Rx For the Magnavox 8800 Series

Discussion in 'Tube Audio' started by dcgillespie, Apr 24, 2018.

  1. dcgillespie

    dcgillespie Fisher SA-100 Clone Subscriber

    Messages:
    9,622
    Location:
    Ball Ground, GA
    When I posted my original thread on the 8600 series, I had absolutely no idea of the interest it would generate in the AK tube community. Then when AKer Kidmoe asked me to go through a 9300 series model he'd donate for the cause, the undying love for Magnavox console amplifiers was again confirmed and reinforced. The listening pleasure these amplifiers bring is absolutely undeniable. Well, the Kid has been at it again, this time donating up a totally stock, unmolested 8800 series amplifier for analysis, dissection, pondering, and general "what if" thinking, original tubes and all. It arrived here back in November of last year, and as before, I told him I didn't know when I'd get to it, but that I would in fact get to it. That effort has move along in the slow lane as the holidays and life events all took stage front and center, but it has moved along, and now to the point that I can get a thread started. Along the way, a lot of information has been gleaned.

    Initial Observations -- stock unit

    Others will certainly know more than I, but it would appear that the 175/185 6V6 amplifiers were Magnavox's first stereo vacuum tube push-pull amplifiers, and the 9300 series (6BQ5) was the last, with the 8800 6V6 series sandwiched in between these two models, almost certainly acting as a transitional piece while manufacturing converted from the larger octal tubes, to the smaller 9-pin miniature tubes. As such then, the 8800 is most certainly the bottom feeder runt of the Magnavox vacuum tube stereo litter. Consider:

    1. Power Supplies: Both the 175 and 9300 units have power supplies of higher voltage and current capability, employing heavy duty rectifier tubes and a choke for B+ filtering. The 8800 series utilizes a smaller medium capability rectifier tube that is operated near maximum current limits with R/C filtering only. The 8800 also uses the same filter cap as used in the 8600 Single Ended amplifier series. The end result is that B+ capacity and filtering is quite limited in the 8800 series, versus that of its more capable siblings.

    2. Operating Conditions: Magnavox is well known for designing their equipment for extended tube life. In the 8800 series, this was pushed to the max by operating the output tubes at a mere 20 mA of quiescent current each. With each output tube operating with a plate dissipation of less than just 4.5 watts then, the unit operates with high output stage distortion (on the edge of cross-over distortion), which is readily visible on a 1 kHz sine wave scope display at anywhere near approaching maximum power output (6.76 watts RMS per channel into 4Ω, both channels driven). This operating condition was no doubt driven by the limited power supply capability these units had. But by using an inherent characteristic of the 6V6 and a common cathode resistor for all four output tubes, the unit could be kept safely away from producing gross distortion, and so was "good enough" for its application. This will be discussed in more detail later on.

    By comparison however, the 175/185 series used an identical biasing scheme, but operated the tubes at a much more appropriate operating point to minimize output stage distortion. And, with a higher B+ voltage, it also produced this lower distortion with greater power output as well. In the 9300 series, the same biasing scheme was also used. And while a reasonable operating point was used in that unit as well, resistor based cathode bias in general is hardly an optimal way to bias the 6BQ5 family of tubes when operating in Class AB, and therefore formed one of the legs upon which improved performance could be produced from that unit. However, with greater power output available in the 9300 over the 8800, the biasing scheme worked well enough at the typical power output levels that console duty would dictate. So with the 8800's operating conditions as well, it is again at a notable disadvantage compared to the other two series of amplifiers.

    3. Design Effort: On top of the previous comments, the design itself was almost surely an exercise in achieving a minimal part count. While this would generally be true of any design built to a price point, it seems to have been taken (again) to the max in this unit. While this unit uses different OPTs than that of the 175/185 series, they are no doubt electrically very similar with regards to response characteristics. In the 175/185 however, there was an effort made to stabilize the HF resonances of the OPTs to produce good transient and frequency response. The sensitivity of the design was maximized as well. In the 9300, the choice of tubes helped to maximize sensitivity, and reasonable HF stability was achieved on a case by case basis based on the differing installation scenarios. In the 8800 however, sensitivity was sacrificed, and HF transient and frequency response was basically allowed to run uncontrolled as will be seen in the performance data gathered. The one sole component used to act as a limiting agent for the HF response characteristics is shown as optional on the Sams schematic, and was not installed in my unit here. Once again then, the 8800 is the step child of the family.

