ECC99 push-pull build with JCM800 preamp - the JCM Eight Hundredths

[PeanutNore]

This is my recently finished second build - I've been obsessed with self-split dual triode output micro amps for a while now, and last year I built a 12AT7 Firefly in a pedal enclosure that I've been trying different preamp arrangements in.

I wanted to try something with more gain and a full tone stack, and enough power to be able to put it to good use, because the 12AT7 / 12AU7 output just can't drive enough bass to sound mean.

It is a JCM800 preamp tweaked very slightly (to my own tastes) feeding an ECC99 dual triode in self-split cathode biased push pull configuration with no phase inverter. The output transformer is a 22.5k plate-to-plate replacement part for a Blackstar HT-5, which runs the same power tube, and the power transformer is a replacement part for a Fender Excelsior. It turns out it's a bit excessive and it runs with 500v on the plate, but I've been stress testing it for the past week and the JJ ECC99 I've got in it does not seem to mind one bit. It doesn't have a center tap for the 6.3v heater winding, so it had a pretty bad hum until I made a virtual center tap with 110 ohms from each side of the winding to ground. At the same time I was adding that bit I replaced the power LED with a lamp.

I'm kind of blown away by how well this thing turned out, but I leaned heavily on robrob's site (his Bassman Micro and Deluxe Micro designs, his JCM800 explainer, and his bias calculator) and I guess with those resources plus the experience of having built a self-split mini head before I was equipped to pull it off.

Guts:


Schem:


Sound clips:
1
2

[KevinOConnor]

Hi Guys

Just as a point of interest, self-split output stages can be built using regular power tubes as well.

With the right speaker cabinet or two, a 1W amp can easily be too loud.

Have fun

[J M Fahey]

Very good idea and build, congratulations

Just a personal comment, to my ears it sounds somewhat bass heavy, I see you skipped the preamp volume bright cap and the second attenuator brightening cap.
Maybe you want to kill the angry bees in a can, cool, but in any case maybe you could try the old 1959 Super Lead bright channel bass-cutting trick (which is not the same as boosting highs) and replace first .022 coupling cap with what they used: .0022 .


In any case, it´s your choice and sounds killer as-is

PS: just curious, what speaker did you use to record?
Maybe what I hear is a bass heavy speaker.

[PeanutNore]

just curious, what speaker did you use to record?
Maybe what I hear is a bass heavy speaker.

That could be it - I recorded it using a 1x15 cab with an Eminence Delta 15 (so basically a bass cab).

I was kind of surprised listening to the recordings because it sounded much more trebly when I was in the room playing it than it did on the recording. Might also be the mic or my interface. I tuned the amp though for an open back 2x6" cab that I built using a pair of PA speakers that turned out really nice.

The bright cap on the gain pot was removed on purpose from the beginning because I don't like it. The bright cap on the second attenuator I removed while troubleshooting an issue I had with that stage and I never put it back because I ended up enjoying it the way it was.

Maybe I'll make a recording of it with the cab that it's actually meant for.

Anyway, thanks!

[PeanutNore]

self-split output stages can be built using regular power tubes as well.

I was just thinking about this earlier. The self-split design doesn't have anything to do with the pair of triodes (or whatever) being in the same bottle. My understanding of it is that the tube with the input signal connected to the grid is also serving as the phase inverter for the tube with the grounded grid, and the inverted signal from the first tube's cathode enters that second tube through the shared cathode connection. I've thought about trying it with a pair of EL84s.

[eschertron]

I've thought about trying it with a pair of EL84s.

Now I'm thinking about this too. (put head-hurting smiley thing here)
I can imagine having a design where I'd have only 1 triode available and being forced to choose between a recovery stage and a cathodyne PI. If the recovery stage is enough to generate at least 2x the desired grid drive voltage, it would be enough to make a self-split output work. Seems obvious once it's been pointed out to me

edit: head still hurting. What happens when the driven side of the self-split pair goes into cutoff? That limits the drive to the slave side, right? So a self-split pair can't be used in AB? Or at least won't sound like class AB? Dang. Now I'm going to have to read Kevin's books!

