The 300B Super Amp
A 300B single ended A2 triode amp. NOT TESTED!
This amp is directly derived from the 211 High Voltage amp, please refer to its article for design choiches.
Here’s a 300B version. It can run from a 450V supply and a ±150V supply. It’s better if the 450V supply would be tube rectified (or, in general, delayed a bit), and the ±150V one should be absolutely SS rectified and very fast. This is because we need negative grid bias BEFORE that the 300B starts to conduct.
The output transformer is a standard 4500ohm/8ohm unit, it needs to be of very high quality (>40H), and possibly very low secondary winding DC resistance. I suggest a power rating of 20W, because the amp above when driven hard could easily do 16W with 1.4Vrms in, if you choose an “hard” operating point for the 300B (85mA). Otherwise if you want the 300Bs to last longer, choose 75mA, and you’ll get 12W out with 1.3Vrms in. Those values are taken assuming 5% distortion. Obviously, distortion at 1W is under 1%.
However, please note some tricks: take a look at the bias pot. I could have done with a single pot without all those resistors: instead they’re there to limit the bias range (more negative wouldn’t be an issue, but lesser would), so if you screw it up with the trimmer, you can be sure you wouldn’t destroy your 300Bs, at least not immediately. Also, R16 is there in the case the trimmer’s wiper looses contact (oxidation…).
Please note that this is (like the 211 amp) an UNTESTED design. And not only: there are high voltages and expensive parts, I suggest prototyping it only if you have experience and proper tools.
And this is another 300B design example. This one is using a cascode frontend, feed by a CCS (battery biased, by Pimm) and that 56k resistor, that implements “partial” feedback (you can learn about it here). The usual mosfet driver, although with cathode bias on the power tube, and some cathode feedback from the output transformer. As you can see, it’s only a spice schematic, so measurements cannot be accurate, and must be intended as a simulation only, not reality. Probably it has to be optimized when prototyped.
10W at 1.5% maximum THD in the whole audio range, 1.15% at 1kHz.
Over 5W at 2% THD when the load drops to 4ohm.
9Hz -3dB point without using big film/foil caps or bad sounding electrolytics, by using negative feedback instead.
Damping factor: 4.4 at 1kHz.
The EL36 PP amp project
The EL36. Interesting tube: “Horizontal deflection in TV-sets”. It’s a very common power beam tetrode, with top cap, and 10W rated plate.
Most of the time, astonishing performance come from underrated and sightly forgotten objects. That’s true in high-fidelty, especially true with tubes.
It took quite some time for a strange one to do this strange experiment: let’s strap the EL36 in triode mode.
Here are the results. Not bad linearity for an easy obtainable tube.
So I decided to make a PP amp to suit my needs. I don’t have high efficiency loudspeakers yet, so I needed about 10W. EL84PP can give me up to 5W in triode mode, I wanted more.
I found this EL36, and tried some simulations with it, on a PP output stage: 10W almost all in class A, with unmeasurable distortion on 8k and 300V B+.
The output stage. A tube amp really needs to be designed from the end to the beginning, or from the power stage to the first audio amplification stage. Here, in the power stage, power supply requirements are dictated, and also about drive voltage and drive power. EL36 are not easy to drive in triode mode, due to the lowish gm: in the quiescent point choosen above, they need at least 35Vrms or more to drive them to full saturation. This eliminates the possibility to have wimpy driver stages: hey, I once drove an EL84 with an EL84, so no I don’t like wimpy tubes.
The gain stages. I wanted the least amount of coupling capacitors in my amp: that’s a quite exciting desing challenge. I tried various tubes (in simulation only) for the first and the second stage: at the beginning I thought of a grounded cathode stage driving a long tailed pair, to get the splitted phase outputs. This way, some interesting combinations resulted, but they all suffered from poor drive capability at those limited supply voltages: I really don’t want to make another supply in the amp.
So after a while I settled for a LTP first – then gain configuration. While using another triode section this way, I gained two interesting features:
I can enclose even the gain stages in the feedback loop (if it’s necessary), because I can inject negative feedback signal in the grounded input of the LTP.
also I can get much higher drive voltages for the output tubes, with some carefully chosen tubes and operating points.
