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

[DrGonz78]

Well to start I bought the power amp portion of a 1959 SF-183 Zenith Radio. That amp has two EL84 tubes for the output section with cathode bias and a 5U4GB rectifier. I have installed the power transformer into a chassis and currently testing it out. I got a well made chassis on Fleabay to house the whole thing. Actually, I am pretty sure the chassis was a left over one from Mission Amps. It has those really nice EL-84 converter kits installed and if you remove them you could install 8-pin sockets without much effort too.

Anyway I am still trying to decide on the exact amp that I want to design or clone. I really want the design to be something original, but this will only be the second amp that I have built. One thing is for certain is that I am going to spend a lot of my time understanding the design process. In other words, whether I build semi-original or complete clone, I want to understand the theory of tube design much more than I do now. I have searched long and hard online looking at many amp designs that use a pair of EL84's.

So my first question is a matter of recommendations of a tube amp that uses two EL84's? That is a pretty general question but I am interested in other builders recommendations.

My next question is in regard to building the power supply section of the amp. The Zenith PT has a 640vct secondary rated at 150ma, 5v for rectifier, one 6.3vct and another 6.3v without CT. I am playing around with PSU2 Duncan program but I find it is a bit finicky and confusing. The Zenith SF183 amp has 345v on the plates and 300v on the screens of the EL84's. I have the old EL84 tubes that were in this thing and will test them out on another amp to make sure they are working. The older tubes from what I have read can take the higher than max voltage ratings better when compared to newer ones today. Anyway it is the current draw that is a tube killer right? Still I would appreciate any guidance and knowledge on how to design the power supply section of this amp. Any explanations of how to use the PSU2 program would be beneficial too. Thanks. BTW I have attached some of the schematic for the power sections of the SF183 amp.

[Chuck H]

What sort of tone do you want? Or what sort of tone do you usually use? Or, if you already have that covered, what sort of tone would you like to ADD to your stuff? EL84's in that voltage range and tube rectified are perfect for certain vintage designs. Most have a nice chimey clean tone and the classic "creamy" overdrive. That doesn't mean you're limited to such designs, but it's a natural go to if you're interested in it. Then there's the "18 watt" (and their better sounding counterparts, the WEM Dominator) amps with their grittier, more aggressive tone. I have a design that's not unlike a half power Trainwreck Liverpool if you'd like a more saturated lead sound and a lot of control from the guitar. It really depends on what you want. Will you be buying a speaker for the project or should this build be tailored for the speaker/s you already use with other amps? You're voltages are perfect. Right in there for most design options.

[olddawg]

Check out the Ceriatone layouts, they have several variations on the 18 watt Marshall 1973. You might want to consider using an SS rectifier and bring up your B+ a bit. There's also the Vox AC15. Lots of options.

[Chuck H]

Well... I have a 2xel84 amp running a very hot AB1 with a 600VCT PT and a "sag resistor" about equal to a 5u4 and it has 355Vp. That Zenith has a 117V primary. If we correct for this and plug it into Duncan PSUD2, which accurately predicted my amps Vp, it tells me that with a 5u4 you can expect 380Vp. With a silicon rectifier you could expect about 445Vp. My experience with el84's is that they don't mind current as long as the plate volts aren't too high. Even at 355Vp, with the plate potential so high, I can't clip without too much crossover distortion unless I bias beyond max. Lower voltage and higher current to get to max is, IMHE, better for el84's if you're going to clip them. This build may even require the zener voltage drop trick depending on what design is chosen.

EDIT: Looking at the schematic I'll guess this is a class A push/pull design. But I probably wouldn't bias that hot at 380Vp. Which means less current draw and therefor even higher Vp. Like 390. I don't much care what the schematic says anymore. So often the numbers are erroneous. I'd trust PSUD2 and expect 380 to 390Vp. Then I'd prepare for dropping some voltage with zeners. Since you'll get plenty of sag from the 5u4 you probably wouldn't want to add any more resistance. OR... Remove the adaptor plates and use 6V6's. They can do 390Vp and sound better doing it than el84's at that voltage. The OT primary is probably more suitable for this than the PT is for el84's at this point. You could go any way you want to though. Nothing too tough to manage any way it goes.

[DrGonz78]

Thanks for all the great recommendations so far! I am not sure what my tone is these days since I barely play much lately. More or less I am doing this to learn more about other tone discoveries. What I love most about guitar tone is that from the 60's-70's rock. Been looking at a lot of different tone stacks and what not. I will post something on a preamp that I am working out later. Also, we can shoot the breeze about tone stacks as this thing progresses. As far as speakers go I am building this into a head so I will be able to hook it up to all sorts of cabinets. I have a two 10" combo cab, 12" cab and even a nice 1969 15" Jensen for a rainy day.

