1. Win a Broadcaster or one of 3 Teles! The annual Supporting Member Giveaway is on. To enter Click Here. To see all the prizes and full details Click Here. To view the thread about the giveaway Click Here.

The EF-Bomb: A unique EF86 preamp with ECC99s as output tubes

Discussion in 'Shock Brother's DIY Amps' started by laird, Jul 30, 2009.

  1. laird

    laird Tele-Holic

    Posts:
    741
    Joined:
    Jul 8, 2009
    Location:
    Palm Harbor, FL
    Warning: This is a long story, but a fun one if you like tinkering around with amps and burning your fingers with soldering irons!


    Preface:
    Once upon a long time ago I built my own version of an AX84 Firefly and fit it into an old Frontman 15 cabinet. It was okay but not good enough for me to want to play that amp. So one day I decided I wanted a true low-power Marshall and built a 1959/1987 type design into the same Frontman 15 cabinet... That turned out to be one CRAMPED workspace!! To me, the real magic of the JMPs and JCM800s was the quad-output section in the 100-watt models. I wanted to get the same richness but in a package that's ACTUALLY quiet enough to run at home (unlike an 18-watter). So I decided to use 4 triodes (2x 12BH7A) in push-pull for the power section.

    A pic of the Mini Marshall

    It sounded very good, better than any 5w-or-less Marshall type clone I've tried, but not as good as I wanted. After trying countless tweaks and bias changes and grid stopper changes I decided to order some JJ's ECC99s and give them a shot. On their spec sheet it suggests using them for low-power push-pull output stages, and I hadn't found anyone on the net who really put them to the test and published their results. The sound was amazing, actually amazing! It roared like a 100-watter but without the bleeding eardrums, with the richness and swirl that only a quad-output can provide. With just one ECC99 the push-pull sounded every bit as good as two 12BH7As, but two ECC99s was just heavenly!

    But the amp gave me some trouble. The B+ wasn't as high as I wanted for a JCM800 preamp, the chassis was way too cramped, I bozo'd the layout, and as a result the noise floor was too high for me to get good recordings.

    So this little amp sat around for a while, looking cool with its knobs that all go to 11, but it didn't get much play. I built a true PTP AC15 and fell in love with the magic of a good EF86. But my AC15 didn't have a tone stack on the EF86 channel and I just HAD to know what it'd sound like with a full tone stack. So I gutted the mini Marshall and set out to create a unique low power amp that would showcase the tone of the EF86 and the dynamics of the ECC99 quad-triode power section. And I was going to do it with 100% spare parts I had laying around. Here's how the amp I affectionately call the "EF-Bomb" came into being...

    The Preamp:
    It all starts with the EF86. Well, almost. The EF86 is very sensitive to input strength so I decided to place the volume before the EF86. I also didn't do any RF attenuation on the input because (1) I like the air traffic controllers and Cuban AM radio stations around here, and (2) I didn't want to suck out any of the signal.

    From the EF86 I dumped straight into a 12BH7A cathode follower. This keeps the tone stack's load off the EF86 so it can keep running in its happy little perfect world. 12BH7As, if you haven't heard of them before, are basically beefed up 12AU7s with super huge long plates. They're excellent current drivers and perform well in no-gain situations such as cathode followers without changing the signal as much as 12AX7s do.

    The cathode follower drives the tone stack, which is a Fender tone stack with a few value changes. I actually used a 1Ma pot for bass but jumpered it with a 470k resistor (hey, this was a 100% spare parts project!).

    Phase Inverter:
    Next came the phase inverter. I decided to use a concertina PI instead of the typical long-tail because I didn't want the added gain and coloration that LTPIs produce. The operating range of the 12BH7A's grid voltage is huge (16v possible swing at the voltages I'm running) so I know this PI can take anything and everything being dished out and split it without distortion. Plus I had a trick up my sleeve...

    Post-PI Current Driver:
    Now this is where I really wanted to break the mold. Reading around various amp building forums I saw a number of people talking about running current drivers into the power tubes but I hadn't seen anybody actually DO it. The concept sounds good - keep a nice strong signal pushing the power tubes and they'll in turn run a more consistent current into the output transformer. On paper it should give a richer, clearer, more developed sound at lower volumes. So let's try it!

