Interesting Reverb, and a question

[Bob-I]

Normster happened on this one tube reverb using an ECL82, pentode Triode. I like the idea and I'm thinking of using it on my next amp, but as I was drawing up the schematic I noticed something that concerns me.

Look at the screens and cathode of the pentode. The screen filter goes to the cathode, not ground. Now the issue I see is that the screen is a 450V cap and the cathode is only 50V, plus there's no balance resistors.

Has anyone seen something like this before?

[MarkB]

I must be missing something - not for the first time. The screen to cathode cap blocks DC from getting from the screen to the cathode, and it's rated to deal with screen DC voltage. The cathode bypass cap is rated to deal with the DC voltage at the cathode. I'm guessing you could just send the screen cap to ground as in a usual pentode stage. Is there someting I'm not seeing?

[VacuumVoodoo]

ECL82 is my tube of choice for reverb drive. You can connect screen grid cap both ways, doesn't really matter. Screen grid resistor in this circuit seems rather too large, should be in the range 2-5 kohms. Also, put a 2.2nF/400V cap across transformer primary.

The pentode in ECL82 should see 8k plate load for best performance so choose your reverb transformer accordingly.
The triode tends to be microphonic so caution is required when using it as reverb recovery.

It's a good reverb configuration that sounds more like classic stand alone reverbs. Go for it.

[Bob-I]

The screen to cathode cap blocks DC from getting from the screen to the cathode, and it's rated to deal with screen DC voltage.

No, that's a filter cap, removes AC ripple.

[MarkB]

Just as I understand - the screen cap is not there to remove power supply ripple - it's there to keep the screen voltage stable under signal conditions. So it is an AC filter cap, but I wouldn't call it a ripple filter as the term is commonly used. And since the top of the cathode cap is AC ground, it is actually grounded.

Rereading, I think I understand this time your concern about the lack of balancing resistors. Hmmm.... I need to think about that.

[Bob-I]

Just as I understand - the screen cap is not there to remove power supply ripple - it's there to keep the screen voltage stable under signal conditions. So it is an AC filter cap, but I wouldn't call it a ripple filter as the term is commonly used. And since the top of the cathode cap is AC ground, it is actually grounded.

Makes since.

Rereading, I think I understand this time your concern about the lack of balancing resistors. Hmmm.... I need to think about that.

Even with balancing resistors, 300V / 2 = 150. That cathode cap is 50V.

I'm thinking of just grounding that cap. I don't see the purpose for connecting at the top of the cathode.

[oldhousescott]

The tube and the cathode resistor are your balancing resistors. Idle bias current through the cathode resistor sets the DC value at the top of the cathode resistor and consequently at the junction between the screen cap and cathode bypass cap. Even if the bias current was 10mA, that would only be 7.5VDC at the top of the cathode resistor. I think the values are fine as is.

[MarkB]

The tube and the cathode resistor are your balancing resistors. Idle bias current through the cathode resistor sets the DC value at the top of the cathode resistor and consequently at the junction between the screen cap and cathode bypass cap. Even if the bias current was 10mA, that would only be 7.5VDC at the top of the cathode resistor. I think the values are fine as is.

Sounds good to me.

[Bob-I]

I posted this question on the 18 watt forum also. From the answers I got there it was obvious that no one actually looked at the schematic.

One response was insteresting in that the designer of this amp apparently didn't like it and scratched the project. Someone posted another schematic showing an ECL83 reverb. I like this one better and it removes the issue of the screen filter.

[img1353]http://www.drtube.com/schematics/marshall/2046.gif[/img]

[VacuumVoodoo]

Well, I'd say this one's a bit weird. btw, it's ECL86 not 83.
Ponder the configuration of Dwell and Volume pots. Adjusting dwell changes volume, you have to compensate with volume pot.
Reverb return level pot configured "backwards" as a reostat will interact with volume control. ok, this we can live with.

Reverb recovery gain is too low, 10k non-decoupled cathode resistor there makes no sense to me. 1.8k and 1uF cap would seem more apropriate.

Regarding screen grid decoupling capacitor, take a look at ECL82 or ECL86 pinout and see what pins are closest to triode's control grid. I would recommend to decouple the screen grid in any case.

Consider trying this:

Brake connection between dwell pot wiper and 22k resistor sitting on top of vol pot.
Take a 22nF/400V capacitor and connect it between that 22k resistor and ECL82/6 pentode plate.

