• WARNING: Tube/Valve amplifiers use potentially LETHAL HIGH VOLTAGES.
    Building, troubleshooting and testing of these amplifiers should only be
    performed by someone who is thoroughly familiar with
    the safety precautions around high voltages.

Solid state versus tube rectifiers in a PSU

Jump to Latest
andyjevans

andyjevans

Member
Joined 2003
artsandmedia.com
#1
What's the general preference between solid state and tube power supplies?

I've been building power supplies recently, and for some reason I prefer a tube rectifier like 5U4 for the 2a3 output stage. But paradoxically I prefer a solid state PSU for the 10Y driver stage with Wolfspeed 1200V Schottly diodes. The same diodes on the output stage sound flat and lifeless. I've tried this a few times with the same results. I'm not complaining though it does mean 2 supplies. Any thoughts welcome.

What are your preferences, and why?
 
#3
Yes - output is 303V to the 2a3 stage. Same mains transformer, same choke, same capacitors.

Could be the current - 120mA for the 2a3s, 16mA for the 10Ys.

Also, the 2a3 stage drives the speakers so the damping factor etc comes into it. The driver stage only drives the output stage.
 
Last edited:
#4
I have not used a vacuum tube rectifier since about 1996.

I use some HEXFRED diodes, and plan on using more.
Slightly expensive, but less than a quality vacuum tube rectifier, and they will never fail in my amplifiers.

I think that overall performance is best with Choke Input B+ filters.
Ground loop noise is far better, the spectrum of choke input B+ filter consists of only very low frequencies; versus the low and especially the high amplitude very high frequency spectrum of Capacitor Input B+ filter.
The ground loop transient current is far larger with cap input filters, with much faster rise time of the current, not to mention the additional heating of the power transformer secondary due to I squared x secondary DCR with a cap input B+ filter.

And, B+ cap input filter current transients can reflect onto both the primary and all the filament windings too.

Be real careful of the orientation and spacing of the B+ choke input filter, to the output transformer.
Do not use a magnetic steel chassis.

On the average (pun intended), choke input filters are better.

Most of my amplifiers have less than 100uV of hum and noise at the 8 Ohm load.

Your mileage may vary.

Just my opinions.
 
Last edited:
  • Like
Reactions: prairieboy
#6
The sonics is usually the last thing I think of with rectifier decisions, because there are so many other decisions to decide first both pros and cons. The rectifier might matter, but it is so far upstream from the signal path (music) that does pass through the power supply. That I think the quality of the last filter capacitor probably matters more that the rectifier three times removed upstream in the Pi filter. Assuming a traditional amp, that doesn't involve Ultrapath or anode CCS which isolates the PS more. Assuming a 1950's basic Pi filter amp.

1) Tube rectifier ----
pros - nice slow start to the B+; softer rectification as far as ringing and reactance with transformer and 1st capacitor; looks cooler;
cons - big voltage drop; heat; reliability; not as stiff as diodes saggy(? newby unsureness); best with choke input so even more voltage loss; doesn't like a big C1;
2) Choke input ----
pros - better regulation; easier on the diodes tube or SS; eliminates all the hash generated by C1 in a cap input supply; many say sounds better;
cons - no voltage gain; limited to constant current amps due to unavailability of swinging chokes; critical inductance must be assured, but that is easy if load is constant;

Maybe the best compromise is your favorite SS diodes with choke input topology. That way you don't pay the huge voltage drop cost of a tube on top of the no-voltage-gain of choke input. These together is a huge cost in efficiency. Probably an even better choice is choke input but with a very small .15 - 1.5 uF input capacitor. This will up the voltage loss just a little, yet not enough to turn it into a cap input topology. The small cap up front will also smooth out choke to diode reactance(? newby unsureness). Another possibility is hybrid tube/SS, but that only re-introduces the voltage drop cons again, but it does add the slow B+ and protects the amp if the tube fails.

