2A3 SE

2A3 SE amplifier using Italian C-Core output transformers

Inner parts:

  • C-Core output transformers (prim. winding resistance 170ohm)
  • Italian Power transformer (310-0-310V unloaded voltage)
  • Hammond 158L chokes (15H, 420Ohm)
  • Spirit filament transformer 250-240-230-0V to 0-2.5V 0-2.5V

Tests for presence of blocking distortion. The idea behind this test is fully described in Merlin Blencowe’s book “Designing high fidelity valve preamps” (page 160). The grid-stopper resistor in this case is 1k and the grid-leak is 270k.

Now, testing with grid-stopper 20k and the grid-choke 1000H. The oscillations as predicted by LTSpice.

There is a post with a similar simulation and conclusions, here: https://www.audioasylum.com/messages/tubediy/171226/grid-choke-vs-grid-resistor

Difference in the frequency response for 1k (grey lines) and 20k (color lines) grid-stoppers. Larger values for grid-stoppers reduce the bandwidth from top, but also make it more difficult for the blocking distortion to happen.

Test using a grid choke 1000H (5.5k resistance, grid-stopper 20k) instead of the grid-leak resistor of 270k. The resonance peak of +10dB is at around 7-8Hz.

Modeling of distortion using LTSpice:

Modeling of the driver stage using Dmitry’s software

Some tests of two channels with grid-chokes for THD

Test of the low frequency

The following schematics helps reduce the 100Hz, 200Hz and so on using the trick from here http://diyaudioprojects.com/mirror/members.aol.com/sbench/humbal.html

In the case of 2A3 it was possible to reduce only 100Hz with 6-8dB and a bit modified schematics (see the second peak for the left (up) channel). For the first peak the 100 Ohm potentiometer was not always adjusted. That does not influence the injection of 100Hz and above into the grid of 2A3.


16-11-2019 New experiments with the grid choke and 2.2uF coupling caps. The gray color shows the response with the choke and the green one corresponds to 270k grid leak resistor:

Testing blocking distortion with the grid leak resistor of 270k and burst input of ~50Hz, the displayed measurements are on the grid of 2A3 without any grid-stopper resistor.
Testing blocking distortion with grid choke and burst input of ~50Hz, the displayed measurements are on the grid of 2A3 without any grid-stopper resistor.