Have fun

[Bob-I]

Well, I'd say this one's a bit weird. btw, it's ECL86 not 83.

Ahh... that's a bit hard to read on the schem.

Ponder the configuration of Dwell and Volume pots. Adjusting dwell changes volume, you have to compensate with volume pot.
Reverb return level pot configured "backwards" as a reostat will interact with volume control. ok, this we can live with.

On this last schem there's no dwell control at all, variable or fixed. I'm a little puzzled by that, no grid load, connection from the volume to the reverb driver grid through a grid stopper.

Reverb recovery gain is too low, 10k non-decoupled cathode resistor there makes no sense to me. 1.8k and 1uF cap would seem more apropriate.

I haven't looked at the ECL specs yet, but I was wondering about the same thing. I actually drew 1.5K /5uF on my schem.

Regarding screen grid decoupling capacitor, take a look at ECL82 or ECL86 pinout and see what pins are closest to triode's control grid. I would recommend to decouple the screen grid in any case.

I'm confused, what does the triode's control grid have to do with the pentode's screen.

I plan to decouple the screen with 1.5k/22uf then run a 100R screen resistor. What's that look like?

Consider trying this:

Brake connection between dwell pot wiper and 22k resistor sitting on top of vol pot.
Take a 22nF/400V capacitor and connect it between that 22k resistor and ECL82/6 pentode plate.

Have fun

I'm really thinking of simplifying this like in the Marshall 2046 I posted. The de-coupling caps are getting out of hand. Right now my filter string looks like...

2x100uF series -> choke -> 22uF screens -> 12k -> 22uF PI -> 10k -> 22uF preamp -> 1.5k -> 22uF reverb return

In addition there's another tap off the first 2x100uF caps into a 1.5k -> 22uF for the reverb driver.

That's 8 filters. I can't see any way around it but I'm thinking that 10uF might be enough for the reverb de-coupling.

[VacuumVoodoo]

Ponder the configuration of Dwell and Volume pots. Adjusting dwell changes volume, you have to compensate with volume pot.
Reverb return level pot configured "backwards" as a reostat will interact with volume control. ok, this we can live with.

On this last schem there's no dwell control at all, variable or fixed. I'm a little puzzled by that, no grid load, connection from the volume to the reverb driver grid through a grid stopper.


I see now the 2nd pot is a tone control, albeit treble only. Volume pot doubles as dwell. Pentode grid is tied dc wise to gnd through volume pot. Works ok as long pot doesn't fail.

Reverb recovery gain is too low, 10k non-decoupled cathode resistor there makes no sense to me. 1.8k and 1uF cap would seem more apropriate.

I haven't looked at the ECL specs yet, but I was wondering about the same thing. I actually drew 1.5K /5uF on my schem.

Regarding screen grid decoupling capacitor, take a look at ECL82 or ECL86 pinout and see what pins are closest to triode's control grid. I would recommend to decouple the screen grid in any case.

I'm confused, what does the triode's control grid have to do with the pentode's screen.

Pentode screen pin is next to triode cathode pin, it presents risk of capacitive coupling between the two if not decoupled.
I plan to decouple the screen with 1.5k/22uf then run a 100R screen resistor. What's that look like?

Consider trying this:

Brake connection between dwell pot wiper and 22k resistor sitting on top of vol pot.
Take a 22nF/400V capacitor and connect it between that 22k resistor and ECL82/6 pentode plate.

Have fun

I'm really thinking of simplifying this like in the Marshall 2046 I posted. The de-coupling caps are getting out of hand.

There is a thrifty supply decoupling scheme where tubes whose plate signals are out of phase with each other share common filter cap. Can be a bit tricky wiring but works fine in most cases.

[tictac]

The schem with the ECL83 look like the reverb driver circuit Eric Barbour designed for Svetlana 6BM8. In that configuration the driver is typical of pentode wiring and works well.

I've used both the 6BM8 and the ECL86 (6GW8) reverb circuits and I like them. Use the 6GW8 for amp with a B+ 350 and less. The 6BM8 for Higher B+

[BobW]

Assuming the Plate were at 450V worst case, and the screen to cathode cap (22uF) was shorted, you still have a voltage divider with the 30k and 750 Ohm resistors. This places a max of 10.9 VDC on the cathode. The values are fine.

[dB King]

Bob-I, I have build this amp, if you look on the 18 watt site in the gallery you will see pictures of the Amp. Is sound o-kay a little underpower however.
Let me know if you have any questions