I am not 100% set on the EL84 designs and can shift my thinking towards the 6v6's too. Later on those EL84 adapters will no doubt come in handy either way. I was running the PSU2 designer using a 5Y3GT and got very similar results. I did some recent reading on the zener diode voltage drop trick, but have not really found a schematic that uses that method. Will research more on that one and looking down on this thread page already gives some good links to look at. I think Old Dawg is right about using silicon rectifier, especially if I go the route of the 6v6's. Still there is something alluring to me about old technology and tube rectifier sag. I am also considering getting a different OT for this design. I like AC15, Marshall 18 watt, and the Train Wreck amps(Express Versions) and have looked at those in some detail.

This all might go a little slow for me as I am absorbing so much information and there is really no way to retain it all at once. This is a great start and I will check in soon with some more questions I am sure. Thanks

[Chuck H]

There may be an easier way to implement the Zener voltage drop circuit than the typical ENORMOUS WATTAGE zener mounted to a heat sink that may melt the solder joints in actual use. I haven't tried it yet, but I do have some preliminary experience to support my idea and I intend to test it (whenever Mouser get's my stuff off @#*&ing backorder).

You can use a resistor in the circuit that is valued to drop the desired voltage as it will be at the highest current draw (read that again if it didn't sink in ) Then parallel it with a zener that is valued at the desired voltage drop. The resistor will carry most of the load and the zener only needs to handle whatever voltage the resistor isn't already dropping. So if, say, the resistor drops the desired voltage at max current of, say, 40V, and you have a parallel zener with a ZV of 40V, but under low load the resistor is only dropping 30V, the zener only needs to manage 10V instead of 40V. Why is this significant? Because resistors that can handle high wattage are cheaper and easier to obtain than zeners. AND... Zeners in DO package (much easier to implement than chassis mount) are available up to 5W. You can even do a series string to avoid chassis mount zeners. You may question whether the zener carries more load once it's reverse voltage is met and the resistor becomes moot. I did too. About my experience...

I use a circuit for some of my amps that parallels a zener with the bias resistor of cathode biased amps. I wanted to know whether the resistor or zener would be carrying the load once the zener voltage was met so I tested it. The zener will only ever carry what the resistor doesn't. LTSPICE simulations seem to bear this out as well. Not that many voltage regulation circuits don't already bear this out. Nothing new under the sun, right? Since big resistors and 5W zeners are eminently cheaper and easier to implement than a single high wattage zener we should be taking advantage of it.

[DrGonz78]

So after some thinking I am going to stick with my original plan of using the EL84's for the output. Besides then I won't have to rename the title of the thread or confuse someone that reads it later down the road. Also, I have another project after this one all lined up ready to go. That one is an old 1947 Temple G-722 radio chassis I got a few weeks ago. It has a pair of 7C5 loctal type tubes for the output and from what I have gathered those tubes are basically old 6V6's with different pin outs. So the next amp will be that one!

I like the idea of dropping the B+ voltage using the zener trick. I have read about others talking about this before but have to admit I did not understand it very well. I get it now much better. I see that you can use a mosfet in conjunction with a zener diode and two resistors to bring down the B+. Also, I see the idea of daisy chaining of 4-6 zeners in the same way using the CT of HT secondary windings to ground(onto negative side of first filter cap). This idea is making much more sense to me now and I would like to get the hands on experience doing something new too. Chuck, I get what you are saying about using a resistor to carry the weight of the load and that must help dissipate some the heat on the zener. Still re-reading that last post several times to get it sink in better.

So we are saying the B+ will be somewhere between 380-390Vp and is that without all the tubes loaded? All the variables in designing a power supply for a tube amp seem to be like opening a can of worms. I plan on really getting a fuller understanding of all these concepts as I go forward on this project. For example, I get confused on how to figure out how much the B+ will drop with all the tubes loaded in the amp. The very "generic" schematic that I working on at the moment includes one 12ax7 for gain stages, one 12ax7 for PI and the two EL84's. I have the 5U4BG rectifier and I have plenty of 5Y3's. Will using a 5Y3GT lower my B+ a tiny bit compared to a 5U4GB? According to the PSUD2 it drops about 10v. Also, is the PSU2 referencing the voltage drop of newer type rectifiers? I have old stock 5Y3's that I could put in this amp to at least bring the voltage down a tad bit more.

I will post what I have of a schematic by tomorrow or so. I am starting out really basic with the design and possibly others here can help by giving me some pointers along the way. Thanks again for the feedback as usual.

[Chuck H]

That 380 to 390V would be the loaded voltage as far as I can tell.

[The Dude]

Just a small point, but nonetheless something to consider. If you plan to stick with cathode bias, the cathodes will be elevated from ground allowing for a slightly higher B+. Nothing substantial, but I thought it should be noted.

[Chuck H]

Just a small point, but nonetheless something to consider. If you plan to stick with cathode bias, the cathodes will be elevated from ground allowing for a slightly higher B+. Nothing substantial, but I thought it should be noted.

