Building EL-84 push/pull amp out of Zenith Radio Parts

[olddawg]

Of course it is subjective, but I have seen very little difference in whether a 2 x EL84 amp has a tube or solid state rectifier in sound, touch, sag or whatever. The B+ and other design and parts choice (especially speaker) factors of course make a difference. Using a SS rectifier takes less space, produces less heat, is much more inexpensive, and more reliable. The Zener string method for B+ works. People usually mount them zig zag on a perf-board and elevate that to breath.

[Chuck H]

There won't be 10V less on the other side of the zener unless the current through the resistor multiplied by the resistor value comes to more than 10V. When the resistor has 9V across it so does the zener (they are in parallel). At 9V the zener is non conducting which means all the voltage drop is down to the resistor so there will be resistive sag. When the zener starts to conduct the voltage across the resistor is clamped to 10V so there will be no more resistive sag.

Ach! Of course you're right! "I" must have been suffering from "resistive sag". The way you describe it is exactly how I've been using them. I should wait until after my first cup of coffee to post from now on.

[DRH1958]

Great advice, this is how I do it, zig-zag and leave an extra set of lugs in case I want an extra diode. In addition to elevating them is also leaving the leads of the diode long as this also adds a small amount of heat dissipation.
In response to whether there is much sound difference, I DO know there is quite a difference between SS, 5AR4, 5U4, and 5Y3 in voltage sag. This is not debatable. It's simple ohm's law. More resistance, more voltage drop. Whether the sound changes I suppose is in some debate. It seems unavoidable to some extent since the voltage change on the plates gives tubes a different sound. I suppose it comes down to how fast the rectifier can recover, so filter cap values will play a role too.

EDIT: I suppose I should have said "how fast the power supply can recover" above. Very interesting thread. Near and dear to me because EL84s are my favorite tube and even though sometimes reasoning fails us, it still is thought provoking and that is always good!

[Dave H]

The way you describe it is exactly how I've been using them.

That's what I thought because it's similar to your zener across the EL84's cathode resistor trick which works great.

[Chuck H]

Correction to any above info regarding Zeners and resistors!!!

Dave H was very good to point out that I conceptualized the circuit action incorrectly. Which is odd because I've managed it correctly before. Anyway... Thanks Dave for keeping things accurate

In the good Doctors circuit it looks like a 100 ohm resistor parallel to each 10V zener is about right to keep the zener from switching and still put the most load on the resistor during any point of operation. Likewise a 150 ohm resistor for each 15V zener.

Sorry to any and all for the above confusion.

[DrGonz78]

Okay so I will be dropping lets say 60 volts from the B+, from 390v to 330v. That figure is just a starting point for now but it seemed like a good place to start. Anyway I am trying to figure out how to really understand the PSU designer program and just need some advice on that too. I will attach a couple of example files of what I have started on so far. I had to play around with the program to include the expected voltage drop and I am using a 5Y3GT for the rectifier tube. One thing I am confused on is how to accurately determine expected current draw on the B+ secondary taps. I have started looking at tube data sheets to gather some information and it is starting to make more sense. I am sure I will have more questions on that matter soon.

The other thing involves the PSU designer software and whether I am truly using it correctly. I realized tonight that there are stepped current taps that you can add in each section. I am guessing that the current taps are there to include in each power stage as a way to model the current draw between each power supply sections tube compliments.

Also, I understand plate and screen for the power tubes voltage stage pretty well so far, I hope . However, I am not certain on the phase inverter and preamp gain stages plate voltages, both will be 12ax7's. One thing that I was guessing on was whether the phase inverter supply will have a higher B+ plate voltage than the preamp gain stages. The plate voltages on the 12ax7's in my PSU files are 280v for PI and 240-250v for the 1st & 2nd gain stages. Of course that B+ voltage is what hits the plate load resistor and will be lower on the actual plate of the preamp tubes. I did a lot of reading on bias of 12ax7 preamp tubes, still reading more.

[Chuck H]

Gonz, I can't exhibit the files for some reason.

On another note, I spent some time at the bench yesterday playing with zeners and resistors. Results are inconclusive because I wasn't able to pin down what was happening enough to know what relevant measurements to take. But here goes...

