Winfield Hill's 100W, 10MHz, 1000V/uS AMP-70
Update August 2021
Finally got around to complete my two builds in a nice enclosure. I found that both had an output DC offset of some +1.2V. Short of adding a small DC servo, the best quick solution is adding an 180k resistor from the +15V supply to the input summing node, as shown in the pic. With this mod, the DC offset drops to about 50mV.
This section is about the building of Winfield Hill's AMP-70 project. AMP-70 is a test and measurement amp, that looks innocent enough as a 100W audio power amp, but with a large signal bandwidth of minimum 5MHz! This amplifier has been designed by Winfield Hill (of Art of Electronics fame) for electromagnetic field experiments, but is also a great audio power amp. Some of us have also use for it driving power supply regulators with a combination of DC and AC to measure things like PSRR and Zout as a function of frequency. AMP-70's 5A, 50V ouput capability will cover pretty much all audio power supply regulators out there. Consider this as a build thread - I will expand and amplify the page as I and others progress in completing, testing and using this amplifier.
05 June 19 - Starting with the power supply subsystem. Mounted all supply caps, did the mod for the bipolar DC-DC converter (see List of known issues). Mounted all DC-DC converters. Tested all OK.
14 August 19 - Complete the first amp, up to and including a single output pair (Q21, 26). Running at +/-30V which is the maximum my current lab supply supports. Current consumption is 300mA which is nexactly what I calculated: 40mA each for the Q3.4 and Q12,13 chains; 80mA through the bias transistors Q20, 24; 60mA through the drivers Q19, 25; 80mA through the one output pair Q21, 26. There is an extra 350mA from the pos. supply for U1 DC/DC converter and downstream converters.
1kHz sine wave looks good, gain is ~10 x which is as expected: input opamp gain is 2, (R81/R80), main amp gain ~5, R02A/R01. But there is a ~6MHz low level (200mV pk-pk) oscillation visible at the output.
Modified the amp as follows to try tro cure it:
1. Connected the side of C01 that was connected to the output, to the emitter of Q25, as recommended by the German Tigris group;
2. Connected 33pF across the opamp feedback resistor R81, and connected the free end of R82 to gnd with a strap as foreseen;
3. Connected a zobel of 20 Ω and 0.1nF across the amplifier output;
4. Inserted a 10 Ω base stopper directly at the base pins of Q19, 25 (also recommended by Tigris);
5. Isolated the TO220 devices mounting plate from gnd, and let it float; and connected it to Vout through a resistor to 'bootstrap' the output devices tab capacitance;
None of these sequential changes had any impact on the 6MHz low level oscillation. Any hints welcome!