Beh ... da qualche parte bisogna pure partire. Tanto vale usare formulazze che ho postato (che poi sono vecchie come il cucco).
Come abitudine io misuro il ripple residuo e verifico che non ci sia motorboating (instabilità a bassa frequenza). Per quest'ultimo serve un oscilloscopio, per il primo basta un voltmetro rms.
Quando ho messo insieme i pezzi della PSU dell'attuale amp (a valvole single ended, per PP le cose sono diverse) mi sono tolto anche la curiosità di vedere come variava la tensione nell'ultimo cap con l'amp in funzione e con un segnale di 50, 100 e 200Hz in ingresso per ampiezze crescenti. In linea di massima se la cella è criticamente smorzata la perturbazione dovuta alla modulazione della finale è minore ... cioè la PSU fa il suo dovere mantenendo la tensione costante al variare della richiesta di corrente. Per celle sotto-smorzate la tensione ai capi dell'ultimo condesatore oscillava molto di più. Ecco uno stralcio degli appunti che avevo postato su
AA
Spero possa esservi utile. Credo di aver postato anche qui da qualche parte i risultati, ma non ricordo.
Case #1
Prim: 220V, 2.6Ohm
Sec: 700V-0V-700V, 180Ohm
866A half wave
10H/80Ohm
50uF Motorrun
4H/65 Ohm
150uF Motorrun
Q11=1/DCR*SQRT(L/C)=1.2 (1st LC neglecting core losses and 866 DCR)
Q12=1/DCR*SQRT(L/C)=2.5 (2nd LC)
RC12=9.75ms (last cap time constant)
Hum: <5mV (much less)
Or case #2
10H/80Ohm
200uF Motorrun
Q2=0.62
RC2=73ms (last cap time constant)
Hum:560mV
Case #2 was quite close to a critically damped LC cell.
So, what’s going on during the experiment? The tube is sucking current from the reservoir capacitors and they are recharged at the same freq (100Hz) through the chokes and the tranny. How can the tube feel the PSU? Well, I am no expert, but the fluctuation in B+ can be interesting… that’s what you look at with PSUD.
FINDINGS.
In case 1 you can see (herr … the pic is not that clear … I need a new digital camera) the B+ shows a 350mVrms 100HZ triangular wave. Going down to 14Hz that wave gets bigger: 3Vrms. It goes slowly to zero increasing the freq as expected.
http://img95.imageshack.us/img95/6277/100hzlclc5tq.jpg
In case 2 that perturbation in masked by the hum. I included a pic taken at 105Hz to show clearly that the hum is modulated. BTW the effect on B+ is lower now: the hum+perturbation is 650mVrms at 100Hz (NB: the hum is 560mV here). That perturbation is smaller than case #1 at all freqs. Maximum effect on power supply was found to happen at 3Hz: 2Vrms circa.
http://img88.imageshack.us/img88/7223/105hzlc3tq.jpg
CONCLUSIONS.
The old rule, make the LC critically damped, does work. I’ll follow it.
I’ll also strictly observe the other rule to tune the LC at freqs (much) lower than the classical audio band limit 20Hz.
And, why not?, I’ll also follow the suggestion by Thorsten Loesch … make RC not greater than 40ms. Ok 73ms was the best compromise I could get with the irons/caps I have.
** we are made of nothing but voices echoing from the past
and pointing their fingers to the future showing the direction. we will eventually follow that line **
"i was looking back to see if you were looking back at me to see me looking back at you"