• WARNING: Tube/Valve amplifiers use potentially LETHAL HIGH VOLTAGES.
    Building, troubleshooting and testing of these amplifiers should only be
    performed by someone who is thoroughly familiar with
    the safety precautions around high voltages.

EL509 (JJ) or PL519?

Hi all,

I decided to have a go at a design from Bob Danielak, a screen driven direct coupled EL509. Screen Driven, DC Coupled, Single-ended EL509 Amp

An externally hosted image should be here but it was not working when we last tested it.


Now old EL509's are hardly available, but Askjanfirst has the PL519. And JJ came up with something they call EL509.
What's the JJ like? I'm inclined to go for the PL519 (I like the top cap), but that would mean an extra filament transformer.

Thanks ;)
 
The JJ EL509 claims a 700 volt rating on the screen grid. If this is true it will be useless for screen driven applications.......Can you explain me why?

There are a few of us here that have played with screen driven amps. Notice that the common element among all of the successful designs are they all use sweep tubes. My experience is that the lower the maximum screen voltage rating, the easier it is to drive.

The screen grid has almost as much control over the current flow in a pentode as the control grid. Like the control grid it is possible to wind the grid wires with a fine or coarse pitch, and the spacing between G1 and G2 can be varied. A fine pitched G2 winding will exert more control over the current flow as will a close spacing between the grids. Both of these things will also cause G2 to intercept more electrons and therefore draw more current. Placing G2 in the shadow of G1 is another tube makers trick to reduce G2 current.

Most of the typical audio tubes like the 6L6GC, EL34, KT88 and so forth have G2 optimized for operation at voltages from 350 to 500 volts. This implies that G2 will have some control over the tube current, but a large change in G2 voltage is needed for a big change in tube current. Attempting to screen drive a KT88 requires about 500 volts Peak to Peak of drive voltage. The linearity under these conditions sucks too. I have tried it.

Most sweep tubes have a maximum G2 rating in the 175 to 250 volt range. A small change in G2 voltage can make a big change in plate current. A 6BQ6 or a 6AV5 can be driven to 80 watts in P-P with about 150 volts Peak to Peak of drive and the linearity is very good.

Further investigating of the JJ specs leads me to believe the 700 volt G2 spec is not real. Look at the spec for the original EL509. The 700 volt G2 spec is for a non conducting tube. The maximum voltage spec for G2 in operation is 250 volts. I don't know what the real spec for the JJ tube is. If the tube can operate with 700 volts on G2 then it will take more drive voltage than you can make. If that spec was blindly copied from the original with some info lost in translation (which is what it looks like) the JJ tube may actually work. I have not tried one, and the JJ EL509 does have a reputation for blowing up so it may actually be a real sweep tube.
 
My experience is that the lower the maximum screen voltage rating, the easier it is to drive.

The actual characteristic that makes tube easy to drive via g2 is high mutual transconductance of g2. In case of PL509, it is some 5,7...6,0 mA/V at Ug2 values 160 to 190 volts ( ...and Ua=400V,-Ug1=30 V).

Philips PL509 datas contain three sets of plate characteristic curves with different g2-voltages (160V, 175V and 190V).

The JJ EL509 curves are given only with 160 V g2-value, but one can see that it is equal to Philips PL509 with this respect.

As a reference for 6L6GC the g2-transconductance is some 0,9 mA/V only
and would therefore need few hundreds of volts driving voltage if g2 is driven.
 
I decided to have a go at a design from Bob Danielak, a screen driven direct coupled EL509.

If I were building this amplifier, I would also make there a chance to (fine) adjust the Ug1 of the EL509. This means that the bias circuit would be a combination of cathode bias and fixed bias.

After reading the whole story from the link of Bob Danielak, I have the belief that the bias value for the output tube has been decided mostly by simulation only, and therefore the final results can be essentially improved by adjusting the most important component of the amplifier, namely the output tube.

