Ok. Just now got 339 Vac iddling and 329 Vac at max power and from 439Vdc to 413Vdc at first node cap please. This difference shows a 580 mA AC difference in current over 17.2 ohm HT winding from iddling to max. It is clear the PT cannot support the nominal current the power stage asked for. It drops to much and will overheat.
Last edited:
The current draws from power tubes shows 100mA around for max power. That meant a 60mA difference from iddling 40mA state. For 6 tubes it mean 360mA more than iddling state. The voltage dropped at secondary side is 10Vac which, in respect with winding resistance of 17.2 ohm meant a 580mA AC difference. For sure I have an measurement error somewhere but I don't get it where it is ?
I think I cannot get the right measurement measuring the current directly from voltage reading over the winding resistance because the winding is not a matter of resistance but more impedance meant. I should place a series resistor to measure the current instead... so that's the point my measuring errors occurred I think...
The sense resistor is the proper way of measuring the current. With a true RMS reading meter. But suppose your current rise is 580 mA at full sine wave power. Are you listening to music that way? If so, maybe you do need a bigger transformer. But if your power needs are more typical, you will need maybe 1/3 of that current rise to support your average listening level. Measure the current draw under music conditions, at a level where the clipping has just gone to the intolerable point. Only you know where that is. Could be just audible, or at party levels and pretty distorted. You know how you will be using this amplifier. Make sure your transformer is sized to be able to handle that comfortably and continuously and it will be fine. Your output power measurement (for comparison and benchmarking) should always be at full throttle sine wave, but hi-fi use is never that demanding. If it’s a guitar or DJ PA amp, typical use can be with the controls dimed and you design the power supply accordingly.
With tube amps, the transformer is usually sized larger than for solid state because of the higher quiescent current - in your case something like 290 mA. It’s always that high, then you account for the rise. With sand amps, quiescent is negligible and practically all of it is current rise.
Class A amplifiers are easy to build a power supply for. The current draw won’t budge under normal use so that exercise has been done for you (and not left to the reader).
Supply dropping from 439 V to 413 at full sine wave is perfectly reasonable and expected. You may measure 425 or so at the onset of objectionable clipping.
With tube amps, the transformer is usually sized larger than for solid state because of the higher quiescent current - in your case something like 290 mA. It’s always that high, then you account for the rise. With sand amps, quiescent is negligible and practically all of it is current rise.
Class A amplifiers are easy to build a power supply for. The current draw won’t budge under normal use so that exercise has been done for you (and not left to the reader).
Supply dropping from 439 V to 413 at full sine wave is perfectly reasonable and expected. You may measure 425 or so at the onset of objectionable clipping.
Ok. Did a test with new PT. primary: 230V , 1.2 ohm DCR, secondary 4 ohm DCR, 1.5A rated I get the follow measurements: mains 233V , secondary 363Vac, first cap 483Vdc iddling with 6x54mA = 6x2.7mA scr = (308mA) plates + 58mA screens and the rest of circuit = total iddling current 366mA.
At onset clipping 36Vpk/4ohm- 160W output- the voltage drop at first cap from 483V to 470V and in the screen node cap from 476V to 456V. My only concern now is screens dissipation: dimmed with 12.5V over 470ohm stoppers meant 26.6mA x 444V= almost 12w - no good, should be reduced by half...and the start-up problem of course
At onset clipping 36Vpk/4ohm- 160W output- the voltage drop at first cap from 483V to 470V and in the screen node cap from 476V to 456V. My only concern now is screens dissipation: dimmed with 12.5V over 470ohm stoppers meant 26.6mA x 444V= almost 12w - no good, should be reduced by half...and the start-up problem of course
Last edited:
Regulated tends to be better, rather than using dropping resistors. Most of us cheat and use mosfets, or use a lower voltage tap on the supply that’s the right value without regulation or dropper resistors. Sand-o-phobes can make a regulator out of tubes. It doesn’t need to be perfectly regulated - you just want to avoid wild swings with signal level which introduce IMD.