With just a resistor in the HV rail in a similar amp to Gonz's I measured the voltage drop. In this case I used a 150R and saw 13V dropped across it. Then I put a 12V zener parallel, reverse bias and the voltage went up, but disproportionate to the 1V I expected.?. I didn't write it down but it was something like 8V. I noticed that not only had the rail gone up past the dropping resistor, but so had the output of the rectifier.?. What happened, as far as I can tell, is that the zener relieved the resistor and minimized current through it. This lower current created a voltage rise relative to the HV winding source resistance. This doesn't happen in my SPICE simulations, which demonstrate the expected voltage drop of the zener value when the drop across the resistor is higher. The amp is closed up again for now, so I'm not running further tests at the moment. But this unexpected additional dynamic surely complicates the idea of paralleling zeners and resistors in the HV rail. For now I would say to skip the resistors then as I didn't get enough data to predict results.

But... This discovery might be valuable in creating a more accurate rectifier tube simulation. With the zener reducing current in a parallel resistor and the HV rail actually increasing with the lower current that may be something like the two curve dynamic that Dave H was interested in. It required more experimenting by someone that better knows what to look for. Definitely interesting and IME impossible to recreate with SPICE because of some missing component property in the simulation.

[DRH1958]

Your first node will have all the tubes' draw, the second has the screen current and the two 12AX7s, the next node only the first preamp tube plus PI tube and last only the first 12AX7.
The data sheet gives 46mA for max plate current, 11mA max screen current for each EL84 and the PI 12AX7 will be the same as two 12AX7 triodes. The current for each triode is about 1mA so 4mA total for both 12AX7s. So the first node will be 46 X 2=92mA(plates), plus 11 X 2= 22mA(screens), plus 4mA for the 12AX7s, so 92 +22+4=118mA for the first node, 22 +4= 26mA for the second and 4mA for the third and 2mA for the last node, assuming you have a node for each section.
You will have higher voltage on the plates of the PI because you have to subtract the voltage on the cathodes from the plate voltage to get your working voltage value to determine your voltage swing going into your EL84s. Let's say you have 300v plate and 40v on the cathodes. You will only have 260v to work with for figuring your voltage swing going to your EL84s, so the plate voltage usually is higher.
Since the preamp tubes' cathodes will be at roughly 1.5v, you don't need higher voltage for that reason.

[Chuck H]

The zener diode didn't go to ground?

If it was simply across the load resistor (parallel), I don't think they work that way.

Ha ha! Well they do SOMETHING.?. The sharing of properties and how it affected current, and therefor sag at the HV winding source resistance surprised me. More research needed to see if there are useful applications.

A zener in series with the HV, not to ground, and installed in reverse bias will drop HV by the zener voltage. In spice simulations, including the condition of a source resistance, adding a resistor in parallel doesn't change a thing until the voltage across the resistor is below the ZV. And then the resistor just takes over. This is what I expected, but not at all how it happens.

[Jazz P Bass]

Sorry, I deleted my post after doing some research.

In a regulator circuit, a zener can be used to drop the over high voltage that the regulator could not stand. The wattage must be calculated.



In this example a 20V/ 5W zener is used.

So I guess the question is what was the parallel resistor doing to the zener (or not doing).

[Dave H]

Anyway I am trying to figure out how to really understand the PSU designer program and just need some advice on that too. I will attach a couple of example files of what I have started on so far.

You need to set up PSUD2 to look someting like the attached zip. The voltages and currents are at idle. To see how much sag it will have at full power substitute the values given in post #38

J20.zip

I1 is the plate current for 2 x EL84, I2 is 2 x EL84 screen current, I3 is PI current and R5 is to simulate the first 12AX7 plate current.

[Dave H]

With just a resistor in the HV rail in a similar amp to Gonz's I measured the voltage drop. In this case I used a 150R and saw 13V dropped across it. Then I put a 12V zener parallel, reverse bias and the voltage went up, but disproportionate to the 1V I expected.?. I didn't write it down but it was something like 8V. I noticed that not only had the rail gone up past the dropping resistor, but so had the output of the rectifier.?