    4. Output Transformer Impedance: I'm not sure what process Sams used to determine the OPT primary impedance of the transformers used on the 8800, but it could not be more wrong. It's not even close, or anywhere near the ballpark for that matter. The part #320295-1 OPT used on the 8800 chassis has a 50:1 turns ratio, indicating that with a 4Ω secondary load, the reflected primary impedance is 10K Ohms, which would be entirely correct for this application. The 6K Ohm specification that Sams indicates becomes yet another of their classic schematic errors.

    5. Sams Schematic C4: Shown as a permanent installation in Channel 1 and optional in Channel 2 on the Sams schematic, it's installation no doubt had to do with the cabling and characteristics of the preamp used in the various console application scenarios. It's installation does act to control somewhat the otherwise run amuck HF response, and the HF transient response (somewhat) as well -- but even still, with its installation, both could hardly be considered as yet properly controlled. Do note this however: If you operate the stock amplifier into 8Ω speakers, then the installation of C4 in both channels is mandatory to prevent the formation of parasitic oscillations on medium power (and higher) frequencies below 100 Hz.

    4. Performance (operating from 115 vac, which produces normal operating voltages):

    A. Power Output @ 1 kHz:
    At 4Ω: Individually: 8.27 Watts RMS. Both channels driven: 6.76 Watts RMS.
    At 8Ω: Individually: 5.28 Watts RMS. Both channels driven: 5.12 Watts RMS.

    B. Full Power Bandwidth (one half power) based on individual channel performance:
    At 4Ω: 37 Hz - 11 kHz.
    At 8Ω: 41 Hz - 15 kHz.

    C. Frequency Response (Ref: 1 kHz) without C4 installed:
    At 4Ω: @ 20Hz: -.65 DB, @ 30Hz: +1.0 DB, @ 20kHz: +1.2 DB, @ 47kHz: +6.5 DB, @ 62kHz: -1.0 DB.
    At 8Ω: @ 20Hz: -.70 DB, @ 27Hz: +2.4 Hz, @ 20kHz: +.80 DB, @ 55kHz: +9.2 DB, @ 72kHz: -1.0 DB.

    With C4 installed:
    At 4Ω: @ 20kHz: +1.0 DB, @ 30kHz: +2.8 DB, @ 36kHz: +3.4 DB, @ 49kHz, -1.0 DB.
    At 8Ω: @ 20kHz: +1.4 DB, @ 30kHz: +3.7 DB, @ 42kHz: +7.3 DB, @ 57kHz: -1.0 DB.

    D. Sensitivity:
    At 4Ω: 1.20 vac rms for maximum power output.
    At 8Ω: 1.10 vac rms for maximum power output.

    E. Total Harmonic Distortion at 1 kHz at:
    At 4Ω: At 1 db below 6.76 watts rms: 1.4% (2.4% at the onset of clipping).
    At 8Ω: At 1 db below 5.12 watts rms: .8% (unchanged at the onset of clipping).

    F. Stability: Unloaded to .05 uF.

    G. Negative Feedback:
    At 4Ω: 11.9 DB
    At 8Ω: 16.3 DB

    H. Internal Impedance:
    At 4Ω: 1.26Ω
    At 8Ω: 1.30Ω

    A few pics:
    Below: In progress: Basic components checks, setting the unit up for proper AC power (connecting the bucking winding), and removing the needless wiring to the Tuner Power Plug.
    SAM_2398.JPG

    Below: The unit operating for initial testing.
    SAM_2383.JPG

    Below: Oscillation into 8Ω at 90 Hz (without C4):
    SAM_2397.JPG

    Below: 10 kHz Square wave into 4Ω without C4:
    SAM_2382.JPG

    Below: 10 kHz Square wave into 4Ω with C4:
    SAM_2381.JPG

    So there you have it. 5 or 6 watts of power depending on what impedance speakers you're using, potential parasitic oscillation, HF transient stability that's anything but damped, and 1 kHz distortion that leaves you wondering if it's that high at 1 kHz, how much higher must it be at more demanding frequencies? At any rate, sandwiched between the 175/185 and 9300 series, the 8800 series -- by comparison -- is hardly Magnavox's best effort with these types of amplifiers, leading me to believe it was considered to be a transitional piece.

