EF86 Screen Voltage

[GregS]

I just completed an amp design that involves 2 12AX7 stages in front of an EF86 pentode stage in the pre-amp. I had intended that the screen voltage be about 10 volts lower than the plate voltage. However the reverse is what has happened with the screen voltage being about 10-12 volts higher than the plate at idle.

What are the implications of this condition? Does the tube act like a triode instead of a pentode? I have some very good sound coming from the amp so maybe this isn't such a bad thing - although the tone I'm liking may be being created elsewhere in the amplifier.

Thanks for any input on this.

Greg

[Merlinb]

It will still operate as a pentode, and it is an acceptable way to run a preamp pentode. The large screen dropping resistor prevents large amounts of screen current flowing, which might otherwise occur in a power pentode when the screen voltage is higher than the anode. If you like it, stick with it!

[Wilder Amplification]

That's because the EF86 only draws 0.6mA at idle, which means the screen resistor isn't dropping it very much at all and the OT primary DC resistance is dropping more voltage to the plate than the screen resistor is dropping voltage to the screens.

To get a 5 volt drop out of that you'd need about an 8K screen resistor. Or you could do an 8K resistor between the plate and the screen nodes (assuming you're using a choke) but this will also increase the drop to the rest of the amp, which you can adjust other resistor values to compensate for.

[tubeswell]

Hi Greg

Have you got a schematic you can post?

[GregS]

Thanks for the replies.

It will still operate as a pentode, and it is an acceptable way to run a preamp pentode. The large screen dropping resistor prevents large amounts of screen current flowing, which might otherwise occur in a power pentode when the screen voltage is higher than the anode. If you like it, stick with it!

Ok. That's good to know. Right now I'm not sure how much this stage is affecting the overall sound. I'll look at it more closely. I will probably try a larger value screen supply resistor as well to bring the screen volatege down and see how that sounds.

Jon, the EF86 I'm using in a pre-amp stage, not the power amp. I'm using 2 6v6's in the power amp and the conditions there worked out as planned.

Tubeswell: at the moment my schematic is a hand drawn mess with many revisions. Were you interested in the EF86 part in particular or the amp in general? I'll be away on vacation for the next while so it'll be a couple of weeks before I can get back to this. In the meantime I can give you some details on the EF86 stage. If you have Merlin's book it's based on the example circuit with variable screen bypassing. I used a 75K plate load resistor and a 330K screen supply resistor. I started with a 820R cathode resistor and later switched to a 1.2K. The cathode is bypassed with a 25uF cap. The screen bypass cap is 100nF. My HT supply voltage is 283V. With the 820R cathode resistor I measured 125V at the screen, 113V at the plate and 2.32V at the cathode. With the 1.2k cathode resistor: 143V at screen, 131V at plate, 2.98V at cathode.

I can post a schematic when I get back if you're still interested in this which brings me to another question: Is there a software program I can use to draw one directly on the computer rather than by hand?

Thanks,

Greg

[tubeswell]

Hi Greg

I don't know how much help I can be with the 125V scr voltage (since I couldn't find an EF86 tube characteristics chart for an HT of 280 with a scr of 125V) and even tho' I looked at ch 3 in Merlin's book, I'm confounded if I can draw a estimated set of grid curves for a 125V screen voltage.

But I did manage to do a DC load line on the Philips datasheet for 75k load line at HT = 280V scr + 140V, which is close enough to your other example. I couldn't get the loadline to show a bias of ~-3V with 1k2 rk, (it was more like 2k2 rk) and the plate voltage was about 175V. So much for theory huh?

However since it all a bit quick'n'dirty, I drew another DC load line for 47k, and it goes through the 0V Vg curve better. With a 680R rk it looks like it'll bias at ~-2V with a plate voltage of just over 140V. (Actually come to think about it, I'm not sure how much help any of that is at all.) Interested to hear what others have to say

[soundmasterg]

I've always read that for traditional pentode design, you should aim for a screen voltage that is .3 to .4 times lower than the plate voltage, and thats what I've done in my projects and it works pretty well. That said, you see the Vox AC15/AC30 circuit with the EF86 where the screen is often higher than the plate, so it will work that way obviously. I think as long as you're within the specs of the tube, and it sounds good, roll with it.