Is the resistor between the rectifier and first filter cap? If so the current pulses charging the capacitor will make the meter show strange voltage readings. The peak voltage across the resistor could be a lot higher than the average voltage read by the meter making the zener incease the B+ voltage by more than expected.

[Chuck H]

Is the resistor between the rectifier and first filter cap? If so the current pulses charging the capacitor will make the meter show strange voltage readings. The peak voltage across the resistor could be a lot higher than the average voltage read by the meter making the zener incease the B+ voltage by more than expected.

The zener was between a sag resistor and the first filter cap. I'll look for that next time I open it up. I did measure my voltages to ground and not from one end of the resistor to the other. I sure looked like strange behavior to me.

[The Dude]

...On another note, I spent some time at the bench yesterday playing with zeners and resistors. Results are inconclusive because I wasn't able to pin down what was happening enough to know what relevant measurements to take. But here goes...

With just a resistor in the HV rail in a similar amp to Gonz's I measured the voltage drop. In this case I used a 150R and saw 13V dropped across it. Then I put a 12V zener parallel, reverse bias and the voltage went up, but disproportionate to the 1V I expected.?. I didn't write it down but it was something like 8V. I noticed that not only had the rail gone up past the dropping resistor, but so had the output of the rectifier.?. What happened, as far as I can tell, is that the zener relieved the resistor and minimized current through it. This lower current created a voltage rise relative to the HV winding source resistance. This doesn't happen in my SPICE simulations, which demonstrate the expected voltage drop of the zener value when the drop across the resistor is higher. The amp is closed up again for now, so I'm not running further tests at the moment. But this unexpected additional dynamic surely complicates the idea of paralleling zeners and resistors in the HV rail. For now I would say to skip the resistors then as I didn't get enough data to predict results.

But... This discovery might be valuable in creating a more accurate rectifier tube simulation. With the zener reducing current in a parallel resistor and the HV rail actually increasing with the lower current that may be something like the two curve dynamic that Dave H was interested in. It required more experimenting by someone that better knows what to look for. Definitely interesting and IME impossible to recreate with SPICE because of some missing component property in the simulation.

Chuck,

A "shot in the dark" here, but if and when you get back into your testing there, could you measure the ripple both with and without the zener? When the zener begins to conduct, it effectively lowers the resistance of the parallel resistor/diode combination moving the filter on that "node" closer to the first "node". I'm wondering if maybe the resultant increase in B+ is simply due to slightly better filtering of the supply overall.

[Chuck H]

Whenever I get back in there I plan to take more extensive measurements for analysis. Maybe not by me.?. But I'll post them and hope that some of the more educated persons here will be able to see what's happening. This is my own personal play/practice amp so I'll never leave it apart for long and I don't tear it open more than necessary. On that note...

I was using a 150R "sag" resistor in the circuit to both add that rectifier tube dynamic and reduce voltage. Plate voltage with this circuit was 360. The tone was very bouncy and dynamic with ample bottom end, but...

I always wondered what the amp would sound like with a more reasonable resistance and a zener drop. So I currently have a string of three 13V zeners in there and a 68k series resistor... Plate volts now are 337 and the sag resistor value has been reduced. The results confirm all my experiences with EL84's...

The tone now has that classic chime. There is a characteristic "ring and hang" to the treble that is so cool. Like the amp really "gets under the notes" and hangs them in the top end. This is a classic characteristic of el84 tubes at this voltage, but... The bottom end is gone! So far it seems that if you want to get bottom end performance from these tubes it only comes at the expense of their characteristic goodness.

I'm still on the fence. I may try a different speaker to revoice the overall tone for the new response character. I may also try removing a diode to bring the voltage to about 350 and hope for the best of both worlds (really long shot I think). I'm really screwed at this point because I want the bouncy, dynamic tone with a bit of bass thump, but I also want that "get's under the note" effect and the classic el84 intermodulation distortions. I don't know if both are possible in the same circuit but I'm sure as hell going to try and tune this thing to find out

The story so far is that el84's perform one way at high plate volts and another at lower plate volts. Both valid. Both great. But can I get a bit of both and will it be the best of both worlds or too little of either trait??? I'll report when I find out.

When I have the amp opened I'll tack solder some things and try to get some meaningful measurements of the zener/resistor interaction too.