![[-]](https://theultimatetone.com/images/collapse.png "[-]")
Yesterday, 07:28 PM
Hi Kevin,
I made some progress tonight and the results seem to show that the excessive full-output voltage drop is not simply due to the tolerance of the PS pot or R5.
The amp is being tested with all four EL84s removed and PS fully clockwise. The excessive drop was present both before and after I added the low-end PS3 trim, so that trim does not appear to be the cause.
The originally fitted components were:
- PS pot: CTS 1MA, measured 990k end-to-end
- R5: removed and measured 326–327k
- R23: previously lifted and confirmed 36k
- Added PS3 trim currently set to 176k
As an experiment I replaced R5 with the lowest value 3W 330k resistor I had, which measures 323k, and temporarily replaced the 990k audio PS pot with a 1.1M linear pot. The linear taper was only used for testing the full-CW endpoint.
With 1.1M PS pot + 323k R5, EL84s removed and PS fully clockwise, I measured:
Point Voltage
W1-1 / B+ input 381.7V
W1-2 / OT centre tap 330.9V
Voltage lost through plate regulator 50.8V
R5 PS1 end 38.1V
PS2 / pot wiper 38.2V
M1 gate 38.18V
M1 source 31.92V
M1 drain 382.1V
M2 gate 337V
So changing from the 990k pot / ~326.5k R5 combination to 1.1M / 323k only reduced the full-CW loss from roughly 55V to 50.8V.
There are a couple of things that look significant:
PS1, PS2 and M1 gate are effectively identical at full CW, so the pot is reaching its high-end correctly and the command voltage is getting through R1 to M1.
M1 is also clearly operating: gate 38.18V, source 31.92V, so VGS is about 6.26V; its drain is essentially at B+.
M2 gate is 337V, while W1-2 is 330.9V, so the gate is only about 6.1V above the regulated plate output. M2 is therefore still dropping the entire 50.8V between its input and output.
R5 also seems to be behaving sensibly. With 381.7V at its B+ end and 38.1V at its PS1 end, the 323k resistor has about 343.6V across it, corresponding to roughly 1.06mA.
Given your simulation examples, I would have expected the measured combination of approximately 1.1M PS pot, 323k R5 and 36k R23 to be considerably closer to the desired full-output condition than this. The change in pot/R5 value has moved the output slightly in the expected direction, but clearly does not account for the remaining ~51V loss.
My next proposed experiment is to temporarily parallel the 323k R5 with approximately 10M, giving an effective R5 of about 313k, close to the value in your example that produced roughly a 4V drop. I have not done that test yet.
At this stage it looks to me as though the excessive drop may be arising further into the M1 / QP driver / M2 gate-drive section rather than simply from the PS-pot/R5 tolerances.
Do these voltages suggest a particular node or component in the M2 driver section that you would check next?
Best,
Steve
I made some progress tonight and the results seem to show that the excessive full-output voltage drop is not simply due to the tolerance of the PS pot or R5.
The amp is being tested with all four EL84s removed and PS fully clockwise. The excessive drop was present both before and after I added the low-end PS3 trim, so that trim does not appear to be the cause.
The originally fitted components were:
- PS pot: CTS 1MA, measured 990k end-to-end
- R5: removed and measured 326–327k
- R23: previously lifted and confirmed 36k
- Added PS3 trim currently set to 176k
As an experiment I replaced R5 with the lowest value 3W 330k resistor I had, which measures 323k, and temporarily replaced the 990k audio PS pot with a 1.1M linear pot. The linear taper was only used for testing the full-CW endpoint.
With 1.1M PS pot + 323k R5, EL84s removed and PS fully clockwise, I measured:
Point Voltage
W1-1 / B+ input 381.7V
W1-2 / OT centre tap 330.9V
Voltage lost through plate regulator 50.8V
R5 PS1 end 38.1V
PS2 / pot wiper 38.2V
M1 gate 38.18V
M1 source 31.92V
M1 drain 382.1V
M2 gate 337V
So changing from the 990k pot / ~326.5k R5 combination to 1.1M / 323k only reduced the full-CW loss from roughly 55V to 50.8V.
There are a couple of things that look significant:
PS1, PS2 and M1 gate are effectively identical at full CW, so the pot is reaching its high-end correctly and the command voltage is getting through R1 to M1.
M1 is also clearly operating: gate 38.18V, source 31.92V, so VGS is about 6.26V; its drain is essentially at B+.
M2 gate is 337V, while W1-2 is 330.9V, so the gate is only about 6.1V above the regulated plate output. M2 is therefore still dropping the entire 50.8V between its input and output.
R5 also seems to be behaving sensibly. With 381.7V at its B+ end and 38.1V at its PS1 end, the 323k resistor has about 343.6V across it, corresponding to roughly 1.06mA.
Given your simulation examples, I would have expected the measured combination of approximately 1.1M PS pot, 323k R5 and 36k R23 to be considerably closer to the desired full-output condition than this. The change in pot/R5 value has moved the output slightly in the expected direction, but clearly does not account for the remaining ~51V loss.
My next proposed experiment is to temporarily parallel the 323k R5 with approximately 10M, giving an effective R5 of about 313k, close to the value in your example that produced roughly a 4V drop. I have not done that test yet.
At this stage it looks to me as though the excessive drop may be arising further into the M1 / QP driver / M2 gate-drive section rather than simply from the PS-pot/R5 tolerances.
Do these voltages suggest a particular node or component in the M2 driver section that you would check next?
Best,
Steve
