Op Amp Null Tester

This project started out as a test to see if I could measure the fine differences in the many types of op amp chips. I had already detected some variations when testing the LM358 against other parts but I wanted to be able to take a closer look at some of the devices commonly used in effects pedals.

I decided to design a null tester to use as the tool to give me a better look at the very low level details of the circuits.

The photo to the left shows the first version of the null tester. The power switch is at the top, and a multi-turn potentiometer is at the bottom to adjust the null. There is also a multi-turn trimpot for fine adjustments. It is the blue rectangle near the lower left edge of the pcb.

The basic idea is to send a wave from a signal generator into the null tester, and combine it with the same signal that is passed through a device under test (DUT) but inverted in polarity. This method cancels anything coming from the input jack, including the test signal, hum, and harmonics from the signal generator leaving only the distortion products generated by the DUT, which are amplified (+40db) by the OPA1612 and then meaasured.
If the op amp is extremely low distortion, then there are little to no distortion peaks in the output signal, as shown in the example down below. Note that the cancellation of the input signal is more than -50db, which is a reduction of 99.68% of the input signal. The graph shows the 1kHz signal at -10db but there is +40db gain on the output stage so the 1k test signal is actually at -50db.

I expected better rejection of the test signal but there are a several reasons why it is not more. First, although I used good quality metal film resistors, I did not match the resistor values as closely as I should have. This led to a small error in the signal strengths that reduced the cancellation amount. In the next version, I will take more time to match the resistor values closely. Also, during the testing, I noticed that the depth of the cancellation drifted as the circuit warmed up. If you compare the graphs of the test results, you can see that the ua1458 cancellation is better than that of the LM4562. These were the last (ua1458) and first (LM4562) chips tested over a period of about 15 minutes. The null trims were adjusted at the start of testing and not changed during the process.

Finally, in reading information on other audio null test devices, I found that slight changes in phase can have a big impact on signal cancellation. I was somewhat aware of this before starting, and designed the circuit with the same number of passive components of identical values in each signal path. However, I did not measure the capacitor values and there may have been a very tiny phase shift that degraded the depth of cancellation. I have seen schematics of other null testers that allow trimming of the capacitor values to minimize phase shift. In the end, this does not seem to be worth the effort and -50db was more than adequate since everything but the main test signal disappeared into the noise floor.

The null test circuit was powered by a pair of new alkaline 9v batteries, as was the device under test. This worked just fine for the first round of testing but in the end I decided to change the power supply for the next version so that the mixer/gain circuit is powered by two 9v batteries but the DUT is powered by a separate 9v battery to better simulate the conditions commonly found in a stompbox. It also allows me to try depleted batteries or other power sources for the op amp being tested.

When, I build the next null tester, I will match the resistors more closely, and probably check the values of the capacitors as well (why not). During the test, I will allow the circuit to warm up for at least 15 minutes before making final adjustments of the null trims. I am not sure about this, but I may adjust the values of the summing resistors to lower noise, though I do not want to go too low because some of the common chips are not capable of driving heavy loads.

The new null tester will actually be the third version of the pcb. The second version pcb had changes in the null trim, and was ordered but never used since I had already revised the design to include a different power source for the DUT and thought that was important enough for a new board revision. Another change in the latest version is to not use an smt op amp for the gain stage. This allows me to use a socket so that the gain chip can be switched out. I have some smt-to-dip adapters on hand that I can use if I need to try a chip that is only available in the smt package.

Share this article with your Twitter followers: Tweet