• 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.

Is there a 9-pin mini equivalent of 6AU6?

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
Thanks guys. I have the tubes. My problem is knowing what to do with them.

I have lots of brand new 6J9P and 6J9P-E, so that's what I'll use. These are very close to 6688 or E180F. High gm, but don't require the gobs of current that 6J52P or D3a like.

I have lots of nice 6L6GC, so I'd like to use those instead of EL34. Besides, that's the tube O. Schade wrote about, right?

I've been looking for someplace to read how to figure the parts values to get the feedback right. I haven't had a few hours to sit and puzzle it out yet. I've looked for a step-by-step I could follow, but I'm having a hard time. Everything on the innerwebs seems to be for different situations than mine. This thread looks useful -- http://www.diyaudio.com/forums/tubes-valves/177754-plate-plate-feedback-measurements.html

This discussion gets fairly complex, fairly quickly. Will require more concentration than I can give right now...
http://www.diyaudio.com/forums/tubes-valves/177680-schade-feedback.html

I was able to throw the proposed circuit into LTspice and see what happens. Quick and dirty only. I found that if I want 120V on the 6J9P screen, the shared screen resistor needs to be 56k ohms. That's far off from the 100k Doug suggests for 6AU6.

I have 22k wirewound resistors to use for the 6L6 screen to grid/6J9P plate load resistors. Doug's schematic shows 30k, elsewhere he mentions 33k. I also have 35k WW, but LTspice shows the 6J9P plate voltage going too low if I use that value.

- Is 22k a reasonable value to use for 6L6 screen-to-grid R if I'm using 6J9P for the LTP? How about 27k4? I have those too, but the plate voltage goes down below 100V if I put those in.

- If the 6J9P plate resistors are 22k, what value should I use for the 6L6 grid leak resistors? Would 100k be reasonable?

- How does Ia = 9mA and Isg = 3mA sound for each 6J9P?

________________________________
 
The best article on Schade feedback is the one on Tubecad. Might take a bit of searching though.
Its a complex arrangement to conceptually grasp. The feedback resistor effectively turns your high impedance grid into a low impedance node which is one of the reasons why a constant current driver works best. The other difficult concept to grasp is that the the feedback divider is made up by the feedback resistor against the combined parallel impedance's of the grid leak resistor of the final, the load resistor of the driver and the impedance of the driver. The last element, the driver impedance is the reason why a pentode driver is preferred because it is a constant value rather than a variable value with a triode. This means that a pentode driver gives us a fixed feedback ratio and hence a fixed amount of feedback. A triode has a variable impedance with signal so gives us variable feedback with signal.
Many people (including myself) underestimate the demands placed on the driver and under specify this critical component. Some people have even gone to the lengths of using something like a EL84 as a driver to ensure that there is powerful drive into the low impedance node.

If you choose a sufficiently beefy pentode driver which can cope with the low impedance output grid node, you basically just need to calculate the parallel impedance of the three elements described above and choose a suitable feedback resistor to give you the feedback ratio you want. Because the parallel impedances tend to calculate out to a rather high value (10's K ohms typically) this is why the feedback resistor is generally very high and multiples of the calculated parallel value.

These are the essential, though there are other important things going on as well, but you can safely ignore them. I would lash something up and run it through spice to see how it flies. Unfortunately the signals between the input and the output cannot tell you much since they are dominated by currents and can look very distorted compared to either the input or output signals.

Shoog
 
Last edited:
if were rongon, i will build the amp as designed at first, and then after getting the feel of how the thing works, move on to befier pentodes...

as i understood it from the postings of PRR here, reason why smaller plate currents were chosen in the old days was because of psu filtering...

ecaps have come a long way since then, they have shrunk in size while getting higher capacities, and beefier triodes requiring bigger plate currents are more practical nowadays...
 
