Power Amp- Current Build

[silverfox]

 "This project has gone through a stage of design that is no longer applicable to the current build. Due to mismanagement on the part of Silverfox, irrelevant posts are removed."

[Ed_Chambley]

I am not sure if a JJ is internally connected EL34 or not.  It doesn't matter.  Niether pin 1 or 8 should be grounded in fixed bias amps.  Look at any schematic on a fixed bias el34 and you will see pin 1 and 8 connected.  For consistency, both tubes should have pin 1 and 8 connected.  If you are using a 1 ohm resistor, you can simply use the lead of the resistor to do so and ground the resistor lead on the other end.

If you put in a single Mullard el34 with pin 1 and 8 not sharing a connection you will get a pricy surprise.  A 5881 can be connected the same way.  Then there is no need to concern yourself which tube you put where.  If pin 8 is grounded in any way, you will have a cathode bias t0 ground with no resistance.  The amp will play, but the tubes will fry.

Also, IMHO combining 5881 with el34 make for strange bedfellows considering the differences in OT requirements.  It "will" work, but not efficiently.  A supro with 6l6 or 5881 has a 5k9 OT in cathode bias.  I have purchased many Mercury Magnetics OT for supro and the are all 5K9 and el34 are looking for 3k4.  The mismatch is not really a problem, but the el34 will not sound the best.

Quick story.  I had a customer build a 5E3 and he said it sounded great but after a couple of hours the distortion would get very nasty and finally the amp volume would fluctuate.   He had inadvertently grounded the cathode.  I was not sure what he had done, but when I turned on the amp the PT got hot.  Anyway, I cut the ground and placed the correct resistor from pin 1/8 to ground.  No more tube overheating and PT runs cool.

I think you missed what Sluckey stated.  He is correct, I can assure you.  When he chimes in on something I am doing, I do what he suggest and it always works, always. If the idea is to have the mid power of the el34 and the roundness and breakup of 5881, one or the other will suffer simply because of the OT mismatch not to mention the differences in idle bias needed.  I am not saying not to do it as it may be something you like, but I would consider separate cathode bias for each tube or if fixed dual bias.

[HotBluePlates]

My understading is pin 1 is internally connected on an EL34.

Pin 1 on the EL34 is the suppressor, and is not internally connected, according to Mullard and Svetlana.

You're thinking of the 6L6 and 6V6. These tubes don't have "suppressors" and the data sheets call them "beam pentodes". They have beam-forming plates which perform the function of the suppressor and side-step European pentode patents. As you see from the data sheets, the "third grid" (which is the beam plates in these tubes, but would be the suppressor grid in a pentode) is internally connected to the cathode.

It is working. With pin #1 unconnected on the EL34-L output tubes.

 It doesn't need to be fixed. I just find it odd that pin 1 is unconnected and it sounds fine.

You can drive a Ferrari with 4 flat tires, press the gas and it will go. But if you want to run laps at the track, you'd better get good tires on it.

Connecting the EL34's pin 1 (supressor) to pin 8 is simply a handy way to get it either at cathode voltage or ground (if the cathode is grounded for fixed bias). You could just run a wire from pin 1 to ground.

What happens if you don't? Well, now you have a tetrode instead of a pentode. The tube will still work, but you're guaranteed to have less output power. You won't notice the difference until you try to crank the amp all the way up (a full explanation is fodder for a different thread; suffice to say that tetrodes haven't been common audio output stages since maybe the 30's).

An EL34 must have the suppressor grid (pin 1) connected to something.

Before you think my statements contradict Sluckey's, the problem with tetrodes is when you do crank that amp up, the screen grids are gonna draw a lot of current that should have gone to the plate, and the screens may exceed dissipation and melt. Maybe you'll avoid catastrophic failure, but either way, you won't get full performance at high output from your EL34's.

... I followed the schematic and as you can see, no pin 1 is shown. This is a schematic that Tubenit drew for me ...

It's a drawing, not the Bible.

Tubenit had to make his own tube symbols for the program used to draw that schematic, and his amps often use 6L6's or 6V6's. He doesn't have to worry about externally wiring pin 1, or even depicting a suppressor grid or beam plates in his schematics.

I know he didn't intentionally lead you astray. I'm sure he simply used the schematic symbols he had available and assumed if you use EL34's you'll know what to do. Pins 2 and 7 aren't on the schematic, yet you connected those to the filament winding.

For what it's worth, pin 1 is a recurring source of confusion. Pin 1 on the old metal-encased tubes often connected to the metal shell (see the RCA 6V6), which was almost universally grounded. Glass-encased versions came out, and there was no longer a need for pin 1 to be connected to the metal shell for grounding. Maybe the pin would be there, connected to nothing. Maybe the pin would be connected to the metal base of the tube (like some 6550's and others). Maybe the pin wouldn't even be installed.