    There is help for these things however, and all without breaking the bank. Of course, to allow the 8800 to really compete against anything in its (potential) power class, a new OPT will be needed. Luckily, there's an answer for that too -- if you are so inclined to go that far.

    For now, the unit has undergone my usual approach to these projects: One channel remains completely stock (as much as possible), while the other is modified to investigate all the potential opportunities available. Those will be discussed next time.

    Dave
     
    Last edited: Apr 24, 2018
    Dandy, Schmidlapper, faber12 and 8 others like this.

     

    Please register to disable this ad.

  2. RUmad

    RUmad AK Subscriber Subscriber

    Messages:
    914
    Location:
    New York
    Somehow I can’t wait to see how this goes, although it will likely mean I’ll be rebuilding my 8802 for the second time in a year. BRING IT ON! Thanks Dave.
     
  3. Spacey37308

    Spacey37308 AK Subscriber Subscriber

    Messages:
    82
    Location:
    Fort Wayne, Indiana
    Thank you for starting this. I have an unmolested unit in my closet. Can't wait to see what you've come up with for this series.
     
  4. Dswankey

    Dswankey Super Member

    Messages:
    3,608
    :lurk:

    These types of threads are what make AK great!!!

    Kudos to @dcgillespie for sharing his time and knowledge.

    I'll be keenly following this thread.
     
  5. gadget73

    gadget73 junk junkie Subscriber

    Messages:
    36,419
    Location:
    Southern NJ
    This is the one with the 6CA4 rectifier, yes?

    The chassis looks awful similar to the 9300. Wonder if its the same stamping, just with different rear tube socket knockouts.
     
    RUmad likes this.
  6. RUmad

    RUmad AK Subscriber Subscriber

    Messages:
    914
    Location:
    New York
    Pretty sure you’re right Gadget. In fact I’m pretty sure I even bolted a choke on my 8800 in the same mounting holes as my 9300.
     

     

    Please register to disable this ad.

  7. bobabode

    bobabode AK Subscriber Subscriber

    Messages:
    9,704
    Location:
    SoCal
    Very cool Dave. Many thanks to you and to Kidmoe for supplying the testbed.
     
  8. Kidmoe

    Kidmoe AK Subscriber Subscriber

    Messages:
    321
    Location:
    Southeast Mass.
    Curious to see if the 6CA4 survives. I would think 4 6V6 tubes, properly biased, would be exceeding the current limit of that rectifier at idle, not to mention at elevated listening levels.
     
  9. drtool

    drtool It might get loud In Houston Subscriber

    Messages:
    2,197
    Location:
    Houston Texas
    See, this is the meaning of life. New toner cart for my printer. I'm ready. I have a friend with one of these. And reports are, it does not play nice.:banana:
     
  10. danrclem

    danrclem AK Subscriber Subscriber

    Messages:
    1,644
    The 8803 and possibly others were used in 1963 Magnavox consoles which I think was the last year of their tube consoles. They were used in some of the BOTL consoles and the TOTL consoles were SS in 1963.

    I didn't realize the 175/185 amps were better than the 8800's but that's good to know.

    I'll be following this one too.
     
  11. dcgillespie

    dcgillespie Fisher SA-100 Clone Subscriber

    Messages:
    9,622
    Location:
    Ball Ground, GA
    Modification Considerations

    1.
    Like the previous two Magnavox efforts, a direct connection to my 121.5-122.0 vac line produced 7.0 vac to the heaters, which is excessive. As before, using the Tuner heater winding in buck fashion with the primary produced excellent results with 6.43 vac being delivered to the heaters. Not using a buck winding or variac will produce a little more power output, but the cathodes of your tubes won't like the excessive heat none too much. Excessive heater voltage can boil off cathode material and potentially deposit it on the control grid. This causes the the grid to then become an emitter and drawing reverse grid current, which permanently ruins the tube, forever making it unstable to use. Using the Tuner heater winding as a bucking agent resolves the problem in all but the worst utility locales.