    Each side of the phase inverter goes through a .01uf coupling cap and into a 12BH7 triode, biased more for current than gain. After initial build I was getting some strange behavior and it took me a while to track it down. For some reason the triode connected to the cathode-side of the phase inverter was getting to some sort of grid-bias condition and it was causing the two halves of the signal to be way out of balance - like 90%/10%. I tried multiple tubes and had the problem no matter what type of tube was in there. Running a 100k ground reference resistor brought the two sides to the closest balance and solved the problem.

    Output:
    And finally the signal passes .01uf coupling caps, a 470k ohm ground reference and hit a 1500 ohm grid stopper on each phase before going into the 4 ECC99 triodes. They are cathode-biased with a 330-ohm 3-watt resistor bypassed by a 22uf/35v cap. It goes into a Hammond 125C output transformer at 5600 ohms and out to 4-ohm and 8-ohm secondary taps. I'm using a little Jensen C8R which does a better job than expected.

    Power:
    I used a Hammond 269EX for the power supply; the same one that powered the original Firefly. If I was spending money on the project I would've chosen a higher voltage secondary to get a higher B+. The ECC99s and 12BH7As can operate at 400v so there's definitely more potential headroom available.

    The Result:
    As it is it's a surprisingly clean amp, and the tone at low volume is fully developed and very musical. Once that volume knob gets up around 7 or 8 it has this big, Big, BIG overdrive sound that leaves me wondering how a single 8" speaker with 5 watts of signal could possibly be making all that music! With a bigger speaker it sounds even better, but the tone can start to get a little bassy. That's why the "high" input is a high-pass filter, to reduce the low end when using a larger speaker.

    Lessons Learned:
    Don't build an EF86 into a combo, ever. EVER. EVER!!! I did it only because I already had the chassis loaded and ready to go, but the natural microphonics of all EF86s are a problem. In this design the only attenuation between the EF86 and the speaker is the tone stack, and because of that every microphonic nuance comes through the amp at full volume. I have tried quite a few variations to remedy this but in the end I ALWAYS get some degree of rattle when playing with the internal speaker. Using an external speaker is MUCH better and microphonic rattle only becomes an issue if the amp is mic'd and going through a PA.

    Get a decent output transformer. At first I tried a cheap no-name P/P OT that Antique Electronics sells, and it got TROUNCED by a tiny 3-watt Hammond 125A!! The 125C sounds great and it definitely doesn't break the bank.

    More Power Filtration is a GOOD thing. I tried by starting with 8uf caps like some Vox designs (but of course I didn't use a choke) and the hum was pretty bad. I put in bigger filter caps and it made an immediate difference.

    Links, Pics, Etc:
    Schematic, as my prototype is built
    Suggested Layout if building from scratch
    Some pics of the build

    I need to put some sound clips up. If I spent half as much time playing as building I might actually start sounding good. Anyways, they'll be coming soon and then everybody else can laugh at my playing too. :p

    I hope this inspires some of you, or maybe give you something to build if you're bored. I'd love to hear any suggestions... This is my first truly self-designed amp and it has been quite a humbling learning experience!

    -Laird
     
  2. Natstrat79

    Natstrat79 Tele-Meister

    Posts:
    339
    Joined:
    Jul 28, 2007
    Location:
    Iowa
    Cool project! This thread is definitely more fuel for the fire as I have really being itching to build a small sub 5watt amp that uses either 12au7, 12bh7, or ecc99 as power tubes.
     
  3. celeste

    celeste Friend of Leo's

    Age:
    59
    Posts:
    2,905
    Joined:
    Jun 24, 2006
    Location:
    Maryland
    Cool project.

    A few observations and suggestions. Microphonics becomes less of a problem as you lower gain. You can lower the gain of the EF86 to around 100 and mostly beat the microphonics but still have all the pentodey goodness. Add a gain stage in front, to recover the lost gain, and you can over drive the pentode. Try adding a 250-500K pot below the .1uf screen cap, it will give you control over some squishy compression.

    Putting a volume pot before any gain cuts the signal where it is most vulnerable to noise.

    One of your "current driver" girds is floating, every grid need some kind of a ground reference. And what is the purpose of the current driver, the 12BH7 cathodine can move some current on it's own. Blocking distortion is still an issue because of the coupling caps. I am running 12AU6 based CF's into DC coupled 6N6P's with .3W grid dissipation, but you are not pushing your '99's hard at all.

    As an experiment to add richness, try converting 2 of the ECC99's to fixed bias and run them cold, run the cathode biased hot.
     
