HOW TO BUILD THE MJR7.
Part 1. Adding the components.
Part 2. Parts List.
Part 3. Setup and testing.
Part 4. Photos of stages of construction.
This amplifier is not a commercial design, it is just for the benefit of DIY enthusiasts, and is recommended only to those with some previous experience who have already successfully built other electronic projects.
I have added some of the specifications at the end of this page. These need to be updated sometime, most are from earlier versions, but all should still be reasonably accurate.
The information on the original MJR6 and MJR7 pages, plus additions for the MJR7-Mk2 are mostly relevant to this final version, and only changes and additions are covered here. The MJR6 page includes distortion extracted using the nulling method with a speaker load and a music signal to demonstrate that these designs have no audible distortion in normal use. The MJR7 has even lower distortion, for example the 19kHz + 20kHz intermodulation product at 1kHz is about 19dB lower.
Now there is an additional 220uF capacitor plus two diodes and a resistor which slow down the switch-on to keep the output pulse low enough for use with most high sensitivity speakers. For use in a bi-amp or tri-amp system with active crossover filters the pulse also needs to be small so that direct driven high frequency drive units are not damaged. The circuit addition is a simplified version of an idea suggested to me a few years ago. With the modification added the switch-on thump becomes almost inaudible. With the original MJR6 the output pulse was 4V peak, this has now been reduced to 1V. There is no audible thump at switch-off.
I would prefer to leave out the 12V gate protection zeners, but I doubt whether the mosfet internal zeners have adequate current rating. With a negative output into a low impedance or shorted output the zener current would only be limited by the 300R gate resistor, which is why the zeners are placed after the resistors, and relatively high 300R values are used.
There are possible alternative, but so far untried, input stage transistors, the Toshiba 2SA970BL and 2SC2240BL, which appear to be good substitutes for the Hitachi 2SA1085E and 2SC2547E. The 'BL' suffix is the high gain group. Alternatives for the Sanyo 2SC2911 and 2SA1209 are the Toshiba 2SC3423 and 2SA1360, which again are untried. Based entirely on my own experience Toshiba devices are more likely to be faked, maybe this is because they are more popular so more attractive to substitute remarked cheaper devices, or maybe I have just been unlucky with these.
Some of the component values have been changed (29 Jan 2009). This is to ensure easier availability, for example a 7R5 resistor has become 6R8 because I wanted to use a 3watt component, which is not so easy to find in the old value.
The 470uF and 220uF could be rated at 100V to enable use of a higher supply voltage. Increasing the 4400uF (2 x 2200uF) to 100V would then be a good idea although it normally operates at half the supply, to ensure safety under fault conditions or when adjusting the output voltage. Increasing the supply close to 100V the average sinewave power into 8ohms increases from 30W to over 80W.
Before deciding on the higher power option I suggest checking carefully what output level is really needed before worrying too much about power rating. There are many people happily using 5 or 10 watt class-A amplifiers who feel no need for more power. Remember the logarithmic loudness effect, doubling the power is nowhere near a doubling of percieved loudness, and the difference between 30W and 80W is not as dramatic as we might expect. I wrote a piece about the problems of high power, which include thermal compression, while high sound levels can cause both temporary and permanent hearing loss. The temporary effect can be a raising of the threshold of hearing by as much as 40dB, so using high levels in order to achieve a 'high dynamic range' could have the opposite effect. The MJR7 is designed for quality rather than quantity, and if much higher power is really needed there are other more suitable design approaches.
CLOSED-LOOP GAIN AND PHASE RESPONSE.
