The subject is an Ultra Simple Symmetrical Amplifier to drive 4-8 ohms speaker. The maximum power is limited by the single output transistor pair and the class A heat dissipation. Therefore its use is limited to small power requirements (<=30W rms). The specific implemented circuit presented here is operated in class A until about 20 to 25 Wrms and then goes into AB until 27W rms before clipping when operated at +/-24Vdc with bias current of about 1.2A. The clipping is soft with rounded corners and is mainly due to the relatively low feedback factor (about 34db). The square law effect of Mosfet allows for more class A operation with the same bias current than with a bipolar output amplifier. The design topology is similar to the Pass F5 amplifier (that is similar to the Solectron amplifier which uses Lateral mosfet for output) but uses different transistor types and other parts. The input stage uses jfet transistors or bjt and the output stage linear Lateral mosfet transistors designed for Audio. These transistors have more linear performance than vertical mosfet (as used on the F5 amp). Because of their low VGSoff characteristics and lower transconductance the use of Lateral mosfet in this topology requires an additional buffer stage which in reality is a voltage shifter stage to allow more gain from the input stage.
The thermal compensation of the amplifier is done using thermistors in conjunction with resistors for positive and negative sides respectively. For fail safe consideration, a thermal switch (temperature set point of 60 deg C) is also added on each channel heatsink. This thermal switch interrupts the 120Vac line of the amplifier.
The power amp THD harmonics are dominant 3rd and 2nd with higher ones very low in comparison.
Other Class A amplifiers with low feedback factor, for example, are JLH 10W (mainly Single Ended output and partially push-pull at highest power) or Hiraga with Symetric topology as the USSA. I have already built an Hiraga Super Classs A 20W and another variant with mosfet output. Since I do not have these amplifiers anymore I cannot truly compare with the USSA but my memory tells me that, even if they were good sounding, I was not as pleased with the sound as I am now with the USSA.
This project is for medium skilled builders for the USSA versions 1 to 4 since it uses an F5 amplifier pcb thus it requires modifications to be performed to add extra components and wired connections. However, USSA-5 has a dedicated pcb thus more easy to mount. Note that the pcb of USSA-5 can also be used to mount USSA-3 and USSA-4 with some parts removal and substitution.
1.1 Main Differences between USSA versions
1.1.1 USSA version 1:
This is the 1st version. The sound is transparent and sound stage is deep. Dynamics are good but not extreme. Bass is expressive where you can hear the detailed sound of bass strings. The sound is very natural as opposed to be exaggerated and cause no listening fatigue.
1.1.2 USSA version 2:
This version has the damping factor increased to about 20 compared to 15 for the USSA ver. 1.
This is first achieved by reducing the degeneration on the output mosfets. The mosfet source resistances (R15, R16) value change from 0.25 ohms to 0.1 ohms. Because of that, one resistor value used for temperature compensation is also changed accordingly. Indeed, R9 and R10 values changes from 2k2 to 1k5 ohms. And secondly, the CCS current is reduced to 4.6ma.
The bias current will need to be re-adjusted with P1 and P2 (therefore DC offset too at the same time) otherwise it will be too high compared to original bias current value.
You will also notice that the initial bias current at power up of amp will be less high compared to USSA version 1 and may take slightly longer time to stabilise to the target bias value set point. There is slight improvement on the bass response depending of speakers where it can be more “tight”.
1.1.3 USSA version 3:
This version has jfet inputs, mosfet as voltage shifter/buffer stage and Lateral mosfet double die as output transistors. It has no CCS since it uses jfet property to set current in input stage.
The double die mosfet and buffer stage allow for enhanced output current thus more power and control in difficult load.
This version is more detailed for sound. The bass is a bit more tight and controlled and bass guitar is very natural. Vocals and cello are sweet sounding. Damping factor is more than doubled compared to version 1. Refer to builders evaluations below. Also It includes a second harmonic of THD adjustment with a sigle pot on the pcb. Refer to section below.