I've been going around and around on this too, I think I'm settling on this compromise over voltage loss, reactance noise/heat, I know I lose the slow B+ advantage but since it's choke input I'm not slamming the inrush anymore either so there is a win there:

1) State of the art HV Schottky
2) Choke input topology
3) With a small film input capacitor .15 uF to 1.5 uF (small so as not to wander into cap input territory).
4) A DC Link final capacitor bypassed by a small film capacitor
5) LCRCLC (trying to involve a second choke if possible)
6) With capacitors ascending in uF across the whole filter for example (L - 60uF - R - 100uf - L - 150uF) with the middle R conveniently being used as my trimmer (within +/- 10%).
7) Heater elevation divider, which also serves as my bleeder.
8) Optionally: Quasimodo snubbers, fuses between the AC and the diodes
 
Last edited:
  • Like
Reactions: Hipocrates
#12
Semi-con and HV (high vacuum) rectifiers act differently in the transition region between on and off, and the capacitances are different and differently variable with voltage/current. All rectifiers cause di/dt spikes though their transformer winding, which is coupled into other windings and (somewhat) coupled forward into their DC supply. These noise spikes can be dealt with by including damping Zobels across the secondary windings.

Excellent discussion of the issue can be found in Linear Audio issues 5 and 10, by Morgan Jones and Mark Johnson, resp.

All good fortune,
Chris
 
  • Like
Reactions: DCE1198
#13
Windcrest77 said:
The sonics is usually the last thing I think of with rectifier decisions, because there are so many other decisions to decide first both pros and cons. The rectifier might matter, but it is so far upstream from the signal path (music) that does pass through the power supply. That I think the quality of the last filter capacitor probably matters more that the rectifier three times removed upstream in the Pi filter. Assuming a traditional amp, that doesn't involve Ultrapath or anode CCS which isolates the PS more. Assuming a 1950's basic Pi filter amp.

1) Tube rectifier ----
pros - nice slow start to the B+; softer rectification as far as ringing and reactance with transformer and 1st capacitor; looks cooler;
cons - big voltage drop; heat; reliability; not as stiff as diodes saggy(? newby unsureness); best with choke input so even more voltage loss; doesn't like a big C1;
2) Choke input ----
pros - better regulation; easier on the diodes tube or SS; eliminates all the hash generated by C1 in a cap input supply; many say sounds better;
cons - no voltage gain; limited to constant current amps due to unavailability of swinging chokes; critical inductance must be assured, but that is easy if load is constant;

Maybe the best compromise is your favorite SS diodes with choke input topology. That way you don't pay the huge voltage drop cost of a tube on top of the no-voltage-gain of choke input. These together is a huge cost in efficiency. Probably an even better choice is choke input but with a very small .15 - 1.5 uF input capacitor. This will up the voltage loss just a little, yet not enough to turn it into a cap input topology. The small cap up front will also smooth out choke to diode reactance(? newby unsureness). Another possibility is hybrid tube/SS, but that only re-introduces the voltage drop cons again, but it does add the slow B+ and protects the amp if the tube fails.

I've been going around and around on this too, I think I'm settling on this compromise over voltage loss, reactance noise/heat, I know I lose the slow B+ advantage but since it's choke input I'm not slamming the inrush anymore either so there is a win there:

1) State of the art HV Schottky
2) Choke input topology
3) With a small film input capacitor .15 uF to 1.5 uF (small so as not to wander into cap input territory).
4) A DC Link final capacitor bypassed by a small film capacitor
5) LCRCLC (trying to involve a second choke if possible)
6) With capacitors ascending in uF across the whole filter for example (L - 60uF - R - 100uf - L - 150uF) with the middle R conveniently being used as my trimmer (within +/- 10%).
7) Heater elevation divider, which also serves as my bleeder.
8) Optionally: Quasimodo snubbers, fuses between the AC and the diodes
Click to expand...
If using SS diodes in a full wave configuration on a CT power transformer with choke input, you can use that old trick of a TV damper diode connected to the CT. It will drop volts - but you're gaining some with the small cap prior to the choke - but will give a soft start. Damper diodes are cheap and tough.
 
  • Like
Reactions: Windcrest77
#14
  • Thank You
Reactions: Chris Hornbeck
#15
andyjevians,

I am not sure if they were Vishay; but they were 1200V and 8A, same as the Vishay.
The vendor no longer has them, but I should find some and purchase a few while they are still available.

Silicon HEXFRED and Silicon Carbide are obviously different semiconductors.
But I bet both of them work very well for B+ service.
 
#16
jcalvarez said:
Can we include SMPSes in the discussion? They are solid state after all. I'm interested to hear the pros and cons.
Click to expand...
It was my intention to confine this to linear power supplies otherwise it would get very diffuse and scattered. SMPS is a different technology in many ways, and typically for low voltage applications rather than tube B+.

It's my intention to try HEXFREDs as well.
 
You must log in or register to reply here.
Top Bottom