Sounds cool, doesn’t it? Here’s a chance to build your own amplifier – a simple design with outstanding sound quality. We are offering the schematic absolutely free, and we think you will be surprised at just how good the sound quality of this amp is.
Let’s Go Inside
There are four tubes, two per channel:
- KT88 (or KT90 or KT120)
Now, the 12AX7 is not considered the absolute premier front-end tube, but that is largely because it attenuates high due to its higher plate resistance, which forms an LC filter with the capacitance of the following stage. This “feature” causes loss of detail, a problem that was neatly and easily remedied by using a cathode follower (extremely low input capacitance) as the next stage.
The KT88 output tube is triode-connected (screen grid tied to plate through a 470 ohm resistor) and makes a pretty nice output tube.
There are several innovations in this circuit that will be described in detail later in this article, but the upshot of it all is that this is an amp that is clear, detailed, full and crisp, and with a soundstage that makes the music sound as if you are right there as it is being played live.
This was our first attempt at a sellable hi-fi amp. However, there were two minor problems with the original design:
- We couldn’t provide enough auto bias adjustment for the triode-connected KT88 output tubes without adding a sound-quality-degrading bypassed cathode resistor. Pentodes are much less reliable than true triodes such as a 300B or a 2A3.
- In geographical areas with remarkably high line voltage (as unlikely as that is) there was the potential for problems with the power supply and the output tubes.
While both of these problems are easily overcome for a home designer willing to fix the biasing problem by adding a biasing potentiometer and adjusting it manually, it is much less user-friendly, and therefore, something we prefer not to do as a rule. If excessive line voltage is a problem, it can be fixed by adding a resistor in series with the primary of the power transformer, or using a variac to power the amp.
There are several innovations in this amp. One is the use of a cathode follower directly coupled to the output tubes. This keeps the tubes from needing continual bias adjusting since changes in biasing caused by grid current are very low — the total impedance of the tube’s grid circuit is around 3K ohms by virtue of the cathode follower’s inherent self-adjusting.
The biasing voltage is also achieved rather differently than in most amplifiers. As a choke input filter is used, the average of the sine wave coming in is provided to the plates of the tubes, instead of the capacitor charging up to the voltage of the sine wave’s peaks, as would be the case in a capacitor input filter. However, this extra voltage from the peaks of the sine wave coming in hasn’t disappeared—it’s still around somewhere, and can be found across the choke, as might be expected. By putting the choke in the negative lead of the plate supply instead of the positive (as is the norm) and rectifying the voltage developed across the chokes (see the schematic for an idea as to how this is done), a negative voltage can be obtained.
Another advantage with this scheme — not only will the negative voltage increase with an increase of plate voltage, thus helping to adjust the bias to reflect changes in line voltage, but, due to the internal resistance inherent in every choke, a certain amount of auto biasing will occur. The biasing voltage will increase with an increase in plate current.
There are three other things that you will find differ from most amplifiers:
- We only used around 60% of the total plate current of the KT88, rather than 90% of the available swing as is standard in most tube amplifiers. In other words, we are not trying to push the tube to its limit; rather, we are sacrificing some power output to achieve good sound.
- We used a diode to bias the input tubes. We did this to get rid of the cathode biasing resistor and associated bypass capacitor, with its slight degradation of sound quality.
- The single coupling capacitor stage has an unusually high RC constant compared to what one might find in a typical tube amplifier. The reasons for this are described in our article on film capacitors.
In the original circuit we used a single Auriacap. This was before we discovered sound quality could be improved by paralleling an Auriacap with a Solen — a modification that might be well worth trying.
One more thing. We did not use a standby switch in the original design. We had a hard time finding an affordable switch that was actually rated for use at the high voltages employed. A standby switch may be added by the user, however, if and as desired. Just note that the reliability of such a switch can be dubious if the selection is not made carefully.
- To adjust the bias, add resistance to either of the points labeled with an X. Adding resistance in series with R10 will increase the tube’s current, while adding resistance in series with R11 will decrease the tube’s current. Add resistance as needed. Better yet, a potentiometer can be added by connecting each end of the resistor element to one of the two points labeled X, and then attaching the wiper of the potentiometer to the biasing line (the line with 2M of resistance heading to the grid of the second half of the 12AX7).
- The circuitry below the dashed line is for only one channel; duplicate this for the other channel.
- The output tube, by the way, should ideally run at about 90mA; adjust to achieve this current level. To read the current, either add a small resistor in the cathode of each tube and measure across this (maybe adding measuring points outside the chassis?), or (carefully!) measure the voltage drop across the output transformer’s primary and the primary’s resistance (don’t add external connecting points to here). In either case, use Ohm’s Law.