Easiest thing in the world to do. You kill the king.
Take your string of zeners and shove a source follower after it. Use a gate stopper of 100 ohms. Clamp vgs with a 15 volt zener, and vgd with a reverse bias UF4007. Put a power resistor in the drain circuit calculated so VDS falls to zero at whatever current you want to start limiting at. RC network between the zener string and gate stopper if you need to get rid of zener noise for powering a preamp - unnecessary for a screen supply.
Sone designers prefer the screen to track the main B+, having it sag by the same %. This tends to maintain the same Iq as the supply varies when fixed vg1 is used. In that case, just use a resistor divider in place of the zener string. For those that go all tube, a cathode follower is used instead of a mosfet. It gets a little messier, as you usually need an isolated heater winding for that tube. Triode/pentode or dual triode combos can be used if you go to the trouble of doing real regulation with an error amplifier.
Take your string of zeners and shove a source follower after it. Use a gate stopper of 100 ohms. Clamp vgs with a 15 volt zener, and vgd with a reverse bias UF4007. Put a power resistor in the drain circuit calculated so VDS falls to zero at whatever current you want to start limiting at. RC network between the zener string and gate stopper if you need to get rid of zener noise for powering a preamp - unnecessary for a screen supply.
Sone designers prefer the screen to track the main B+, having it sag by the same %. This tends to maintain the same Iq as the supply varies when fixed vg1 is used. In that case, just use a resistor divider in place of the zener string. For those that go all tube, a cathode follower is used instead of a mosfet. It gets a little messier, as you usually need an isolated heater winding for that tube. Triode/pentode or dual triode combos can be used if you go to the trouble of doing real regulation with an error amplifier.
I use 47 and 68V zeners, with 12 and 15 V for trim (And other uses). Tolerance can be an issue, often a 68V is over 70. And you do have to watch dissipation in the zener and the pass device. If you get 306V when you were shooting for 300 it’s close enough. It is the simplest way to make an arbitrary screen voltage without custom iron.
...well, I used 51V 5W Imax 95mA. 7pcs. Expected voltage: 7x51=357V. But real voltage with 25mA through = 385V. That meant divided by 7, 55V. Okay, still acceptable.
Now, real deal in circuit : Input 470V, 1K series resistor, only 45mA through , meant 425V regulated (the screens draw 1.5mA x 6= 9mA total) at 45-9= 36mA through diodes. So my 51V zeners regulate 55V at 25mA and 60V at 36mA despite the fact are max rated for 95mA...
So, for around 40mA I think something like 30V will fit better, and work cooler of course.
I cannot limit the current more my screens draw 12mA each at 380V around so 72mA total. 1k series resistor fit well for this purpose please.
Those zeners seems works correctly rated for 1W around. I don't get it why they rated as 5W if not proper work? Even at 2.5W it smoked and are hot as hell...
Now, real deal in circuit : Input 470V, 1K series resistor, only 45mA through , meant 425V regulated (the screens draw 1.5mA x 6= 9mA total) at 45-9= 36mA through diodes. So my 51V zeners regulate 55V at 25mA and 60V at 36mA despite the fact are max rated for 95mA...
So, for around 40mA I think something like 30V will fit better, and work cooler of course.
I cannot limit the current more my screens draw 12mA each at 380V around so 72mA total. 1k series resistor fit well for this purpose please.
Those zeners seems works correctly rated for 1W around. I don't get it why they rated as 5W if not proper work? Even at 2.5W it smoked and are hot as hell...
Last edited:
I really don't understand how usable this zeners are. I did a jig with no less than 13 pcs of 30V / 5W. 480V input 1K series limiter. There are 7mA to screens and 40mA clamped through diodes . I expected 390V output but real output is more 430V around. The voltages over diodes are 33.5V around. There are still not usable at 1.3W ?