[nickb]

A fun project for sure. I love the simplicity. Good job. I reckon the output is 400mW clean

It got me thinking. The self split can only ever do class A. Also with, of necessity, a small value of resistor in the cathodes it means the currents in the two halves are not equal. On paper I experimented with a constant current source in place of the 1.5K. This improved things a lot. If I also set the current to 19mA or so ( max ECC99 Pd) and changed to load impedance to 48K I could get 1.4W out. You could get 45k from your current transformer if you connected a 16 ohm speaker to the 8 ohms tap at the expense of a little loss on the low end. If you leave it at 22.5K the power out is 900mW. If you want to play this is what I used for a current source:



The transistors are nothing special, 40V will be fine, HFE>50. Q1 does dissipate 350mW so pick something suitable. The BC337 shown is 625mW at Tambient = 25C. A bit beefier wouldn't be a bad thing.

[PeanutNore]

So a self-split pair can't be used in AB?

It only works in class A as far as my understanding goes, but admittedly I don't know anything about why it works. I know that it works, and a tiny bit about how it works (i.e. signal passes from cathode to cathode) but I have no idea what's actually going on at the cathode under an AC signal.

[J M Fahey]

That´s it.
If you drive them externally, you´re the boss and can set bias and drive as you wish, so pick any Class you want, between pure Class B and pure Class A.
Now if your only choice is driving the second tube by driving the first one , it´s clear that the second current can go "up" the same amount the first one goes "down" and vice versa, common_grid/cathode_driven amplification does not provide current gain by definition.

So best/full drive means that #1 and #2 idle current must hover halfway between full and 0 ... classic Class A bias.

Yes, using a transistor constant current source is the best and purists shouldn´t get angry as that, since it does not affect how the amp behaves or distorts.

Those who don´t want "sand" there might add a HV negative supply (or at least between -50 and -100V) and connect a large cathode resistor there or, now that I remember, since you have 500V +V , you might drop that in a useful way by adding a large cathode resistor so 50 to 100V drop across it at desired idle current.

You will need an extra biasing resistor, between cathodes and the large extra one , for self biasing, and refer grids to that one .
It will probably be around the 1k you have now but proper value will come from datasheet graphs.

In a way, you are turning that power stage into a "power" LTP.

EDIT: my earlier post disappeared, so I repost:

I recorded it using a 1x15 cab with an Eminence Delta 15 (so basically a bass cab).

That´s a Bass speaker if I ever saw one , to boot it´s a very flat, thick cone, huge coil PA type.
A Legend 15" or even more an old 15" Jensen (C/P 15N) would have been way brighter, so that explains it.

I was kind of surprised listening to the recordings because it sounded much more trebly when I was in the room playing it than it did on the recording.

Oh it always happens and it´s natural; human ear loses A LOT of Bass sensitivity al low volume, so things sound thinner/tinnier than they actually are.

Everybody complains about buzzy/tinny resistive attenuators ... fact is that they are FLAT by definition; it´s human ear which is not.

Might also be the mic or my interface. I tuned the amp though for an open back 2x6" cab that I built using a pair of PA speakers that turned out really nice.

I remember an old demo about supply voltage controlled attenuation, a.k.a. power scaling [tm] I guess, and variations, and I was *amazed* at the *unvarying* sound timbre and quality when tubes were attenuated from, say, 50W (quoting from memory) @ 450V to, say, 1 or 2W at incredible 50V or less.

Now that was *recorded* sound, microphones and preamps do not have a "loudness/sound contour" problem, but I´m certain people in thatbroom found it thinner and thinner.

Just thinking aloud, you *might* add a bass cutting switch as I suggested earler and lebel it: "playing" (flat) / "recording" (some bass cut).

Again, congratulations on the build, you went way ahead of Firefly and current projects.

[Zozobra]

That sounds pretty awesome man - have you scoped it for power?

I'm currently playing around with an ecc99 design for an amp, although I'm driving the power stage with an AC coupled cathodyne rather than self split. It looks a little something like this:
http://i.imgur.com/gvThfas.jpg

I'm intend to try out both a bax/james and bandmaster tonestack to see which I prefer in full fat mode. I'll then tweak in the bridged t mid notch to taste. It's currently set to cut quite heavily at 600 Hz but I may drop that down to 350-400 Hz.
Oh, I forgot to draw the ground on the ecc99 cathode resistor but hopefully that is an obvious enough goof.

[PeanutNore]

That looks pretty cool. I can never leave anything well enough alone for very long and I can see myself at some point in the future replacing the cathode follower with an IRF820 source follower and using the triode it frees up for a cathodyne PI. I'm sure I'd get more power out of it by a good margin. I don't have a scope (I keep meaning to build a voltage divider probe for my PC sound card though) so I have no idea how much power it puts out now, but subjectively it sounds a little bit louder than my Blackstar HT-1, and quite a bit louder than my self-split amp with an ECC81.