    2. To get a handle on distortion and improve power output, more B+ is required from the power supply. The original power transformer powered an AM-FM tuner, which could be counted on to pull at least 30-35 mA of B+ at 250 vdc which is available without the tuner connected. The tuner heater winding would have supplied about 2.5 A of current at 6.3 vac as well, of which about 1.5 A is available since the winding is now being used for bucking service. Finally, if the power supply is converted to SS operation, not only is more B+ made available, but another 6.3 vac at 1A is made available as well. Finally, the unit is not supporting a cabinet power indicator lamp which provides incrementally more available power as well. Collectively, this adds up to about 25.4 more watts of power that the transformer was specified to deliver over an above powering the amplifier alone.

    The original amplifier proper draws about .09 A of B+ current at 255 vdc or 22.95 watts. The amplifier chassis audio tube heaters we'll forget about since the plan is that the original tubes would remain as they were. We'll also forget about ripple charging current since that was present in the original application as well. Therefore, both of these current requirements are a constant, allowing them to fall out of the equation. This means that the B+ power available from the transformer for the dual amplifiers is 22.93 original watts + 25.4 estimated Tuner/SS rectifier watts, for a total of 48.33 watts. If we guesstimate a new B+ of as much as 300 vdc, then than means that we have a total of 161 mA of current available to power the amplifiers, meaning that the output stage quiescent current can basically be doubled, and the transformer will still be supplying no more continuous power than it did in its original application.

    This exercise isn't perfect, but is close enough to suggest then that so configured, the extra current draw needed to make the amplifiers really stand up and sing is available, with no particular concern for the health or life expectancy of the power transformer in doing so. As an ultimate double check, the amplifier was operated with the stock audio circuitry, but with a conventional C-L-C filter and SS rectifier. Additionally, the Tuner heater winding was configured to buck the power applied to the primary winding, and a bleeder resistor network was added at the output of the B+ filter (OPT CT connection) to bring the total B+ current draw from the power supply up to 161 mA. After four hours of operation in an open 71-72F environment with this scenario, the transformer was found to operate at 139-140F at the hottest point on the transformer that was not influenced by other heat producing elements (top of outside side or rear of transformer). This is very well within what is considered a safe (and even normal) operating temperature for the transformer, and little higher than the temp others report that their transformers operate at in the basically stock design (sans Tuner), with a couple of CL devices used to drop the AC voltage applied to the transformer (134F).

    3. Coupling caps: The value of (Sams) C5/C7 sets the LF response of the amplifier. The original value was set such that in combination with the NFB loop, LF response was actually quite flat down to 20 Hz, falling off like a rock below that point. Down to 20 Hz, the loop could counteract the loss created by the time constant of the C5/C7 coupling circuit and OPT, but below that point, it could not, so response falls off quickly. This was necessary to prevent record warp and rumble from saturating the OPT and robbing all the power in the process. If vinyl is not your thing however, then the value of these caps can be doubled for improved LF performance for those wishing to keep the original design. This happens because increasing the size of these caps means that more NFB will remain in place in the lower registers. This in turn works to keep the output impedance of the amplifier low at low frequencies, which then provides for improved speaker damping. The improved damping produces a more defined bass reproduction -- rather than the very bloated, flabby bass that accompanies a high output impedance amplifier. Increasing the size of the coupling caps then doesn't provide for any increased LF response, but provides for better loudspeaker damping with bass frequencies. Do not be too tempted to increase the size of these caps much further than suggested so as not to promote saturation of the OPT on heavy bass transients.

    4. Output Transformer Performance: With the smallish OPT employed, every effort will be needed to maximize its performance. This would include the use of an output tube DC Balance circuit to maximize LF performance, and an AC balance control to maximize distortion reduction performance.

