Reverb driving methods

[tubeswell]

I am trying to think of an easy way of getting more 'verb before I have to change the reverb transformer on my latest DIY build (schematic on this thread http://music-electronics-forum.com/t14515/ ).

I've got a parallel triode 12AT7 reverb driver. The plate resistance (ra) is 11k on a 12AT7, so paralleling the triodes effectively does what to the plate resistance? (doubles it? or halves it?)?

The reverb transformer I'm using is a 22921 that runs 15k Pr to 8R sec. 22921 Fender Reverb Driver Transformer

The problem I have is that there is not enough signal being driven into the parallel triodes' 12AT7 grid(s) to make the reverb sing in a decent way. I have determined this because if I use a 12AX7 in V1 there is more reverb than it I use a 12AY7 in V1.

So I was wondering if I would get a better result using one half of the 12AT7 as a driving stage for the other 1/2 of the 12AT7, and then drive the RT off the 2nd 12AT7? (Or as an alternative, maybe using a JJ ECC832 with the 12AX7 1/2 driving the 12AU7 1/2, which would in-turn drive the RT?)

How feasible is that idea do you think? (anyone?)

Cheers

[JoeM]

Maybe a 6BM8 as a reverb drive?

[tubeswell]

Maybe a 6BM8 as a reverb drive?

Yep I've been thinking about that, but that would mean changing the RT wouldn't it?

[Chuck H]

How many volts do you have (at a normal clean setting for the amp) driving the parallel 12at7 now??? And by your guestimate, how many volts do you need?

Chuck

edit: Have you tried increasing the value of the 500pf cap feeding the grid? How about decreasing the 12at7 bias resistor? Heating up the bias will make the grid more sensitive, to a point.

[tubeswell]

How many volts do you have (at a normal clean setting for the amp) driving the parallel 12at7 now??? And by your guestimate, how many volts do you need?

Chuck

edit: Have you tried increasing the value of the 500pf cap feeding the grid? How about decreasing the 12at7 bias resistor? Heating up the bias will make the grid more sensitive, to a point.

Hi Chuck

I measured 435V at the plate(s) of the 12AT7 at idle. At the moment there is a 2k2 cathode resistor bypassed with a 25uF cap. I 'guess' I could go down to 820R(?) but I haven't seen a static anode characteristics graph for a paralleled 12AT7 driving a Reverb Transformer, so I have no idea what the 'safe' bias point is especially at that plate voltage (although the graph would be something between a 12AT7 preamp tube graph, but with a reactive load like an output tube graph I guess). The 435V is powering 2 triodes at the same time', so the current would be divided evenly into both triodes. So at the moment that's (not quite) the same as a single triode at 217V with a 4k4 cathode resistor (isn't it?)

I just got word that the copy of Merlin's book I ordered is (finally) arriving in a few days. I wonder if he covers this topic in it?

I haven't tried a bigger coupling cap yet.

[txstrat]

How about decreasing the 12at7 bias resistor? Heating up the bias will make the grid more sensitive, to a point.

I agree. Depending on how much/little signal you get from V1 you might need a more sensitive grid on the 12AT7.
You can get down to 680 ohms with no bypass cap (see Deluxe Reverb II schematic here: http://www.schematicheaven.com/fende...ereverb_II.pdf note the 560pF cap @ 3KV between plate and cathode) or 1K with a 25/25 cap (see 6G16 schematic here: http://www.schematicheaven.com/fende...6g16_schem.pdf). Either would leave you with a decent change in gain of this stage.

[pdf64]

'paralleling the triodes effectively does what to the plate resistance? (doubles it? or halves it?)?'
Halves it, ie ra = 5k5.

'The 435V is powering 2 triodes at the same time', so the current would be divided evenly into both triodes. So at the moment that's (not quite) the same as a single triode at 217V with a 4k4 cathode resistor (isn't it?)'

No, it's the same as a single triode with 435V on the plate and a 4k4 cathode resistor.

With 435V on the plate, I think you might run into going over spec on plate dissipation if the reduce the cathode resistor below the BF standard. SFs, with a 1k, 680, or 470R shared cathode resistor ran the plates way over spec, something like 2 watts per plate (1 watt max).
Fender got away with it, so try it yourself, but use a JAN / CV4024 etc 12AT7 with a chunky plate.
I would prefer to cascade the 12AT7 sections, use a 1k5 cathode resistor for each cathode.
I prefer to leave the reverb driver cathode resistor unbypassed, as it makes it much harder to overdrive (which gives a horrid reverb sound), and should increase the output impedance, so should better track the tank impedance, and get better constant current drive.
Accutronics advice is to drive the tank hard, but avoid the driver circuit clipping.
The extra gain of the 1st 12AT7 section (now cascaded) should more than make up for the lost gain in the driver by leaving the cathode unbypassed.

Or how about using an ECL82 or ECL86 (think of a single 12AX7 section and EL84 type pentode in one 9 pin tube), to solve this problem?
'Doh - damn US / European tube names - 6BM8=ECL82'
As you aren't aiming for maximum ouput from the driver stage, even with a non ideal load for the 6BM8, there would still be more power than with paralleled 12AT7 sections.
Peter.

[tubeswell]

Thanks for putting me right of the parallel triodes pdf64. I recollect now that I knew the correct answer a couple of years back but time fogs the memory. So 1/2 a 12AT7 cascaded into the other 1/2 will be okay to drive that RT huh?