Next is the final complete amplifier gain and phase response from 1Hz to 10MHz with a 600R source impedance and 4R load. The gain falls by 0.5dB at 20Hz and 20kHz. The -3dB low frequency point is about the same with 8R or 2R load, which shows that the output coupling capacitor has little effect, a further benefit of including it inside the overall feedback loop. The low frequency response is determined almost entirely by the input capacitor, and can be extended if required by increasing the 2.2uF value. Further increase could do more harm than good, the -3dB point is already at 6Hz, and such low frequencies will be inaudible, but could still cause both amplitude and phase modulation in the speaker. Another danger is that the signal amplitude prior to the output capacitor could be excessive with very low frequencies, and this capacitor would also need to be increased. With an 8R load the 4400uF specified should still be sufficient even with a 4u7 input capacitor.
(The apparent phase jump is just an AIM-Spice effect, anything more negative than -180 being shown as positive.)
Even ignoring the effects of AIM-Spice this phase plot doesn't look at all nice, but see what happens if we use a linear scale for both frequency and phase and only look at 1kHz to 20kHz:
This shows that the phase shift is almost perfectly proportional to frequency. This will have the same effect as a constant time delay, (about 3.7usec.) which can have no audible effect, all wave shapes remaining unchanged. The total phase nonlinearity of the low-pass filtering between 1kHz and 20kHz appears to be around plus or minus half a degree. The plot still looks reasonably linear if extended to 30kHz, becoming noticeably curved only at higher frequencies. The high-pass resulting from the 2.2uF input capacitor is not equivalent to a constant time shift at low frequencies, but the resulting phase error is only 10 deg at 35Hz, which will almost certainly be insignificant compared to typical phase effects of speaker and room responses. The phase error at 35Hz can be reduced to 5deg by using a 4u7 input capacitor.
I specified an output inductor 0.4uH for my amplifiers, estimated by measuring the voltage drop when in series with a resistor. A more recent check compared the voltage drop across the inductor with that across a 0.5ohm resistor at the same current at 100kHz, and this suggests that the value is about 0.55uH.
When I checked using the formula
L (uH) = D2N2 /(L + 0.45D) for N turns, dia. D, length L (metres).
this gave a value 0.8uH. This is no problem, if correct this will give better stability margins than my simulations with 0.4uH suggest, and anyone who has used my description of 13 turns 1.2mm dia. enamelled copper wire with inside diameter 1cm will have the same value I used. (D in the formula is the diameter at the centre of the wire, not the inside diameter.) The measured impedance gives a damping factor over 100 at 20kHz for an 8ohm load.
The inductor should be air-cored, with no other components close to its ends. The occasional practice of winding the inductor round the damping resistor may be a bad idea, the magnetic properties of resistors are generally not specified.
The inductor value I used was intended to give stability with capacitive loads up to about 4uF. For a general purpose amplifier any combination of speaker and cable could be used, so we have no certainty what is the worst case load we need to design for, and my own choice of 4uF pure capacitance is fairly arbitrary and maybe a little over-cautious. In practice 4uF was not actually the 'worst case' because with the original circuit there was instability with small capacitances around 2nF, which could be the capacitance of unterminated cables and was therefore a real problem. This has been solved in the more recent updates.
The following section is included to explain how the output network was determined. This is an area of design which deserves more detailed attention rather than the common approach of just adding a 0.1uF in series with 8R across the amplifier and a series inductor of a few uH in parallel with 8R. The idea often mentioned is that these are designed to ensure the amplifier sees a resistance similar to the nominal 8R of the speaker over the whole frequency range, but this is virtually impossible to achieve unless the speaker has a known and very simple impedance characteristic. What really matters is stability with any reasonable load, and this is considered next.
Further thought and a few simulations were needed to find a better choice of output network, and some of the more interesting results are shown next. They are all, in effect, small-signal simulations which ignore nonlinearity, the only aim being to determine stability margins.
The conditions for stability can be different for a nonlinear system, but the only serious nonlinearity in this application is at clipping, and this is checked here by looking at what happens when the cascode stage starts to clip and therefore its output impedance falls, but again this is a linear 'small-signal' approximation so not entirely correct. Analysis of nonlinear systems is notoriously difficult, and simulations not entirely dependable, so testing of a real amplifier when clipping into various loads is still a good idea.
