The OPA2677 that I have been using so far is very convenient and I have used it a lot, but the limit is 500mW, and I have outputted 800mW, but it was dangerous if the heat dissipation conditions were not properly maintained. Therefore, I tried an experiment this time because there is a device that can output more power in a current feedback type op amp format. The major differences compared to the 2677 are: 1) Power supply voltage is 32V (±16V) 2) Slew rate = 8000V/us 3) Thermal shutdown function The disadvantages are that the 2677 comes in two pieces per package, but the 3491 comes in one piece per package, so while one 2677 was enough, two 3491s are required. The circuit configuration is the same as the 2677, with a BTL configuration, that is, two circuit outputs are connected via bridge, and twice the voltage output is obtained compared to the push-pull output method, and four times the power. Originally, the BTL circuit was created when car stereos first appeared on the market as a method for outputting high power from a 12V battery voltage, and while home speakers are 8Ω/16Ω, most car stereos use 3Ω/4Ω speakers. Enough of the introduction, let's move on to the main topic. There are two circuit diagrams, Circuit Diagram A and Circuit Diagram B. Circuit Diagram A is for when you want to connect with an input of 50Ω, while Circuit Diagram B allows you to freely set the input impedance, but the lower the input impedance the more stable it will operate, and with the same gain settings there is almost no difference in frequency characteristics. Additionally, the output circuits are matched so that both circuits output 50Ω. 　　　　 　 　　　　 The coupling coefficient of the input/output transformer determines the F-characteristics, so I created an FB-801 with a Trifilar and Φ2.6 with 5 turns (5t in terms of the number of turns on the core, 4 turns counting the outside of the core). Now let's run a simulation. 　　　　　 　　　　　 To test the actual device, I designed a pattern. I had a manager I know etch the PWB for me. The important thing here is that in order to get an output of more than 1W with this device, it is not possible to test it using a perforated board for testing. The board is double-sided, with the copper foil on the other side dissipating heat, and the aluminum plate provides the heat dissipation conditions. 　　　　 　　　　 　　　　 　　　　　　　　　　　　　　 I tried to check the characteristics with both ends of 50 Ω by applying power, but the oscillation was unstable with circuit diagram B, so I tried applying power with circuit diagram A. As a result, the thermal shutdown function was activated at a continuous output of 1.2 W or more, the signal was cut off, and the temperature repeatedly went up and down, repeating the ON/OFF operation. The heat dissipation resistance from the heat dissipation pad to the aluminum plate was too high. Maybe my processing method was bad, but in the end, I had no choice but to make a prototype with a double-sided through hole.