I put another 2k2 in parallel with the existing 2k2 and it did give the tube some more gain. I also upped the coupling cap from the mix control to the first grid of the DC pair from .01 to .02. But it sounds a wee bit muddy when the dwell, tone & mix controls are all maxed out.

At the mo' I'm running a long-plate 12AT7, rebranded for Sheldon, but I've got some NOS TESA and GE (JAN spec) 12AT7s I could try again, altho' they are shorter plates and I have noticed that the gain isn't as high.

I also tried a Telefunken ECC83 Long Plate in V1 and that adds a bit more drive to the reverb. As you up the vol control the reverb gets quite a bit deeper.

The PT probably would handle a 6BM8, but I'd have to change the RT, so I might exhaust those other possibilities first.

[Tubenit]

Increasing reverb

I am wondering if you're reverb increases if you bypass the reverb tone control? Perhaps you need a larger value reverb tone pot? You could check that out by adding a 220k resistor to ground on the ground side of that pot and see what happens?



I use a one tube 12AX7 reverb successfully in all my amps and am pleased with it. However, I insert the reverb at a different juncture in the topology.
I usually have the dwell pot (1M) at around 50-75% & the reverb pot at around 3 ......... and have plenty of reverb.

I have this reverb topology on all 3 of my amps.

With respect, Tubenit

[pdf64]

tubeswell, there would be no problems with using a regular reverb transformer with just 1 section of a 12AT7 driving it, or a (suitably biased cold) ECL82 / ECL86. Or even a single section of a 12AX7, as Mesa and Tubenit have found.
Always check that your static plate dissipation is under the max specified. Just multiply the plate to cathode voltage by the cathode current. I fear a 1k1 shared cathode resistance may be overcooking the plates (what's the cathode voltage?).
Also, is that coupling cap to the DC pair needed? Just take the grid directly from the Mix control wiper, and put the 1M ground reference after the 0.1uF cap of the dry buffer. Then you will get less opportunity for blocking, and no signal loss in the Mix control, not that there would be much anyway.

[JoeM]

Yep I've been thinking about that, but that would mean changing the RT wouldn't it?

I dont think you even have to change the RT, I'll have to look for the article, but IIRC the circuit works with the Fender stock RT.

[Chuck H]

You can still increase the value of the 500pf cap driving the 12at7 grids. Volts is volts... All it's doing is setting up a voltage to shake the springs. I don't think the reverb would get "muddy" or anything if you increase that cap value to 1000pf or 2200pf, and you'll get a lot more drive.

Chuck

[tubeswell]

Cascaded reverb driver with 12AT7

Okay I did a schematic of how I think this should work. (Ver. 12e)

Seeing as how the parallel triode 12AT7 in Fender BF amps was driven at about 440V with a 2k2, and after the feedback from pdf64, I have put the single triode at 4k7. I'm not sure tho whether bypassing it would be better - I guess I can try it both ways. What do you others think? (The RT primary zed is 15k)

I think the first stage is about right (100k plate resistor? and .01uF coupling cap? - whatever I can adjust those as necessary)

If this doesn't do the goods, I'll maybe try a JJ ECC832 or changing everything to a 6BM8.

I had another qn about the recovery stage - at the moment the grid leak resistor is 220k (as per the BF type recovery stages). Wouldn't I get less attenuation of the recovery signal by increasing this to 470k or 1M? - I note that the 6g15 recovery stage has no grid leak resistor - what gives here? is grid current not as much of an issue in the recovery stage as it is in other stages? (maybe because of the absence of a preceding driving stage??). Surely any excessive build-up of grid current could cause bias runaway regardless???

I will also try some changes to the tone control as tubenit suggested, (but I haven't put these on the schematic yet

(Edit - Hmmm I've just noticed that I wired the RT primary with the Red wire going to the plates of the paralleled 12AT7, and the blue wire going to the B+. Should that be the other way around? surely it wouldn't make any difference?? - The green secondary is going to the pan and the black secondary goes to ground)

[Merlinb]

I had another qn about the recovery stage - at the moment the grid leak resistor is 220k (as per the BF type recovery stages). Wouldn't I get less attenuation of the recovery signal by increasing this to 470k or 1M? - I note that the 6g15 recovery stage has no grid leak resistor - what gives here?

The reverb tank can act as the grid-leak path- unless you unplug the tank, which is why you need an extra resistor there!
However, 220k is waaay too big, it just adds LF noise. 47k is more than enough and will reduce noise. Making the grid-leak even bigger won't get you any more signal- the reverb tank has very low output impedance remember.

If you're having trouble getting enough drive for the tank then you'll definitely want to bypass the driver's cathode resistor.

[pdf64]

Reverb driver cathode resistor - partially bypassed might be a good compromise, say 470 or 1k in series with a cap, across the cathode resistor. So try that as well.
1st 12AT7 gain stage is fine, though you don't need the 500pF coupling cap/1M grid resistor anymore - just direct link to the 270k mixer resistors. Or you could take it from the output of the 'dry' buffer. That will take some load off the 270k mixers, giving more vol and treble.
Make the 220k resistor (reverb driver grid) into the Dwell control, and the coupling cap to 2n2 (like the 6G15).
Accutronics advise that pan outputs are loaded with 220k//2n2 (rather than 1M).
That 0.01uF cap from the Mix control wiper isn't needed. And the 1M grid resistor could be moved to the 'dry' buffer output (=dry end of Mix control).
In the interest of stability, 10k or 22k grid stoppers on the 12AT7 socket would be good practice (I use them on every pre-amp gain stage). Peter.