    5. Output Tube Characteristics: Unlike the previous two Magnavox amplifiers, this amplifier uses 6V6 tubes, which are beam power tubes. This type of tube is very different from the EL84 power pentode tube employed by the previous two amplifiers, specifically in that the screen grid current starts and remains comparatively low as power output increases. This is due to the aligned grid feature of beam power tubes that power pentodes do not employ. This feature makes beam power tubes more appropriate for cathode biased circuits, since the reduced influence of screen grid current on the total cathode current means there will be less change in bias voltage generated in cathode bias configurations with increasing power output. This is further enhanced by the use of a 10K primary impedance OPT, which mates well with 6V6 tubes when operated with resistor generated cathode bias. With these tubes, this load, this bias configuration, and a 300 volt B+ supply, it should be possible to develop 10 watts RMS with low distortion.

    6. Driver Stage Configuration: Magnavox sticks with the paraphase phase inverter design in the 8800, that served them so well over the years. Simple and economical to implement, but hardly known for its ability to maintain balance over the years or remain in balance when the inverter tube is changed out. That means it is not the best design to achieve the best performance (i.e. lowest distortion) out of the stock OPT. Additionally, if the output stage is to include a DC balancing system within a cathode bias configuration, then the existing phase inverter design will need to be changed out of necessity anyway -- this because the output tubes will operate with a small positive voltage on their control grids with a DC balancing system in place, which would upset the operation of the stock paraphase design. Once again, using a modified version this time, a floating paraphase inverter design can resolve this issue, and still easily accommodate an AC Balance control as well.

    With these thoughts then, a modification has taken shape that improves power output, distortion, and transient response to a very worthwhile degree. For now, a couple of teaser pics of the evolving amplifier:

    Below: Now outfitted with a full set of Bias, DC Balance, and AC Balance controls, a choke and SS power supply (but no output connectors yet), power output and distortion have made very worthwhile improvements. Power output is now 10 watts RMS at 4Ω midband with both channels driven, and distortion has dropped by a factor of 10. Even with both channels driven, THD at 1 kHz at 10 watts RMS power output now hovers around 0.25%. Sensitivity is also improved, now requiring just 0.75 vac rms to develop full power output. Also note that a full set of test points have been added so that all controls can be adjusted without requiring access to the bottom of the unit to do so. A power switch, pilot lamp, and a couple of hole caps round out the top of chassis changes -- so far.
    SAM_2414.JPG

    Below: In addition to the other improvements, thanks to a new NFB and HF stability design, the improvement in HF transient stability is readily noticeable. The upper trace is the original design producing a 10 kHz square wave into 4Ω (showing no damping at all), while the bottom trace is the modified design under the same conditions. In the modified design, the waveform quickly settles down to a flat top after the transient wave front producing excellent stability. The waveform from the original design improves slightly over this presentation with the addition of C4, but is still well under-damped for all intents and purposes, and therefore offers little improvement other than the elimination of parasitic oscillations when using an 8Ω load as noted earlier. The square wave presentations here both represent the same amplitude relative to the fundamental frequency. However, the uncontrolled peaking of the original design when presented with a transient signal (music) steals power output, while the uncontrolled damping muddies detail presentation.
    SAM_2403.JPG

    Next up, details of the new design. But first, I need to finish installing it into Channel 1.

    Dave
     

     

    Please register to disable this ad.

  12. AlTinkster92

    AlTinkster92 AK Subscriber Subscriber

    Messages:
    3,211
    Location:
    NC
    :lurk: Gotta watch this, great thread Dave..
     
  13. dcgillespie

    dcgillespie Fisher SA-100 Clone Subscriber

    Messages:
    9,622
    Location:
    Ball Ground, GA
    Got the old Channel 1 wiring removed, with the new output stage wiring installed, tested, and adjusted for proper Bias and DC Balance.
    SAM_2421.JPG

    A closeup of the driver/inverter stage wiring for Channel 2.
    SAM_2420.JPG

    Dave
     
    s-petersen, kirkendoll and bobabode like this.
  14. dcgillespie

    dcgillespie Fisher SA-100 Clone Subscriber

    Messages:
    9,622
    Location:
    Ball Ground, GA
    Phase I -- Finished

    One of the really neat things about the Magnavox console amplifiers, is that from their humble beginnings, they come with so much potential for improvement, that can be had with a little bit of work, but not a whole lot of cost. It is really only in the 8600 series that an output transformer change is required to help unlock its potential, and this because the power level being dealt with is so limited to begin with. In that case, every watt counts. The idea that every watt counts is true with any amplifier, but when you've got more of them to play with, then an absolute (output) transformer change to realize significant improvement becomes less important. As a result, with the 9300 series for example, significant improvement was still had from the initial modifications offered, even if not taken to the ultimate step of replacing the outputs with Z-565 transformers.