In this kind of technical articles like this of Bob Danielak, I would like to also see a reasonable amount of test results, not only the criterias for the design. I would be interested in distortion values at low, mid and high frequencies at few power levels. Now there was nothing mentioned.
However, the result for listening test was mentioned.
I could respect such, but only together with real test results.

As an outcome, I have an impression that very much effort has been put into the desing and simulation, but the final results were not worth to be published.

Then a question arise in my mind: Is this amplifier worth to be built ?


But look at this:

An externally hosted image should be here but it was not working when we last tested it.



Svetlana has published very interesting curves. This is called "zero-bias"- or "Hi-mu"-triode connection.
As you can see the amplification factor is huge, some 150. Thus in this connection the required driving voltage is relatively-, or actually very low.

This operation however works best with quite high anode voltages, say 800 volts, but at the same time the anode current is low and this makes the working enviroment of output transformer much easier.

I quickly calculated that with 10 k anode load (and Ua= 800V /Ia= 40 mA ) some 20 W output power would be achieved with driving voltage less than 10 Vrms. It is important to notice that driver stage must be capable to drive grid bias- and drive power due to grid current of EL509.

I have built prototypes (and tested) with this principle by using the tubes 811A, GU50 and 807.
 
The actual characteristic that makes tube easy to drive via g2 is high mutual transconductance of g2.

Exactly. Mutual conductance is the ratio of voltage change at the input terminal to the current change at the output terminal. It is rare to find specs for G2 transconductance. An easy generalization that seems to hold up well is the max G2 voltage spec. I have seen tubes where this doesn't hold though. These are the tubes where you can violate the G2 voltage without blowing it up. The 6LW6 is one. Datasheets for this tube alone are scarce, let alone one with any actual data in it. I believe the max G2 spec is 275 volts, but I have run them to 400 volts without glow. This tube will work in screen drive, but it takes about 400 volts peak to peak of drive. To remedy this a few of us are experimenting with driving G1 and G2 at the same time.

It looks like JJ just merely copied selected information from the old datasheets for their own datasheets. Since they changed the base and removed the plate cap from the tube, what is inside the glass? It is impossible to tell from the datasheet. The only way to find out how this tube will work is to test it.

I have worked with this circuit before. I modified it in a similar manner as most of my screen drive circuits though. I used a mosfet for the follower and a 12A*7 for the input tube. It works well with the usual selection of cheap American sweep tubes. Don't tkink I tried the 6GK6 since they are not cheap.

It is my opinion that screen drive is best suited for push pull amps since many sweep tubes behave in a highly linear manner at very low idle currents when screen driven. This allows for a low static dissipation and plate efficiency approaching 80%. These advantages are lost in an SE amp.
 
G1 + G2 drive looks highly non-linear in that Svetlana curve plot...

Caveat: I have not built the Danielak amplifier! However, having read the documentation, it seems it's been designed via simulator. He does give an Vp/Ip/Vg2 plot but people have independently measured this under the same conditions and reality seems to be nothing like the idealized plot generated from a simulation model. A fellow from germany named Tom Schlangen, who has greatly indebted this and other forums with numerous plots of pentodes connected as triodes, has done several measurements of the EL509/519, as far as i know only NOS, and failed to get even close to the results expected by bob Danielak. Severe tetrode kinking is visible in his plots, although linearity at higher Vp seems to be there. Since I have used Tom's data on numerous occasions, i have no reason do doubt him, given that I would not be building an amp using Bob Danielak's schematics.
 
Hi ilimzn,

the plots you are refering to still are avaiblable here: Tom Schlangen's Homepage - Tube DIY Pages and the screen driven [E|P]L509 plot is available here: http://www.tubes.mynetcologne.de/roehren/misc/el509bpt_sd_33v.pdf.

The screen driven EL509 plots shown in the original article from Bob D. are not plots of a real tube, but a simulation indeed. This was not made clear in his the article he wrote as an "application report" for Svetlana. Honi soit qui mal y pense.

