Low current oscillator designs

[KaiZX]

oscillator cd4007

I originally posted a question here. **broken link removed**

That question was asked in relation to a project I'm doing that requires breadboarding. As it turns out, I also need a clock generating circuit for a different, ASIC design project I'm doing. Because this time I'm doing ASIC design, I do not have the option of using a crystal oscillator. Rather, I need an oscillator only FETs and passive elements (resistors and capacitors).

There're two goals I'm aiming for: low current draw (preferably a few uA), and robust design across a fairly wide Vdd range.

I've already investigated the possibility of a ring oscillator, but I've ruled out that possibility. The main reason is because of Vdd instability. So instead, what I'm doing is a bastardized version of the LMC555 timer, except with some additional modifications. I'm using are simple diff-amps as comparators (this is in direct relation to a post I made earlier about T-spice analysis **broken link removed**). I also have a buffered stage seperating the output from the RC feedback, to prevent the output from being loaded down by the resistor.

The main current drain is the voltage-divider network. In an IC package they're fine, but in my current design they can drain up over 0.1mA, and that's totally unacceptable. My quick fix is increasing the resistor values. My comparators also draws fairly large current.

Are there any ideas for improvements to my ideas? I'm also definitely interested in alternative design ideas.

Thank you in advance.

 

[manish12]

mosfet rc oscillator

what is the your desired range of freq.?
use CMOS chip for low power .
but as far as the relation is concern about is that “power directly proportional to the freq.”

use dual threshold techniques , sub threshold techniques , stack of series transistor , reduces the size of transistor or best use SOI models that will bring the power low and increases the speed.

Many more techniques are there to get power without any power – just joking.

 

[KaiZX]

low current oscillator

I'm looking at a fairly low range of frequency. The clock I need to generate is 2MHz, so I'm looking at a range between 1-3MHz.

 

[unkarc]

cmos rc oscillator

Hi,

Have a look at these circuits, carefully:

https://www.linear.com/pc/downloadDocument.do?navId=H0,C1,C1010,C1096,P9790,D9620

**broken link removed**

https://www.aldinc.com/pdf/fet_11120.0.pdf

**broken link removed**

**broken link removed**

I guess the CMOS circuits current consumption will be higher at 2-3MHz than it is in the above circuits but in case of the circuits with discrete components it is worth tinkering I think.

regards
unkarc

 

[butterfish]

sub ua low area oscillator design

whats the frequency accuracy you need?
if argin > 1MHz+-3%, a cmos RC oscilator can be used.

 

[KaiZX]

low current design pdf

Hi. Thanks to everyone who have answered so far.

One of the ideas that I'm playing around with is using a Schmitt trigger as basis for an oscillator. For those of you who have the Baker-Lee textbook, take a look at the chapter on "Special-Purpose Digital Circuits" where he talks about the Schmitt trigger.

It's an interesting idea, but it seems however that the Schmitt trigger he uses draws quite a bit of current. Does anyone have suggestions for how to modify his design?

 

[unkarc]

mosfet oscilador

Unfortunately I do not have access to the book you refer to. Can you scan/upload the circuit how it is built or show a similar schematics in the web?

 

[KaiZX]

74hc14 oscillator crystal

Sure. The figures are actually online on the book's website. Here's the link:

**broken link removed**

The figures I'm speaking off are Figure 18.3 (Schematic of the Schmitt Trigger) and Figure 18.8 (the Schmitt Trigger as an oscillator). The design draws a lot of current due to the way he switches between Vdd and GND voltages, I believe.

 

[unkarc]

low current rc oscillator

Thanks, I see. Unfortunately I am not so good at CMOS MOSFET design in computer, sorry for this (I assume you simulate these circuits in a software?). Perhaps there are width and other parameters to choose for the MOSFETs that restrict possible drain currents but still insure a minimum transconductance (gm) just to maintain oscillation.

I remember the CD4007 integrated circuit which include three p and n channel MOSFETs, all I can suggest is trying to build an RC oscillator from them, but I do guess the current consumption will tend to higher than you would expect around 2-3MHz. Here is a data sheet on the IC: **broken link removed**

Or have a look at a 74HC14 hex Schmitt trigger, https://www.fairchildsemi.com/ds/MM/MM74HC14.pdf OOPPs, I just see its logic diagram is the same you refer to in the book Fig 18.3! Aware of that?
But if you tried this IC also in practice, then sorry...

rgds, unkarc

 

[KaiZX]

74hc14 ring oscilator

No, I never knew that. I'm guessing the book probably referenced the 74HC14 design or a paper that detailed such a design that Fairchild then used.

In any case, I was trying NOT to do an RC oscillator, due to issues with resistor performance over temperature (they're tricky in ASIC design and I'm not too sure how to tune them). However, it looks like I don't have a choice, because the design from the book simply draws too much power.