As my thread on the 8600 series of Magnavox amplifiers was winding up, I was contacted by fellow AKer Kidmoe, who was willing to donate a completely stock 9300 series unit for the purpose analyzing, dissecting, and otherwise going over every aspect of the design in this series from Magnavox. The Kid came through too as not long after we talked, a package showed up containing a well packed 9300 series amplifier, complete with (by all appearances) the tubes as he originally received it. I told him I didn't know when I'd get to it, but that I would make good on the deal, and so the saga starts now. However, this one will be handled a little differently.
Unlike the previous effort that was presented as virtually a completed project from the outset, this one will move along in real time -- which is not to say real slow, but move along on a piece by piece basis, running along in tandem with various other projects, including the holidays that are virtually upon us. With this first entry then, a base line of stock performance will be presented, along with a highly recommended (but simple) modification to this amplifier, if not already performed. As the thread unfolds then, the various sections will be addressed until an optimized design emerges. Going forward, the amplifier will serve as a test cell for future ideas for these amplifiers.
As before, to offer the most to the most, all modifications will stay within the confines of the existing chassis, and the basic topology of the design (i.e. a single small signal tube and push-pull output stage per channel with common power supply) will remain intact. Also, the original power transformer will remain in place as well. In other words, the thing has still got to resemble and represent the essence of what the original 9300 series was designed to be. Beyond that however, pretty much everything else is fair game. Those of you who want to go outside of those constraints have at it, but all who follow are encouraged to participate.
With that then, what arrived was a well preserved 9302-00 unit, that required little effort to get up and running again. The can cap reformed itself within a matter of a minute or two, and the tubes -- most of which were still original -- all tested quite strong -- a testament if anything, to Magnavox's desire to ensure long tube life in their products.
Before powering up, the OPTs were measured and in fact found to be wound for a 4 Ohm load, as is in keeping with most Magnavox products. The primary impedance of both transformers was measured at 7628 Ohms when the secondary is loaded with 4 Ohms. Components were also tested, with all the important ones within their rated tolerance to permit testing for base line development.
Power was first applied with the rectifier tube removed, so as to take a reading of the heater voltage applied to the tubes. At 7.0 vac, that is excessive, and is certainly enough to reduce tube life, being a direct result of today's higher AC line voltage levels. Fortunately, as with the 8600 series project, this amplifier included the Molex plug to power a separate tuner/preamp chassis, which meant an extra heater winding is available for buck service to tame the high line voltages. For those with these amplifiers who have not utilized this winding to buck the AC line voltage, I highly encourage it for the health of your tubes, and the power transformer as well.
The modification is easy to make:
1. Disconnect the brown and brown w/tracer transformer leads from the Molex plug.
2. Disconnect (or isolate) the black transformer primary lead.
3. Connect the black and brown w/tracer transformer leads together. There should be enough terminals on the nearby T-strip to dedicate one for this purpose.
4. The brown transformer lead then becomes the new transformer lead to apply one side of the AC power to (the red/black being the other).
This modification produces a heater voltage of 6.40 vac, and a B+ at the OPT CT connection of 318 vdc -- all very much in line with the original design center operating voltages with the unit operating from a 121 vac line at my locale.
Of course, it is assumed that you have already configured the unit so it will power on without the need for the external tuner to do so. If not, that information is readily available from many places, and simple to do. In converting it to stand alone operation, an on/off switch is usually added, and a 2 Amp Fast Blow 125 volt fuse should be added as well -- things I have yet to do to the development unit.
For the purposes of all the performance tests run then, they were performed with the transformer buck connection in place, and all the wiring removed from the Molex plug. Otherwise, it was absolutely bone stock. This model also includes the balance control onboard, so it was adjusted to provide an equal resistance from each outside terminal to ground, while the hum balance control was centered.
Since many use this amplifier with the original OPTs, but with 8 Ohm speakers, information is given for both loading conditions. The results are as follows:
1. POWER OUTPUT in RMS watts, both channels driven, average both channels (presented in a 4 Ohm/8 Ohm format):
40 Hz -- 5.29/4.65
100 Hz -- 7.84/8.2
500 Hz -- 8.70/7.80
1 kHz (ref) -- 10.24/9.25
5 kHz -- 8.12/7.51
10 kHz -- 3.40/5.28
Comment: From this you can see that power is concentrated in the mid-band as would be expected, with the amplifier honestly rated as a 20 watt stereo amplifier, 10 watts per channel, with a power bandwidth of basically 45 Hz to 10 kHz (half power points).
