Alembic f2b Cathode Follower ...

[lars-musik]

Quite a while ago, I build my first tube project - the Alembic f2b clone as proposed by Lorcan here: and here .

It worked, I guess, but the output was HUGE. It was supposed to replace the pre-amp in our bassist's Trace Elliot AH250, but absolutely no chance to make these two amplifiers communicate in an orderly fashion. So slammed a Master Volume pot before the output but this one had to be turned down all the way and the sound really left me wanting.

Now one or two years passed by and I learned one or two things and I think I'll give it another try. I bought all the expensive stuff like transformator and unusually large caps and I want to put them to good use.

First thing: It's likely that there's an impedance mismatch problem. On the Alembic site it says that "the output impedance of the tube circuit is fairly high". Further I learned that a cathode follower helps with that.

So I tried to put together a schematic with a master volume and a cathode follower. But: Where would I place master volume? Will the output still be too high? Is it good this way( it is basically the schematic from Moosapotamus and Merlin's cathode follower plus Lorcan's PSU:



PDF:
https://www.dropbox.com/s/suwx86uiko20zi1/AlembicF2B_CathodeFollower.pdf?dl=0

??

If you all agree to the schematic any time soon, I'll try to do a pcb layout and just rebuild the whole thing.

Thanks, Lars

[PRR]

1Meg pot at the output-jack is probably not good.

Since you have a 22K fixed there, why not a 25K pot?

But the gain of this thing IS "huge" and it may need further experimentation.

Real tubes don't use PCB. Build it on tag-board so you can re-re-wire.

[teemuk]

Buffer should be last in signal chain. No sense to buffer a 1 MEG volume pot because the resistance of that potentiometer will practically negate buffer's existence: The output impedance is everything from buffer's output to pot's wiper, in case of 1M pot, easily several hundred kilo-ohms.

Yes, this circuit has moderately high gain. But it also has very high losses due to tone stack and volume control immediately following the first gain stage. First stage practically just helps to keep noise at sensibly low levels when ample amount of attenuation gets introduced to the signal right after first gain stage. Second gain stage probably just "recovers" some of gain lost in total.

This Fender preamp is not known for overdriving and definitely not known for "high gain" tones. It's just too "lossy" for that overall and nominal signal levels at the "high gain" stages are very, very low.

[Tubebass]

The Fender circuit from which the Alembic f2b is copied uses a 10 k midrange pot. A 250 k pot in this position will reduce the loss of the tone circuit, which will make the output of the preamp much greater. It will also make the bass and treble pots almost ineffective.

[Gus]

You can remove C10 and then C1 for less gain OR you can add a series resistor in series with C10 and/or C1 to adjust the gain

Looks like you are missing the ground connection to the bridge rectifier

Add a few ohms after the diode bridge before the 47uf for less diode turn on off conduction "ticks", try uf4007s.

You can elevate the heater supply with a voltage divider(the voltage divider can do the same thing as the 470k) off the B+ and light an LED at the ground side

You can direct couple(maybe a series resistor) the CF to U2.1 plate and use a single cathode resistor Gets rid of some parts

You might want to try the circuit without the heater LED circuit, if you want a led run it off the B+

Get rid of R22 and drop the volume control resistance 

[Tubebass]

Factory manual with schematic.

http://www.alembic.com/support/F2B_manual.pdf

[Phoenix]

Hi lars-musik.

This is how I'd do it. Take the output from the cathode of the first second gain stage. This will give you the benefits of a cathode follower (low output impedance) without needing to add an extra tube stage. It will also limit the maximum output swing to roughly twice the cathode bias voltage ~3V peak-to-peak, which is a good level for coupling to other equipment. You'll also avoid having to attenuate the signal which degrades SNR. Notice I've also switched the order of the grid leak and grid stopper resistor on the input stage. It still has the same low-pass frequency response, but now does not attenuate the input signal and degrade SNR. I also believe that the 250k mid pot is an error, and should be 10k or 25k. A 250k pot will act essentially as a volume pot with very little change in frequency response, and will make the treble and bass controls essentially useless. Schematics from other sources show it to be 10k.
I've also made some recommended changes to your power supply, the filtering you were using is MASSIVELY excessive for a circuit with such low current requirements. Even what I have shown here is massive overkill. There were also some minor errors in your power supply, C9 and C101 were simply in parallel, I think you meant to decouple between the two gain stages, which I have included, but is probably not required as the stages are out of phase and unlikely to cause each other trouble. I also changed the location and value of the bleeder resistor R102 to the reservoir cap where it more properly belongs, and the value so that the time constant is <30 seconds for safety.
I've omitted the heater circuits for clarity.

Hope this helps, and let me know if you have any questions.


Edited to correct error

[PRR]

> 10 k midrange pot. A 250 k pot in this position will

Good catch. And the no-mid-knob forms use 6.8K or 15K here. 250K will, as you say, give "no" tonestack action or loss until the 250K is in the crack between zero and one.

I also note that Lars has omitted Fender's usual Hi/Lo input jackery. Leo put that hi/lo stuff on for a reason (high output pickups). And numbers from a bass-amp similar to the one cited suggest that this may be a case where hi/lo jack is warranted.

And as Phoenix says: Fender put the 1Meg on the jacks, 34K (2x68K) jack to grid. If it IS 34K, the order matters little. At 68K it is a 0.6dB difference of signal, thus signal/hiss ratio. Not worth ripping-up, but worth doing right from scratch.

S1 "hi cut" wants about 10Meg across the contacts or C2 to control POP (more like "tik").

Take the output from the cathode of the first gain stage.

That's unity-gain, and we "probably" want some voltage gain between a guitar and a power amp (but see below).

We also want some controls and tone-shaping.

