can you check if everything is ok with el84 schematic please

[Trileru]

Hi all. I came here from the fullrange forum because after I settled for a Dallas II enclosure for my Fostex FE206E I decided to finish my EL84 pp amp that I started over a year ago. Can you please check the schematic to see if everything is ok please? Maybe if you can suggest some fine tunings I would appreciate. My current setup is as follows. I have a torroid for supply, one secondary at 240V 400mA (for both channels) and 2x6.3V 6A for all the heaters. Output trannys are custom made by someone from a local forum. They are HUGE and stay real cold after hours of music. He told me he did them at 6k6. They don't have ultralinear taps. only 2 cables output for 8ohm speakers. The caps are WIMA mks4v for 0.22uF and mkp4 for 0.1uF. If not appropriate I'm willing to swap them for something better. The 220k pot at input is junk, I will swap it asap. The 100uF electrolytic are junk as well. Cheapest I could find, only ones I could find in local stores in fact. My dillema is as follows. I wanted to know if it is worth it to apply global feedback to my setup. I applied it locally to preamp stage. And if not, is it worth it to keep one lead from the output wires connected to ground?

Tubes I use is one 6n1p-ev for preamp/PI stages, and 2xEL84 for output tubes. (one channel ofc).

My readings are:

Preamp 6n1p:
Anode: 82 V - 5.1 mA
Cathode: 3,7 V - 5.2 mA

PI 6n1p:
Anode: 298 V - 0.55 mA
Cathode: 26.2 V - 0.55 mA

Top EL84 (in scheme):
Anode: 308 V - 38 mA - 11.7 W
Cathode: 11.2 V - 41.48 mA

Bottom EL84:
Anode: 310 V - 37 mA - 11.47W
Cathode: 11.3 V - 41.85 mA

Measured the gain factor as well (100hz sine wave 205mV AC measured input from soundcard - M-Audio FastTrack Pro)
without global feedback: at bottom grid 391mV AC -> 1.9X gain factor
at top grid 391mV AC -> 1.9x gain factor ( here I had some problems, it seemed to oscillate from 390 to even 631, but mostly stayed in the 390-410 region. The upper point measured was dead stable at 391. maybe someone could explain what is wrong here)

with global feedback applied: at bottom grid 314mV AC - 1.5X gain factor
at top grid 311mV AC (same problems, oscillates till 405mV, but mostly 310-320)

*later edit*
top or bottom grid refers to the grids of the upper or lower el84 in the scheme


I found that global feedback reduced a lot of high frequencies and amplified the bass alot, but not in a very pleasant way, and reduces the volume pretty much. Everything at max and I can't tell that there's and amp at full volume somewhere in my house. Without GNF it's pretty loud but ... even that not very much. What am I doing wrong here? 6n1p has an amplification factor of 35. The output should be about 10W, so with Fostex FE206E I should have a pretty decent loud system. Also bass distors a bit less with GNF, more profound but the overall volume is lower so If I could have the mid+high from non-GNF with the profound bass from GNF all at a higher volume would be perfect, I could source higher quality components and start looking for a case for it

Forgot to mention the voltage from ABC:
A: 328 V
B: 326 V
C: 318 V
Voltage varies constantly 3-5 volts, sometimes after few hours I measure the Anode for EL84 and drops to 300V. Electrolytic caps from PS are junk as well, cheapest I could find, only ones I could find as well. Rectifier as well. I got some buzz in the speaker that you can notice without music at about 1 meter from it. But given the fact that the whole amp is spread on a breadboard I don't wonder In fact is very good that with all the wires and possible ground loops and crappy caps I don't have more. So I am open to suggestions, fine tuning and what other ideas you may have. I pretty reluctant to changing the structure of the stages at this moment but with a very good argument I may do it. I also have the following preamp/pi tubes that I can use: 6n2p-ev and 6n6p. For testing purposes I put a fostex speaker into a classic rectangular vented box of 134 liters with 10x10cm round vent. Only to hear the amp, I will put it into a Dallas II enclosure in future.
As well I have a pair of EM84 vu-meters that I would love to add to this amp, but only after everything is perfect, and I have enough juice to power everything without affecting the sound quality. Thank you and waiting for your input.

[MarcelvdG]

The circuit around the first valve looks very strange, particularly the two diodes at the cathode and the 700 kohm resistor to the grid. The feedback frequency compensation capacitor also has an unusually high value (usually produces a zero at some supersonic frequency, this one kicks in at 320 Hz or so). Could you explain the purpose of these unusual things?