On a similar note, in manufactured designs it often looks like EL84's are biased grossly over their max. But this isn't always the case. Because EL84's are often run cathode biased AND in/or very near class A, AND there is a relatively low bias voltage requirement, AND the plate has a relatively low dissipation limit... It's easy to overlook the screens as part of the whole picture. An EL84 biased at 13.5W is just touching 100% plate dissipation due to the screen current addition. In a nut shell... An EL84 in class A or even a hot AB1 in cathode bias will often look to be over dissipating at idle even though they're not. I love EL84's!!! They're electronically tough as hell. You can beat the tar out of them and they come back for more. It's unfortunate that all the current offerings are so mechanically frail. That's the only drawback to this tube type I've had to reconcile.

[The Dude]

I love EL84's!!!

As do I. Most of my favorite amps are EL84 amps.

[DRH1958]

To answer your question about the 5U4 vs. the 5Y3, there will be around 15v less with the 5Y3 depending on how fully loaded the supply is. The 5Y3 will obviously be more fully loaded than the 5U4. According to Merlin's Power Supply book, a 5Y3 averages 300 ohms plate resistance and will supply 100mA and a 5U4 averages 150 ohms for plate resistance and will supply 275mA. I have an amp with the same tubes as you propose, except I have a 12AU7 reverb driver an I lose between 20 and 25 volts between the two tubes.
Another item to remember is max. reservoir cap size, although I don't know if this is adhered to very much. A 5Y3 is spec'd at 20uF and a 5U4 is 40uF. This is for when the tube is used at max. current and the least limiting resistance is used, which should be spec'd by the data sheet.
Agree with the Dude, EL84s rule! They can be chimey like a Vox or snarly when pushed with little NFB. I would hunt down some vintage Amperex or similar brand. Black plate Sylvanias are great too. If you want, you can also go with 7189s for higher voltage cushion. They are rated for 400v plate and 300 screen I believe. Pinout is the same.

[Chuck H]

The biggest problem with EL84's is their fragility (IMHO). There are a lot of brand shootouts pontificating the virtues of tone, but the truth is that they all sound pretty good. Finding one that'll last more than 300 hours in a combo amp is the challenge. I love my little EL84 combo but I probably won't build any more EL84 combo's. I recently got into a "lot" of "slightly used/test new 6P14P-K tubes. It's a Rusky EL84 that's stable against vibration. So far so good. They sound great. Only time will tell if they rattle apart, but for now I'll recommend them. Hard to find NOS. Hard to find used with any sort of description that inspires confidence. I took a chance on a lot of 16 tubes and ended up with 6 good matched pairs, another matched pair that did something a little funny, so I'll be watching them whenever they're installed, and two tubes that died as soon as I put the screws to 'em. Not bad overall. Calling it 6.5 matched pairs () for $72 bucks is $11 a pair!!! I don't care that they're "slightly used" since they have a 3000 hour rated life and my combo destroys tubes in under 500 hours every time! If their "vibration resistant" rating holds true I should get more use out them than many tubes that cost much more.

[DrGonz78]

I am intrigued I must admit. The 6P14P-K tubes sounds really interesting and I found a seller that has feedback from buyers stating the tubes worked great. It comes out to $2.25 a tube for 24 or in a lot. So now I am thinking about ordering up some of these tubes that take the higher plate voltage. Once DRH recommended the 7189 tubes I then found myself stumbling on the 6P14P-K tubes and then I read Chucks next comment. This seems like the proper type tubes for the amp I am making!! Cool. Let you know if I buy them. Thanks for the recommendations

[Chuck H]

A little out of my depth, but I will say this about my observations regarding plate voltage and most EL84 types (including 6P14P's and 7189's). They seem to prefer higher current and lower voltage and don't perform analogous with higher voltage and lower current. They will produce more watts at higher voltages but then they don't transition into clipping without copious crossover distortion at only slightly more output than a lower voltage and higher current. That is, there seems to be a minimum current to produce a good cutoff transition almost regardless of plate voltage. This is probably part of the reason so many 18W homebrew amps have buzz issues. You can just increase current until the crossover is under control, but you'll be over dissipating the tubes. Which also happens often as we've discussed.

I don't know exactly why this should be the case, but the NOS market offers many different 9 pin skinny pentodes with different pinouts that all have a max plate and G2 voltage of 250. Perhaps the EL84 types were developed as a higher performance version in this niche but certain design limitations were never adequately addressed.?.

IMHO a good plate voltage for any EL84 type (including the versions rated at 400Vp) for guitar amp performance, assuming clipping them is the goal, would be 335 or less. Regardless of the higher plate specification. It doesn't do a guitar amp much good to make an extra watt or two if excessive crossover distortion develops at the onset of clipping.

My personal amp has a Vp of 355 to 360 (depending on my AC mains at the moment). I'm using a couple of tricks and biasing at 90% to get the crossover distortion under control and it's working fine. But that could have been avoided with 25 fewer plate volts and I'd advise anyone with the option in the development stage of a project to observe this and avoid the usual pitfalls that commonly plague EL84 homebrew's.

JM2C