MJR7 Constructor's Page
Test Results and Photos. Last update 08-Jan-2013.
A few of those who bought boards for the MJR7-Mk3 have said they will do their own measurements. An Email address is included at the end of this page for anyone who would like their test results or photos included here. I have included my own construction of the Mk5 version near the end of this page.
The first results to arrive are from Peter Schröder, and he has carried out more detailed tests than my own. Some of his results are shown in the first four diagrams. The most reliable measurements are usually intermodulation tests which measure frequency components not included in the original test signal, so an ultra-low distortion signal is not needed. The first plot is CCIF 19kHz + 20kHz intermodulation, and the result is near to my own measurement of -109dB (0.0003%) at 300mV input. The other measurements are generally higher than my own, but still reasonably good. The 1kHz spectrum includes more high harmonics than I found, but all well under 0.001% so not a serious concern. I used a test signal about 6dB lower, and high harmonics can fall faster than low order as the signal is reduced, but even so this is something I will try to check further. The spectrum also reveals some supply breakthrough at 60Hz and its harmonics. My guess is that this is picked up from the magnetic field of the transformer. If it was direct supply line breakthrough I would expect this to be strongest at 120Hz, not 60Hz, for a full-wave rectified supply.
Photos.
The first two photos are again from Peter Schröder, who has used a wooden enclosure. In locations with no serious interference sources this is no problem, I have used wooden enclosures myself in the past. I particularly like the big heatsink. This is not essential, but keeping the amplifier cool will certainly improve long-term reliability, and it looks good too.
The next two photos are an amplifier built by by Peter Bergmann, who has used his own board design and a metal case. A heatsink is mounted inside the case. The mosfets can survive at high temperatures but other components, particularly the electrolytic capacitors, would have a shorter lifetime if allowed to get hot, so ventilation holes have been added to the base, back and top to maintain adequate air flow and keep the interior cool. An earlier version had problems with supply harmonics being picked up on one channel, but this was improved by a change to the wiring layout, with the positive and negative wires from the supply capacitors now bound together to minimise current loops. The excellent distortion measurements can be seen in a RightMark Audio Analyzer test report.
The next photo is from Paul Bentz, who has also used his own board design. The amplifier is part of a tri-amped speaker system. The MJR7s are used for the mid and high frequency drivers, while the bass is provided by a Rod Elliot design included in the same case. This looks very impressive, with good solid construction. Paul mentioned that accidentally shorting an output blew the fuse, but otherwise did no apparent damage. The lateral mosfets are hard to destroy, but it is still reassuring to know that the simple fuse protection is effective.
The next photo is not a final construction, just a first evaluation, and is interesting because it does have a problem. The channel on the left works fine, but the right channel initially oscillated around 10MHz. The Mk3 boards are not made as mirror images as in the more recent Mk4 version, so differences between channels are certainly possible. The single star earth point I have always recommended has been used, but still layout seems the most likely explanation for the problem. The constructor of this example tried a few experiments, and found one way to restore stability, which is to add a 100n capacitor from an earthed point on the board direct to the power supply earth, and this is visible in the photo. This suggests that the original longer route from board to supply earth was responsible for the instability.
One possible solution I would suggest in this sort of layout is to use two star earths instead of one. One earth point can be at the supply capacitor negative, and all the high current output stage earths can be taken to this point, so that the path lengths are kept short, and as far as possible the positive supply leads can follow close to the same route to avoid current loops. The existing star earth can then be used for all the other earths, and should then be connected via a single thick wire to the supply earth. This approach is still effective in avoiding earth loops, but keeps the high and nonlinear currents well away from the sensitive input wiring. The speaker output socket earth leads could follow the paths of the amplifier output wires from the sockets, and then connect to the supply earth to prevent the output current also having a large loop.
Even with this less than ideal layout the results of distortion tests were still good at 1kHz:
0.0003% at 2.83V into 7R
consisting of -109dB 2nd harmonic, -120dB 3rd, -138dB 4th, -133dB 5th.
