Simplest way to generate 2ch clock 180 degree phase shift with 40% duty cycle 500kHz

Thread Starter

emil.wingblad

Joined Jan 27, 2022
5
As the title describe. What do you think is the simplest way to generate 2ch clock 180 degree phase shift with 40% duty cycle and 500kHz frequency without any mcu/serial config input? Using two 555 cmos timers seems not like to optimal solution.
 

panic mode

Joined Oct 10, 2011
2,112
that may be the simplest way but it is not the best way, because if you need to change clock rate, things fall apart.
if the duty cycle and phase shift are to be maintained regardless of used clock rate, you would need to use different approach. one simple solution is to use oscillator, counter and decoder.

in this case counter could be 4017. decoder could be two OR gates (one for each channel).
1st channel would be set by counter outputs 1,2,3,4 (that is 40% since 4017 is decade counter)
2nd channel would be set by counter outputs 6,7,8,9.
 

panic mode

Joined Oct 10, 2011
2,112
then you may use 555s but it will need 3 of them. one to generate ch1 40% duty cycle, one to add small delay, one to create ch2 pulse...

if using decoder as mentioned above, oscillator would need to run at 10x faster than output and that would bring it to 5MHz so 555 will likely not be an option (at least not good one). fortunately there are many alternatives and ... why not use crystal oscillator? OR gate can be made with 4 diodes and a resistor so that only IC is 4017
 

Irving

Joined Jan 30, 2016
2,546
I know you said no MCU but an ATTiny85 8pin MCU will generate two biphase outputs. Using timer1 off the internal 64MHz clock divided by 128 gives 500kHz and setting the dedicated dead-time generator prescaler to 1 and the dead-time count to 13 sets the low-high and high-low dead-times as 203.1nS, giving a duty cycle of 39.8%.. A single 8-pin chip - it doesn't get much simpler or more reliable...
 

Ian0

Joined Aug 7, 2020
4,863
If you want exactly 40% then @panic mode 's suggestion can't be improved upon.
If you don't need that level of accuracy, then a SG3525 will work.
IR2153 is another possibility - the HIgh-side driver will also work as a low-side driver.
UCC3808 will also work.
 

Thread Starter

emil.wingblad

Joined Jan 27, 2022
5
I know you said no MCU but an ATTiny85 8pin MCU will generate two biphase outputs. Using timer1 off the internal 64MHz clock divided by 128 gives 500kHz and setting the dedicated dead-time generator prescaler to 1 and the dead-time count to 13 sets the low-high and high-low dead-times as 203.1nS, giving a duty cycle of 39.8%.. A single 8-pin chip - it doesn't get much simpler or more reliable...
Yes, go with a MCU solution would normally be my first thought also. But for this application, no external software load will be available so a pure analog/passive solution is required for this one. :)
 

Thread Starter

emil.wingblad

Joined Jan 27, 2022
5
If you want exactly 40% then @panic mode 's suggestion can't be improved upon.
If you don't need that level of accuracy, then a SG3525 will work.
IR2153 is another possibility - the HIgh-side driver will also work as a low-side driver.
UCC3808 will also work.
+-2% is okey for the duty cycle. Thanks for the recommendations. See that some of the requires a bit more voltage than I´ve available which is 3.3V.
 

BobTPH

Joined Jun 5, 2013
4,769
Out of curiosity, why no micro? One 6-pin chip and a capacitor would be the entire circuit.

Saw your post as I was typing. Firmware is loaded in a micro once, as you build the board. How is that a problem?

Bob
 
Last edited:

Thread Starter

emil.wingblad

Joined Jan 27, 2022
5
Out of curiosity, why no micro? One 6-pin chip and a capacitor would be the entire circuit.

Saw your post as I was typing. Firmware is loaded in a micro once, as you build the board. How is that a problem?

Bob
Because of production procedure as it is now. For this (hopefully) quite simple signal generation, a passive solution would be the best in this case. :)
 

Ian0

Joined Aug 7, 2020
4,863
See that some of the requires a bit more voltage than I´ve available which is 3.3V.
Now you tell me!
So, make the 555 oscillate at 1MHz with a duty cycle of 80% (it might just do 1MHz, otherwise try a 74HC14). Connect the output to a 74HC74 set to divide by two (Connect NOT-Q to D). AND each output with the 555 output using a 74HC08.
You can get single gate versions such as 74LVC1G14, 74LVC1G74 and 74LVC2G08 if it needs to be small.
 

panic mode

Joined Oct 10, 2011
2,112
i know you said "the simplest...". there are so many ways to do this. for example even good old logic ICs. chip with quad gates and Schmidt trigger would do the job too - one gate to form an oscillator, one to invert the signal and remaining two gates with RC element to get the duty cycle... (just saying).

but if precision and stability are required, i would avoid anything that uses RC oscillators.
 

BobTPH

Joined Jun 5, 2013
4,769
Because of production procedure as it is now. For this (hopefully) quite simple signal generation, a passive solution would be the best in this case. :)
I have no idea what that means. Do you think the chip has to be programmed every you time you power it up? What is not “passive” about it?

Bob
 

AnalogKid

Joined Aug 1, 2013
9,956
One CMOS hex inverter.

3 gates form a crystal oscillator.

1 gate to buffer and square up the signal.

Asymmetrical R-C delay circuit to create the 40% duty cycle.

1 gate to buffer and square up that signal

1 gate to invert that signal for the 180 degree output.

ak
 

eetech00

Joined Jun 8, 2013
3,107
Here's my suggestion.
NAND gates are CD4093B.
I could have made a T-FF with the remaining CD4093B's (whole circuit using only one chip) but I think the CD4013 will provide better symmetry.
Uses two chips including the CD4013B. Probably run better at 12v.

1643466357318.png

1643437228774.png
 
Last edited:

AnalogKid

Joined Aug 1, 2013
9,956
If you look at the internal schematic of a 4013 (my NS datasheet is dated 2002), the Q and -Q outputs come from the input and output of an inverting gate. Thus, I don't see any advantage over replacing the 4013 in your schematic with another 4093 gate.

I had the impression that better frequency stability was required, which is why I mentioned a crystal oscillator approach in #15. Otherwise, I would have gone with a 4093 oscillator, You do not need two diodes, but you get better temperature tracking of the duty cycle with two.

ak

ps. You have two different part types labeled U1.
 
Top