6DJ8/ECC88 in Dumbleator Sounds good

[talbany]

So I just finished this Dumbleator and was in the middle of rolling some tubes through discovered this..
Since I wired this one for DC filaments I thought I would try an ECC88. I picked up a bunch of these out of some old Test scopes (Tecktronics) All really nice tubes Telefunkins/ Amperex Holland Bugle Boys so just for grins popped one in and surprisingly worked fine and could make the argument that it sounds just as good if not better IMO than the 12AX's I was rolling. For one thing the 88's are dead quiet with little to no noise or hum and microphonics. Sound wise impressions were a very clear open top end and fat sounding low end!..I also found I didn't need to use my bright switches as often. Although the gain factor (u33) is lower than an AX I didn't find much of a volume difference or gain setting going between the 83 and the 88 could be do to the very high transconductance ratings on the 88. Anyway this is just a first impression and want to give it some time. I am sure the setup is not exactly optimal for this tube type as normally I would use a lower value plate resistor and possibly a different cathode set up. So any of you HiFi guy's out there that have had some experience running this tube any input on the optimal setup before I run all the curves and do any modifications I would be curious to know any opinions (good or Bad) . So far my voltages are 103V DC on the plates and around 50V of the CF (buffer side) I know this tube is also widely popular for headphone or phono preamps in some high end stereo amplifiers as well as Microphone pre's
What do you think

E88CC-TK-Tube-Data-Sheet.pdf

CathodeFollower.jpg

Tony

[norburybrook]

is there any advantage using these tubes Tony over the regular Ax7 other than if you have a box full of them?



M

[Mr. dB]

6DJ8/ECC88 has a lower output impedance and more current drive ability than a 12AX7.

[talbany]

is there any advantage using these tubes Tony over the regular Ax7 other than if you have a box full of them?



M

Yeah! It's quiet less noise like Mr dB mentions the lower output impedance affects frequency response = more treble= a clear top end without having to use the bright caps=less harshness. More Current /Drive= punchier tighter low end and more headroom= not as compressed as an AX. These tubes are known to make excellent CF buffers too Like I said I don't have it set up optimally so I want to tweak it a bit however so far it's promising alternative!

[norburybrook]

is there any advantage using these tubes Tony over the regular Ax7 other than if you have a box full of them?



M

Yeah! It's quiet less noise like Mr dB mentions the lower output impedance affects frequency response = more treble= a clear top end without having to use the bright caps=less harshness. More Current /Drive= punchier tighter low end and more headroom= not as compressed as an AX. These tubes are known to make excellent CF buffers too Like I said I don't have it set up optimally so I want to tweak it a bit however so far it's promising alternative!

great, this is looking like it could be the next evolution in the Dumbleator. do these tubes need DC filament power ? can you use the standard 6.3v AC and rectify it to DC to work?

I have one sat on the shelf waiting to be built here, I'm watching this with interest.

M

[Mr. dB]

is there any advantage using these tubes Tony over the regular Ax7 other than if you have a box full of them?



M

Yeah! It's quiet less noise like Mr dB mentions the lower output impedance affects frequency response = more treble= a clear top end without having to use the bright caps=less harshness. More Current /Drive= punchier tighter low end and more headroom= not as compressed as an AX. These tubes are known to make excellent CF buffers too Like I said I don't have it set up optimally so I want to tweak it a bit however so far it's promising alternative!

Should be less sensitive to cable length and capacitance too?

[rootz]

I can’t really see the benefit with respect to output impedance perse. From the top of my mind the cf drives a 250k pot feeding the output. Worst case the pot is half up, thus 2 x 125k parallel —> output impedance 62.5k. Full up the pot is out of the equation and thus the output impedance of the cf will dominate. Which will be low with a well adjusted Ecc88. Correct me if I’m wrong on the preceding thoughts please. If I am correct, you could drive a lower value pot more easily.

On the lower noise you’re getting Tony: is this because of the lower gain of the recovery stage (I’d guess it is around 20 now, not 60, 70 or thereabouts) or do you get a disproportionately lower noise level?

[talbany]

I can’t really see the benefit with respect to output impedance perse. From the top of my mind the cf drives a 250k pot feeding the output. Worst case the pot is half up, thus 2 x 125k parallel —> output impedance 62.5k. Full up the pot is out of the equation and thus the output impedance of the cf will dominate. Which will be low with a well adjusted Ecc88. Correct me if I’m wrong on the preceding thoughts please. If I am correct, you could drive a lower value pot more easily.

On the lower noise you’re getting Tony: is this because of the lower gain of the recovery stage (I’d guess it is around 20 now, not 60, 70 or thereabouts) or do you get a disproportionately lower noise level?