The safe bet is to connect pin 1 either to ground or to the tube's cathode. If you plug in EL34's, you make the proper connection for the suppressor. If you plug in 6550's with a metal base, the base either gets grounded or is at a reasonably low voltage. If you plug in modern 6L6's, it doesn't do anything but also does no harm. If you plug in a metal 6V6 or 6L6, the grounded metal shell offers some shielding.

... I grounded the center tap of my bias circuit and connected the other two leads to a full wave bridge circuit. ... this caused the circuit to malfunction- overload due to a ground loop somewhere; don't know where.

A bridge rectifier can never have a grounded center-tap. The center-tap is what caused the short, and in a high voltage supply will usually destroy the bridge and/or transformer winding.

Grounded center-taps are for regular full-wave, non-bridge, rectifiers.

[silverfox]

In the interest of clearing up the Topic- Anyone opposed to me removing some of the old post. Since this is now another rebuild but I don't think it should be listed as another new power amp build?

I'll just summarize everything prior to Ed_Chambley's post as follows: "This project has gone through a stage of design that is no longer applicable to the current build. Due to mismanagement on the part of Silverfox, the previous irrelevant posts are removed."


Silverfox.

[PRR]

> bridge rectifier can never have a grounded center-tap.

It can, *if* you want plus and minus supplies relative to ground.



This is *the* standard way to make bipolar power for DC-coupled transistor amplifiers.

It is nearly zero use (and usually a stinky mistake) for tube amps. Main exception: Analog computers with DC response.

------------------------------

EL34 pin 1 *must* be connected to something. It *is* internally connected to G3, and EL34 G3 is not (as in some other tubes) connected to Cathode.

Yes, it will work (idle and small-signal) if pin 1 is floating. As HBP says, full-power operation is uncertain. Get swarms of secondary electrons bopping between plate and G2 and back, G3 near cathode keeps them from bopping TOO much.

In my brief tests, pin 1 G3 only needs a resistor, perhaps as large as 470K, not a hard short, to be stable at full power on a dummy load. I never measured any significant G3 current.

Pin 1 G3 connection is NOT why you were unable to trim G1 voltage. Something is not connected like you think it is (was).

[silverfox]

PRR Said- "Pin 1 G3 connection is NOT why you were unable to trim G1 voltage. Something is not connected like you think it is (was)."

Whatever it was, I had cut the existing dual bias circuit completely out and rebuild it nice and neat on a separate board to determine the proper component values; and then that started to get messy. Was unable to come up with a circuit that would adjust below 42 mv for the bias so I scrapped the entire build. The separate bias winding tap on the power transformer was reading 52 VDC.



Full power question: The May 1st posting has the schematic I'm using now. It has voltages listed on it so I should be able to more easily diagnose problems if they arise.

I don't know what level signal it will take to drive this amp to full power. Being a stand alone design, I would think a mixer or effects pedal will do so. ???

Silverfox.

[HotBluePlates]

Full power question: The May 1st posting has the schematic I'm using now. ...

I don't know what level signal it will take to drive this amp to full power. ...

Unfortunately, it doesn't list bias voltage for the output tubes, making the calculation uncertain.

I looked up some EL34 amps, and it seems that with 450v B+ for the output tubes, a bias in the range of -30 to -35v is typical. Let's assume you will need -35v as a worst-case (even though EL34 plate voltage is only about 400v).

From EL34 grid-to-grid, you need 70v peak (35v for one tube, 35v for the other, based on assumed bias voltage) to drive the tubes to full power.

Without calculating anything for the long-tail, its gain from one output to the other is typically the same as a single 12AX7 stage alone. We'll call that a gain of 55, and assume it needs to provide 70v output.

70v/55 = 1.27v peak, or about 0.9v RMS. That's a good healthy line-level for mixers/outboard gear, or the output of a preamp.

I don't know what level signal it will take to drive this amp to full power. Being a stand alone design, I would think a mixer or effects pedal will do so.

A line-level output from a mixer would do it, an effects pedal almost certainly won't.

It's not really "standalone". The input to the master volume says, "connected to preamp" so it assumes you have a preamp to develop the line level input to the phase inverter.

[silverfox]

drgonzonm said: "I checked out the JJ EL-34 cutsheet, and it does indicate one grid(screen) is wired to pin 1. I believe as others have stated, pin 1 and pin 8 should be wired together for this PP application"

www.jj-electronic.com/pdf/E34L.pdf 

I wired this as shown in the schematic posted above- AX84 site. It appears then that there will be a potential difference of whatever the bias voltage is between the two grids.