The high feedback used is not itself a problem, instability normally only occurs at the loop unity gain frequency, not at frequencies where the loop gain is high. One problem is that the unity gain frequency is not fixed, partly because the series resonance with the output inductor and a capacitor load can reduce gain to unity over a range of frequencies for different capacitances, but also it can change over a wide range near clipping. Some designs can have short bursts of high frequency oscillation when coming out of clipping, and even if this has no audible effect there could be interference with other equipment, so it needs to be avoided.
The problem then is to prevent the excess phase lag round the feedback loop approaching too close to 180 deg at the same frequency where the gain round the loop is unity, and achieving this for any reasonable load and any level of clipping. There are inevitably different opinions about what loads should be regarded as 'reasonable'.
The next diagrams are phase shift and gain round the feedback loop with a 4uF load, and without the 100n + 1R across the load. The first has a 7R5 damping resistor across the inductor.
( The apparent discontinuity in the top phase plot happens because AIM-Spice only plots phase from -180 to +180 degrees, and anything more negative than minus 180 becomes a positive phase, so the sudden jump is of no real significance. A 1V input to the feedback loop was used, so the voltage in the lower plot is equal to the loop gain.)
Although the phase exceeds 180 deg. from 40kHz to above 100kHz the gain only falls to unity slightly above this range, so it may just avoid instability, but the safety margin is far too small. To improve the stability margin to a more acceptable level the damping resistor can be reduced. In the next diagram it is 2R2.
This is better, the gain falls only to 2, at a frequency around 120kHz, with about 90 deg phase shift, and is about 10 when the phase shift reaches 180 deg.
The phase lag still exceeds 180 deg over a range of frequencies. Where this may become a problem is near clipping where the gain can reduce to unity in the frequency range where phase exceeded 180 deg. Fortunately with the method of high frequency compensation used the phase shift also reduces near clipping, and to simulate this effect we can reduce the impedance at the output of the cascode stage, which will be the initial effect as we approach clipping (which is why we want the driver stage to clip first before the output stage). Reducing this impedance from its normal value of 200k down to 3k, again with the 4uF load, gives the following result:
This impedance reduction is sufficient to reduce loop gain to unity around 120kHz, but the maximum phase shift falls to 120 deg around 90kHz. At the 120kHz unity gain frequency we have a 90 deg phase margin, so clipping is no problem in this case.
The problem of high capacitance loads may be solved, but reducing the damping resistor can make things worse at higher frequencies where the instability found in the original MJR-7 with loads around 2nF may still remain. This was caused by the 2nF having a series resonance with the output inductor close to the unity gain frequency. Adding 100nF across the load prevents the total capacitance being as low as 2nF and so avoids this problem but replaces it with another. Adding an inductive load there will then be a resonance with the 100n, and the result could still be capacitive at the originally problematic frequency. The next diagram is with the 100n added together with a load of 0.05uH
This shows large gain and phase effects above 2MHz, and in combination with other phase shifts, e.g. from the input stage, which has been ignored here, there could still be a stability problem. It may be highly unlikely that any speaker plus cable would have a pure inductance of 0.05uH at 2.5MHz, but it is easy to prevent the problem. The solution is to add a small resistor in series with the 100n to damp any resonances. 1R was found to give adequate damping, 0R5 or smaller values still gave a noticeable notch in the phase response. The result with 1R added is shown next.
The final version, as shown in the circuit diagram, has phase shift round the feedback loop which only reaches 180 deg for capacitive loads 2uF or more. The next diagram shows the result with a 2uF load added.
Other simulations were tried, and here is a summary of the more useful results:
With inductor 0.8uH to match the calculated value all stability margins were improved, so if this is the correct value this is no problem.
With source impedance increased from the 600R used for the above results there was some reduction in stability margin, but not enough to be a problem.