2. FREQUENCY RESPONSE: (average both channels)
20 Hz -- -7.0 db
45 Hz -- -1.0 db
1 kHz (ref) -- 0.0 db
10 kHz -- +2.0 db
20 kHz -- +3.5 db
30 kHz -- +3.8 db
45 kHz -- -4.5 db
Comment: Coupling caps are often increased in value in this design, which will certainly help on the low end of things, but will do nothing for the rising response on the high end.
3. HUM AND NOISE (input shorted) -86 db below 10 watts each channel, hum adjustment set for lowest noise. The addition of a temporary bottom plate improved this figure by 2 db.
4. NFB: (measured) 12 db (average both channels)
5. PHASE INVERTER BALANCE: (average of both channels)
Open loop (NFB disconnected) 82.9%
Closed loop (NFB connected) 87.1%
With this large of an imbalance 1 kHz THD was well over 1% at 1 db below 10 watts (7.8 watts), so no further distortion tests were made, as they would all rise from this point. Clearly, phase inverter balance is a point of opportunity in this design.
6. SQUARE WAVE RESPONSE: With a 10 kHz waveform, rise time is very slow, with significant overshoot, and ringing across the wave top.
7. STABILITY: Any practical value of capacitance would not cause instability when loaded into a 4 Ohm resistive load. Under conditions of no load however, less than .01 uF connected across the output terminals would cause sustained oscillation. This is not inconsistent with other power amplifiers of more prominence, but for maximum stability, the ringing should be equal on both the top and bottom of the waveform.
Pics include:
A closeup of the transformer connections to provide buck action.
The complete underside before testing. The only modifications are the removal of the Molex wiring and creating the buck connection.
Ready to start testing.
A 10 kHz square wave into a 4 Ohm resistive load.
The left channel has only a .005 uF cap connected to it, with any more capacitance causing it to spill over an oscillate.
So there you have it. Other things can certainly be tested such as internal output impedance, channel separation, and the like. But this will do for kicking things off. We'll start getting into it next time.
Dave
Unlike the previous effort that was presented as virtually a completed project from the outset, this one will move along in real time -- which is not to say real slow, but move along on a piece by piece basis, running along in tandem with various other projects, including the holidays that are virtually upon us. With this first entry then, a base line of stock performance will be presented, along with a highly recommended (but simple) modification to this amplifier, if not already performed. As the thread unfolds then, the various sections will be addressed until an optimized design emerges. Going forward, the amplifier will serve as a test cell for future ideas for these amplifiers.
As before, to offer the most to the most, all modifications will stay within the confines of the existing chassis, and the basic topology of the design (i.e. a single small signal tube and push-pull output stage per channel with common power supply) will remain intact. Also, the original power transformer will remain in place as well. In other words, the thing has still got to resemble and represent the essence of what the original 9300 series was designed to be. Beyond that however, pretty much everything else is fair game. Those of you who want to go outside of those constraints have at it, but all who follow are encouraged to participate.
With that then, what arrived was a well preserved 9302-00 unit, that required little effort to get up and running again. The can cap reformed itself within a matter of a minute or two, and the tubes -- most of which were still original -- all tested quite strong -- a testament if anything, to Magnavox's desire to ensure long tube life in their products.
Before powering up, the OPTs were measured and in fact found to be wound for a 4 Ohm load, as is in keeping with most Magnavox products. The primary impedance of both transformers was measured at 7628 Ohms when the secondary is loaded with 4 Ohms. Components were also tested, with all the important ones within their rated tolerance to permit testing for base line development.
Power was first applied with the rectifier tube removed, so as to take a reading of the heater voltage applied to the tubes. At 7.0 vac, that is excessive, and is certainly enough to reduce tube life, being a direct result of today's higher AC line voltage levels. Fortunately, as with the 8600 series project, this amplifier included the Molex plug to power a separate tuner/preamp chassis, which meant an extra heater winding is available for buck service to tame the high line voltages. For those with these amplifiers who have not utilized this winding to buck the AC line voltage, I highly encourage it for the health of your tubes, and the power transformer as well.
