PSU RC filter vs. VR regulator tube

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
First, the amp is a 2 channel P-P using 60FX5 pentodes.
I have a Triad N51X (120-120 isolation transformer)
The 'Q' point is 7mA per channel, and at full signal it's 70mA.
I want B+ to be 156V (150V + 6V cathode bias).

I am having trouble understanding the design of the power supply.

If I use an RC filter, I can get the correct 'Q' voltage, but when the amp is at full signal the B+ voltage will drop to 130V because the current is also flowing through the series resistor.

I've been reading posts on VR tubes, but they also have this series resistor.

I've tried using Duncan's PSU Designer and LTSpice, and they both confirm that the voltage will drop under full signal load.

However, this is a common design, so what am I missing?
 
It will only work if the VR tube can handle in excess of 70mA. The VR tube acts the same way a zener does. It clamps the voltage by pulling current through the series resistor.

Say you select your resistor to provide 80mA of current with 150V across the VR tube. 7ma goes through your output tubes, and 73ma goes through the VR tube. Now on a peak signal swing, the tube draws 70mA, and the remaining 10ma goes through the VR tube.

It is necessary to run 10mA through the VR tube at peak demand of the output tubes to insure the VR tube stays in regulation.

Under these conditions there will be a constant 80mA through the resistor, and your voltage will remain stable.

Problems: VR tubes generally are only good to about 40mA.
 
@TheGimp, thanks for the explanation, it looks like VR tubes are out of the question.

@azazello, I have a bridge rectifier connected to my transformer secondary. So I get ~166V. I need to drop 10 volts at 15mA, so R ~ 660 ohms. However, at 150 mA, R would drop 100 Volts.

Even 660 ohms seems extremely high, I must be missing something.

Is the current needed at full swing supplied by the caps? I would think the current at full signal would be pulled through the PSU series resistor.
 
Administrator
Joined 2004
Paid Member
I don't think the extra 10V is that big a deal , you could slightly adjust the cathode resistance to maintain a quiescent current of 7mA at the higher plate voltage.(Or allow it to creep up slightly if dissipation is OK) Determine whether or not the plate dissipation is within acceptable levels at idle and you should be fine.
 
This (your conundrum) is why the Good Lord made inductors (chokes), Eddie. Significant filtering (and more over, 2nd order), without attendant unwanted voltage drop.

Use a RCLC configuration, and you get your cake and eat it too. The rectifiers → resistor → capacitors significantly cuts reverse recovery noise. The R doesn't need to be very large, either. It should drop about 3% to 5% of the total supply voltage at peak continuous current. You probably know your formulæ, so this should be easy.

The C (first one) is mid-sized, to not stress out the rectifier array. 40 μF to 100 μF is fine. The L however needs to be significant: one might think that the 'higher the better', but actually like the capacitor, modest values mitigate sudden current draw slumping. 10 H to 50 H range. Then the next capacitor is big. How big? Pretty physically big! For a 150 volt supply, I would likely spec something between 2,200 μF to 10,000 μF. Simply put, besides power-supply ripple reduction, the larger these reservoirs, the more they stop momentary power-supply sagging on peak music-program waveform excursions. Don't need to be ridiculous about the size, but bigger is better.

All this you can easily model in LTSpice.

As to regulation in general? It is even more awesome compared to passive methods (like resistors or inductors) at either series modulating the final reservoir capacitor, or shunt-regulating the same. Awesome in that with a sufficiently ripple-reduced pre-regulator front end, the regulator will reduce ripple to nearly unmeasurably low levels and keep output voltage nearly constant. Magic devices in some regards. Microsecond reaction times, too.

Lastly, and entirely as opinion more than hard science, I would venture that one can simply eliminate the heavy/costly choke in lieu of a CRC pre-regulator power supply arrangement plus regulator. And the CRC doesn't need to be very robust. 47 μF, 330 Ω, 470 μF.

GoatGuy
 
Why are you worried about 10V in 160V ish? Your mains supply may vary nearly as much as this, and there is no need to define valve bias to this level of accuracy. In the valve era 20% variation was the norm, and 10% was regarded as high tolerance.

I've measured my mains, 123V RMS at the fuse box in the basement, 122V RMS upstairs in the room this amp will be used. But your right, that value could change at any moment for any reason.

I'm running the tubes at the maximum plate voltage (150V). I guess I could run them at 160 with 6V cathode bias. I'm a 50 year old engineer, I've always designed "within specified limits". Old habits :)
 
Well if you like to design within tolerance, then ± 10% is easily within that tolerance. You might want to use a regulator scheme … to keep within the tolerance (they're very good at that). Burn a few watts in the power regulation department to keep solid, stable, line-cord independent volts supplied to the tubes.

But, in the same measure, I advocate not using specific-voltage negative bias on the grid side! One of the most useful constructs invented in the 1920s was "cathode bias" utilizing a resistor to 'raise the cathode' a few volts making the ground-referenced grid relatively negative. The genius of the approach is that the ΔV adjusts itself to the operating condition of the tube, accommodating small variations in manufacturing tolerances, runtime-degradation and so on.

Just saying. Set your voltages nicely, remove hum from the DC side aggressively, but also choose stage topologies that are self-adjusting over time and component ageing.

GoatGuy
 
I'm using a solid state bridge rectifier, and 160V is only 6% above the limit, so I think I'll just use some filter caps (I have a bunch of 330uF 350V caps), cathode bias the tubes, and call it day.

It's just a small 4 tube 4 watt amp, not hifi, used with an MP3 player or connected to my computer. Not worth fretting over the power supply.
 
Seems you have got where you wanted to go, but aside from your RC voltage drop, remember the power transformer winding itself has resistance, and so the AC voltage leaving the transformer also varies with current draw.

The max values in the tube book are not brick wall numbers, and the above mentioned 10-20% tolerances were perfectly standard anyway. But a 150v tube won't disintegrate at 151v. In the world of guitar amps, Fender ran 6V6 tubes in their Deluxe, Deluxe Reverb, and other models at a full 100 volts over the RCA max spec. These amps are still working just fine 60 years later. They do not have a reputation for eating power tubes.

You are building more or less a table radio, no offense. You describe it as small, low power and not hifi. It will not stress your tube.
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.