I'm curious about what C7 does in your schematic. I've seen that cap in other designs but I can't figure out what it does. Also it looks like you've got your first grid leak resistor connected to the sleeve instead of the tip of the jack.

I'm definitely going to follow this and see how much power I'm leaving on the table by not using a PI, although I did mean for this to be a low volume bedroom amp and I probably shouldn't fix things that aren't broken.

[Zozobra]

C7 is a treble snubber. It tames brightness/fizz, although 680p is quite a large value so this will go right into the mids to add a bit of bass boost. It'll almost certainly be adjusted to a smaller value (100-220p) as I intend to use a filtered crossline as a sort of presence control. It's quite common to see a snubber in high gain designs on the 3rd stage to kill fizz and keep the resultant intermodulation distortion from sounding nasty. And yes, that's just the cackhanded way I draw the input jack. I know it's wrong and needs the grid leak but I forget and draw it that way...

There are some cool designs for MOSFET cathodynes on here btw. A little more fiddly to pull off than a source follower but by all accounts they sound good.

[robrob]

Very cool little amp. I'm looking forward to some new sound clips through a guitar speaker.

[KevinOConnor]

Hi Guys

The self-split output stage can be built using any tube type or combination of tubes. The key to its operation is the shared cathode resistor which is UNBYPASSED. This assures dynamic balance of the circuit even though hum balance at idle is not assured - other measure might be required for that. As with most things "differential", the balance is greatly improved with balanced signal inputs, but that defeats the whole self-split motif. Besides, this is for "tone creation" not hifi.

With a pair of EL-84s, there is the usual advantage of low grid drive voltage and easy attainment of 10W class-A, maybe a little more. With two 12A_7s you can have any preamp gain up to modern metal and enough drive for the output stage.

Octal power tubes can be used in pairs or dissimilar pairs provided the latter are idled at whatever is safe for the lower-rated tube. So, a 6V6 and KT-88 could be used together and would have a rich tone that is a composite of the characters of each type. All bog standard stuff for anyone who has read my posts, FAQ or books since '95. The drive requirement is higher for these tubes so at least one gain stage must be devoted to power amp duty.

It should be obvious also that the tubes in the self-split can be operating in different modes quite easily; one tube can be wired as tetrode and the other as triode, or one in UL and the other pentode - any combination you like, or switch both tubes in unison between different wirings.Again, all standard stuff for how we've done things the past couple of decades.

In none of the above variations does one need to stray from the simplicity of a grounded Rk. Yes, an active current source will provide better balance of the circuit BUT in doing so it is CHANGING TONE of the circuit, too. Anyone who builds circuits knows this - sims do not show the whole picture. The mere fact that there is an improvement in balance tells you right off even without building it and listening to it that the circuit has changed. many hifi guys use active current-source biasing and have found to their dismay that the "sound" of the CS telescopes through the tubes. Remember two things here: first, that a CS has its own frequency response and this will be laid over that of the output stage; second, that the voltage gain from K to A is the same as from G to A, so that frequency shaping from the CS is magnified. This is why the three-terminal voltage regulator used as a CS is so completely terrible here - only a well-designed discreet CS can have any hope of being acceptable.

You can retain some of the simplicity and most of the sonic character using a low negative rail and slightly larger Rk. Going to extremes with a high negative rail will alter the sound. For most hobbyists and builders, the negative rail is one of the things they were trying to avoid and they might as well go to fixed-biasing if a negative supply is to be used - but that would defeat the self-split, too.

Have fun

[PeanutNore]

Octal power tubes can be used in pairs or dissimilar pairs provided the latter are idled at whatever is safe for the lower-rated tube. So, a 6V6 and KT-88 could be used together and would have a rich tone that is a composite of the characters of each type. All bog standard stuff for anyone who has read my posts, FAQ or books since '95. The drive requirement is higher for these tubes so at least one gain stage must be devoted to power amp duty.

It should be obvious also that the tubes in the self-split can be operating in different modes quite easily; one tube can be wired as tetrode and the other as triode, or one in UL and the other pentode - any combination you like, or switch both tubes in unison between different wirings.Again, all standard stuff for how we've done things the past couple of decades.

So if I wired the OT to give me a primary Z of ~45k plate to plate, I might be able to actually pull off using a pair of EF86 for a power stage, or even a single EF86 + a 12__7 triode of some type?