amplifier measurement

After the repair the amplifier was measured. The substantial parameters such as distortion factor, rate of rise, Slew rate, range and damping factor are to be examined.

Automatic website translation by


Kandel im November Kandel mit erstem Schnee im Jahr

At the beginning of November fell the first snow in the Black Forest, is it time become the amplifier to be measured


Distortion factor measurement of 1 kHz


from special interest the distortion factor of the output stage is natural. Use finds a 3580A audio spectrum Analyzer, which can represent railways to maximum volume range of 90, i.e. he can measure a harmonious distortion factor of approximately 0.003%.

Klirrfaktor Rechts 1 kHz 1,6Vpp 4 Ohm kalt Spektrum des linken noch kalten Kanals. Regulierungspunkt der Endstufe ist direkt am Ausgang
Amplifier had a temperature of 10 degrees Celsius when switching on on. A distortion factor with 1.6 V peak peak based on 4 ohms of the right channel. Apparent the quiescent current necessary for low distortion factor did not in-regulate itself yet. Achievement 0.08 W rms.

the left channel with same conditions 1 kHz, 4 ohms and 1.6 V of point point at the exit, already better, has also a few minutes of heating up behind itself. Horizontal dissolution 1 kHz/DIV. The highest line correspond 1.6 Vpp each further 10 railways less. Achievement 0.08 W rms.

Measurement directly of the exit of the output stage, near the point of adjustment on that the regulation works. This load resistance was used. Voltages with high levels by means of 10:1 probe, the smaller directly 1:1.

Spektrum bei 10 Volt peak peak Spektrumanalyzer berührungsloses Thermometer

Amplitude of right increased to 10 V pp, 4 ohms, 1 kHz, output stage warms up thereby to 40°C (ambient temperature 15°C). Achievement 3.1 Watts rms, current amplitude 1.25 A

Amplitude left increased to 10 V pp, 4 ohms, 1kHz, output stage concerned meanwhile at 36°C.  Achievement 3.1 Watts rms, current amplitude 1.25 A

to measure the radiator box temperature with a contactless thermometer leaves itself. It has an inserted red laser diode as a goal. At dark surfaces it works well

Spektrum am rechten Vorverstärker Ausgang Spektrum am Vorverstärker Spektrum KA907 bei einem Volt Spitze Spitze

Spectrum based on the right Pre Out exit, the 10 Vpp measurement. The distortion factor changed temporally still. K3 is with -79  dBV, that is about 0,011% distortions for the third harmonious one here.

Spectrum directly based on the left Pre Out exit of the vorverstaerkerstufe. Heard to 10 Vpp the measurement. The preamplifier produces here no harmonious one within for dynamic range of the Analyzers of approx.. 90 railways.

Spectrum shows right exit with only 1 V peak peak. Harmonious distortion factor is approx..  0,005%. The second harmonious one is to be still recognized. Achievement 0.031 Watts rms, current amplitude 0.125 A

Spektrum bei 1Vpp linker Kanal Spektrum zeigt breitbandige Störungen Spektrum des rechten Kanal der Endstufe bei kleiner Ausgangsspannung

Spectrum left channel with 1 Vpp, outside of the dissolution of the Spektrumanalyzers. the position of the volume control marks -36. By the small energy dissipation the output stage has itself cooled down on 23°C. Achievement 0.031 Watts rms, current amplitude 0.125 A

Spectrum left channel with 100 mVpp. The harmonious ones are not a problem, with the small tension appear increasingly stray effects etc. below 1 kHz. The long persisting Analyzer Sweep was centrically broken off. Achievement 0.00031 Watts rms, current amplitude 0.0125 A

Spectrum right channel with 100 mVpp. No more harmonious ones clearly recognizably, on the right of and left of the primary wave a few disturbances and noise. Achievement 0.00031 Watts rms, current amplitude 0.0125 A

Signalquelle ist sauber

distortion factor a measurement requires a clean signal source. The picture shows the spectrum of the signal source with a level of 100 mVrms. Exactly this source with the adjusted level served as source for the past measurements. The preamplifier entrance AUX is for a level of approx.. mVrms laid out, the signal corresponds to 150 thus for instance to a 2/3 rejection of the entrance amplifier. To the Veringerung of the output voltage the volume control was down-turned, corresponds to the behavior of the user. One could have set theoretically also the volume control to full-scale and the output voltage over the level of the Einggangsspannung put on to reduce be able, the result very probably even still better would have become, the proceeding to application in the reality would however not correspond. The signal source shows two small peaks, K2 and K3, the harmonious source scarcely over 90 railways, thus about 0.004% harmonious distortion factor. To see are a few disturbances caused by the frequency and their harmonious one, they also lie in the first left division.

