-> The schematic and transformer set are no longer available for DIY construction, anyone interested can now purchase SB Phoenix. <-
One day I receive this message:
I recently bought an amplifier that uses KT88 tubes in single ended class A, I did not build it myself but purchased it already assembled by a third party. Long story short, at first in my system the sound seemed nice, a bit particular, but not bad. Then a friend lent me an ARC sp3 preamp and I liked it even more. Then the fatal mistake: I took it to my friend’s house and in his system, connected to Klipsch La Scala, it literally sounds awful, like an intercom. A trauma! I would like to ask: would it be possible to make it a good sounding amp? For the record, the unit uses JJ tubes. My system: Klipsch kg4 speakers, Carver c6 preamp, Aiwa ap2500 turntable with Dynavector 10×5 neo; Dual turntable with Shure v15III + Jico sas. My friend’s system: Klipsch La Scala, Diego Nardi 2a3 power amp, Dual turntable with Shure m91 + Jico sas. Obviously I can send you photos and further details. Thank you in advance and best regards.
It was clear that this was one of the many “impresentabili”, a KT88 single ended in ultralinear mode driven by an ECC88, a PCB of dubious origin mounted in a wooden box patched together with 2 Ikea handles and transformers also of unknown origin with no labels. But how did this device actually work? Because besides the appearance maybe… On the oscilloscope, on a resistive load, the simplest test of all at 1 watt output, it produced a sine wave like this:
In yellow the generator signal, in light blue the amplifier signal which shows a strongly compressed half-wave, and below you can see the related spectrum analysis.
A distortion of 2.4% at 1 watt on a resistive load. The main reason for such high distortion is the use of the ultralinear connection in a single ended configuration, something I already discussed in that article. Also because of the ultralinear mode, the maximum RMS power output was only 6 watts per channel. Measuring the bandwidth I wanted to see how the transformers performed — could they be reused?
Better to forget it… How did the circuit behave on a reactive load? Here the biggest problems came out… Let’s look at a simple 1kHz sine wave at 1 watt on a reactive load…
It looks clipped?! But it is a strange clipping… it is not clipping! It is a combination of several circuit design problems. This bitten-off waveform remained even when lowering the amplifier volume to just a few milliwatts. It was completely incompatible with the reactive load, which more or less represents what the circuit sees when connected to a real speaker… The spectrum analysis is absolutely disastrous, with 10% total distortion at just 1 watt RMS. I understand the owner who described it as an intercom, but if I measured a real intercom I am not sure which one would win.
The damping factor is not any better with a DF of 1.6… the frequency response on a reactive load shows this clearly…
The rest of the measurements I took are superfluous, pointless. In the end very little could be salvaged from that unit: a pair of Mundorf parts, the power transformer, and a few other bits here and there. The schematic and transformer set are no longer available for DIY construction, anyone interested can now purchase SB Phoenix.
The input stage is made with an unusual (but not too unusual) ECC85 VHF double triode, generally used in FM receivers in tube radios and also found in some Luxman amplifiers such as the 3600. The tube is loaded with a transistor-based constant current source using the classic MJE350. This choice was made to keep everything running with a single driver tube, without adding more cascaded stages, while still obtaining a decent output impedance and a fast, detailed, bright-sounding stage. A ECC83 loaded with a resistor would have been much warmer sounding, and the customer asked for a fresher sound, so the transistor CCS fit perfectly. Instead of the ECC85, the more common ECC81 can also be used, as it has almost the same electrical characteristics, you just need to respect the different filament pin connections. As output tubes you can install KT88 / 6550 / KT90.
The output stage is a pure pentode configuration, fixed-bias with servo control, meaning there is an active circuit that automatically regulates the bias, which is not the same as self-bias where a resistor and capacitor are used, with all the reactance of the capacitor between the cathodes. Of course there is NFB as you must include it if you want appropriate damping of the speaker.
The power supply section uses the classic CLC cell with a real choke and not a MOSFET, generous capacitance and good capacitors. The screen grid voltage is obtained with a circuit called a voltage divider, which allows a lower voltage than the input according to the ratio set by the resistive divider, but it outputs at low impedance… In practice, when feeding G2 with a voltage obtained from a classic RC cell, the voltage is not stable because when the audio signal exceeds certain limits and G2 starts drawing current, the voltage across capacitor C gradually drops while the signal persists and recharges when it stops, causing memory distortion consisting of shifting the tube’s operating point according to the “past” signal. The voltage divider, even though it is not stabilized (because it follows the input), has a very low impedance and is not significantly disturbed by G2 current draw…
The original power transformer could still be used, while I supplied two new SE6K-KT88 output transformers and the choke. Here is the customer’s assembly:
What measurements did this circuit achieve?
Power: 9.8 Watt RMS
Damping factor DF: 5.6
Bandwidth 10Hz -0.2dB / 40kHz -1dB
THD @ 1 watt on resistive load: 0.37%
How is the 1 watt sine wave on the reactive load???
And the distortion on the reactive load? Still at 1 watt: 0.067%. Yes, because an amplifier is used to drive a speaker, not to be listened to on a resistor… Distorting little on a resistive load is easy, on a reactive load it is much harder. Knowing how to design a circuit also means making it perform better on a real loudspeaker and not only on a resistor just to show off a single measurement.
Square waves at 100Hz / 1k / 10k…
The triangular wave showing perfect waveform symmetry
And the frequency response on the reactive load
How does it sound? The customer who built the premium version with SB-LAB transformers tells us: The power is more than enough, at 10 o’clock the room is already full of sound. The harmonic content of string instruments is remarkable. The question comes naturally, but what the %$£§ have I been listening to until now? Pardon my French… I have never heard such sweetness, detail, and refinement. I am practically listening to my Tannoys for the first time! It is incredible, there is not the slightest trace of harshness, the sound is a caress, even at high volume. The soundstage is breathtaking, it feels like having an orchestra in the room, it never tires, bass and mids are wonderful! But again, what shocked me the most is that you do not feel sound pressure, I turn the volume up and the sound becomes bigger, not louder. Never happened before! My compliments. It was a long journey but it was worth it.