With mosfet capacitances doubled to simulate the dual die types or parallel pairs there was no serious problem, but reducing total quiescent current below 100mA should then be avoided.
The small negative resistance output impedance at very low frequencies was investigated and found to be a maximum of 0.09 ohms around 6Hz. Increasing the 10uF in the feedback network increases this value, so is not a good idea. Reducing the 10uF increases the effect of the output capacitor a little.
There is one 'problem' which will occur to some extent with any amplifier using an output inductor and overall negative feedback. In the next diagram the upper trace shows the gain at the output before the inductor with a 2uF load. The small peak at 180kHz is no problem, but the lower trace is the gain at the input of the output stage. (This is a linear approximation with unlimited signal levels, so it is not what would happen in reality where clipping and other sources of nonlinearity will limit driver stage output.) The 75V peak here is rather more worrying, and with a 60V supply there will be severe clipping with a 1V input. In reality we would hope never to encounter a 1V input at 180kHz with a 2uF load, so there is no need to worry too much. The cause of the problem is the series resonance of the inductor and the capacitor load which load the output stage with about 0.1 ohms, and then the driver tries to apply 75V to the output stage in an attempt to produce 3V across this 0R1. The output current would then be 30A, so there would still be a problem even if there was no driver clipping. There are ways to reduce the peak, for example reducing the 2R2 damping resistor, but this may cause other problems if reduced too far.
Note that the MJR-7 has a second-order low-pass response because of the 470p input filter and the 10p across the feedback resistor (the 330p at the input does not affect the closed-loop response, but it is an essential part of the loop stabilisation). This, giving -0.5dB at 20kHz, helps by keeping the output down to 3V at 180kHz. Some published designs have less input filtering, often starting at a much higher frequency, and in some cases using small output inductors with 8R damping resistors. The above problem, if it ever really is a problem, could in this case be far more serious.
The original MJR-7 layout worked well, but had fairly big current loops in the output stage. One way to reduce such current loops is to use a double sided board, but with the mosfets mounted on an aluminium heatsink bracket bolted to the board this would be more difficult. I wanted to simplify construction as far as possible, so the solution chosen is to use wire links for the high nonlinear current paths. This may look a little untidy, but neat layout is never one of my highest priorities.
Distortion measurement.Harmonic distortion, measured at 300mV rms input.
The 20Hz figures are not very accurate, maybe plus or minus 10dB.
Intermodulation from input 20kHz plus 21kHz, each at 150mV rms, the 1kHz distortion component was measured at -109dB relative to the output from 300mV.
Distortion: Input 1kHz at 300mV, distortion just appears above the noise level. Second harmonic measured at -116dB (0.00016%). Third harmonic at -129dB (0.000035%).
Distortion: Input 7.5kHz at 300mV. Although this looks a little spiky the only audio frequency component is the 15kHz second harmonic, measured at -106dB (0.0005%).
INPUT IMPEDANCE: 10.4k
VOLTAGE GAIN: 25dB (x17.8)
FEEDBACK LOOP GAIN: 80dB at 1kHz, 65dB at 20kHz.
SIGNAL TO NOISE RATIO, 20kHz bandwidth 1V rms input, source impedance 600R, calculated ratio -114dB. (With a higher source impedance the amplifier gain falls and output noise reduces.)
OUTPUT POWER WITH Vs=60V: 30W average sine-wave power into 8ohms.
DAMPING FACTOR: This is included to demonstrate that the output capacitor effect is not only eliminated by the feedback loop, but the output resistance even becomes slightly negative. The damping factor is an almost entirely unimportant specification. The resistance of the speaker voice coil in series with the amplifier output impedance limits the level of speaker damping possible, so provided the damping factor is more than about 20 further increase makes practically no difference. The values measured for an 8ohm load are:
380 at 1kHz
820 at 30Hz
1500 at 20Hz
Negative resistance -0.01ohm at 15Hz, -0.06ohm at 10Hz.
