18Watt.com

Merlinb wrote:

His solution for getting more chime out of it was to give the tube its own filter cap

Good advice- that goes without saying. Pentodes have zero PSRR.

This I did....8 uf. Kinda small, I know, but I was initially going for a dead-on AC30/4 vibe....

rjgtr wrote:P.S. I found your posts helpful anyway.

I'll try not to let it happen again.

My memory says that London had a lot of trouble with spurious oscillations when they first tried the EF86, and that's why they now have the "extra" resistor. To my mind, if you want to try for a different sound with a cascode circuit, try an EC91 running into an EF86 wired in triode mode (yes, it is hungry on the heaters). Or for another variation, there are a lot of sharp cut-off pentodes lying around that sound very different if you use them as triodes, and I don't think they get enough attention, especially as they are cheap!

interesting......

Well, I connected a 1M resistor in parallel with the .1uf cap on the EF86 grid (forming a divider with the 1M resistor from the filter cap to the grid itself...) and this took the voltage from about 120v down to 77v or so on the grid, and RAISED the voltage on the plate from 125v to about 172v. Interesting....anyone know why that happened? I wager it makes sense, but I'm a bit too tired to process.... Also, I tried running about 2M on the grid (no resistor to ground...just increased resistance to the grid) and this yielded similar voltage results. What exactly is going on with that? I'm sure if I knew more about Ohm's Law, etc., it would make sense, but I figured that 2M in front vs. a 1M/1M voltage divider would yield different results...

However, the voltage change did, in fact, make the EF86 smoother, softer, and more articulate, albeit a bit quieter overall. I do like that effect, but actually prefer to run it like the AC30/4 normally does, with the 220k on the plate and the 1M on the screen. This keeps the channel warm and thick, but with enough articulation to let my SG speak clearly enough. Still though, it is good to know of some options for running an EF86...

I added an EF86 and cathode follower (and some other stuff) to my brother's Silvertone 1484 and I ended up going with a 220k plate resistor, a 2k2 cathode resistor with a 25uf cathode cap, and a 1M3 screen grid resistor. I read somewhere that you should be shooting for a screen voltage that is about 30% to 40% what the plate voltage is. So after experimenting with it, I found that you get a cleaner sound if you do this, and as you bring the screen grid voltage closer to the plate voltage, the sound gets dirtier. Its a balancing act though because the plate voltage doesn't stay stationary as you're playing with screen grid resistor values. I finally settled on 106v on the plate and 92v on the screen, and the supply voltage is about 340v.

Greg

soundmasterg wrote:I added an EF86 and cathode follower (and some other stuff) to my brother's Silvertone 1484 and I ended up going with a 220k plate resistor, a 2k2 cathode resistor with a 25uf cathode cap, and a 1M3 screen grid resistor. I read somewhere that you should be shooting for a screen voltage that is about 30% to 40% what the plate voltage is. So after experimenting with it, I found that you get a cleaner sound if you do this, and as you bring the screen grid voltage closer to the plate voltage, the sound gets dirtier. Its a balancing act though because the plate voltage doesn't stay stationary as you're playing with screen grid resistor values. I finally settled on 106v on the plate and 92v on the screen, and the supply voltage is about 340v.

Greg

Interesting... Thanks Greg! Your plate and screen voltages are closer to what a standard AC30/4 arrangement produces. I find it interesting that you settled there. When I take my circuit and input the "Matchless" value resistors, I get a great disparity between plate and screen voltages, much like what happens when I add a dropping resistor across the .1uf cap on the screen. I'm glad to be learning about the options present for an EF86, and especially what those options produce in terms of tonality and response characteristics. Many thanks again to all who have contributed here...

I originally had the stock AC30/4 values, but with the supply voltage I had for the EF86, (actually closer to 325v) the screen voltage was higher than the plate, and I didn't like that, so I went with a larger screen grid resistor. I have a cathode follower direct coupled to the EF86, with a tone stack and volume control, and a gain stage using 1/2 of a 12AY7, so the gain level the EF86 was putting out was just fine.

Greg

soundmasterg wrote:I originally had the stock AC30/4 values, but with the supply voltage I had for the EF86, (actually closer to 325v) the screen voltage was higher than the plate, and I didn't like that, so I went with a larger screen grid resistor. I have a cathode follower direct coupled to the EF86, with a tone stack and volume control, and a gain stage using 1/2 of a 12AY7, so the gain level the EF86 was putting out was just fine.

Greg

Interesting... What were your plate and screen voltages before you made the changes?

dehughes wrote:...and this took the voltage from about 120v down to 77v or so on the grid, and RAISED the voltage on the plate from 125v to about 172v. Interesting....anyone know why that happened?

Lower screen voltage compresses the gid curves, = lower quiescent anode current, but bias voltage doesn't drop proportionately because it also handles screen current via the cathode, hence the increased anode voltage.