    With the 8800 series, the performance improvement to be had without going to the expense of a transformer change is just as significant. What starts life as a what can surely be described as a "get by" design in so many ways, can really have a very nice polish put on it, with the attendant performance improvements that brings. Granted, the amplifier is still clearly limited by the characteristics of the OPTs, which no amount of design trickery can overcome. With the design as originally offered however, the transformers were hardly the only limitations of the design, there in providing the opportunity for significant performance improvements to be had. Like the 9300 series project then, what the modifications offered here do is to take the amplifier to the point where further improvement will in fact require a transformer change. If that is your goal, nothing offered here is lost towards that effort. For many however, these modifications alone will provide a very enjoyable and worthwhile outcome.

    The modification approach taken with this amplifier takes on a somewhat different approach than the previous two projects did. This is largely because of the different output tubes employed, and the loading conditions used. Because of the use of Beam Power output tubes and the higher primary impedance OPTs, the use of cathode bias is very practical in this amplifier, and offers little disadvantage over the use of fixed bias operation. In addition to these elements, thanks also (in part) to changing to SS rectification and a (clearly) willing power transformer, power supply regulation from quiescent (no signal) to full power output in both channels simultaneously is quite excellent, being on the order of better than 2% -- and this in spite of maximum power output being improved by 48% in both channels! This is really excellent power supply performance. As a result, even the application of EFB™ in this case would gain little in the way of any performance return. The only real loss that the use of cathode bias represents in this case is the loss of power output due to the voltage drop created across the cathode bias resistor network, which amounts to about 0.8 watts RMS per channel when operating into a 4Ω load. The only way to eliminate that would be to create a negative bias supply -- and while that could be done, that starts to infringe rather significantly on the topology of what the 8800 represents. In view of the limited benefit that fixed bias operation provides in this case then, it was decided to stay with the original bias scheme used -- but upgrade it to allow for full Bias and DC Balance capabilities.

    That brings us to the basic building blocks used for the modifications presented:

    1. As with the past two projects, use the Tuner heater winding to buck the AC power to provide for proper heater voltages for the tubes. This is always the best way to achieve voltage reduction as it offers the lowest impedance in the process of providing the reduction.

    2. Convert to SS rectification as discussed for increased B+ voltage, and the elimination of what would otherwise clearly be a current limitation were the original rectifier tube still employed. Even with the original design, the output of both channels could be seen to pulsate on a scope as the limitations of the rectifier tube were exceeded. The conversion to SS rectification completely eliminated that poor social behavior. The increased voltage and current available from the modified power supply provides for both increased power output, and reduced distortion at the greater power output available.

    3. Add a choke that passes all output stage current to eliminate the output artifacts due to minimal B+ filtering.

    4. Increase output stage quiescent current to 40 mA per tube, which provides for the lowest distortion operating point for the output stage.

    5. Provide adjustments to balance the amplifiers both statically (DC) and dynamically (AC) for maximum natural distortion reduction from the push-pull connection, and to maximize the the performance available within the capability of the OPTS employed.

    6. To provide for a DC Balance control in the output stage, it was necessary to modify the design of the phase inverter circuit. Once again, since a floating paraphase inverter configuration is much more desirable performance wise than that of the simple paraphase design that Magnavox used, the event of modifying the inverter to allow for output stage DC balancing was also used to convert the inverter to the improved performance of the floating design as well.

    7. Retain the stock level of NFB employed to retain the overall sonic character (i.e. warmth) of the original amplifier, while using new NFB and HF stability networks to achieve excellent supersonic frequency and transient response, together with good HF stability performance.

