Help, triangle generator

[amdxjunk]

I have posted waveforms and a schematic of a triangle
generator I have put together, it is not working properly.
Please see them at abse under- Help, triangle generator.

Mike

 

[Larry Brasfield]

"amdxjunk" <nojunk@knology> wrote in message
news:6c578$42ac6df1$18d64e32$7250@KNOLOGY.NET...

I have posted waveforms and a schematic of a triangle
generator I have put together, it is not working properly.
Please see them at abse under- Help, triangle generator.

The LM324 datasheet shows the amplifier slewing at
about 0.5 V/uS. The first stage in your circuit, used as
a comparator (U1A), appears to slew at 10V/50uS,
or 0.2 V/uS if your article regarding pin numbers is to be
believed. So that makes me suspect something differs
between what you've built/probed and what you've shown.

Also, the (alleged) waveform at pin 3 does not look like
the weighted sum of waveforms at pins 7 and 1, as can be
expected of a resistor summer. This too makes me suspect
a difference between what you did and what you claim.

Looking at the schematic, it appears to have been designed
to keep both U1B and U1C out of saturation, but without
enough margin to account for switching delays due to the
LM324's slowness. I would reduce R7 to keep U1C from
hitting its limits.

As for the waveforms you get not matching what you see at
http://www.cpemma.co.uk/pwm_erg.html ,
those are clearly simulation results. The LM324 has a lot of
crossover distortion and will never look that good in real life.

At 5 KHz, I expect you will hear that circuit's effect upon
the fan motor. Be prepared in case you dislike that noise.
If you want a higher frequency, you will need a faster part.
The LM324 is one of the slowest normal-power op-amps
around. That circuit is not specifically designed for that
op-amp, so you should be able to switch as long as the
pot trim range is matched to the op-amp's swing. (And
that is mainly a nicety, not a hard requirement.)

HTH.

--
--Larry Brasfield
email: donotspam_larry_brasfield@hotmail.com
Above views may belong only to me.

 

[Fred Bloggs]

amdxjunk wrote:

I have posted waveforms and a schematic of a triangle
generator I have put together, it is not working properly.
Please see them at abse under- Help, triangle generator.

Mike

The circuit is fundamentally flawed in that the slew rate required for
the integrator U1B- which is dV7/dt= +/- (Vcc/2)/(47K*C) or +/- 1.3V/us
which is about 2.5x the typical LM324 performance ( and even that is
theoretical because V7 will not hit the rails), so it will not keep up
in other words. Then throw in the 12/0.5V/us= 24us for U1A to slew
between the rails and storage time effects for coming out of saturation,
and you have similarly substantial errors relative to your ~4KHz there.
The TL084 will overcome these problems ( better) and should work with
the values shown, except get rid of U1C and replace R8+VR1 with an
adjustment that goes from (1/4-3/4)*Vcc (that would be an R8A+VR1+R8B
type thing), and switch the input polarity of U1D to get the same
sense-might try adding a smidgen of hysteresis there to make things a
little more snappy.
Have fun.

 

[Fred Bloggs]

Fred Bloggs wrote:

amdxjunk wrote:

I have posted waveforms and a schematic of a triangle
generator I have put together, it is not working properly.
Please see them at abse under- Help, triangle generator.

Mike



The circuit is fundamentally flawed in that the slew rate required for
the integrator U1B- which is dV7/dt= +/- (Vcc/2)/(47K*C) or +/- 1.3V/us
which is about 2.5x the typical LM324 performance ( and even that is
theoretical because V7 will not hit the rails), so it will not keep up
in other words.

My error- the S.R. is 1.3mV/us and this is well within the LM324 capability.

Then throw in the 12/0.5V/us= 24us for U1A to slew
between the rails and storage time effects for coming out of saturation,
and you have similarly substantial errors relative to your ~4KHz there.

Your period should be on the order of 7.5ms or 130Hz.

You obviously have used a component that is wildly off the schematic
value. You do know that C=100n is 0.1u or 104, and 47K is
yellow-violet-orange?

 

[amdxjunk]

the integrator U1B- which is dV7/dt= +/- (Vcc/2)/(47K*C) or +/- 1.3V/us
which is about 2.5x the typical LM324 performance ( and even that is
theoretical because V7 will not hit the rails), so it will not keep up
in other words.

My error- the S.R. is 1.3mV/us and this is well within the LM324
capability.

Then throw in the 12/0.5V/us= 24us for U1A to slew
between the rails and storage time effects for coming out of saturation,
and you have similarly substantial errors relative to your ~4KHz there.

Your period should be on the order of 7.5ms or 130Hz.

You obviously have used a component that is wildly off the schematic
value. You do know that C=100n is 0.1u or 104, and 47K is
yellow-violet-orange?

Thanks all,

I noted one of the posts said it was running about 5Khz
and it was supposed to be running about 125 hz.
So I found I had installed a 1n where it the schematic called for a100n..
So who uses anyway I put in a .1uf and it works fine.

Thanks again.