Which OpAmp for buffering a CV distro Multiple?

Sebo · Joined: Apr 27, 2007 Posts: 564 Location: Argentina

Hi:
I'm about to breadboard a simple set of buffers (op amps in voltage
follower configuration) to set up a multiple that I will use to distribute the
1V/Oct output of my MIDI to Cv interface to my oscillators. So I need some
accuracy. Should a TL072 (or TL074) will be enough or I need something
better (with a lower offset / lower drift)?
Maybe a LF412 or LM358 or LF442 or CA3240, or something else.
Also I don't want to use an output resistor, so the output of the opamp
should be short circuit proof.
Thank you.

bubblechamber · Posted: Sat Apr 11, 2009 6:04 am Post subject:

i sould probably try the TL074 first and see how it worked. the LF4xx will probably be better, but it might be overkill for CV mults.
david

Sebo · Joined: Apr 27, 2007 Posts: 564 Location: Argentina

Thanks bubblechamber!!!
I will try TL072 first (I used 8 pins sockets), and if aren't good enough will
try the LF412.

urbanscallywag · Joined: Nov 30, 2007 Posts: 317 Location: sometimes

Isn't the LT1013 popular in low offset applications?

Dave Kendall · Posted: Sat Apr 11, 2009 11:52 am Post subject:

urbanscallywag · Joined: Nov 30, 2007 Posts: 317 Location: sometimes

He has a funny post somewhere about building a precision CV summer for less than a Venti coffee at Starbucks. Laughing

Tim Servo

The LT1013 isn't THAT expensive. Mouser carries the TI version for a little over one dollar (part #595-LT1013CP). Not bad for a precision low offset dual op-amp.

Tim (mmm... Starbucks...) Servo

Sebo · Joined: Apr 27, 2007 Posts: 564 Location: Argentina

I can't find LT1013 locally now, but I remember it was really cheap. I don't
think it worths to make an international order for a couple of opamps.
Thanks again!!!

Sebo · Joined: Apr 27, 2007 Posts: 564 Location: Argentina

Hi:
After all this time I finally do the circuit in a perfboard. First I tried TL072 but
for pitch CV the error was too big, up to a semitone. I replaced the opamps
with LF412 and they worked perfect, I don't have any noticeable error.
Just letting you know.

Edit: Some pictures...

Multiple (Buffered) Front.jpg

Description:

Multiple (Buffered) Front

Filesize:

102.54 KB

Viewed:

305 Time(s)

This image has been reduced to fit the page. Click on it to enlarge.

Multiple (Buffered) Back.jpg

Description:

Multiple (Buffered) Back

Filesize:

115.96 KB

Viewed:

341 Time(s)

This image has been reduced to fit the page. Click on it to enlarge.

johans121 · Posted: Thu Aug 13, 2009 9:50 pm Post subject:

Good job Sebo!

numbertalk · Posted: Fri Aug 14, 2009 8:52 am Post subject:

Hi Sebo,

Sorry to go a bit off-topic, but I've thought of building simple utility circuits like this for myself but there are a few things I'm still foggy on. What value did you use for the bypass caps? Also I notice you only have 1 cap per IC - sorry if this is a thick question but how exactly do you have those wired up - one lead to + rail and the other lead to - rail? This issue is something I'm unclear on - I've seen circuits have 2 per IC - one between + rail and ground and another another between - rail and ground. Do all chips need this or are more stable chips ok without it?

Which also makes me curious about something else I'm not quite clear on - what are the cases where you do and do not need a cap in the feedback path of the op-amp (between the output and inverting input) to correct HFO/stability issues in the signal path? Are some uses of the op-amp more susceptible to this than other uses (the voltage follower maybe not having these kinds of problems)? Or is it that this is a more stable chip?

Also are those 22uf elec caps you used for the power supply filtering?

Thanks.

Sebo · Joined: Apr 27, 2007 Posts: 564 Location: Argentina

Hi numbertalk:
The bypass caps are 100nF ceramic, but lately I'm triying to use multilayer
ceramics.
They are wired from +V to -V as the chip is not connected to ground.
The use of bypass caps is a good practice in general, there are various threads
here about it, and some papers around the net that could explain it better than
me, but bypass caps helps to isolate the chip from noise in the power rails,
also act as a little reservoir of energy, so when the chip needs a little more there
are a source handy.
The cap in the feedback path is to avoid high frequency oscillation, and yes
some chips are more suceptible to that than others. I think they are not needed
in buffers because the cap acts as a high pass filter that lower the gain of the high frequencies,
and in a buffer the gain is one (that's why I didn't use them here).
Yes the electrolitics are for power supply filtering, but I'm not sure if I used 22uF
or 10uF (I don't have the module here to check), but anyone will work.
Hope this help.

numbertalk · Posted: Fri Aug 14, 2009 6:41 pm Post subject:

Hi Sebo,

Thanks for all the info! This is very helpful.

Uncle Krunkus · Posted: Fri Aug 14, 2009 10:55 pm Post subject:

Yeah, I remember someone (I think it was Ray Wilson) saying that the feedback caps are mainly needed once the op-amp is providing a gain of *10 or more. This will vary, I s'pose a gain of *9 might need a little settling. Anything from 10pF to 100pF IIRC. You won't hear any difference until they get bigger than 100pF, but I s'pose anything will take the edge off a transient. Which may or may not be okay.

daverj · Joined: Jul 07, 2009 Posts: 28 Location: USA

It's better to use two bypass caps, one between each rail and ground, rather than a single cap between rails.

A single cap between rails will reduce noise between the rails, hopefully preventing the chip from oscillating. But it does nothing to reduce the coupling of that noise to the signal.

Even though the chip doesn't have a direct connection to ground, the signal you are feeding through it is a voltage relative to ground. Any noise in the power pins relative to ground can get coupled into the signal.

Modern op amps have fairly good isolation between the signal and power pins, but it's not perfect. And the higher the frequency of the noise on the power pins (relative to ground) the more of it will get through into the signal.