Testing Millenium Amp

 Testing of Millenium Amp

Tested: 21 Mar 2026

When the amplifier was first built in 2000, some test were done and the record was not kept.  The sound feel right and no work was further done.

Now in 2026, I decided to do a proper test with the following objective

1.  To tune the frequency performance.

2.  To determine the overall voltage gain

3.  To measure the hum level.

Setup for the test

Input Test Signal

Normally, for a proper test, a signal generator is needed to set the input signal.  However, I do not have a proper signal generator, so I decided to tap the 500mVp-p square wave signal from my oscilloscope.  The signal is passed through a voltage divider to reduce the fixed signal level to 92mVp-p.  This was arbitrary chosen because of the use of 3.9K and 1K resistor

 R1 = 3.9K, R2 = 1K, Vin 500mVp-p, Vout = ~102mVp-p

Measured value is 92mVp-p signal is to be fed to the amplifier input.

Output Load

For the output load, a resistor of 8 ohm load is used to replace the speaker.  This is to ensure a consistence load impedance over the audio frequency range. 

Output level is determined by measuring the voltage at the load resistor 

Tuning the frequency performance

Test was done with 1Khz square wave.  Slight overshoot and undershoot of about 15% but no excessive ringing.  This should exhibit a slight peak in the higher frequency response.  

The overshoot can be reduced by more feedback at higher frequency by increasing the value of C3. 

With C3 changed to 1000p, the overshoot is corrected. 

   Channel L 2.35Vpp

   Channel R 2.3Vpp

                    

Overshoot and undershoot is resolved.   Closer observation shows about 2% overshoot/undershoot.

Measured Values

The following peak to peak values are recorded with input of 92mV.

All voltages recorded at peak to peak values in volts.

                   U1A                   U1B                       U2B              U3A      U4A          Output (8ohm)  

Pin         7            6            1            3              2            5            3            3                      

Left        0.092    0.27*      0.24       0.24*       4*          4            36.5       36.5*      2.75

Right      0.092    0.27*      0.24       0.24*       4*          4            34          34           2.6

Left - U3A, U4A (JJ E34L)

Right - U3A, U4A (Svetlana EL34)

Slight difference in output because of difference in output tubes

Close loop Gain computation of each stage

U1A  = 0.27/0.092 = 2.93 

U1B = 0.24/0.27 = 0.88

U2A/U2B = 4/0.24 = 16.67

U3A/U4A (left) = 36.5/4 = 9.125

U3A/U4A (right) = 34/4 = 8.5

Output/(U3A+U4A) (left) = 2.75/(36.5+36.5) = 1/26.7

Output/(U3A+U4A) (right) = 2.6/(34+34) = 1/26.7 = 1/26.15

Trace U1B pin 1 and pin 3 in opposing phase at 0.24Vp-p

Trace U2B pin 2 and 5 in opposing phase at 4Vp-p

Note that the feedback has lowered the gain of U1A.

The difference in output tubes for U3A/U4A between left and right channel has also affected the gain of the last output stage.  Left was tested with JJ E34L tube and the right is tested with Svetlana tubes EL34.  JJ tube shows a slightly higher gain than the Svetlana tubes. 

From Output/(U3A+U4A) ratio at 26.15 to 26.7 matches the transformer voltage ratio of 25 for Hammond 1645.

Overall Gain of the system

The actual system was tested with input of 92mVp-p and the output delivered is 2.38Vp-p on left channel and 2.3V on the right channel.  The close loop gain is 25.9 or 28.3dB on left channel and 25 or  28dB.  Since this signal of 92mV, with output tube translation of more than 34V (8% of full translation), it is likely in class A and about to transit to B.  

Close loop Gain (L) = 2.38/0.092 = 25.9 or 28.3dB
Close loop Gain (R) = 2.3/0.092 = 25 or 28dB

Deducing the open loop gain, given the feedback factor B = 0.02482

Open Loop Gain (L) = 25.9/(1-0.02484*25.9) = 72.6

Open Loop Gain (R) = 25/(1-0.02484*25) = 66

Notice that the gain stage 1 measured earlier is a closed loop gain with feedback, what is the open loop gain without feedback.

Find the open loop gain of stage 1,

AOL = A1*A2*A3*A4*A5

A1(L) = AOL/(A2*A3*A4*A5) = 72.6/(0.88*16.67*9.125/26.7) = 14.4

A1(R) = AOL/(A2*A3*A4*A5) = 66/(0.88*16,67*8.5/26.15) = 13.84

A1 gain is within expectation since the voltage gain of U1A (6189) is 20.

Hum of the system

The hum level is also measured at the output with maximum volume with the input shorted.

The trace below shows the left and right channels.

Left  @20mV/div

Right@20mV/div

The 50 Hz hum level is 16~20mVp-p.  The mean value is below 5mV.  Right channel shows a slightly higher hum level.  

When tested with the speaker, it is almost not detectable at 6" distance.

It must be mentioned that some of the noise are from the earth connection.  Even when the set is not powered up, there is already some level of common mode noise from the earth connection.  Of course, I could lift off the ground line to have a better measurement without the earth interference, but I did not do that given that the actual set being made of metal, should be earthed for safety.   Perhaps, I should add a Y-capacitor to filter noise from the earth path.  I could do that in future.

I am not sure if this level of hum is reasonably good.  However, to achieve this level of hum, a lot of emphasis is placed on following proper grounding rules which I will discuss in future as a separate topic.