[Merlinb]

I don't know how much help I can be with the 125V scr voltage (since I couldn't find an EF86 tube characteristics chart for an HT of 280 with a scr of 125V)

On page 69 of my book there is a set of curves at 130V. Close enough?

[tubeswell]

On page 69 of my book there is a set of curves at 130V. Close enough?

Hi Merlin, does it matter what the plate voltage is when you draw the grid curves for a particular screen voltage? Or are the grid curves for a particular screen voltage always the same, no matter what the plate voltage is?

[Merlinb]

Hi Merlin, does it matter what the plate voltage is when you draw the grid curves for a particular screen voltage? Or are the grid curves for a particular screen voltage always the same, no matter what the plate voltage is?

Strictly, when you transfer the data points from the mutual characteristics onto the anode characteristics, you should do it at the same anode voltage as indicated on the mutual characteristics graph (usually 250V). Then you can use your judgement to estimate the curvature of the grid curves/knee.

[GregS]

Hi Tubeswell,

Thanks for your efforts to help with this.

But I did manage to do a DC load line on the Philips datasheet for 75k load line at HT = 280V scr + 140V, which is close enough to your other example. I couldn't get the loadline to show a bias of ~-3V with 1k2 rk, (it was more like 2k2 rk) and the plate voltage was about 175V. So much for theory huh?

I think the reason could be that the grid curves vary considerably depending on the particular tube your using. I'm using a new JJ Electronics EF86. I based my design on the anode characteristics graph in Merlin's book with the screen volatage of 130V. Since my actual results were significantly off the theoretical ones I concluded that the grid curves must be slightly less compressed for the tube I'm using than for the one that was used in the graph. Also, something I missed when doing the design was that when using a 75K load rather than the 47K used in the example the load line would pass further below the knee of the curves than planned for. Using a lower screen voltage should compress the curves and compensate for this. I will try replacing my 330K screen resistor with a 470K and see what happens.

Greg

[GregS]

After changing the screen resistor from 330K to 470K the new voltage readings are: HT 292V, screen 122V, cathode 2.56V, plate 160V. Again the plate resistor is 75K and the cathode resistor is 1.2K.

These relative voltages seem more normal to me for a pentode than they were before with the screen higher than the plate. However, I can't tell any definite difference in the sound of the amp either way. If anything it has even more gain than before. Right now I'm getting into power amp distortion with the volume at 1 so I'll definitely have to increase the attenuation between some of the stages.

Greg

[tubeswell]

Looking at the Mazda EF86 datasheet circuit values for HT of 300.

For Ra = 100k, Rg2 = 390k , Rk = 1k; 129V plate, 62V screen, 2.4V cathode
For Ra = 220k, Rg2 = 1M, Rk = 2k2; 194V plate, 53V screen, 2.42v cathode

Therefore I'd say, if you want a lower plate voltage without gain, increase the plate resistor further and increase the screen resistor, or increase the cathode resistor

[GregS]

Actually I'm ok with the way it is now. I've always had way too much signal going to my power amp in all of the configurations I've tried with the EF86. Acutally I wasn't trying to solve the "too much gain" problem with the changes to the EF86 stage - just trying to get the relative voltages more in line with normal pentode operation.

I have to put some attenuation after my pre-amp to limit the amount of signal to the power amp. I intend to learn to use my scope to see what's going on with the signal throughout the amp. That should be fun!

Greg

[Alexander]

While you're adjusting part values, you may want to drop your cathode bypass cap value down to 2.2uF or 4.7uF, and try a plate coupling cap value of .0022- .0047uF... I'm assuming you don't want bass response down to 10hz! Pentodes require much smaller values than triodes. More HF response usually equals more tasty harmonics ;-)

try it you may like it!