Distortion factor measurement of 10 kHz


Signalquelle 10 kHz Spektrum 10 kHz bei 1 Volt RMS Spektrum an 3 Volt

Spectrum of 10 kHz signal source for  distortion factor measurement

Spectrum left channel at 4 ohms with 1 V rms. achievement  0.25 Watts rms, current amplitude 0.35 A - 0.03% harmonious K3

Spectrum with 3,16 V rms. achievement 2.5 Watts rms. current amplitude 1.11 A - 0.15% harmonious K3

Spektrum am 10 Volt rms

Spektrum an 60 Volt peak peak


Spectrum with 10 V rms. achievement 25 Watts rms. current amplitude 3.53 A - 0.1% harmonious K3

Spectrum with 60 Vpp = 21.2 V rms. achievement 112 Watts rms. current amplitude 7.5 A - 0.05% harmonious K3


During the hard requirement of the frequency of 10 kHz the distortion factor becomes more visible than with 1 kHz. A tribute to the small becoming open loop reinforcement. Starting from a certain achievement level the harmonious portions do not worsen any longer, it can much achievement be pulled. The harmonious distortions do not achieve here with 2,5 Watts rms the better results as with 112 Watts rms, who would have that thought.


Cross modulation and frequency disturbances

The disturbing and noise level in the spectrum particularly with low output voltages are also effects of the stray effect of the power pack, noise and mixing products. Is not amazing, around this to improve would be enormous efforts at the screen and DC voltage sieving necessarily. The expenditure in addition is well-known-measured some more at additional mechanics and electronics. An additional expenditure the customer from the large mass are only unfortunately only rarely to pay ready.

A simple attempt demonstrates the kanaltrennung already reached:

  1. clamp e.g. right loudspeaker box in the switched off condition

  2. Volume controls on very quiet place

  3. Amplifiers switch on

  4. , tuner or Aux put on music signal

  5. left signalkabel take off

  6. Volume controls slowly fully wind up

  7. in the left channel now a very quiet cross modulation of the right channel will be audible

  8. Volume reduce, switch off, loudspeakers again attach

  9. other channel in reverse test

  10. Phonoeingaenge test

With an ideal amplifier one would hear nothing of the other channel with this test. I tested this also, sound the whole in such a way: as a very quiet box, with which the bass automatic controller is fully untwisted fully back and the altitude controls. Why? these ouple mechanisms work with zunehmder frequency ever more effectively, since parasitic koppelkapazitaeten with rising frequency in their impedance decrease and thus at conductivity win, speak that over ouples happen with higher effect. Also with existing galvanic interconnection by line inductances the effect of the interconnection rises with higher frequencies. By the bases of complex ac technique, wellentheorie and the behavior of parasitic capacity and inductance these effects are explainable. Complete remedy creates here only one reduction of the ouple mechanisms, which would be connected with high expenditure. I have with this test much over ouples was probably also therefore heard, there a very efficiency-strong horn loudspeaker for testing used and in connection with this high performance amplifier with  completely untwisted volume result in that an audible over-coupled signal. Despite all the words over linking and disturbances, a noise is hardly perceptibly even with fully untwisted volume.


Rise time and Slew rate at 4 ohms


The rise time and the Slew rate are a simple method for the evaluation of the "electrical speed" of an amplifier, if one may define so generally comprehensibly, the ability as fast it its tension in the exit to change can. For measurement are sufficient a fast square-wave generator or pulse generator and an oscilloscope with sufficient range.

Were used:

  • 3325A with rectangle a rise time of approx.. 20 LV

  • 7904 Tektronix 500 MHz oscilloscope 

  • 7A19 vertically amplifier 500 MHz 50 ohms of inputs

  • 7A26 vertically amplifier 200 MHz  1 Mohm input

  • 7B92A dual time cousin

  • more compensated 10:1 probe with 350 MHz range


Rechtecksignal 1kHz 10 MHz Rechtecksignal

was tested with 100 mVs 1 kHz square wave signal, the lower jet is lengthened in the rising flank on 500 ns/DIV.

a right signal of approximately 10 MHz shows, the rate of rise of the generator is smaller 20 nanoseconds, which should be fast as source enough.  Measured with the 7A19.

