close as possible to 50% so I used a symmetrical circuit and with 1% timing components it was really close. I guess I should have stressed this point more in my original question.
If there is a way to make this other oscillator circuit give a symmetrical output without more gates please tell me. The duty-cycle is more important than the frequency which only has to be a nominal 25kHz.
There is a maximum of two gates available spare for the oscillator.

MikeMI: Schmidt inverters are necessary for the rest of the circuit, the oscillator is to be built from the two remaining spare gates so I really would like to use them.

Les: This is exactly the kind of idea I was hoping someone would have. I will definately try this and let you know the outcome.

 

[JimB]

If there is a way to make this other oscillator circuit give a symmetrical output without more gates please tell me.

I am not sure about "without more gates", but if you make the oscillator run at twice the required frequency and then divide by two using a flip-flop, you will get a 50:50 mark-space ratio.

JimB

 

[audioguru]

A CD4047 is a Cmos RC oscillator, a digital divider that produces perfect 50:50 duty-cycle and it has two complementary outputs. Its power supply can be from 3V to 18V. A 74HC4047 might be available for a higher output current and will work from a supply that is from 2V to 6V.