measuring OP-AMP gain-bandwidth

[mike zias]

I have a reasonably good lab set-up.  I am measuring the gain-bandwidth of some op-amps.  I have a 741, 358, and ancient 709.  On my 741 I get a 23 db/decade slope.  In theory this should be 20 db/decade.  I am confining most of time to 20 and 40 db gain.  How close to 20 db/decade is a real 741 op-amp?

[TimFox]

Over what range of frequency did you measure that slope?
(e.g., 400 to 40,000 Hz for two decades)
The slope will be less at very low frequencies and higher at very high frequencies.

[mike zias]

my 40 db gain was at 8.53 khz and my 20 db gain was at 75 khz.  Input voltage at 0.01 vrms.  However, I notice my signal generator output (HP 3311A) has a distortion on one side of the sine wave.  I have been using RMS voltage measurements so I thought the distortion would not be a big factor.  I have a HP 3581A wave analyzer, but it only goes to 50Khz.  I may try that and see what slope I get. 

[mawyatt]

Might try using 6.02dBV/Oct instead of 20dBV/Dec. Has advantage of finer granularity for spotting where slope changes.

Best

[magic]

3dB is a lot, you should be able to easily see it on a scope.

I got sensible results consistent with datasheet specs using the circuit below, which operates the chip at -1 closed loop gain and artificially reduces open loop gain by 80dB. Apply frequency of interest to in, measure out and err amplitudes (or phase if you care) and your open loop gain is V(out)/V(err)·R3/R4. Try not to put too much capacitive load on err, or maybe reduce R1,R2 resistance if the time constant is getting uncomfortably long (but note that the feedback network loads the output, as usual).

[Terry Bites]

https://www.analog.com/en/resources/analog-dialogue/articles/simple-op-amp-measurements.html

741s only exist because people keep buying them by accident. The 70s are long gone.

[Randy222]

I guess you need to define what the test spec is. Is -3db where you call it a start or end ?

[Kleinstein]

The -3 dB point wout be for the unity gain in the follower circuit.

For the gain bw product one is interested in the range with -20 dB/decade slope a little below the unitiy gain BW.  In most cases a frequency roughly a factor 10 to 100 below the GBW is used.  So measure the gain there and than calculate frequency times measured open loop gain.  Data-sheets sometimes give the frequency they actually use (e.g. 1 MHz for fast op-amps).