That means to me that, I have to connect both grids to the same side of the 1 Ohm resistor. The schematic shows pin 1 grounded.

Fox.

[HotBluePlates]

I wired this as shown in the schematic posted above- AX84 site. It appears then that there will be a potential difference of whatever the bias voltage is between the two grids.

Bias voltage in a fixed-bias amp is applied to pin 5.

The AX84 schematic shows a 1Ω resistor (written as "1R"). Since it's there to help you monitor tube current for to adjust bias voltage, the voltage difference between the cathode and G3 (pin 1) will be 10's of millivolts at idle, and probably less than 200mV or so at full power.

[silverfox]

The tube sockets are wired as shown in the diagram.

I understand the bias feed is on pin 5.

Preliminary information: The JJ EL34-L data sheet shows pin 1 and 8 are not connected. The schematic listed above shows pin 1 connected directly to ground and pin 8 connected to ground through a 1 ohm resistor.

Question: This being the case, will a difference in potential between the two screens have any effect on the output of the tubes?

I'm tending to agree with "drgonzonm" that I should wire this different then shown in the AX84 schematic.

silverfox.

[silverfox]

My mistake. I had confused the screens and cathodes.

Yes I wondered if a potential difference between that grid, and the cathode would influence the output? I see now that would not be the case since the (grid/screen) would be at ground.

I just looked at the tube diagram again. I'm going to go back and read basic valve theory again and probably start a topic on how this output tube works because as I look at it, it is very confusing. Why? because the first thing that comes to my mind is when the input signal appears on the control grid it should simply travel to the B+ and melt the tube or something like that.

I don't expect an answer on that last part. I'm going to go reread tube basics and then I'll start a thread.

Silverfox.

[HotBluePlates]

Yes I wondered if a potential difference between that grid, and the cathode would influence the output?

You need to go back and review what makes a pentode a pentode. The front matter of an RCa tube manual is a good place to read this (digital copies can be downloaded here), or review one of the general tube electronics books here.

Short answer: No, as long as the suppressor is substantially lower voltage than either the screen grid or plate. You'd ideally like it to be at cathode voltage or less, but as long as it is lower than the lowest voltage the plate swings to with maximum current draw, it should still do its job.

... when the input signal appears on the control grid it should simply travel to the B+ ...

The control grid is a helix of wire not physically connected to anything. The input signal can't travel anywhere.

The control grid simply creates an electrostatic field that influences the current stream moving from cathode to plate.

Even though a solenoid operates electromagnetically instead of electrostatically, it's probably useful for you to visualize how it works. Current in the solenoid coil creates a field that makes the solenoid plunger move. There's no connection between the coil and plunger (or from control grid to cathode or plate) but the field that is created when it is energized has influence on the plunger.

The input signal on the control grid creates a roughly similar field that influences plate current. It just does it based on influence of electric charge field rather than influence of magnetic field.

[silverfox]

First I should point out I attempted to modify the topic and only managed to mess it up if you are following the old history. I tried to get rid of the confusing post in the beginning but was unable to do so for anyone but myself. So now it is more confusing. Sheesh.

As for the amp- I got it going this afternoon after swapping the output transformer leads and finding a missing ground connection.

This is my first build that has sounded fantastic. Initially the bias was up to over 100 mv but due to the fact I was monitoring the bias voltage on the pin 5 feed, I didn't realize that until later. It still sounded better than any I had built. Once I got the bias circuit range resistor correct and understood that increasing the bias voltage drove down the idle tube current, things took off. No more hot power transformer and the amp really growls on low frequencies.

The lows literally shake the house, the highs are audible and there is a real clarity to the tone. I'm using an amp emulator but the tones I'm describing now came from only using Reverb, a Metal Zone distortion setting, slight Delay and a scooped mid range EQ.

http://ax84.com/static/corepoweramps/50W_PP/AX84_50W_PP_Poweramp_Schematic.pdf

The power supply is different and I don't know if that is an improvement or not. I also split the bias into two feeds that are powered from a separate bias tap on the output transformer. The supply caps consists of 100uf 350 VDC caps in series instead of the 50uf 450 VDC shown in the AX84 version. Not sure if that improves bass response or not but there is no sag all the way up. The closest I've heard to the kind of bass I can get was a Bugera 6262. Perhaps the .1uf coupling caps are responsible for that.

I'm finished with this for now and will move on the the Univox 1221 build.

If there is any interest in pictures or clips I'll post.

Thanks for all the patience and help with this project.

Silverfox.