Rechteckantwort bei 15 Volt gedehnte Rechteckantwort

Rectangle answer of the amplifier to 1 kHz signal. Amplitude 15 V at 4 ohms.  As well as no overshooting. Measurement with 7A26. Achievement 56.2 Watts  of current amplitude 3.75 A.

rising flank of 1 kHz rectangle answer long. Existing overshooting, here tries to follow the amplifier the input signal, in addition he needs a finite time, in order to be able to follow the Orginalsignal again with high accuracy.

1 kHz Rechteck stark gedehnt Rechteck 10 kHz

rising 1 kHz rectangle flank strongly long up to 200ns/Teilstrich. The jet is already lengthened around the factor 250 temporally by the oscilloscope.

the rectangle frequency increases to 10 kHz. Well visibly, the flanks round, derivably from it the range of the amplifier are highly however finite. An even transient behavior at the positive and the negative flank.

From the pictures results a rise time 10% to 90% in the order of magnitude 1µs at 4 ohms takes one the range of zero crossover as basis for the maximum Slew rate, then about 10 Volt/300ns result in the lengthened picture. On one microsecond that is standardized 34 Volt/µs at 4 ohms one may however assume the amplifier would show faster values at 8 ohms and with full amplitude.


Rise time and Slew rate without load


Anstiegsflanke im Leerlauf fallende Flanke

rising flank of a 80Vpp 10 kHz square wave signal without load at the left channel. In the steepest piece of the flank about 130 Volt/µs are read off.

falling flank 10 kHz of the rectangle without attached load resistance at the left channel. In the steepest piece the calculated slope amounts to approx. minus 120 Volt/µs.

positive Flanke

fallende Flanke

positive flank of the 80Vpp 10 kHz square wave signal without load at the right channel. In the steepest piece of the rising flank approx.. 125 Volt/µs.

falling flank 10 kHz of the rectangle without attached load resistance at the right channel. In the steepest piece of the falling flank approx. minus 115 Volt/µs.

In the no-load operation compared to the 4 ohm enterprise is larger the Slew rate. The positive Slew rate is minimum larger in both channels than the negative. The two channels in the comparison are equivalent fast to each other. The falsifying delay of the rate of rise by the Messquipment is to be regarded with this order of magnitude of the Slewrate still as small. 200 MHz vertical amplifier in connection with the probe has about a rise time of approx.. 5 LV for this regarded steep section in that the Slew rate one computed. In addition come still approx.. 10 LV for the generator within this range, in sum all at the most 15 LV. Thus can be regarded perhaps rounded up e.g. the measured 130 Volt/µs to 140 Volt/µs. Even that is not counted yet completely correctly, the calculation runs not additive, but I remember an equation the geometrical means following-end - however hairsplitting is, the measuring equipment is faster with these diverted times around orders of magnitude than the test specimen in this case.

The rising and the falling flank are nearly identical.


Amplitude response and break-even factor in the audio frequency range


The amplitude response can be determined with reduced accuracy particularly over far frequency range very easily with an oscilloscope and a sinus-wave generator. For precise measurements in the audio frequency range is the measurement more with difficulty, possible for example with a Precision AC circuit analyzer and a generator with constant output voltage, thus can a break-even factor be determined.

Messwerte Dämpfungsfaktor Verstärkungsfaktor

the measured values, among themselves compound up thereby, read off from the DMM, the table not too long becomes. Measured values for frequencies from 20 cycles per second to 50 kHz. Noted the input voltage at the AUX input right channel, output voltage without load and the output voltage at 4 ohms load. The output load causes a changed current flow in the transistors, thus the open loop and the frequency response of the amplifier, this change change dynamically have a changed output voltage to the consequence. The input voltage was constantly placed, and constantly along-noted, since the generator cannot keep the amplitude constant over all frequencies 100% like it would be theoretically necessary for this measurement.

the linearity of the adjusted amplification factor shows. It was adjusted by the position volume of the automatic controller at will to for instance an amplification factor of 100. The blue curve shows the no-load operation, red under load, with which the output voltage is reduced. The spannungsverlust can be justified with a dynamic internal resistance or also to output resistance mentioned. By standardisation on the attached load resistance one receives the break-even factor.