0.00045% at 2.83V into 3R5
0.00015% at 4V with no load
0.0003% at 4V into 7R
0.0005% at 4V into 3R5
The next pictures were sent by Anton Yankov, and are his own design circuit boards. These are double sided, and use a number of surface-mount components, which can be seen on the underside view. This makes smaller board size possible, which can help reduce unwanted stray inductance and capacitance which could adversely affect stability. The large electrolytics needed for the MJR7 limit the maximum possible size reduction. I have made my own designs single sided with wide tracks and through-hole components to make them suitable for constructors with a range of abilities. I have rarely tried using surface mounting components myself, soldering such small parts needs a steady hand and good eyesight.
Next is my own most recent construction using the latest and probably final version, the MJR7-Mk5. This uses a single board for both channels and a star-earth point on the board, which simplifies the resulting case layout and wiring. There are just the two pairs of speaker wires and the positive and zero supply lines to connect to the board, plus the input cable, which I attached via a terminal block. The transformer was taken from an old Cambridge ATAC3 amplifier, but gave a supply voltage slightly above the 63V rating of the capacitors, so I added a pair of rectifier diodes in series with the bridge rectifier output to the capacitor positive terminal, and this reduced the voltage to a safer level. The supply smoothing capacitors are a parallel pair of Epcos B41231-A8129-M , each rated at 12,000uF +/-20%, 63V and 6.83A. A total as high as 24,000uF is not essential, but this type were available from Farnell (UK) for only £2-32 each (+VAT), product code 1839291. The case is recycled from an older design I built over 30 years ago.
The only earth connection to the case is via a metal mounting post at the star earth point on the board. I never use supply earths, but for commercial designs this is only permitted provided adequate double insulation is used, and anyone choosing to omit supply earthing needs to be aware of the possible danger and have the necessary experience and ability to make everything safe. My own construction should not be taken as an example of the correct way to achieve this. I strongly recommend connecting all the equipment to the mains supply via an 'earth leakage' cutout device for extra protection.
Next is an extremely well made version of the Mk5. This is from Paul Bentz, who also built an impressive version of the Mk3, shown earlier on this page, as part of his tri-amped speaker system. My own layouts are always simple single sided boards with wide tracks designed to be easily built by less experienced constructors, but without that constraint there are definite advantages to using double sided boards, including the reduction of current loop areas and neater layout, as seen here.
Next is another version of the latest Mk5 amplifier built by Forr, using almost the smallest case size possible. I usually advise keeping the transformer far away from the amplifier board, but with a good transformer with a low external field this may not be essential. Test results can be seen in an article:
L'amplificateur de Mike Renardson à transistors Mosfet
If like me your French is somewhat limited sections of the text can be copied and pasted into Google Translate.
The test results, I am happy to see, agree well with my own measurements.
Another application for the MJR7 is as an upgrade for an existing amplifier, and in the next example, built by Chuck Hicks, the starting point is a Dynaco SCA-80Q. This conveniently has a single 72V supply, and enough space in the case for the new amplifier board, as seen in the following photo. Some care is needed to avoid earth loop problems, the MJR7 is intended to have a single connection to the case via the central mounting post, but the Dynaco has its own earth link to the case near the supply capacitor. After solving an initial hum problem Chuck reports that there is now no audible noise or hum. The Dynaco uses a 250k volume control, which could be a problem for some power amplifiers, but the MJR7 is designed to maintain stability and low noise with a high source impedance, and although the control needs turning closer than usual to its maximum level the performance should be just as good as with a lower impedance control.
The next photo illustrates one of the greatest advantages of adapting an existing amplifier, which is that it would be difficult to achieve the same impressive appearance. Although commercial amplifier designers may fail to attain the level of performance of some good DIY designs at a similar price level, one thing they usually can be relied upon to do well is to make something that looks good. I adapted an old Cambridge ATAC3 for the pre-amp in my own system, but I think the Dynaco shown here looks far better.
