In my previous post, I showed how to build a square wave floating above ground using an operational amplifier. I originally did my test at 5 Khz, which worked quite well but gave a nasty humming sound.
The signal must therefore be above the hearing range; and I have settled on 25 Khz.
Now, if you studied electrical engineering, you would have learned about the “ideal” amplifier. Software simulations often implement it as well: infinite gain, infinite bandwidth & infinite input impedance are the main characteristics. On paper all operational amplifiers perform the exact same task, but in reality they are all very, very different. So for my VFD filament power supply, I ordered a bunch of op-amps and put them to the test.

Testing them is quite simple: all I have to do is slot them into the female headers. Since all op-amps have the same input/outputs, it makes my life a lot easier and also make sure that the only parameter that changes during the benchmark is the op-amp itself.

The jury is out!
The “too slow”: LM2904
LM2904 clearly struggles at 25Khz. It’s way too slow and shows a very high rising time: not suitable for this application. Next!
The “wtf”: TSH24, TS524 & MC33079

On paper TSH24 is amazing: 15 V/μs slew rate, 15 Mhz bandwidth… Except I never got it to work! This is the actual output of this amplifier and I am not joking. I honestly thought I did something wrong with this one, so I got a 2nd unit to make sure: same result. I have no idea what’s going on but this op-amp is unsuitable.
In the same vein, TS524 also gave me an unreadable signal:

And to finish off this bunch of weird results, this is the MC33079 –very, very similar to the TS524:

It seems that all these “hi perf low noise op-amp” are not suitable for the job. I honestly don’t know why though, but I’ll just leave it at that!
The “too weak”: OPA1679
Texas Instruments’ OPA1679 “Low-Distortion Audio Operational Amplifier” gave me pretty good results at first. As you can see below, the signal is very clean, with a very low rise time:
However, as soon as you attach a load to it, it collapses:
This tells me that despite the signal being buffed up by the 100 Hfe gain transistor, the op-amp does not supply enough current. It is a weird case as it’s supposed to provide up to 50mA per channel. I tripled check my soldering job; but no: this op-amp simply won’t cut it.
The “winners”: LF347B, TL074, TL084 & TS274
All the other operational amplifiers performed as you would expect. Choosing the right one is a fine balance between cost and performance. I eliminated the LF347B because it has an unwanted frequency response with a spike on the rising edge, clearly visible below:

In the end, I settled on ST Micro’s TS274 “High performance CMOS quad operational amplifier”. It is capable of driving 60mA per channel which is no laughing matter when it comes to op-amps; and the generated signal is stellar:

In part 3 we will build this circuit in a more permanent way!