Merlinb wrote:

dehughes wrote:...and this took the voltage from about 120v down to 77v or so on the grid, and RAISED the voltage on the plate from 125v to about 172v. Interesting....anyone know why that happened?

Lower screen voltage compresses the gid curves, = lower quiescent anode current, but bias voltage doesn't drop proportionately because it also handles screen current via the cathode, hence the increased anode voltage.

I'm going to have to read that answer a few more times....but thank you!

great thread!

This thread got me curious so I converted the normal channel to cascode and I really like it! It is a fatter sound than the parallel with good complexity.

Yeah, this thread got me 'a tinkering....

So, I've done a few things:

1) Wired up my EF86 stage just like an AC30/4.

2) Wired up my EF86 stage just like a DC30.

3) Wired up my EF86 stage with a 220k plate resistor and a 1M screen resistor with another 1M screen resistor to ground, in parallel with the .1uf cap (attempting to replicate the London EF86 voltages).

For reference, here is that handy-dandy chart so wonderfully stored on this site:



My question now pertains more to what is going on with a pentode when the screen drops in voltage but the plate stays the same.... I noticed that going between the AC30/4 and DC30 options the plate voltage stayed at about 126ish, but the screen voltage went from 118v to about 72v, respectively. Now, when putting a 1M across the .1uf cap to ground (and with a 220k plate and 1M screen) the plate jumped up to 172v and the screen was down to about 75v. We've cleared up here why that happens, so my question is more towards what is happening when you have a 125v plate and a 118v or 72v screen...is the former more of a "cold biased" effect and the latter more a "warmer biased" effect? I seem to remember that being mentioned somewhere...

I'm curious mostly because I'd like to have an explanation for what I'm hearing. With the AC30/4 values and voltages I seem to have a somewhat darker but fatter vibe that is not terribly harsh in the upper frequencies, but articulate enough, and very deep and robust in the low end. With the DC30 values I'm hearing more articulation and a bit more harshness in the upper mids, and not as deep and grand a low end. I'm leaning more towards the AC30/4 values and voltages...but I'm going to give the DC30 setup another day or so...

Okay, prompted by this thread, I started tinkering too. I put the AC30/4 front end on my Lite IIb with 6v6s.

Haven't had chance to play it much, but what I've heard so far sounds great.

Measured voltages---- B+ (for this stage) is 260, screen is at 90 volts, plate is at 70 volts!! Seems odd that the plate voltage is so low. I'd swap the tube and see if it changes, but I only have ONE EF86 (NOS Amperex).

Does it matter that the plate is below the screen? Should I be increasing the screen resistor, which I assume will lower the screen and therefore reduce the current and so also raise the plate?

This pentode stuff is just what we didn't need--- more parameters!

Hmm, so I scrounged up another ef86 and some e80f tubes to try.
I get the same voltages (fairly closely) with all of them.

This I find surprising, because I was expecting the screen voltage to be very dependent on tube characteristics--- the fact that it is connected through a 1Meg resistor means that a small difference in the current to the screen will change the voltage by a lot.

SO I changed the screen resistor to 1M2, now the screen is below the plate... still sounds good... actually sounds better to use the "lo" input, on the "hi" input there's no clean (with an SG)....

I used the AC15/30 EF86 values in the reference chart in a Vox AC15 Lite and it was by far the best sounding amp I built (and even played).

I have some parts lying around for a SS recto 2xEL84 amp and I'm really (really!) intrigued by the "cascoded twin-triode operating in Pentode mode".
It would be a real achievement to get that EF86 magic sparkle and balanced harmonics with an "ordinary" 12ax7.
The Aiken posts on cascode are very promising, so if anyone has any pointers on the "cascode in pentode mode" I'm definitely interested...

I used the AC15/30 EF86 values in the reference chart in a Vox AC15 Lite

Did you measure voltages? I'm curious whether your screen came out higher than the plate with those values...

I didn't take any voltages at the time (converted the amp a few years ago for a friend, so I don't have it anymore)
The amp was an electrically exact copy of the AC15 1960 schematic minus the tremolo channel (I believe I grounded that side of the PI). Only the preamp component values were so badly readable that I used the ones on the EF86 reference chart.
What I found somewhat amazing in the AC15 schematic were the small filter caps (8uF). I wonder how much they contribute to the sound.

Sorry it took me so long to get back to this. The voltages on my brother's Silvertone with the added EF86 had the screen higher than the plate by about 30 volts. Plate was around 75v and screen was around 105v. This was with the stock AC30/4 values. After I increased the screen resistor then I got where it is now, with plate at about 106v and screen at 92v. I've read that the screen should be about 30% to 40% lower than the plate for proper operation, but in my case it sounded great where it was so I didn't bring the screen down any lower. If I was going to though, I would probably split the screen value between two resistors, or better yet, a resistor and a pot so I could control the tone even more.

Greg