1.1.4 USSA version 4:
This version is an hybrid version between 1 – 2 and version 3. This version 4 has the input stage with CCS similar to version 1 and 2 and output stage is similar to version 3. The main advantage over version 3 is the use of bjt compared to jfet in input stage since jfet are difficult to match between N and P channels if you do not have many of them.
The double die mosfet and buffer stage allow for enhanced output current thus more power and control in difficult load. This version has a similar sound to version 3 but with a little better bass control.
1.1.5 USSA version 5:
This version is the same general topology as version 4 but has a CFP input stage and an improved CCS using 2 jfet instead of one. This is the latest crafted version of USSA and sound evaluation is reported in builders evaluation section below.
1.1.6 USSA version B:
This version was designed just after version 1 and is similar to version 2 but uses Sanken darlington transistors as output. These transistors are excellent for audio and are not expensive.
1.1.7 F5-USSA:
This version is the same topology as the F5 thus uses the same number of parts and no modification to the F5 pcb required. However, input jfet transistors IDSS value have to be selected more precisely and output transistors are mosfet designed for audio (2SK1530Y/2SJ201Y) instead of IRFP240/IRFP9240 designed for switching. The gain is also 10 (as opposed to 5 for the F5) to allow to be the same as the other USSA versions and not require a preamp to reach the full available power. The damping factor is reduced from about 80 (F5) to about 30. The input current is also slightly reduced. The overall sound result is an improvement (more expressive bass and natural) over the standard F5 amplifier.
Links for USSA5 comments :
http://www.diyaudio.com/forums/solid-state/323778-ussa-5-build-review-post5459762.html
http://www.diyaudio.com/forums/solid-state/96192-post-solid-pics-post5189240.html
http://www.diyaudio.com/forums/solid-state/96192-post-solid-pics-post5031437.html
http://www.quebecdiy.net/t988p50-deepsound-ussa5#17392
USSA3 comments:
http://www.diyaudio.com/forums/solid-state/96192-post-solid-pics-post5345353.html
http://www.quebecdiy.net/t756p125-amplificateur-ussa-3-version-lateral-et-all-fets-conception#17673
http://www.quebecdiy.net/t756p125-amplificateur-ussa-3-version-lateral-et-all-fets-conception#17916
The USSA-3 includes a 2nd harmonic cancellation potentiometer. It allows between -25db to +14db of 2nd harmonic of distorsion compared to 3rd hatmonic
• Gain: 20db (or x10)
• Input sensitivity: 1.8 V-peak for output (20W rms)
• Damping Factor (100 Hz , 8 ohms): [35– 45]
• Output impedance: [0.23 - 0.27] ohm
• Input Impedance: 22K ohms
• Max power into 8 ohms before clipping: <25Wrms (at Supply voltage of +/-23Vdc, 0.9A Bias current)
• Power in Class A (8 ohms): depending of bias current
• DC offset: trimmable to < 3mv
• Rise time (10 to 90%) < 1.3us
• Typical T.H.D. (1 KHz) (mostly 2nd and 3rd):
1w : 0,005% (2nd harmonic about 8db over 3rd)
4w : 0,011%
6w : 0,02%
10W: 0,03%
15w: 0,05%
20W: 0,08%
25W: 0,4% (start of clipping)
Of course this is not impressive numbers since a simple design but have in mind that I do not care so much about THD except that higher harmonics need to be much lower than 2nd and 3rd.
The higher damping factor can be obtained with higher bias , bigger PSU and more accurate load impedance measurement. DF about 40 to 45 is within my target range.
USSA-3 has similar THD results and same potential for power. However, damping factor is eEd to be lower due to topology.
In 4 ohms you get half power in class A for same bias but potentially close to 1,7 time the power in class AB (really depends on your PSU capability) compared to 8 ohms since it is a push-pull configuration.
Fab
Beppe USSA-5 build
The USSA-6 is the first class AB using the USSA-5 front end but follower output. A prototype using a very modified F5 pcb has been built but no dedicated pcb is planned at this time. Refer to FSSA amplifier section for class AB amplification.