[Trileru]

well the 700k resistor represents the Rf for local negative feedback. I red it on Valve Wizard:
"Since the feedback resistor is effectively in parallel with the anode resistor, we should make its value high or the open-loop gain will be reduced, and we would need a larger output coupling capacitor. Rf should therefore be at least three times larger than Ra, and preferably more. If we choose 1Meg we can be sure the open-loop gain will be hardly affected." So I went for 700k.

The diodes are LEDs used for biasing at 3.7V. One red and one green. Do you think that the 0.22uF is too large? Shall I go for 0.1?

[Trileru]

Please bear in mind that this is my first project in the field of electronics. Hence I might not be familiar with many concepts. But I studied the Valve Wizard chapters regarding this setup.

[Tubologic]

Your schematic as shown has many design flaws which must be corrected:

In a split-load phase inverter cathode and anode resistors must be identical and if possible should be closely matched. A 1:10 ratio (47k/4.7k) is obviously wrong.

The bias operating point of your P.S/driver stage is less than optimal: 550 µA (!) is much too low for this application and the tube will work in a very non-linear region with limited headroom and driving capabilities.

Returning the global feedback loop to a CC biasing diodes (or are they LED's ?) is unusal and not a good practice: unpredictable behaviour and distorsion will certainly result. You must use a cathode resistor in this place to return the F.B loop.

A local feedback in the first stage is not really needed if the tube is correctly biased. The distorsion generated by the output stage will be much higher and appears sooner anyway.
With this local feedback you must use two extra capacitors and direct coupling to the second stage is not possible. Why not use the standard and well-proven direct-coupled Williamson/Dynaco/RCA , etc... topology here ?

It is perfectly normal that global feedback will reduce the gain of the amplifier but it should not introduce peaks or dips in the frequency response if well implemented. (on the contrary, GNFB must linearize the frequency response). This is another sign that something is going wrong with your circuit.

I suspect the poor performance you noticed is not due to component selection but poor circuit design. There are plenty of well-proven simple circuits to choose from, no need to re-invent the wheel.

[Trileru]

Well, I noted it wrong there, it is actually 47k not 4.7k. I remove the local feedback and applied global one. Sounds much louder, but still clips at 3/4 volume. But it is way louder. The p.s/driver stage runs at 2.74mA now, but I have another problem. The amplification factor measured in the 2 locations is now about 16. Little less on the upper tube for example i put 60mV 100hz sine wave and got on upper part 1.10V and lower 1.18-1.2V. And the overall voltage jumped about 30V in A B and C points.
Attached the new scheme.

[Tubologic]

Looks better now but there are still two mistakes:

Your feedback signal is shorted to ground by the 100µF cathode decoupling capacitor: you must either remove this capacitor entirely or split the cathode resistor in two parts series connected with the upper decoupled and the lower not. Feedback signal must be connected to the node of these two resistors. Values of these resistors must be computed depending on the bias and feedback needed. Usually the lower (undecoupled) resistor has a much lower value and forms a voltage divider with the feedback return series resistor. The ratio of these two resistors will set the total NFB.

The second stage triode must be either self-biased or biased by direct DC coupling to the previous (input) stage, but not both at the same time. For direct coupling the 330K resistor is no longer needed. You must also check that the P.S triode is correctly biased : the grid potential must allways be a few volts lower than the cathode potential. All this is very empirical, the exact values need to be computed to optimize the circuit parameters.

[Trileru]

I removed the 330k resistor and I added a 57ohm resistor (made from series resistors as I don't have exact values) and connected the NFB between the two resistors. The bias is at 2.2V witch is great for preamp. I will check the biasing of the PI stage, I believe it is the 2.2k resistor that must be played with. I attached the new schematic.
Thank you very much for helping me, it meas a lot!

[Trileru]

I biased the cathodyne at 2.2V with a 1k resistor instead of the 2k2 one. Attached the modifications.

When I apply the NFB the slight buzz that I had disappears completely and at max nothing distorts, and the attenuation in volume is not that bad, but NFB completely kill my highs. It's like I have only mid+low. Sounds like an old radio. If I disconnect the NFB the highs are crisp clear but everything distorts past half volume knob.

[nigelwright7557]

The phase splitter resistors are still not the same.