The first stage LTP should really have low anode voltage, under 100V, to make DC coupling possible. I settled for an ECC88, a tube I know well and has proven to be sufficiently linear at low anode voltages.
I’m also pursuing another objective: avoid the distortion derived from the unbalanced drive voltages to the power tubes. LTP are a little bit unbalanced: some tricks have been developed to avoid this, I’ve choosen one: loading the LTP with a constant current sink in place of the tail resistor. So I can have adjustable current and very high impedance seen by the cathodes: drive voltage for the second triode is not more smaller than the drive for the first one, it’s now very close.
The second stage was designed to give 140Vpp drive to the power tubes (much more than needed, but in good designed amps one must be able to overdrive the output tubes without significable distortion from the driver stages): only some tubes can do that, and only a minor fraction can do this with only 200V B+ (since grid is not at ground potential, but at something less than 100V). I’ve chosen the 5687 because it’s a very sturdy and good sounding tube, another ECC88 could have been employed here but I didn’t want to cascade identical tubes, I could have pursuited harmonic cancellation this way but I found it not worth the effort.
So what’s up now? I’m searching for a matching output transformer! It has to pass 10W of power, and give an 8k load to the EL36s.
Also the CCS on the first stage has to be decided yet. And the power supply too: basic ideas are SS rectified, CLC filter, the usual negative PSU for bias and CCS. I’ll use four trimmers, one for each output tubes: getting a matched pair EL36 has never been easy 😉
New design, a little bit different:
Now all the gain is provided by the ECC83 long tail phase splitter at the first stage. Then the signal is buffered by a cathode follower, obviously DC coupled to minimize cap interactions and phase shifts.
The constant current sink I’m using under the ECC83 is this on the left.
About 4dB of negative feedback loop is applied: however I’d like to increase it but I’m extremely short on gain. As you can see, B+ is a bit low, in fact those are the values I get with a recycled power transformer that I’m using at the moment. The definitive unit will have about 300-320V B+, and at least 280V available to the LTP input stage (so I can raise anode resistors, I tried 150k with 250V B+ but I got bad results).
With some recycled no-name output transformer, -3dB points are 40Hz and 19kHz with feedback applied. I’m planning to do some listening tests to decide what is the influence of global negative feedback in a design like this.
Power output obviously is a bit low (due to low B+), only 5W or so.
And that’s probably the definitive one. Prototyped, played with it a bit, and discarded.
The High Voltage amplifier
A 211 single ended A2 beast. NOT TESTED!
What in the world could point me to design a single ended 211 A2 amplifier?
I frankly don’t know. Prices are outrageous and this isn’t an amplifier suited to beginners. I think I did this only as an interesting game, a design exercise. However, even if the amp haven’t been built and tested, I think this is a right place to start. Power supply hasn’t been designed yet.
The design, as is, is unsuited to drive 845 tubes. But since many raves about this wonderful triode, considering it better than the 211, I’m also working on a 845 version of this amp. For everyone trying to design such amps, I have to say it’s better to start with a 211, then later tweak to accept use of 845. Starting from scratch can be hard.
The Input Stage:
Input stage is a classical E88CC/6922 cascode design. I decided for this, discarding another option I considered (ECC83 in SRPP) because of better linearity, slew rate and bandwidth. In fact, E88CC has its roots exactly in cascode circuits, and the configuration I’m using is pretty standard. Gain (without nfb) is nearly 95x, or 39dB if you prefer, for an output voltage swing of ±150V.
R4 and R5 set the positive reference voltage for the upper E88CC triode, and C2, C4, R6, R14, R15 provide the insertion of power supply noise in the upper triode section, to obtain noise cancellation. This tecnique has to be evaluated on the actual amp: in fact it could give better results just to regulate the +350V supply (LM317 + IRF840 are cheap), but psu noise injection can be useful even to cancellate noise generated LATER in the signal chain (for example, 211 heaters). Obviously R14 should be set to obtain minimal noise at the output of the amp (not at the output of the cascode stage).
Nfb insertion is standard, at the cathode. Use a good electrolytic for C1.
