The 7695 Matrix Amp


When I was asked to give a talk at ETF2003 (, I thought about another Matrix amp. What are the characteristics that make this topology different? How can I emphasize that? Well, the amps play well, that is, they produce sound a lot louder than one would suspect from the power specifications. They are also inherently compact. Since they offer the simplicity of SE design with tolerance for PP transformers. How about a very small amp?

The 7695 (and its twin, the 7754) is an interesting tube. At low voltages it can put out relatively high power. Here's its plate characteristics:

Also, the impedances work out well so that I could even use those compact toroidal power transformers as input and output transformers. This was the tube to use for this design. This would let me build an amp that would put out a few watts with simple power supply in a small enclosure. What I ended up with is a small 200 cubic inch box weighing about 7 pounds containing an entire stereo amp. Cost is about $100 USD, since the "audio iron" is power supply toroids.


The entire power supply and schematic are contained in a single .PDF file. Here it is.


The overall topology of this amp is similar to the others. An input transformer is used to provide the "phase inversion" of the one channel. Then a small signal pentode driving the pentode output tubes. Like the 813 amp, since the "output transformers" has split winding secondaries, series of one winding from each transformer in phase produces "left" and series of one winding from each transformer out of phase produces "right".

The 7695s are 50v, 150mA heater tubes, so two in series require 100V. As it turns out, bias requirements (see plate curves above) of 130v at 100mA is about 12 volts. So by adding the heater current to the plate current allows us to use 2 6 volt small signal pentodes whose heater requirements can be anything from 0 to 150+100+100=350mA. The 6GU5 "shadow grid" pentode was chosen as a high gm part (providing good gain) and it was on sale at Antique Electronic Supply for 50 cents!

The power supply uses one additional Amveco/Talema toroid in an odd manner: one primary winding for "mains", one primary winding for HT (providing about 110 volts DC for the heaters and the B+. The true secondary then aided the B+ (but not the heater part) to increase the B+ to about 150 volts. This was RC filtered to produce the supply voltages.

Thus, 3 relatively compact toroids, one small choke (1.5Hy at 200mA for the parafeed choke) and two small toroids are all the iron needed for the amp. (The whole amp weighs in at about 7 pounds total).

The parafeed choke is ridiculously small. This was done to illustrate an important point of the Matrix amps. Normally, bass frequencies are relatively omni-directional. In fact, on vinyl LP records, frequencies below about 70Hz are mixed monaural to keep the excursions in the groove under control. Since that is true, then there is NO SIGNAL at low frequencies in the parafeed choke, so it doesn't matter what its value is. Thus listening to real program material, there is no loss of bass, but if you were to measure the response with only one channel driven, it would be deficient in low frequency response.

As a side note, the other thing this points out is that the sound quality is GREATLY controlled by only the one output transformer. In the amplifier as built, the sound is not bad at all, but not perfectly refined. However, by changing ONLY T1, T2 and T3 (not L1 nor T4) the sound can be improved a lot!


sensitivity 0.265V produces 1 watt into 8 ohms
Zout about 4 ohms
freq. resp. -0.7dB at 20Hz, -0.6dB at 20kHz, -3dB at 40kHz
power output 4 watts per channel at 5% distortion
hum/noise 750uV

The distortion components vs power for this amp is shown in the following graph:

Note that 5% distortion occurs at about 4 watts. Yet this amplifier filled the hall for ETF 2003.