The modification is easy to make:
1. Disconnect the brown and brown w/tracer transformer leads from the Molex plug.
2. Disconnect (or isolate) the black transformer primary lead.
3. Connect the black and brown w/tracer transformer leads together. There should be enough terminals on the nearby T-strip to dedicate one for this purpose.
4. The brown transformer lead then becomes the new transformer lead to apply one side of the AC power to (the red/black being the other).
This modification produces a heater voltage of 6.40 vac, and a B+ at the OPT CT connection of 318 vdc -- all very much in line with the original design center operating voltages with the unit operating from a 121 vac line at my locale.
Of course, it is assumed that you have already configured the unit so it will power on without the need for the external tuner to do so. If not, that information is readily available from many places, and simple to do. In converting it to stand alone operation, an on/off switch is usually added, and a 2 Amp Fast Blow 125 volt fuse should be added as well -- things I have yet to do to the development unit.
For the purposes of all the performance tests run then, they were performed with the transformer buck connection in place, and all the wiring removed from the Molex plug. Otherwise, it was absolutely bone stock. This model also includes the balance control onboard, so it was adjusted to provide an equal resistance from each outside terminal to ground, while the hum balance control was centered.
Since many use this amplifier with the original OPTs, but with 8 Ohm speakers, information is given for both loading conditions. The results are as follows:
1. POWER OUTPUT in RMS watts, both channels driven, average both channels (presented in a 4 Ohm/8 Ohm format):
40 Hz -- 5.29/4.65
100 Hz -- 7.84/8.2
500 Hz -- 8.70/7.80
1 kHz (ref) -- 10.24/9.25
5 kHz -- 8.12/7.51
10 kHz -- 3.40/5.28
Comment: From this you can see that power is concentrated in the mid-band as would be expected, with the amplifier honestly rated as a 20 watt stereo amplifier, 10 watts per channel, with a power bandwidth of basically 45 Hz to 10 kHz (half power points).
2. FREQUENCY RESPONSE: (average both channels)
20 Hz -- -7.0 db
45 Hz -- -1.0 db
1 kHz (ref) -- 0.0 db
10 kHz -- +2.0 db
20 kHz -- +3.5 db
30 kHz -- +3.8 db
45 kHz -- -4.5 db
Comment: Coupling caps are often increased in value in this design, which will certainly help on the low end of things, but will do nothing for the rising response on the high end.
3. HUM AND NOISE (input shorted) -86 db below 10 watts each channel, hum adjustment set for lowest noise. The addition of a temporary bottom plate improved this figure by 2 db.
4. NFB: (measured) 12 db (average both channels)
5. PHASE INVERTER BALANCE: (average of both channels)
Open loop (NFB disconnected) 82.9%
Closed loop (NFB connected) 87.1%
With this large of an imbalance 1 kHz THD was well over 1% at 1 db below 10 watts (7.8 watts), so no further distortion tests were made, as they would all rise from this point. Clearly, phase inverter balance is a point of opportunity in this design.
6. SQUARE WAVE RESPONSE: With a 10 kHz waveform, rise time is very slow, with significant overshoot, and ringing across the wave top.
7. STABILITY: Any practical value of capacitance would not cause instability when loaded into a 4 Ohm resistive load. Under conditions of no load however, less than .01 uF connected across the output terminals would cause sustained oscillation. This is not inconsistent with other power amplifiers of more prominence, but for maximum stability, the ringing should be equal on both the top and bottom of the waveform.
Pics include:
A closeup of the transformer connections to provide buck action.
The complete underside before testing. The only modifications are the removal of the Molex wiring and creating the buck connection.
Ready to start testing.
A 10 kHz square wave into a 4 Ohm resistive load.
The left channel has only a .005 uF cap connected to it, with any more capacitance causing it to spill over an oscillate.
So there you have it. Other things can certainly be tested such as internal output impedance, channel separation, and the like. But this will do for kicking things off. We'll start getting into it next time.
Dave