criticise my layout

[catalin gramada]

Hallo. Still I tweakle my last build please to criticise my layout if see something wrong. Sort of 100w KT88 bass amp a friend asked for. Thanks



All voltages was recorded with amp dimmed under near clipping border for 120W clean sine output. Preamp first with flat tone control settings at 400 cps
Thanks

[loudthud]

I think the layout is OK with respect to grounding. The one thing I wonder about is running the PI off the Screen node and the effect the Screen node sag will have on the preamp. My gut tendency would be to add an additional RC filter between the Screen node and the PI.

[catalin gramada]

I think the layout is OK with respect to grounding. The one thing I wonder about is running the PI off the Screen node and the effect the Screen node sag will have on the preamp. My gut tendency would be to add an additional RC filter between the Screen node and the PI.

Hi.The reason was to keep voltage as high for max cathodyne excursion as much as grids.. Thanks for you intro.

[catalin gramada]

late: still thinking around and did not found any answer why should be a problem? not get sag over capacitor but over series elements, so I.m a little confused here. have 100u large enough to keep voltage topped up for such current draw. Am I missing some ? I did a test sweeped from 30 to 10000cps thinking low freq. transients will put in dificulty, but is right steady, so why to add a suplimentary decoupling point if I want to keep the PI voltage as high as grids to benefit by max excursion, please ? I can add suplimentary capacitance over existing one but not feel is necesary... should I? Thanks

[loudthud]

I'm going to just make up some numbers here, your mileage may differ. This all revolves around the fact that an L-C filter for the most part acts like an averaging filter with just a slight DC loss from the resistance of the coil.

At idle the main B+ node might have 10V pk-pk ripple going from 450V to 440V. The LC filter averages out the ripple and supply's 445V at the screen node. When the amp is putting out near maximum power the peak B+ will drop a little but the ripple will also increase. B+ might have ripple going from 440V down to 400V. the screen node will sag down to perhaps 420V or slightly lower. The L-C filter is finding the average. You can try beefing up the reservoir cap on the main B+ node to reduce ripple and sag, but that will only take you so far and add stress to the rectifiers and transformer.

[catalin gramada]

Yup. That.s the way it works. Thanks

[catalin gramada]

To complete the picture this is how my power supply HV structure looks like:



Found many 200w amps like Hiwatt,Sound City,Carlsbro used this type of arangement except for protection and stby. I prefered to modify.

[catalin gramada]

Found a better way to solve inrush current and loud startup stbsw. pop in one stroke using one DPDT switch. It still pops but very,very soft using 10ohm/10w resistor in primary side as max limit for my supply. Made it as in drawing below:

[catalin gramada]

Question:
Can be adjusted the rocking piece inside a DPDT switch to ensure one poles row of contacts are constantly delayed relative to second row of contacts ?



Need to modify the hight of contact or the geometry of the arm, please?

[J M Fahey]

Short answer: no.

IF you need that function, it can be (and is) done with *rotary* power switches, the Industrial kind.
A rotating Pertinax disk carries a set of switches, another Pertinax baseplate has rows of contacts, if in same line all switch at the same time, if staggered (displaced) they will make contact at a different moment.

VOX used a combination power/standby switch (hint hint). Rotary of course.

[catalin gramada]

Thanks JMF. Thinking to connect in one stroke but two steps in this order: I:none/none II:1.first row engaged/second row none 2. first row still engaged/second row engaged. It is still important in second step first row to remain continuous engaged and not jump from engaged to engaged. The delay should be 1/50 sec just enough. I.m pretty sure it is a tolerance delay (supposed by msec. order) in all double pole switches due to manufacture imperfection (not absolute syncronise), but want to be sure it realy is, constantly in the same order. I'll take a look at rotaries as you said.

[J M Fahey]

Of course you may build a two relay switch, triggered by a single visible/panel toggle one which actually enertgizes one relay coil straight away and the other *any* delayed time later.
Delay can be as simple as 1 diode + 1 resistor + 1 electrolytic, feeding the second coil, specially since delay needed is short and timing does not need to be *that* precise.
The relay board is internal to chassis, invisible and transparent to user.
You may build this for, say, $10 ... not sure what a custom or unusual switch might cost, even if you find one which matches the specs you ask.

EDIT: it might be even simpler: panel switch connects straight what you need (transformer primary?), slightly delayed relay (fed from rectified 6.3V?) connects later (automatic standby?)

[catalin gramada]

I know what you're talking. That is what I did in 2kw amp builded from scratch few years ago. Startup module was feed it from separate transformer



for a gtr. amp a soft start relay used to fix a standby mode looks pretty fancy, popping-more traditional aproach