Building minimal PSU for LM1875 gainclone

[Pato666]

Hello, I hope you can help me out.

I've been reading lots of posts very useful, but I still missing something (forgive my brain ).

I would like to build a PSU for a bridged LM1875 GC, or maybe I'll use a single LM3886, I'd like to get at least 50W max out of an amp. I'm about to build the transformer, here where i live is very hard to find toroidal, so I'll build an EI myself (i have done it before, was very careless ).

I read these threads: Gainclone Power Supply

LM3886 Power supply

and many mores, this is what i get from them:

Let's say a single LM1875, 25W @ 8ohms speaker, what do i need?

PSU: Vcc + Vee ==> 25V + 25V (according to its datasheet)
let's say Ripple: 2% (just as a given values)
Regulation: 20%
Freq here: 50Hz

Transformer VA ==> Pd * 1.6 ==> 40VA (most recommend 1.5)
Vac Sec = 19Vac + 19Vac
Isec = 2.2A <-- is this correct? or it should be 1.1A ?

Vload RMS = 14.14 V (maximun ratings)
Vload Peak = 20 V
Iload Peak = 2.5 A

Vpp Ripple = 1.414 V

Cap = 17700 uF
Vrating Cap = 36V

My question: is it right this value for current? I'll explain how did i get to these values, but first I'd like to explain what makes me think there is something wrong here:

According to most of the reading i did, using VA = W * 1.6 is a good practice, that is 25W * 1.6 = 40VA for transformer.

If I use a 19 + 19V @ 2.2A transformer CT isn't its VA then 80VA instead of those 40VA? I think it is.

First makes me think that I did something wrong maybe for the calculus i use 19V for transformer, instead of 38V (19+19), because: 40VA/19V = 2.2A

I mean... what if: I want to use two separated transformers 220V/19V and 220V/19V (all Vac) what would be their rating current to fit LM1875 needs? In this case i believe it would be 2.2A each, or i missunderstood what i've been reading?

The problem also appears when I'm tend to use two LM1875 (bridged or not -that will be stereo-), it will be 4.4A then?


Which is my goal? To get the least transformer as i can, I know there is a compromise here, smaller transformer --> bigger caps, I'm just trying to find the most reliable combination between caps and transformers.

Here are the formulas i use:

Data:
POW max [W]: 25
PSU Vcc [V]: 25
RL [ohms]: 8
Vripple [%]: 2
Freq [Hz]: 50
Trafo Reg [%]: 20
Mains Tol [%]: 8

Calculus
VA = POW * 1.6
Vsec = Vcc/sqrt(2)+1.2 (1.2 because of the diodes)
Isec = VA/Vsec (mmmhhhh...)
Vload = sqrt(POW * RL) (I know it should be Z, instead of RL )
VLpeak = sqrt(2) * Vload
ILpeak = VLpeak/RL
Ripple Vpp = Vcc * ripple% * 2 * sqrt(2)
Caps = ILpeak / (2 * freq * Ripple Vpp) * 1000000 (to get units in uF)
Vcap rating = (1 + Trafo Reg) * (1 + Mains Tol) * sqrt(2) * Vsec * (mains voltage/transformer rated input voltage) --> according to AndrewT info, thanks for that.

I have created an excel file for this, and I'd like to correct any errors it may have and then I'd like to share it with you to improve it.

If this problem has been posted before, I'm sorry i didn't see it. Would you be so kind to point me where to find it? Thanks.

And thanks for your time and help.