tubes for the noobs in us all
I split the impedance and reactance pages into two because it was way too much math/theory lumped together. Impedance is now here. Impedance (as a concept) is really not too abstract. It’s once you start involving frequency dependent impedance that things get really messy. Messiness and simplified explanations now have their own pages!
Not long ago I wrote a short post about MC carts and the noise contribution of tubes when amplifying such tiny signals. I focused on step-up transformers as the solution to noiseless amplification, but there is another approach. If you don’t like solid state, stop reading. Ok, now that you stopped reading and checked out the going prices for step-up transformers, you’re back. Good. Don’t worry, this approach uses the tubeyist solid-state device: the JFET.
A cascode is a compound amplifier in a totem pole arrangement. Here’s a great explanation by Valve Wizard Merlin. This allows you to achieve huge amounts of voltage amplification with fairly economic current usage and without coupling capacitors or multiple phase inversions. The driving force in this arrangement is the transconductance of the lower tube. The lower tube and upper tube do not need to be the same, nor do they even need to be the same type of device.
JFETs (junction gate field effect transistors) are voltage controlled devices, just like tubes. In fact, they bias in a very similar way: Rsource in the above raises the n-type JFET’s source voltage above the gate, similar to the way a cathode resistor in a grounded cathode amplifier raises the cathode above the grid. On the other hand, even the lowliest JFETs have a higher transconductance (gm) than the mightiest small-signal tubes. Icing on the cake is that JFETs, properly chosen and cared for, are lower noise devices. As such, they make a great lower device in a hybrid cascode.
The overall gain of a cascode simplifies to approximately:
gm(lower) * Rload
This equation is a simplified expression of the total gain of both devices:
[gm * (Rp + Rload) / (Mu + 1)] * [(Mu +1) * Rload / (Rp + Rload)]
AKA [JFET gm * load divided down at tube’s cathode] * [grounded grid gain of tube]
Rp and Mu are characteristics of the tube upper device. The choice of upper device affects how much of the voltage gain is performed by the JFET by affecting the load it sees. A high Mu and low Rp upper tube (i.e. high transconductance) presents a lower load as divided down at its cathode, thus less voltage amplification by the JFET (and more voltage amplification made up by the tube due to the higher Mu). A low transconductance upper tube does the opposite. But regardless of the tube (assuming an appropriately sized Rload), the overall gain remains the same: ultimately the transconductance of the JFET multiplied by the load on the upper tube.
So where’s this headed? Obviously there’s a full design coming to try out this idea, but the takeaway is that a hybrid cascode is potentially a great way to step up the tiny signals from a moving coil cartridge with very low noise and hand the now-larger signal off to a tube amplification stage without multiple supply voltages, coupling caps, or an expensive step up transformer.
The catch? Cascodes have poor power supply noise rejection and a fairly high output impedance. But there are ways to minimize these factors, too.
Low-voltage (48V), tube gain stage, MOSFET buffer stage, minimal parts, and modest cost. This is your first tube project. Prototype built.
So many options so little time. On the other hand, if I stopped thinking and started building, I might be able to actually try one or two of these.
Upper left:
DC EL84 shunt regulator, output fed to a grounded grid amplifier (non-inverting), then to a cathode follower (non-inverting), then back into the shunt to amplify and cancel any power ripple on the output. A CCS feeds the shunting device to provide a high impedance.
Upper right:
Same shunting arrangement, now fed by a differential amplifier (non-inverting). You can think about the dif amp kind of like a cathode follower driving the cathode of a grounded grid amplifier. Weakness here is the high output impedance of the dif amp and low input impedance of the DC shunt (shunt could be switched to non fixed bias for a higher input impedance though, see next).
Lower left:
Auto-bias shunt (varies with wall voltage) fed by a cascode. The cascode is on the ‘other side’ of the CCS to take advantage of the higher B+ headroom. This will also introduce an un-wanted ripple signal, so the upper triode is configured for 1x gain (Ra = Rplower + [Mu + 1] * Rklower). Injecting the ripple through the voltage divider to bias the upper grid would, in theory, cancel the ripple on the ouput. The ripple we want comes from the output and is fed to the grounded grid lower triode in the cascode to be amplified as an error signal and fed to the EL84. Multiple feedback loops here, may not be stable.
Lower right:
For the iconoclasts, a TL431 in the cathode of the grounded-grid EL84 controls the current. The tube is really here just to protect the low voltage SS part from the high voltage output. A cap feeds the AC ripple from the output to the TL431.
In a world where near-field and headphone listening has become an unstoppable force. Where every DIY builder is bored to death of rational and safe two-stage, single-ended triode designs. Where power supplies have become an afterthought and parts values are just plugged and chugged. Prepare your butt for a new madness. Prepare it for La Luciérnaga…
Note db scale for the predicted frequency response graph: +/- 1 dB from 40-20khz. If that’s not good enough, I’ll show you how to make it even better. There’s mucho tube math coming your way, amigo.
Here’s the worksheet I used to create the above:
Although I love me VR tubes something fierce, they aren’t in current production and I’d like a simple (or well as simple as possible, I guess) shunt regulator alternative for things like phono preamps or line stages (20mA of load current or thereabouts).
This shunt regulator uses an EL84 for the heavy shunt lifting and a 12AX7 differential amplifier (using non-inverting output) to amplify any ripple on the output (which the EL84 ‘cancels’ across Rs). Quick calculations look like about 100V of headroom would be nice to have so rectified 250Vac with the shunt should be good for about 250Vdc regulated output.
Update: Putting the differential amplifier before the shunt will unload it a bit and provide more B+ headroom for higher gain.
Q:Why not build an all-tube MC phono preamp?
I’ve been doing some reading on tubes in shunt regulator power supplies lately (lots of great articles on TubeCAD including this one). I’m planning to incorporate one in an upcoming build. In operation, this isn’t too different from the VR regulator power supply in my Matemático Phono Preamp, but a shunt regulator with a triode would have an adjustable output and might afford even better ripple rejection.
My recent series regulator project is another example of power supply regulation.
Pete Millett’s Starving Student was one of the first amps I ever built completely from scratch. Unfortunately, the 19J6 tubes have become rare (or at least no longer dirt cheap) due to all the bright eyed DIYers scooping them up to build amps. I think the world needs another <50V tube amp for beginners, so I’m designing one. Like the original, it’s an oddball tube with a MOSFET buffer and an off-the-shelf power brick (same brick, in fact).
Millett is one of my personal tube heroes. This is a tribute. Full write up coming soon (and parts values subject to change once tested).
wauwatosa tube factory 