4 x EL84 in UL PP, 470 B+!! is this right

[malcolmfraser]

Hi all, newbie here, with properly a silly question...

Looking at building the attached design, but with solid state PSU. If i have the calculations right B+ is at 470V, which is way above the 300 from the datasheet.

Is this right?, and just a function of the -20V fixed bias, quad EL84s and UL configuration?

Please help, confused ;-)

[Ty_Bower]

What your schematic shows is not PP, it is PPP (parallel push/pull). Not that it makes any difference with regards to B+ limits.

What your schematic does not show are any of the component values, nor the specifications of any of the transformers, nor even the tube types for the rectifier and the driver stages. I can assume the output tubes really are EL84, since the title block of the schematic mentions it.

470VDC for B+ does seem much too high. My instincts agree with yours - I don't think the tubes will handle that much voltage in ultralinear. There are designers who have used such a high voltage (indeed, much higher even!) but those designs were pentode amps and the screen voltage was held much lower.

How did you arrive at that value for B+? Did you try to build some simulation of the power supply? Does your model include a sensible approximation of the circuit's total current draw? You should probably assume the amp is going to pull at least 150mA at idle.

[malcolmfraser]

Thanks for the reply. I used PSU Designer II and the parts list. Full listing of parts below, plus the calibration proceedure (where i worked out the fixed bias rating)

Rated Power Output: 35 Watts
Frequency Response: 20Hz - 20kHz (+/-2dB)
THD: 0.5% (1kHz/35W)
Signal to Noise: 65dB (Input Shorted)
Input Sensitivity: 300mV
Treble Control: +16 ..-20dB
Bass Control: +18 ..-20dB
RESISTORS:
R1,R5,R8,R12,R16 = 1M
R2,R14,R18,R24,R25,R26,R27 = 1k
R3 = 82k
R4,R7,R13,R17,R19,R20,R21 = 100k
R6 = 150k
R9 = 220k
R10,R32 = 2.7k
R11 = 10k
R15 = 470R
R22,R23 = 330k
R28,R29 = 100R/1W
R30,R31 = 10R/1W
R34 = 27k/2W
R33 = 470k
R35,R36 = 10k/1W
R37 = 56k/0.5W
R38 = 22k/0.5W
P1 = 5M pot. lin.
P2 = 1M pot. lin.
P3 = 0.5 pot. audio taper
TR1,TR2 = 100K trimmpot
TR3 = 100R/0.25W wire wound trimpot.

CAPACITORS:
C1 = 0.047uF/100V
C2,C10,C14 = 47uF/16V
C3,C9 = 0.1uF/400V
C4,C5 = 47pF/50V cer.
C6,C7 = 4700pF/100V
C8 = 300pF/50V
C11,C13 = 0.033uF/400V
C12 = 0.033uF/100V
C15,C16 = 0.047uF/630V
C17 = 680pF/100V
C19 = 47uF/63V
C20 = 100uF/500V
C21,C22,C23 = 47uF/500V
T1 = Output Transformer - no data available
D1 = 1N4148
Power Transformer:
Secondary: 2 x 300VAC 240mA
Filament1: 5VAC/2A
Filament2: 6.3VAC/4A
Filament3: 6.3VAC/2A
Tubes:
V1,V2 = ECC83
V3,V4,V5,V6 = EL84
V7 = GZ34
Calibration Procedure:
Make sure all wiring and connections are correct. Remove tubes V3 - V6 from sockets,
leave only V1 and V2 inserted. Short amplifiers input to common ground.
1. Set both trimmpots TR1 and TR2 in center position.
2. Turn on power and check voltage at TP1, adjust with TR2 to ca. -20.0 Volts. Next turn off power
and insert all power pentodes.
3. Turn on power and let the tubes warm up. Connect DVM between TP2 and TP3 (control grids of the
pentodes), adjust with TR1 to zero Volts. Set DVM to 1 Volt scale and reset again to zero Volts.
4. Next connect DVM to common ground and one of the pentode cathodes, adjust with TR2 to 1130mV
(1.13V).
5. Now connect DVM between the V3 and V4 cathodes and adjust with TR1 to zero volts, use lower scale.
6. Finally measure voltage between common ground and cathodes of V3 and V4 pentodes. Should be 1130 mV
+/- 20mV. Repeat steps 3-5 if needed.