Verstärkung in dBV Amplitudengang normiert auf 1 kHz

the amplification factor converted into decibels. A reinforcement of 40 railways corresponds to a 100fachen reinforcement. Visibly as during rising load the output voltage sinks, with an ideal amplifier the output voltage would not change, with changed lastimpedanz. As yardstick the amplitude response cannot be defined e.g. with a nominal load of 8 ohms or no-load operation, each change within the permitted load resistance range on e.g. 6 or 4 ohms should an effect on the output voltage have, everything else of it is a deviation from ideal.

for the improved representation of the linearity of the amplitude response it was standardized. Arbitrarily the measured values were taken with 1 kHz as reference. The measured values with 1 kHz were set to 0 railways, with all other frequencies are related so the deviation to 1 kHz measured value. In the no-load operation the linearity is to be improved distinguished and hardly. With 4 ohms minimum deviations can be recognized, 8 ohms between them would somewhere lie. Regarding the linearity the amplitude response is linear to approx.. 7 kHz, and sufficiently linear up to the end of the audible audio range. High linearity.



by the measured values a dynamic internal resistance can be calculated. It rises on with rising frequency, considerably because the open loop reinforcement with the frequency and load falls.

from dynamic internal resistance the break-even factor can be computed, that is the standardisation on the attached lastimpedanz.


Source text for the production of the graphs, already with the LabView demo version goes that simply and very fast.


Amplitude response and range


In the case of this measurement the range of the amplifier is to be measured. It measured with different output amplitudes of 30 mVs rms up to 15 V rms. measurement once without load and of 4 ohms. For all measurements a constant input voltage was applied by 100 mVrms to the tuner entrance, the first amplifier stage is thus controlled on order of magnitude 2/3. With the Lautstaerker automatic controller was adjusted the output voltage. The attitude of the output voltage happened in each case with 1 kHz, afterwards at generator the frequency was increased with constant amplitude until the AC indicated one to circuit analyzers at the exit around 3 railways small output voltage. The 4 ohms of measurement regards a capacity range up to scarcely 60 Watts rms. With higher achievement I did not want to load the amplifier with these high frequencies, although it would not have given defects thereby him also with high frequencies to to the power limits to float. Also I  did not put on higher frequencies than the range.

However I made amplifier types when testing the volume-ridden with this already once bad Erfarhrungen. Another 907 copy equipped with orginalen output stage transistors should be examined for the range. During approximately 450 kHz (already beyond the range) and high achievement those went to orginalen transistors of the output stage directly broken, the spare types transistors can with the high working frequencies not fast enough be switched off, as consequence of it do not flow a ever larger shunt current from the positive supply into the negative supply inside by the constantly low impedance becoming NPN and PNP. This effect swings itself starting from a certain frequency and load of alone racing fast up and destroys the transistors. The fuses built in the equipment cannot protect from which get broken transistors faster than the safety device release can. The Aufgbabe of the safety device against such errors is also not to be protected, it has protective functions like permanent overload or fire protection function. With the orginalen transistors "can - should - or -" such a thing not to happen, that amplifiers was co-ordinated with the orginalen transistors and compensated, the installation of other copies, those strongly of the specifications that originals deviates endangers the amplifier, in addition the performance suffers from it. Particularly with repairs these important points are ignored by some repair men, since they do not have the originals at the time of the repair available and from ignorance an error to often commit here. Even if the amplifier with the unsuitable copies runs, the performance is worse and in extreme situations (e.g. range test) danger can exist. Owners 907 with unknown quantities prehistory should strive therefore around the correct types. With a repair is to be used on it to the respected originals.


Bandbreite über Ausgangsspannung

Range of both channels without load and at 4 ohms. With very low output voltages the range is much high and falls to a minimum of 160 kHz with approximately 2 V. One measured with 30, 100, 500mV, 1, 2, 3, 4, 5, 6, 7, 8, 10 and 15 Vrms.

to represent the same diagram in linear representation around the measured values with high output voltages better. The range is not in ranges of middle and high power output on constantly high level, also an increase of power output reduced the range. In the no-load operation the range is higher than under load.

Bandbreite über Ausgangsleistung

the x axis was converted into a power output. The measured range covers the range from approx.. 1 milli Watt up to approx.. 60 Watts. Disturbing the minimum is in the range around 1 Watt, an often used range - however even "only" 160 MHz range are for a Hifi amplifier fast measured values.

The interesting at this measurement, it does not show once more an amplifier is a constant thing. Specified sizes as individual numbers to indicate makes a sense under indication of the boundary conditions and attitudes of the measurement.