> filtering you were using is MASSIVELY excessive

Lars' plan is new-age. Big caps are cheap so don't use so many. Old school would be more like four 16uFd and at least 50V total drop in the resistors. This one, because of the small resistors, turns out to be barely good enough. About 100uV of ripple. The first grid has 2uV of hiss, gain of 50, small PSRR, so at the first plate the hum is not far below the hiss. Indeed hum can be higher than hiss (we don't hear it as well).

Changing 1.5K to 4.7K still gives near 330V B+ and ripple low enough to vanish in other tube-noise.

FWIW: Fenders most-all had 1st and 2nd stage on the same B+ node. Is this vital to tone? I do not know. Your plan gives better filtering to 1st stage; but at low VOL settings the 2nd stage can run at lower level than the first. And one B+ node is less wiring (which may have been on Fender's mind).

While there: IMHO, and as seen in other designs, you can NOT get acceptably low hum with 6VAC on PCB. No Twist! My Ampeg tried, and got-away with it only because the open-back speaker had "no" 60Hz response. Epi Valve Jr tried it, and an early revision was DC heat. If I were "forced", I would blob the heater pins on PCB and air-wire twisted pair to heaters. Since this is the 21st century already, 12V DC makes much-much sense, and clean DC routes on PCB with of-course no hum at all. However the low hum "is" part of the old Fender tone.....

> this circuit has moderately high gain. But it also has very high losses

With no load, and VOLume at full for reference, I would pencil it as 50 * 1/10 * 50, or about 250. You may feel that 45 * 1/15 * 45= 135 is more typical; much depends on the tonestack settings.

What do we got? Fenders of this era overloaded their power stage with VOL on 10 and tone knobs nominal with about 20mV. (But these were aimed at guitar use.)

What do we need? OP mentions "Trace Elliot AH250". The latest specs (does not have AH250) show expected input of 78mV (or 245mV), which may be more bass-size. The available insert points are Effect nominal 0.3V, or a DI output (possibly no input here?) at nominal 775mV.

So with a hot bass (likes the "LO" jack), he can go into Effects Return with no preamp, or any near-unity-gain chain. If the axe likes the "HI" jack we need gain of near 4. If a 775mV power-stage input is available, gain 2.6 times higher. To go right into a bare no-controls Power amp (Crown DC300 and such), 1V to 2V is needed, so 2X the gain again.

Range: gain of 1 to gain of 20.

This is lower than I thought, and IMHO rather low for most external preamps into a bare power-amp; yet high for inserting in many lower-level paths.

In the Blackfaces, the second stage is not strained. There IS some strain and "tone" in the first stage, especially at higher pickup levels.

VOL full up, the gain from input through first stage and nominal tone knobs is around 50 * 1/10 or about 5. Adjustable downward with VOL. That sounds like a good fit to go to the TE's Effects Return. But the VOL pot wiper impedance is high, wants buffering. Make the second stage a cathode follower. I am not sure this is what Lars wants, it is semi-easy to try on a prototype. Results would suggest which way to go next.

But the original is well-loved and uses both triodes as gain stages. Perhaps only because that's what they had in 1972 (musical electronics is sometimes trailing-edge). But let's see where that goes. Gain-tonestack-gain will give voltage gain of 135 to 250. This sure appears to be 12 times more than we could want. These stages are designed around 220K loading. If we follow the 2nd stage with a 200K:20K divider, we get nominal loading, reasonable max gain, and reasonable low output Z. If we round this to 220K fixed and a 25K Master pot, loading is still ballpark, max gain is 13 to 25. With audio-taper Master pot, almost any lower gain (to somewhat under unity) is on tap. In first use, set VOL high, set MAST for really-loud, mark it, then use VOL for song-to-song gain changes.


12/9/2015 1:05:14 AM

[Phoenix]

Take the output from the cathode of the first gain stage.

That's unity-gain, and we "probably" want some voltage gain between a guitar and a power amp (but see below).

12/9/2015 1:05:14 AM

Woops, I meant to say from the cathode of the second gain stage, like is shown on the schematic I linked.

[lars-musik]

Wow, I haven't yet digested all the input you all gave! A massive THANKS form my side (no down regulation needed).
You seem to think of everything. I am going to study all your suggestions carefully and maybe I'll even come up with my first tagboard design (for you to help me with)!

[Ben N]

An alternative to the cathode follower that leaves you with more than unity gain in the recovery stage, but lowers the Z-out is the split plate load. If you turn R8 into a pair of fixed resistors totaling 100k, and tap your output (C7) off the junction between them you may find your sweet spot without adding a cf stage or sacrificing recovery. O'Connor talks about this in TUT 1 IIRC. I think what folks do is wire in a 100k pot (rated to handle the voltage) temporarily to dial up the sweet spot, then remove, measure, and select 2 Rs close to the two sides of the pot setting to replace R8. Also IIRC, Aaron Nelson may have done this--on a Bassman??

[Gus]

Looking at the PS section in Phoenix post

R11 can be two resistors and LED in series. 
If the voltage is say 300VDC with R11 100K you have about 3ma

Maybe you want to raise the AC heaters above ground and have an LED

Make R11 a voltage divider total resistance of about 100K
V+ to "top" resistor,  to heater center node, to "bottom resistor,  to LED anode,  LED cathode to ground.  Keep the LED at the ground side
Between the  top and bottom resistors is a elevated voltage.  You select the resistors with ohms law for the voltage you want

You now have a
B+ bleeder resistance to discharge the caps when turned off
Elevated AC heaters
LED

From building low noise tube circuits like a tube microphone and power supplies, I find adding a small resistance between the bridge and first PS caps helps.
I also try to keep any click, ticks out of the heater circuit.