(V = I*R)
Please note that the voltages listed are minimum values; it doesn’t hurt to increase the voltage ratings.
The fuse should be a 1.6A Slo-Blo.
S1: Switch (> 2A, 250V).
ICL1: This is an inrush current limiter. We used 2A 150 cold ohms unit. This unit is important, as it also helps keep the power supply voltage reasonable during warm-up, as well as limiting (as you might expect) current surges upon power-up. Note — add more series resistance to the primary circuit if the plate voltage is too high. (It should never exceed 500V. During normal, warmed-up operation, our prototype’s voltage hovers around 430V–440V.)
C1 .22uf 250VAC film capacitor.
T1 One-Electron BFT-1B.
Rectifier: Any bridge rectifier rated at 600V 1A or better.
L1 and L2: These chokes are in series. We used a pair of Hammond 2.5H 300ma chokes. However, any choke or chokes will do as long as the total inductance is no less than 5H and the current rating is less than 300mA. Add series resistance as needed until the total DC resistance is no less than 85 ohms.
D1: 1000V 1A or better diode.
D2: Biasing diode. Aim for a diode voltage drop of around .75V @ 100mA. This doesn’t have to be precise, however.
D3: 36V Zener diode. Any wattage greater than or equal to .5W will do.
C2 and C3: 500uf 500V electrolytic capacitor. We used JJ brand electrolytics.
C4 and C5: 20uf 400V film capacitors. Higher capacitance values are acceptable. We used MKP audiophile capacitors for these and all the other film capacitors except for C1 and C10.
C6 and C7: 2.2uf 400V film caps. Larger capacitor values are fine.
C8: 10uf film capacitor, with a voltage of at least 6.3V. (That shouldn’t be hard to find.)
R1: 1K 25W chassis mount resistor.
R2 and R3: 2.2K 50W chassis mount resistor. These and the previous resistor’s wattage ratings are very conservative and should run nowhere near their max ratings.
R4, R7: 220K 0.5W resistor.
R5: 50K dual potentiometer. Use one half per channel, or use two separate pots for independent control of each channel. We used a Precision Instruments Mil-Spec pot. Expensive, but excellent quality. Should be well worth the cost.
C9: This is a silvered-mica bypass capacitor to help prevent attenuation of highs at lower volume levels. 82pf, but adjust value (or remove altogether) as desired for the sound you want.
R6: We used a 680 ohm resistor. This is not being used as a “grid-stopper” as it may appear; rather, it’s being used to (hopefully) protect any sensitive electronics on the input in case of tube failure. It helps limit the max current flow.
C10: This is the only coupling capacitor per channel used. We used an Auriacap .22uf 600V film capacitor. Any brand, however, may be used; just mind the voltage rating.
R8: 2M 0.5W. We used two 1M resistors in series, as 1M was the highest available value in the Mil-Spec series of resistors we were using.
R9: 150K 1W. We used 0.5W, and it worked just fine, but that is cutting the wattage rating a little close — it dissipates around .4W. With our heavy-duty Mil-Spec resistors, 0.5W was close enough. With other, cheaper resistors, however, we would recommend a higher wattage.
R10: 1M 0.5W.
R12: 68K. These biasing resistors get you in the ball park. Don’t forget to add series resistances as needed (see Note 2 above).
NE1: A Neon Lamp, type C2A. What? A Neon Lamp? Yes, I know, it’s a bit odd, but this concept of using a neon lamp to ensure no filament-to-cathode voltages are being exceeded works astonishingly well. A Zener has too much (like well over 200pf) capacitance for this job, but a neon lamp has almost no capacitance across it. And by the way, this idea is not original — our Tektronix vacuum-tube-based oscilloscope has all sorts of neon lamps in key places. These little lamps should also last a long time — maybe even a lifetime.
R13: 2.2K grid stopper. See our article on rolling the KT88, KT90, and KT120 for how we arrived at this value.
R14: 470 ohm 2W.
T2: One-Electron UBT-2 (4.8K 110mA primary and 16, 8, and 4 ohm secondary).
Well, there you have it. I hope someone builds this simple but astoundingly good-sounding amplifier. Enjoy!
By downloading this information, the user understands high voltages are involved. The user agrees to assume all risk. The information provided herein is believed to be reliable; however, Eden Audio Design assumes no responsibility for inaccuracies or omissions. Eden Audio Design assumes no responsibility for the use of this information, and all use of such information shall be entirely at the user’s own risk.