Phase response


cannot be measured unfortunately without expenditure. The 907 has different voltage supplies for preliminary stages and output stages, which mass of the output stage and preliminary stage is appropriate with a continuity tester tested on same potential, when connecting the single Ground phase meter becomes these two masses over this second path however hooked up, the result was for changes in the amplitude and partly oscillation - thereupon without this measurement does. The developers imagined with their supplying and mass concept somewhat, which disturb second mass path. From the high range and the qualitative process of the amplitude response cannot be concluded, the phase response might a problem represent, within the audio range hardly a phase shift available will be, and with higher frequencies increasingly hastening after.


Frequency response Phono entzerrer at the Preamp output


Measurement takes place with one constantly amplitude signal generator 3336B , the attitudes 1 kHz, mm input -53,5 dBVrms (0 dBVrms = 1 Vrms), volume automatic controller 95%, based on the Preout -2.4 dBVrms (approx. 1,1 Vpeak).

The measurement shows the left entzerrer frequency response of the RIAA standard characteristic as well as the first amplifier stage (approx. 20dB fixed), by the all signals (Phono also AUX and tuner) continuously. Since also the phase response was measured, 3575A Gain phase Analyzer use found, with the exclusive measurement up to the Preout is possible the phase measurement, since GND push away to GND Preout MM-IN and on same voltage supply.

The RIAA standard characteristic is left somewhat more near described in this:


Moving Magnet Amplitudengang am Preout links Moving Magnet Phasengang am linken Preout

Amplitude response at the left Preout. Moving magnet 47 kohm.

Phase response at the left Preout. Moving magnet 47 kohm

Vergleich Phono Entzerrer mit der RIAA Kennlinie

Linearer Fit durch die Abweichung der Messung vom Ideal

the Preout amplitude response was standardized on the frequency 1 kHz, the reinforcement with this frequency reaches thus 0 railways. Red the measured characteristic is blue, the theoretical ideal characteristic. The deviation is minimal.

the points show the difference of the measurement of the ideal, an only insignificant deviation. Additionally an involution straight line was put by the deviation, which corresponds to more the material values. The deviation is in far frequency ranges below 0,1 railways.

One considers very important when this measurement somewhat: as measuring instrument a Gain phase Analyzer was used, which has a dissolution of 0,1 railways - alone already the deviation lies in the dissolution range of the measuring instrument. The use of the Precison AC DMM would have been importantly for the amplitude response more meaningfully and more exactly, the measurement phase response would however not be not possible thereby.

Additionally to the actually interesting Phono entzerrer knowing LINE lies in series first approx.. 20 railways amplifier stage for all signals. The result shows therefore the sum of the errors out: Phono entzerrer characteristic + 1. Amplifier stage + frequency response error measuring instrument. In sum a long chain and nevertheless a beautiful result. It does not only point on the high accuracy of the measuring instrument, the measurement shows also that the frequency response of the Phono entzerrer corresponds faithfully to the desired values. A desire-fair Phono entzerrer frequency response. On it following a short comparative measurement of the right channel with the left channel showed the deviation amounted to  max. 0.2 railways von Channel R Preout to Channel L at the Preout. Both mm of channels up to the Preout are measured thereby as "directly loud" to designate with all audiofrequencies.


Thermal load


for normal or classical period listener without heistation, the amplifier becomes well lukewarm. In the constant discotheques employment or the like, he should be cooled by sufficiently large exhausts additionally, in order to increase the life span.




Very interesting for the frequency response and distortion factor measurements of HF of suited capacitive load would be parallel to the 4 ohm load resistance. Kapzitive impedances with higher signal frequencies represent a challenge for each amplifier, are it an operation amplifier or an achievement amplifier, since a capacitive impedance in connection with dynamic internal resistance reduces the phase reserve automatic control loop, than consequence of the fact are possible it that an amplifier begins to reciprocate. That small the signal sine can do overlaid hardly visible oscillations to be, which can to up-swing or also suddenly arise as large signal. Depending upon protection concept of the amplifier, this can mean the end of the output stage, even if this amplifier has protecting inductances in suitable place, I without these attempts will do. In practice loudspeakers are not loads, which are minted capacitively, therefore are these tests more than of practical importance, it can from instrumentation however conclusions from the results be pulled.




very good results of measurement, which see to let to be able itself, beautiful amplifier.



It can be that these measurements are faulted or wrong. The representations do not have any requirement on correctness.


Repair amplifiers

Table with replaced spare parts