The best article on Schade feedback is the one on Tubecad. Might take a bit of searching though.
Its a complex arrangement to conceptually grasp. The feedback resistor effectively turns your high impedance grid into a low impedance node which is one of the reasons why a constant current driver works best. The other difficult concept to grasp is that the the feedback divider is made up by the feedback resistor against the combined parallel impedance's of the grid leak resistor of the final, the load resistor of the driver and the impedance of the driver. The last element, the driver impedance is the reason why a pentode driver is preferred because it is a constant value rather than a variable value with a triode. This means that a pentode driver gives us a fixed feedback ratio and hence a fixed amount of feedback. A triode has a variable impedance with signal so gives us variable feedback with signal.
Many people (including myself) underestimate the demands placed on the driver and under specify this critical component. Some people have even gone to the lengths of using something like a EL84 as a driver to ensure that there is powerful drive into the low impedance node.

If you choose a sufficiently beefy pentode driver which can cope with the low impedance output grid node, you basically just need to calculate the parallel impedance of the three elements described above and choose a suitable feedback resistor to give you the feedback ratio you want. Because the parallel impedances tend to calculate out to a rather high value (10's K ohms typically) this is why the feedback resistor is generally very high and multiples of the calculated parallel value.

These are the essential, though there are other important things going on as well, but you can safely ignore them. I would lash something up and run it through spice to see how it flies. Unfortunately the signals between the input and the output cannot tell you much since they are dominated by currents and can look very distorted compared to either the input or output signals.

Shoog

Thanks Shoog, that's very helpful.

(I saw in a different post that this circuit can sound 'thin,' which makes me wonder if that wasn't the sound of the 6AU6 driver giving up into the difficult load...)

I found the Tubecad article and have read through it a few times. It takes me a few readings for things to begin to seep in.

I also found Patrick Turner's discussion on "Balanced Shunt Feedback." I'm finding this very helpful.
miscellaneous-schematics2

I lashed up a working LTspice simulation (which is showing THD numbers that must be too good to be true). Basically, this lash-up is the same thing as in the original E-Linear schematic, but with 6J9P pentodes for the drivers and 6L6GC for the finals. Rsg for the 6J9P LTP is 56k. 6J9P plate load/feedback resistor is 22k. Grid leak on the 6L6GC is 100k.

Voltages and currents come out to about
6J9P plate V = 195V
6J9P screen V = 120V
6J9P cathode V = 1.5V
6J9P plate I = 9.5mA (plate dissipation = 1.85W each)
6J9P screen I = 2.8mA (screen dissipation = 0.34W each)

I can't use a beefier pentode in the LTP than that, since my power supply will be at its limits as it is.
________________________________________________________

Let's see if I understand the basics of this shunt feedback scheme. Using this 6J9P/6L6 E-Linear...

Let's say the 6J9P ra is 100k ohms.
6J9P plate load resistor (RL) is 22k.
6L6GC grid leak resistor (Rg) is 100k.

The 6J9P ra is the series resistance for the NFB voltage divider, correct? That's 100k.

The shunt resistance of the divider is ra || RL || Rg (I think), which is 15k3.

The 'feedback resistance' is 22k (which is also RL for the 6J9P).

Does that look right?

To use Patrick Turner's calculations, I need to figure out the gain for the 6J9P and 6L6GC. I used this formula to try to calculate the gain (Vgain) of 6J9P:

Vgain = gm x Ri x RL / 1000000 x (Ri + RL)

It came out to 270.

For 6L6GC with a 1.25k load (1/4 the total 5k a-a load of a Dyna A470 OPT), Vgain came out to 6. (But the E-Linear circuit takes the feedback off from the OPT screen tap, which I'm sure changes things.)

Does that seem about right?
--

AJT, the reason I don't want to use 6AU6 is that I already have 9 pin sockets in the chassis, and I have a box full of new 6J9P tubes. I could swap 7 pin sockets in, I guess. It's just more work. I have enough 6AU6 pulls in stock to pick out 4 good ones.