MAXIMUM OUTPUT SLEW RATE: 30 V/usec. This is another specification which can be misleading. For example the figure invariably quoted is the 'large signal maximum output slew rate'. If 1V is needed by the input stage to achieve this slew rate, which is possible when using a jfet or degenerated bipolar input stage, and the audio signal has a peak level of 1V, then that slew rate could only be achieved if the output was zero, so that feedback will not reduce the input stage voltage, so in practice the specified slew rate is not achievable at this input signal level. I have seen jfet input op-amps specified with a 10V input step, which may give a more impressive figure, but is not entirely relevant to audio applications. The MJR7 requires only 10mV at the input transistor base to achieve its maximum 30 V/usec output. An amplifier with an apparently more impressive 100 V/usec which requires 1V input to achieve this could only reach 1 V/usec at 10mV input, so instead of being 3 times better it is 30 times worse. This is more than adequate, but it demonstrates that there is more to slew rate than just the large signal specification.
SQUARE WAVE RESPONSE: 'Ringing' with a capacitive load need not be related to stability, it is more often just a resonance with the output inductor. Normally recorded audio transients will not include components at 100kHz, so'ringing at 100kHz' observed with a square-wave is not going to be produced by an audio signal, so it is of little real importance. What appears to be ringing may in some cases just be correct reproduction of a filtered square-wave with only the removal of unwanted high frequency components, and I gave an example of this in an article: Square Wave Testing. Here I include just one square-wave result, which is the output prior to the output inductor with a 2u2 load and a 10kHz signal. There is some slight ripple as expected, but no stability problems are revealed.
CLIPPING: A test I find more useful for stability testing is sinewave clipping with a capacitive load, and here there can be bursts of oscillation or latch-up effects when coming out of clipping to indicate problems. The following is the top of a 10kHz sinewave with a 2u2 load. (The photo is a bit faint because of difficulty trying to photograph a toneburst signal). There is a small glitch when coming out of clipping, possibly caused by the current source biasing capacitor discharging a little during clipping, but not enough to worry about. For a high feedback design this is a good result. Only positive clipping is shown, negative clipping is fairly similar.
Using an output capacitor for speaker protection gives more rapid effect than methods using relays and other similar techniques. If fault conditions cause the output to short to the 60V positive supply rail then for a 4400uF capacitor feeding a 8R load the output pulse will start at 30V, but fall with a 35msec time-constant. Using a circuit to detect such an output and activate a relay to protect the speaker, for example full output at 20Hz has a 50msec half-cycle, so this is probably the shortest time needed to be certain it is a fault rather than high level bass. Add the time taken for the relay to operate and the delay is likely to be far longer than the capacitor protection takes.
High frequency compensation is usually applied using a 'Miller compensation capacitor'. This has an unfortunate effect, that it is not minimum phase because there is feedforward through the capacitor, which can add unwanted phase shift particularly near clipping where stability can often be a problem in high feedback designs. There are simple methods to avoid this, e.g. adding a resistor in series with the capacitor, but I have rarely seen this done. The MJR7 uses a minimum phase compensation method with a capacitor to earth at the driver stage cascode output. The series resistor is in this case to reduce phase shift at higher frequencies rather than to prevent an even greater addition.
The 100n plus 1R at the output are included to improve stability with small capacitive loads which could resonate with the output inductor close to the feedback loop unity gain frequency. An additional benefit of these components is that high frequency interference picked up by the speaker cables will be highly attenuated. The output inductor is of limited help with this because of the low value parallel damping resistor. Any interference still getting past the output network is fed back to the input stage via the 10p feedback capacitor, but is then attenuated further by the 330p from input base to earth, and so is reduced by another factor of 33. Using a high gain input stage could increase sensitivity to interference, but this is avoided by the effective filtering at both output and input.