What Makes An Audio Op Amp An Audio Op Amp?

This circuit tends to be slow, and won't output much current. Adding b-e resistors to the output transistors will speed it up some, at the cost of CTR. This circuit is best for slow HV stuff, like biasing electrodes.

Adding cascode transistors would help speed a lot, but that wrecks the nice simplicity.

Depletion mode mosfets might be interesting additions.

John

LM324's shared the bias current sources among all four sections. So when one section railed or slew limited, it would collapse the current sources and trash the other opamp sections. I tried using 324's as comparators once, just once. Channel-channel interaction made it unworkable. Crappy opamp design.

Bifet opamps, like LF347, make pretty good medium-slow comparators.

The other problem with LM324 (I emphasize LM, because I don't know if anybody else's is better) is that a tiny amount of ESD diode current in the negative direction, on any of the inputs, will spray charge all over the chip and do all sorts of nasty things. Did I mention that this is a crappy opamp?

Thompson is long on strutting and short on facts. I discovered the trick of loading the output to reduce crossover distortion, but my fix was to buy a better opamp. HC4046 charge pump, ditto.

John

The 4046 phase detector problem isn't the same as the 324's. You don't have to turn off the pulldown, you just need to nudge the output slightly one way or the other. That's a minor wart as opposed to a fundamental design flaw like the LM324's bias problem (which I didn't know about--I've never tried doing anything much with 324s).

The metal-gate 4046 is an excellent part otherwise. The HC4046, now, _that_'s a botch. They took a nice 1000:1 VCO design, which was quite pretty in a small way, and turned it into a 10:1 VCO that simply quits if its control voltage gets too low. If you try building a fancy loop with that, you're in for a surprise. I occasionally use the HC part for its phase detector, but never for its oscillator.

Cheers

Phil Hobbs

Reporting from Long Island... bias behavior depended on brand. Whilst at GenRad I banned Motorola 324's from the purchasing list.

...Jim Thompson

The dual 741 was the 747, not 741.

is

from

Just one thing, where do you get the output transistors?

??-/

sure the optos are, but you're using a emitter follower on the high side and common emitter on the lower side. The low side is going to be closer t0 its -200 volts than will be the +200 side.

That's my observation. Just something I noticed, but like I stated before, at 200 volts, who is going to worry?

Jamie

Alas! Always the reality question ;-) ...Jim Thompson

Well, both have a deadband in their transfer functions, and both benefit from a resistor to ground, and both are dumb designs.

We use a phase detector that we sort of invented... I did the outside stuff and Rob did the FPGA insides.

ftp://jjlarkin.lmi.net/Phase_Detector.jpg

The UP and DN pulses are hard FPGA outputs, not tri-state. Rob managed to overlap them a bit so there's no deadband. This also has a mode where it works as a delta-sigma DAC (with U4 on) so we can tune/calibrate the crystal oscillator when it's not being phase locked to an external reference.

John

I don't see that. Neither transistor knows what mode it's being used in. It just turns on.

John

They are part of the optocouplers.

John

That's the Motorola approach, used e.g. in the MC145152. That produces more reference frequency sidebands and requires more parts, but doesn't have a deadband.

Cheers

Phil Hobbs

Putting LEDs in the power rails of an op-amp is a neat trick. I first came across it in the Burr Brown 3650 and 3652 optically coupled isolation amplifiers. The data sheet is dated 1976. These hybrid devices have a pair of optical cavities, each containing an led and two matched photodiodes. One LED is in series with each of the power rails of an op-amp and one photodiode from each cavity is used to close the feedback loop of the driving op-amp. There is a load resistor from that op-amp output to ground. The isolated receive side uses the other photodiode from each optical cavity in conjunction with another op-amp.. There is no feedback across the isolation barrier, but the overal performance was excellent in its day.

The data sheet carefully disguises the details I described above, but I once dismantled a broken one to discover how it really worked.

John

Good grief! He doesn't get it! Astonishing.

John

Usage.

Hope This Helps! Rich

The deadband in the transfer function is a good thing, though, when your application is battery powered. The output, while good for 20 mA when you need it, takes less power in the 'deadband'. LM324 has less than half the quiescent dissipation of the contemporary LM348, with otherwise similar specs.

LM358, most people know, is a dual amplifier variant of LM324, and is just as deadbanded. It's similarly power-stingy.

ote:

else is

from

That's cool too thanks.

I guess the future circuit I have in mind is a replacemnt for a stacked piezo drive. ~ 100V diffrential drive at frequencies up to

George H.

I love the opa134 and 2134. (I want to keep pushing it so someone else will buy it my ~100/year is not enough.)

George H.

=A0 =A0 =A0 =A0...Jim Thompson

Sorry, silly question.

The open loop gain helps reduce input offset voltage?

George H.

=A0 =A0 ...Jim Thompson

=A0 =A0| =A0 =A0mens =A0 =A0 |

=A0 | =A0 =A0 et =A0 =A0 =A0|

=A0|

=A0 =A0 =A0 |

ide quoted text -