The reason I want to use 6L6 outputs is that I have lots of nice vintage ones and I don't have any EL34's, which are kind of expensive.
--

PS - Speaking of filtering... Does a PP UL wired output stage require higher capacitance filtering than a PP triode output stage? This is in a Dyna ST70 chassis, so the choke is a weenie 1H thing. I'll need to use 500V rated capacitors in very limited space, so the most I can throw at it is 50uF - 1H - 50uF (CLC).

The same question can be asked about a pentode LTP vs. a triode LTP. Does the pentode LTP have poorer PSRR than a triode LTP?
--
 
Last edited:
That would work, since I have also have some 6P15P (Russian EL83 equivalent) to work with. However, that means I'll be bumping up against the limits of the ST70 power supply. I'd want to limit the quiescent current draw to something like 280mA (absolute max) for the two channels.

Do you think EL83 would work well with Ia = 10mA and Isg = 2.5mA?

I wonder if the resulting sensitivity will be low enough that I'll need a gain stage up front. If so, Mullard 5-20 topology, here we come (again)...
--

PS - E180F/6688 is 'wimpy'? Pdiss of 3W? Va of 9mA? It has gm of 15mA/V and ra of only 100k, which seems pretty hefty to me.
 
Last edited:
The EL83 looks really sweet with good transconductance figures. It could be run with something like 150V on g2 to spread out its curves somewhat. Looks far more linear than either the ECL82 or ECL80 pentodes.

E180F/6688 is 'wimpy'? Pdiss of 3W? Va of 9mA? It has gm of 15mA/V and ra of only 100k, which seems pretty hefty to me
That looks quite respectable to me. However if you are going to realize anything like that amount of gain you would want the Va to be at least double what it is otherwise you will be hitting distortion before you can feed it any sort of a signal.


I have to say however that the 6AU6 works out fairly well in the plate to plate feedback application. i have used it at least twice and Gary Pimm uses it in his Tabor amp.

EL83 looks very good in lots of ways - probably the best of the bunch. Try modeling - you wont find a model for the EL83 so substitute an EL84 to get an idea.

Even better might be the EL86 which is designed for low anode voltages and sounds excellent to boot (though looking closer getting it to bias up at a reasonable g1 voltage would be a massive challenge - probably best forgotten)

http://www.r-type.org/pdfs/el86.pdf

Shoog
 
Last edited:
I like this idea!

OK, that's the next thing I'll throw into LTspice. It's convenient too, because the 9 pin sockets are already in place.

When used this way, should I worry if the plate voltage on the EL83 ends up lower than the screen voltage?
--
I wouldn't worry to much myself. Lowering g2 is the best way to bring the Va into a reasonable range by bringing it down towards 150V. Some sort of basic regulation might be in order however.

Shoog
 
The EL83 looks really sweet with good transconductance figures. It could be run with something like 150V on g2 to spread out its curves somewhat. Looks far more linear than either the ECL82 or ECL80 pentodes.


That looks quite respectable to me. However if you are going to realize anything like that amount of gain you would want the Va to be at least double what it is otherwise you will be hitting distortion before you can feed it any sort of a signal.


I have to say however that the 6AU6 works out fairly well in the plate to plate feedback application. i have used it at least twice and Gary Pimm uses it in his Tabor amp.

EL83 looks very good in lots of ways - probably the best of the bunch. Try modeling - you wont find a model for the EL83 so substitute an EL84 to get an idea.

Even better might be the EL86 which is designed for low anode voltages and sounds excellent to boot (though looking closer getting it to bias up at a reasonable g1 voltage would be a massive challenge - probably best forgotten)

http://www.r-type.org/pdfs/el86.pdf

Shoog

What about the C3M or one of the other posttubes
 
Quote:
E180F/6688 is 'wimpy'? Pdiss of 3W? Va of 9mA? It has gm of 15mA/V and ra of only 100k, which seems pretty hefty to me
That looks quite respectable to me. However if you are going to realize anything like that amount of gain you would want the Va to be at least double what it is otherwise you will be hitting distortion before you can feed it any sort of a signal.