Possible modifications of the input stage, which *can* (not yet verified) 845 output tubes to be used:
use 5687 in place of E88CC: R16 removed (no nfb), R3 = 1k, R2 = 18k, R4 = 3.9M, R5 = 2M, and the B+ to the input stage should be raised to 600V.
R11= 6.8k and Mosfet supplies ±200V.
Output of the cascode should reach 180Vpk.
The Driver Stage:
What the heck, sand here? Solid state stuff? Didn’t you hate it? 😀
No, I don’t: simply because this is the simplest, cheaper yet maybe “best” solution to drive a power triode with the grid going positive. As you can see from the 211 datasheet, with +50V Vgk the grid current raises to about 30-40mA: very few tube circuits can do this with 70Vrms swings without distorting too much (and without using an interstage transformer, or worse, another power tube driving it, a la Sakuma).
An high voltage power mosfet (the IRF820) is used as a source follower, with a split power supply of ±150V. R8 sets the quiescent current in the mosfet, and in conjunction with R11 it also sets the needed negative voltage for the 211 grid, about -51V in this case, for an anode current in the output tube of 85mA. R10 and that “strange” connection for the voltage divider is to protect the power tube in case of the trimmer’s faliure.
Quiescent current in the mosfet is about 21mA: this isn’t a problem (it’s lower than the necessary current for the 211’s grid), because on negative peaks of the source (so decreasing current in the mosfet) the source follower has only to drive the 200k resistor and the Miller capacitance of the 211’s grid. On the contrary, on the positive peaks of the grid waveform current in the mosfet will increase and so necessary grid current will be available.
The Output stage:
Pretty much standard 211 A2 single ended operation: ~1050V Vak, 85mA Ia and -51V Vgk into a 10k load. With 100Vpk swings on the grid (giving up to +50V positive drive) you can get in excess of 37W with 3.3% second harmonic and 1.1% third. One could even go higher if you set the limit on the “5%” number, however I would rate this as a 30W amplifier assuming low distortion. Output impedance is more or less 2.5ohm, for some speakers some negative feedback (like suggested before) could be easily employed: however, probably some sort of local feedback would be better, avoiding phase shifts associated with the output transformer.
Please note that the output transformer for such a single eneded amp is probably the most important single device, and no money-saving options should be choosen. Simply use the BEST: at least 50-55H of primary inductance and very well insulated (MUCH more than 2kV) windings. Parafeed output can be used, however a good capacitor and a good anode choke probably are more expensive than a single good output transformer.
The Power Supply:
not yet designed… main idea is to go solid state regulated for all the supplies, except the high voltage for the 211, that probably will use a solid state voltage doubler, with standard CLC filtering.
If you have ideas, suggestions or comments please contact me. Also if you would like to build the amp, let me know!
New version: with 845
A grown-up RIAA phono preamp: updated Feb. 2007
It uses only current production tubes (by the way, NOS ones for those types are cheap). As always, metal film resistors and low-ESR switching electrolityc caps are preferred. All resistors can be 5% precision metal film type, 1/4W, except the following: R8 (1W), R7 and R34 (1/2W), R24 (1W). Capacitor voltage ratings should be at least 350V for all capacitors except C3 and C17 (16V or more, good quality low ESR types).
If you cannot find a particular resistor value, just parallel / series them to obtain the correct value.
Of course you can parallel the electrolytics with good film types, and you can insert at the input of the phono stage the loading capacitance needed for your particular MM cartridge.
You should also select ECC88-6DJ8-6922-E88CC (they are all interchangeable in this circuit) for low noise and microphony.
Gain at 1kHz is about 54dB, and the RIAA accuracy is ±0.2dB worst case (with 10% components) from 30Hz to 20kHz. With this high gain, you could also use an MC cartridge, provided you change R4 with a 220ohm or 330ohm resistor, according to the suggested loading of your cartridge.
Output impedance is about 5kohm, it could drive a line preamp or a simple volume pot, you choose. Great to use with those “integrated” amps that are nothing more than a power amp with a potentiometer at the input.
The power supply should about 300V, and obviously regulated. I suggest the IRF840 + LM317 regulator already proposed in “My Preamp” page. Tube heaters should be clean DC, not necessarily regulated.