When thinking about screen driving tubes like xL509 or xL36, one always should be aware how narrow the "tunnel" really is one has to lay the loadline into to avoid negative impedance behavior on anode swing towards lowish voltages, or else the risk of Barkhausen oscillation inevitably will collect its dues... Also, in this regard one should remember that a straight loadline almost has nothing to do with reality, considering reflected complex impedance behavior of a real world speaker at the secondary of the OPT.

Nevertheless, screen driven xL509 designs have been done and did lead to satisfactory results when investing a thought or two on the inherent problematics. F.e, check "synola 509" keywords on Google, although this design is limited somewhat by its driver stage weakness. Anyway, it is a better point to start than with BDs article which is based on nice looking, but inadequate simulations.

Oh, and by the way, Tim de Paravincini designed a xL509 amp for E.A.R. (Esotheric Audio Research). He went to some lengths about keeping this tube under control in a lecture he held at ETF (may have been ETF in 2007 @ Biezenmortel/NL, IIRC).

Regards,

Tom
 
Last edited:
Nevertheless, screen driven xL509 designs have been done and did lead to satisfactory results when investing a thought or two on the inherent problematics. F.e, check "synola 509" keywords on Google, although this design is limited somewhat by its driver stage weakness. Anyway, it is a better point to start than with BDs article which is based on nice looking, but inadequate simulations.

Tom

The Synola:

Synola509-Amp.gif


I must say it appeals more to me than the Danielak version (no hard feelings, Bob)
Driver stage weakness you say? Like?:confused:
 
Hi Miniwatt,

The Synola [....] Driver stage weakness you say? Like?:confused:

Look at the xL509 screen driven plot; near the lower left corner you will find some pink color screen current "mini"-curves I also added. Then have a look at the paralleled ECC82/12AU7 sections CF driver feeding the xL509 screen grid in the Synola schematic and ask yourself if they would be able to source the needed current under any expected xL509 working conditions in this design - preferably in a most linear manner, of course.

Certainly there is room left for improvement in the CF driver stage of the Synola 509, but I can asure you it sounds very good already just using the schematic you pointed to. Hence my recommendation to use it as a starting point.

Regards,

Tom
 
Hi Miniwatt,



Look at the xL509 screen driven plot; near the lower left corner you will find some pink color screen current "mini"-curves I also added. Then have a look at the paralleled ECC82/12AU7 sections CF driver feeding the xL509 screen grid in the Synola schematic and ask yourself if they would be able to source the needed current under any expected xL509 working conditions in this design - preferably in a most linear manner, of course.

Certainly there is room left for improvement in the CF driver stage of the Synola 509, but I can asure you it sounds very good already just using the schematic you pointed to. Hence my recommendation to use it as a starting point.

Regards,

Tom

Thanks Tom, this kind of amp is new to me, so I was going to build it 'breadboard' to experiment and learn.
The OPT seems a delicate matter, I have some 4K7 I planned to use, but this seems rather high..
 
Driver stage weakness you say? Like?.....Certainly there is room left for improvement in the CF driver stage of the Synola 509

The driver is key in a screen drive design since it needs a bunch of transconductance and needs to be able to source screen current. In my experiments I find that the output tube plate voltage will be near zero when the screen grid is at maximum. The screen current peaks are far more than a 12AU7 can source without serious distortion. I have found that the best device for the task here is a mosfet since it fulfills the criteria and in follower application a mosfet doesn't screw up the sound. The SE circuit I used looked a lot like the Synola thing except the 12AU7 was replaced with a fet. The negative voltage on the output tubes control grid sets the idle current.

I find that most screen driven applications are more linear with a little negative voltage on G1. Too much negative voltage requires the peak G2 voltage to go too high leading to screen grid overload on peaks when the plate voltage is low. This is not an issue on music with infrequent peaks, but keep an eye on the screen grid when testing with continuous sine waves at full power.