I don't think I need 200X gain in this application, so if it winds up being only 100X, that'll be fine with me. I'm just thinking that the 10+mA/V gm and Ia of 9mA the 6J9P delivers looks like enough for this application, and I have plenty of brand new ones to play with. They were very cheap.

I figure if the 6AU6 works in this circuit then 6J9P-E should work even better. But as always, I could be wrong.


I have to say however that the 6AU6 works out fairly well in the plate to plate feedback application. i have used it at least twice and Gary Pimm uses it in his Tabor amp.

I can always remove the 9 pin sockets and put in 7 pin sockets. I have the type that fit. But again, if the 6AU6 work well enough, shouldn't 6J9P work even better?


EL83 looks very good in lots of ways - probably the best of the bunch. Try modeling - you wont find a model for the EL83 so substitute an EL84 to get an idea.

I'm looking forward to seeing if the EL83 or EL84 beats the 6J9P in simulation.

(I'd use EL84 tubes, but they're overpriced due to demand for guitar amps, etc.)


Even better might be the EL86 which is designed for low anode voltages and sounds excellent to boot (though looking closer getting it to bias up at a reasonable g1 voltage would be a massive challenge - probably best forgotten)

http://www.r-type.org/pdfs/el86.pdf

...And I have plenty of 6P43P-EV I can use for this too. The thing is, how much plate and screen current does it need to start working well?

How well do EL83 and EL86 work with only 10mA of plate current, and 3mA or less of screen current?

I have a hard limit of about 280mA from the power supply, and these EL-series pentodes are all fairly high-current power tubes. Four 6L6's at 55mA each already burns up 220mA. Add a 0A2's need for 15mA and I'm at 235mA already. I'd rather run the 6L6's with 60mA going through them (less 3HD that way), but it looks like the driver stage is going to take precedence in the current capacity budget.
--

PS - I also have 1:1 input transformers I can throw at this. A pair of Edcor XSM15K-15K waiting in the wings...
--
 
Last edited:
How well do EL83 and EL86 work with only 10mA of plate current, and 3mA or less of screen current?

I have a hard limit of about 280mA from the power supply, and these EL-series pentodes are all fairly high-current power tubes. Four 6L6's at 55mA each already burns up 220mA. Add a 0A2's need for 15mA and I'm at 235mA already. I'd rather run the 6L6's with 60mA going through them (less 3HD that way), but it looks like the driver stage is going to take precedence in the current capacity budget.

EL83 / 86 dont look too good in these conditions... Now you know why Dynaco had this puny little tube at the input. To make the power transformer as small and cheap as possible. Dunakit or some other makes a drop in replacement, i think?
 
EL83 / 86 dont look too good in these conditions... Now you know why Dynaco had this puny little tube at the input. To make the power transformer as small and cheap as possible. Dunakit or some other makes a drop in replacement, i think?

The stock PA060 transformer is adequate. I think it's rated for 300mA, which is just enough for the job. You can tell the tranny's cheap because it's designed to work while making lots of heat due to copper losses.

I wouldn't want to stuff an ST70 chassis with more heat-producing devices anyway. It's too small to ventilate well.

For this build, I think the limit is 12mA combined plate/screen current per driver pentode, and 55mA combined plate/screen current per output pentode. That's a total of 268mA. I can add 15mA for a 0A2 (for screen regulation), which brings it up to almost 285mA. That's the limit.

So that leaves me with the question: Which pentode (limited to Ia+g2 = 12mA) should I use for the LTP driver in this circuit?

I looked at the plate curves for EL83. The datasheet has them for Vg2 = 170V. It looks like it's biased pretty cold at Ia = 10mA, but it does look usable.
--
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.