[Ty_Bower]

I couldn't find any data sheets that showed plate curves with Vg2 anywhere near the voltages you've described. I did put together a quick LTspice model. I can't say how realistic the model will be at those voltages, but it is the only thing I have.

With B+ at 470VDC and Vg1 = -20VDC, at idle I get 32.5 mA plate current and 2.5 mA screen current. Depending on the drop across your output transformer, that will be somewhere around 15 watts plate dissipation and 1.2 watts screen dissipation. That's a bit on the high side for an EL84.

If you tweak the bias voltage just a hair and bring it down to -21VDC, plate current drops quite a bit down to 23.6 mA. This might get your plate dissipation down to 11 watts, which is within spec. Screen current and dissipation drops too.

I think there were some old vintage receiver designs which ran the output tubes very hard. They might have used 7189A for the finals, or maybe they just ate through tubes like candy.

[malcolmfraser]

Thanks for the info, will definatley have a look at doing this. I am ordering the iron this week and want to nmake sure i don't have a pwr tranny that just turns out to be a valve killler.

Have you seen any interesting PPP at around 25-35 watts that should consider?

[kenpeter]

I have to ask why two screens sharing one stopper?
Won't damp any parasitic modes that can ping-pong
between them.

I am still of the opinion that each and every screen,
grid, base, or gate should have 33R to 330R in close
proximity to spoil stub resonances. That would be in
addition to any larger value resistor needed for other
purpose, limiting the currrent or Miller lowpass. Large
value (over 800R) resistors act as open ends for RF,
don't much serve to spoil the local parasitic.

Your value 330R is good, but you need one for each
screen, to break up the dipole stub between them.

Tetrode screens and plates have nasty habit of
slight negative resistance kink in the "flat" curve.
Especially when secondary emissions get absurd.
Negative resistances love to oscillate on any and
every undamped resonance they can excite.

Pentodes have a supressor, but nothing is perfect.
470V how could you not have secondary emission?
You need those stoppers. Especially if you run the
device outside the window of sanity.

Else you might find it bursts into oscillation every
time the plate dips below the screen voltage.
When done deliberately, its called "Dynatron".

http://www.vias.org/basicradio/basic_radio_30_02.html
http://www.vtodorow.com/index.php?op...d=14&Itemid=17

[Mickeystan]

Malcomfraser,
Since your power transformer has a secondary voltage of 300-0-300 VAC and you are using a capacitor input filter, I believe you have an error in your B+ calculation. I would expect the B+ loaded down to be on the order of 350V or so. With no load on this power supply type, I would expect you to see 424V. When you subtract the drop that will come with all tubes in place and the supply loaded through the rectifier tube and the secondary winding resistance losses, it should be more like 350V or so.

Mickeystan

[Ty_Bower]

Originally Posted by malcolmfraser

I used PSU Designer II and the parts list.

Power Transformer: 2 x 300VAC 240mA

4. Next connect DVM to common ground and one of the pentode cathodes, adjust with TR2 to 1130mV.

Just a few key details from your last post...

If you build a 5AR4 rectified cap input supply out of a 600VCT power transformer (2x300) and then draw 240mA from it, you will end up with a B+ somewhere around 360VDC. Assuming the rectifier can handle that load, the resulting voltage is reasonable. It's almost exactly the same voltage used in ultralinear 6BQ5 designs such as the Dynaco Stereo 35 and SCA-35.

If you follow the biasing procedure, you are dropping 1130mV across the 10 ohm bias set resistor R30 (or R31). That's 113mA for a pair of output tubes, or 56.5 mA per tube. Entirely too high for the 6BQ5 - your overall dissipation will be greater than twenty watts per tube! I think I must have misinterpreted the biasing procedure... I will have to go back again and follow more carefully.

[malcolmfraser]

OK, so how embarrassing is that!! Now I really feel stupid!!
When I did the design on the PSU Designer II, I didn’t think about load, just plugged in the voltage in and hit simulate. Dah, clean forget about the load.

[malcolmfraser]

On the bias calcualtions isn't the max cathode bias for a EL84 65mA?