Rootz
Thanks for jumping in here you being a Hifi guy I agree However here is what I notice with respect to the difference in gain (or mu) between the 2 tubes. The 88 is around 30 vs the AX @ 100.. Although when bouncing between the 2 the volume/gain settings on the return pot yields around the same output volume wise (maybe a notch higher on the 88 and run both around 9 O:clock) Now with that being said If I do decide to keep the 88 I would lower the value of this pot. This is a problem with the Dumbleator IMO. with a 250K output the pot never reaches the 1/2 way point, just about all the lators I've built sit at around 9:oclock which is on a 2-35 taper CTS very low and gives a very high series resistance (hence the need for the bright caps) as well as makes for a sensitive pot.. I believe this to be a real flaw in the design BTW. Perhaps Dumble thought this as well and eventually went down to a 10K "send" pot on his Dumbleator II.. See here https://ampgarage.com/forum/viewtopic.php?f=5&t=5961
Anyway if I were to modify this one to accept the 88 I would lower this pot value.
In retrospect and by comparison the 88 has a lower output impedance than the AX with the same value pot No?

ra = the internal plate resistance (62.5K for a typical 12AX7)
mu = the mu of the tube (100 for a typical 12AX7)
dB = voltage gain in decibels = 20*log(V1/V2) = 20*log(61.5/1) = 35.8dB

So to answer your question proportionally and at the same output/Volume levels on the Return pot the 88 has a darker background at idle..
It would be nice to run the Spice Sims on both tubes however don't have the software anymore and it's been forever since I've run that software.

BTW. I actually built an amp with an 88 for V1 that sounded pretty good (Bright amp) and didn't notice much of a difference in gain between that and an AX for an input. Nothing like going between an AY and certainly not an AU which I believe has a Mu of around 30 or so..I'll have to go back and look up how I set that one up been a while??

Thanks!
Tony

[rootz]

That is a lengthy answer Tony. Would love to reply to everything, but only got a lunch break and phone to do so.

The output impedance of the cf is 1/gm. With a 100k cathode resistor on a 12ax7 I believe Zo is around 600 ohms. Low enough to drive a 10k pot imho. No idea what the gm is in a dumbleator cf. maybe higher resulting in an even lower Zo. A 12at7 would also be a good sub for an ax7. Zo would drop even more. A ecc88 would be an excellent choice for a white follower, but that would stray far from the original.

A level pot before the follower would be even better. But that would Mean cascading volumes from master on the Dumble to input vol on the dumbleator

[talbany]

That is a lengthy answer Tony. Would love to reply to everything, but only got a lunch break and phone to do so.

The output impedance of the cf is 1/gm. With a 100k cathode resistor on a 12ax7 I believe Zo is around 600 ohms. Low enough to drive a 10k pot imho. No idea what the gm is in a dumbleator cf. maybe higher resulting in an even lower Zo. A 12at7 would also be a good sub for an ax7. Zo would drop even more. A ecc88 would be an excellent choice for a white follower, but that would stray far from the original.

A level pot before the follower would be even better. But that would Mean cascading volumes from master on the Dumble to input vol on the dumbleator

Thanks Rootz
Yes we've talked about a pot after the master being a good idea there and agree with that. I guess my whole point in that previous post to break it down is this. I've always had plenty of gain and volume on tap available with the Dumbleator as it sits and at times ( too much gain) so I am willing to sacrifice some gain for clarity and punch and above all a less noisy interface the question now is can I improve on this design!
So what are the advantages to using "white follower?

Tony

[rootz]

Many ways to improve the Dumbleator, that is, when you consider some design considerations to be a flaw in the first place. I suppose it has got its charm as is.

A White cathode follower is essentially a unity gain class a push pull configuration. Half the output impedance, twice the current swing (push-pull). 2nd harmonics a bit less than normal CF, higher order marginally less. With an ecc88 low Zo can be reached, around a couple of ohms should be doable. Although nog entirely in class a. Anyway, much more complicated circuit with minimal benefits for a dumbleator I think.

With a level control before the CF, You could stick with an ecc83 if you want, or use any dual triode that suits. With a level pot after the CF, you could easily use a 25k pot, probably a 10k too. Nice to experiment with. With higher gm triodes you could go even lower/drive the pot more easily.

I should have an LTspice file of the Dumbleator. Let me see if I can find that one.

[martin manning]

Some recent discussion on the D'lator send buffer here: https://ampgarage.com/forum/viewtopic.php?f=4&t=32162

[rootz]

Thanks Martin for the link! The question is asked by Gill why Dumble did use such a high value pot after the CF. Maybe because a 12ax7 will not drive a low Z load? The simply isn't enough quiescent current to do so. A 12ax7 will quickly run out of steam driving a 10k or even 25k pot even though Zo is low enough. I think the only proper way to drive a lower load with a ax7 in this circuit would be to raise the quiescent current by lowering the bias resistor of the CF to 680 Ohm. Even then it has it's limitations, but clipping I the CF occurs much later, with much more driving voltage.

Or you could just pop in a 12at7 (or ecc88 for that matter). I think a at7 would require less reworking of the circuit and work good enough for the application, only slightly raising the distortion levels, mainly because of a lower send pot value adding some compression in comparison with a stock Dumbleator. Only a few resistors need to be changed: the dropper resistors and maybe the cathode resistor of the recovery stage.

Added a sim of the effect of cable capacitance on a stock Dumbleator vs one with a 12at7 and 25k send pot.

Anyway, I'm staying away from the actual topic: using an ecc88 in a Dumbleator.

[Bear]

So for optimizing a Dumbleator for use with +4dBU effects and mixers, lower impedance is going in the right direction. On the whole is this moving the right way?

[rootz]

I think that really depends on what you’re after. If you want to drive long cables without using the treble caps, lower value pots are the way to go. The 270p cap on the recovery volume will help, but also make a peak around 3 kHz iirc.

So you want a real Dumbleator? Use the schematics/layouts found in the files section.
You want to tweak? By all means do it. I did not build a Dumbleator myself. Could have though, but chose to go opamp instead. I now have an outboard loop in a Hammond 1950b box that can drive long cables and drive the pi to clipping if I’d ever want to do that.