FWIW, a few comments about SS rectification.
Take the appropriate steps to control switching noise, by employing either snubbed UFnnnn or Schottky diodes.
Inrush current limiters, like the CL150, tame the worst part of start up surges. AAMOF, in an amp with a bias supply, NTC inrush current limiting the B+ is quite sufficient. The instant on bias supply electrostatically shields the O/P tubes against cathode stripping.
Be sure to include inductance in the PSU filter, when a large value cap. I/P is employed. Small conduction angles are intertwined with high order ripple overtones. The choke kills the "hash".
It's definitely not accidental that amps known for superior bass performance employ SS rectified B+. A sag in the rail voltage when peak power O/P is demanded may sound great in a guitar amp. It's (IMO) positively wretched in a HIFI amp.
Take the appropriate steps to control switching noise, by employing either snubbed UFnnnn or Schottky diodes.
Inrush current limiters, like the CL150, tame the worst part of start up surges. AAMOF, in an amp with a bias supply, NTC inrush current limiting the B+ is quite sufficient. The instant on bias supply electrostatically shields the O/P tubes against cathode stripping.
Be sure to include inductance in the PSU filter, when a large value cap. I/P is employed. Small conduction angles are intertwined with high order ripple overtones. The choke kills the "hash".
It's definitely not accidental that amps known for superior bass performance employ SS rectified B+. A sag in the rail voltage when peak power O/P is demanded may sound great in a guitar amp. It's (IMO) positively wretched in a HIFI amp.
Eli Duttman said:
hey Eli,
Good bass performance is not the sole territory of SS. There are many high-vacuum diodes that do quite well if run conservatively, and better still when paralleled...and then there's Hg vapour.
Also, the past resosnat frequency use of a choke is questionable at best. 60 cps fundamental, 120, 240, 480 and 960 is about the end of it. The 1920 cps is an octave past the usual/likely resonance, and thus well into the capacitor-ish behaviour of that iron.
Now a bit of low-capacitance construction techniques, and series connection of two inductors will deliver a lot better high frequency filtering.
cheers,
Douglas
Long time coil winder J.W. Miller is still at it.
Mouser stock # 5900-102-RC is a 1 mH. RFC that saturates at 400 mA. of DC. Combine that RFC with a silvered mica cap. to ground in front of the main inductance and the main inductance will not "see" the highest order ripple components. Good quality capacitance after the inductance should suppress any crud that sneaks thru.My position about Hg vapor rectifiers is well known. No, no, a thousand times no! I want no part of a toxic waste disposal issue. Now, 6AU4GTA damper diodes are another matter.
However, no volume production commercial design I'm aware of uses 6AU4s.
So, where you going to get the negative voltage in the first place?
I have implemented various schema for this scheisse, and it is possible to get the output tube bias and the input tail from a single supply;
So, show me, in detail, how you want to do this. There are exigencies; sometimes another supply is actually simpler than the acrobatics of getting your urge off an existing supply (whose current requirement may suddenly multiply).
Here's my current supply for this circuit:
You got a better idea, by all means enlighten us.
Aloha,
Poinz
I have implemented various schema for this scheisse, and it is possible to get the output tube bias and the input tail from a single supply;
So, show me, in detail, how you want to do this. There are exigencies; sometimes another supply is actually simpler than the acrobatics of getting your urge off an existing supply (whose current requirement may suddenly multiply).
Here's my current supply for this circuit:
An externally hosted image should be here but it was not working when we last tested it.
You got a better idea, by all means enlighten us.
Aloha,
Poinz
I've built the trioded EL34 PP version of Poinz' Machine. Excellent amp. I've tweaked it some, notably cascoded the 10M45s in the tail, level shifted g2 (~32V) instead of the simple resistor and diode to the plate (I worry less about g2 dissipation this way), and run the '34s hotter, 390V @65 mA. That's some yummy sound.... Those 6GK5s are the sh*t... I'll knock up an EL84 UL version at some point. Thanks again, Poinz, for sharing the design.
Geoff the 'nuck
Geoff the 'nuck
If I may ask, why is G3 not connected to ground? wouldn't this be a triode strapped Tetrode? not a Triode strapped Pentode?
I like this a lot, but I think I am going to run the plates at a little more voltage than 390 (I was thinking 450-500) and what Kind of power out should I expect? if its enough, this may double as another attempt to get my Maggie SMGs tubed up. an unforgiving flat 4ohms impedance, and 88db sensitivity.
I like this a lot, but I think I am going to run the plates at a little more voltage than 390 (I was thinking 450-500) and what Kind of power out should I expect? if its enough, this may double as another attempt to get my Maggie SMGs tubed up. an unforgiving flat 4ohms impedance, and 88db sensitivity.
I have g3 connected to the cathode. I didn't really notice much difference either way. If you were to connect it to the anode, I might differ from Poinz here and connect it through its own resistor. There was a post here by TubeLab who found less distortion with g2 connected to the cathode. I found reducing the value of the grid leak on the '34s to 150K or 100K helped. This may entail using 0.33 uF for coupling rather than the 0.22. 450 to 500V on the plates is going to put you deep into class AB1. I think the spec for total dissipation (anode + g2) for an EL34 in triode is 30W. Running at 390V 65 mA (anode + screen), -25V bias, into a 5K primary gets me about 16W out, perfect for 88dB Spendors. I don't know how electrostatics will fare with a zero feedback design...
Guys,
The current demands made on B- and C- supplies are modest. A multi-stage voltage multiplier along the lines of the 2nd schematic shown here will do quite nicely. Construction employing Schottky diodes and a 6.3 VAC trafo yields approx 7.5 V. per stage. Build with LARGE value 'lytics and place "hash" killing RFCs in the line(s) leading away from the multiplier. Multiple rail voltages are a "no brainer".
The current demands made on B- and C- supplies are modest. A multi-stage voltage multiplier along the lines of the 2nd schematic shown here will do quite nicely. Construction employing Schottky diodes and a 6.3 VAC trafo yields approx 7.5 V. per stage. Build with LARGE value 'lytics and place "hash" killing RFCs in the line(s) leading away from the multiplier. Multiple rail voltages are a "no brainer".
alexmoose said:
Hey-Hey!!!,
Look at the EL34 triode curves. Treat the PP out as a SE of half the a-a load for Class A load lines. Let's look at the 400V B+ and see what happens...
400V and the EL34 plate dissipation of 25W leaves us a maximum idle current of 63 mA. So for 6600/3=3300Ohms. V_swing = .063*3300 or 206V. That is swing 206 low as you increase the plate current to 126 mA and swing it up 206 as you cut it off. By the curves I found here:
http://www.mclink.it/com/audiomatica/tubes/el34.htm
we'd need higher load( numeric ), or about 350V of B+ with a 6k6 plate-to-plate load in order to stay Class A.
cheers,
Douglas
alexmoose: with higher voltages you're limited by the dissipation rating of the tube. The higher the voltage, the lower the bias current. If you're building this with driving maggies in mind, you won't want to go Class A: not enough power. Too low a bias current/too high a plate voltage with EL34s sounds like crap. And you'd want to find out what kind of damping the maggies need before you go ahead with a no NFB design. There are probabbly better tube designs out there to drive electrostatics...
Bandersnatch: 6600/3 = 2200 * 0.063 = 139V
Bandersnatch: 6600/3 = 2200 * 0.063 = 139V
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.