  4. laird

    laird Tele-Holic

    Posts:
    741
    Joined:
    Jul 8, 2009
    Location:
    Palm Harbor, FL
    Thanks for the suggestions Celeste!

    One of the things I can't believe I failed to mention was that I was designing this amp to be low-gain, focusing on maximum 'clean' operation. One annoying thing about a lot of low-wattage amps is how uselessly buzzy or farty they become in the last 1/4 of the volume knob. I wanted to drive the EF86 shy of distortion, run all the interim stages within their max voltage swing (at full volume) until the signal reaches the output triodes, where there would be enough gain available to overdrive comfortably but not ridiculously, even when dimed.

    Pushing a 12au7/bh7 at low gain (5x) into the volume control, then lower the gain of the EF86 to 100... That would double the voltage swing at max volume, which would cause the PI to clip in the top registers of the volume control. A grid stopper and a lower-impedance volume control could help bring the voltage down but it would also attenuate some of the highs. It's a tradeoff that might work well. I wish I had an extra triode to test it with. :)

    I tried the squish control on this circuit and it didn't sound good at all. I'm not sure why though. Maybe the 250v B+ is too low for the squish?


    Yeah, that's definitely the weak point noise-wise. The other standard option was to stick the volume after the tone stack, but that's so... normal! And it didn't reduce the noise or microphonics as much as I thought it would. I also liked having some control over the input voltage, the volume control is really dynamic.


    At one point I tried putting a 100k ground reference in on the other grid, but it didn't make any measurable effect. I didn't feel like redoing my layout to fit it so I just kept it out. ;) But you're right, it should be there.

    The cathodyne can move some decent current, but it has some issues:
    * The output impedances of the two phases are very different
    * cathodynes do not respond well to high impedances
    * The plate and cathode behave VERY unevenly to grid-conducting loads (which driven power tubes provide)
    * Its input voltage swing was deliberately set below clipping levels, and since it runs at unity gain its output voltage swing remains below the ECC99's clipping levels.

    An AC cathode follower stage would probably be the ideal tool to resolve this in most circumstances, but that wouldn't provide the additional gain I wanted to overdrive the ECC99s. The low-gain 12bh7 setup provides a balanced, low-impedance load to the cathodyne while amplifying the signal past the ECC99's voltage swing target (when the input signal is strong enough).


    The high current drive of the 12bh7 stage drives enough power to overcome power tube grid conduction within the bypass range of the coupling cap. Since the same value coupling cap is used in the previous stage there is a minimum of signal below the bypass frequency, and what is below the frequency is significantly attenuated. It doesn't completely eliminate blocking distortion, but the gain of this amp design isn't terribly high so it doesn't become a big problem.

    The ideal solution would be to DC-couple to the power tubes. It'd take a much higher B+ to reach the right operating point and I just don't have enough to work with in this power supply. :(

    For the bias experiment, should I convert both triodes on one phase or one triode on each phase? I guess either way could work, just different results...

    Thanks again!

    -Laird
     
  5. fernando

    fernando Tele-Afflicted

    Posts:
    1,141
    Joined:
    Feb 6, 2008
    Location:
    Barcelona, Spain
    Could 12AU7 be substituted without further changes?
    The 12BH7s are not cheap anymore...

    Also, I have some 6N6P, I know they are very similar to ECC99, but just checking, would you change any value to acomodate them?


    This amp looks very interesting, thank you!
     
  6. laird

    laird Tele-Holic

    Posts:
    741
    Joined:
    Jul 8, 2009
    Location:
    Palm Harbor, FL
    Yep, 12AU7s will do just fine as a drop-in replacement. I was a little surprised at how much of a tonal difference V2 and V3 made, and I prefer the 12BH7A's sound. Plus I have a few laying around. :)

    On a quick search on the 6N6P (never looked at 'em before) it seems like they could work. The output resistance is pretty close (2k vs 2.3k), maybe experiment with different output impedance taps on the OT. The big question is what value cathode resistor to use for bias. I'd start higher (500 ohms/10w?) and work my way down until the bias is hot enough.

    Since that schematic, I've made one correction for rev. 5. See the 100k ground reference resistor on the last 12BH7? It should be 470k, and there should be one on each of the signal inputs.

    Right now I'm building a rev. 6 prototype, which makes some major changes in the hopes of lowering noise, increasing compression and eliminating blocking distortion. Unfortunately I forgot to trim a rectifier diode lead, it touched ground and melted my PT. :( So I'm waiting on a new PT before I continue. Rev 6 takes advantage of the higher max voltage of the 12BH7A and ECC99, so I wouldn't recommend a 12AU7 in v3 or the 6N6Ps in the new design.

    http://www.vectorstar.com/laird/music/ef-bomb/ef-bomb r6 schematic.jpg

    -Laird
     
  7. Groovey Records

    Groovey Records Banned

    Posts:
    2,390
    Joined:
    Apr 5, 2008
    Location:
    Desolation Row
    Laird the EF86 is my favorite tone generator I recently read a paper on using one as an output tube!

    I'll let you know when I find it
     
  8. fernando

    fernando Tele-Afflicted

    Posts:
    1,141
    Joined:
    Feb 6, 2008
    Location:
    Barcelona, Spain
    Thank you Laird. I was thinking also in using 12AV7 I have.

    Anyhow, it may be possible to use the 6N6P to substitute both the 12BH7 and the ECC99?
    I've seen it substituted in amps like the Firefly for the power section. Would it work in your EF-bomb?
     
  9. laird

    laird Tele-Holic

    Posts:
    741
    Joined:
    Jul 8, 2009
    Location:
    Palm Harbor, FL
  10. fernando

    fernando Tele-Afflicted

    Posts:
    1,141
    Joined:
    Feb 6, 2008
    Location:
    Barcelona, Spain
    thank you Laird, I have some 6N6P laying around and I'd like to try them.

    Then I thought maybe your amp could be done with an EF86 (even 6J32P :) and 4 x 6N6P!

    But also see how 12AU7 and 12AV7 would work in the pre...
    The 12AV7 is reported as having a very nice high freq response and could do justice to the brillance of the preceding EF86...
     
  11. ThermionicScott

    ThermionicScott Poster Extraordinaire

    Age:
    40
    Posts:
    6,179
    Joined:
    Dec 3, 2005
    Location:
    CID
  12. fernando

    fernando Tele-Afflicted

    Posts:
    1,141
    Joined:
    Feb 6, 2008
    Location:
    Barcelona, Spain
    I was reading the article yesterday.

    The needed impedance of the OT for push-pull is very high though (80k!)
    Maybe a 22.5k pri. OT, using the 4 ohm sec. tap and using a 16 ohm speaker?
    A bit too weird... any good idea for the OT?

    The 10k OT used is too far from ideal (poor freq. response I guess) and you get only 130mW

    Hey, let's start a new thread, this is the EF-Bomb thread, oops!
    .
     
    Last edited: Nov 17, 2009
  13. fernando

    fernando Tele-Afflicted

    Posts:
    1,141
    Joined:
    Feb 6, 2008
    Location:
    Barcelona, Spain
    Laird, any update on the EF-bomb?
    I'd like to build it but need a stable version since I'll be uncapable of debug it or fine tune it other than tweak the tone stack or similar simple tasks...

    thanks
     
  14. laird

    laird Tele-Holic

    Posts:
    741
    Joined:
    Jul 8, 2009
    Location:
    Palm Harbor, FL
    Wow, talk about digging up a thread from the past! :)

    Well, I've done a good bit of experimenting with the EF-Bomb, trying two different branches after this thread was done. Neither of those experiments have turned out as well as I had hoped. The "stable" design is Revision 5, found here. I prefer the sound I get with the lower B+ of the 269EX, and I a smaller cathode bypass cap on the EF86 (.47u or 1u) improves the tonal balance.

    -Laird
     
  15. tubeswell

    tubeswell Friend of Leo's

    Posts:
    2,429
    Joined:
    Jul 1, 2008
    Location:
    NZ
    Hi Laird, you probably have the EF86 set up for too much gain.

    Try this: HT 250, Plate resistor 100k, screen resistor 560k, screen bypass 100nF, cathode resistor 680R, cathode bypass 10uF, coupling cap 4n7.

    With this the plate should be around 100-120V, the screen 70-80V and the cathode 1-1.2V. The gain won't be high, but then who needs it?

    (With this set-up, you might just want to put the vol pot back after the coupling cap, and have 'normal' fender hi-lo inputs)
     
  16. fernando

    fernando Tele-Afflicted

    Posts:
    1,141
    Joined:
    Feb 6, 2008
    Location:
    Barcelona, Spain
    Hehe, yes, I kept the message marked since I'd like to build the amp.
    Thanks for your help
     
IMPORTANT: Treat everyone here with respect, no matter how difficult!
No sex, drug, political, religion or hate discussion permitted here.