    Results:

    A.
    Power Output: 10.0 Watts RMS per channel into 4Ω at mid-band, both channels driven (+47.9%). Under otherwise same conditions, the amplifier will now deliver 6.125 watts RMS per channel into 8Ω (+19.6%). Single channel mid-band 4Ω power output = 11.90 watts RMS (+43.9%).

    B. Full Power Bandwidth: Is now 41 Hz to 14 kHz at 4Ω. While this might seem like a deterioration on the low end from that of the original design, it must be remembered that the bandwidth is defined by the frequency at which power output becomes 1/2 of rated power output. With the modified amplifier, the rated (or in this case: measured) power output is 48% greater than that of the original design. Therefore, the 3 db down power point is greater as well. With greater power passing through the same OPT transformer, the - 3db point will naturally increase slightly which is the case here. At the power level that the original specification was determined to be 37 Hz, the same performance is achieved as before. This clearly shows the limitations of the OPT core size, upon which the only real workaround is to replace it (if desired). On the high end, the bandwidth figure is extended due to the output tubes having greater current capacity to deal with the winding capacitance of the transformer. This allows a higher power output at a higher frequency to be achieved.

    The real benefit with the modified design is that with the amplifier fully balanced now, the bandwidth defining frequencies are produced with notably less waveform deviation than produced by the stock design. The improvement in full power bandwidth performance at 8Ω is commensurate with that at 4Ω.

    C. Frequency Response (ref: 1kHz):

    At 4Ω: @ 20Hz -0.20 db (avg between channels), @ 20kHz -0.5 db, @ 43kHz -1.0 db (avg between channels). Smooth roll-off on either side of these frequency extremes.
    At 8Ω: @ 20Hz +/- 0 db (avg between channels) @ 20kHz +0.2 db, @ 50kHz +1.5 db, @ 63 db -1.0 db. Smooth roll-off on either side of these frequency extremes.

    Notice the much greater control of the supersonic response of the amplifier (greatly reduced peaking) over that of the original design. This translates directly to the excellent HF transient response the unit now produces, as shown in the 10 kHz square wave presented previously.

    D. Sensitivity:

    4Ω Load: 0.75 vac RMS for maximum power output.

    8Ω Load: 0.70 vac RMS for maximum power output.

    E. Total Harmonic Distortion at 1 kHz:

    4Ω Load: 0.28% at the onset of clipping (10 Watts RMS) both channels driven. This is an 88.3% reduction in distortion.
    8Ω Load: 0.55% at the onset of clipping (6.125 Watts RMS) both channels driven. This is a 31.25% reduction in distortion.

    Note that these are full power distortion readings taken while driving both channels. The distortion figures quoted for the stock design were taken at a level of 1 db below maximum power output for the stock amplifier, for the loading condition indicated. Distortion for the modified amplifier at the onset of clipping at 4Ω when individually driven is 0.22%.

    F. Stability: Unloaded to > .05 uF. No LF parasitic oscillations observed when loaded with 8Ω load.

    G. Negative Feedback: 11.0 db when loaded with 4Ω load.

    H. Internal Impedance: 1.30Ω, 4 and 8 Ohm loading.


    Final pics, construction details, a schematic, and some mop-up comments follow shortly, which cover the specific changes to the power supply, how the efforts of this project can be applied not only to its bigger brother (9300), but also to its close siblings in the 175/185 series, and some listening impressions.

    Dave
     
    Schmidlapper, Dandy, drtool and 4 others like this.
  15. dcgillespie

    dcgillespie Fisher SA-100 Clone Subscriber

    Messages:
    9,622
    Location:
    Ball Ground, GA
    Details

    Because of the planning and numerous additional holes required to fit the various adjustment controls, test terminals, and choke, this project was more involved than the previous two Magnavox projects from a standpoint of chassis preparation. Then too, building the modified design into the prepared chassis also makes the underside decidedly busier due to the controls as well. For those who which to omit these controls, I offer a simplified circuit option to use in place of them. However, the inclusion of the Bias and DC Balance controls is highly recommended, because the modified design -- which biases the tubes correctly -- makes them idle at a much higher static current level. At the higher value, plate dissipation is still just 10.3 watts per tube, or 73.5% of the dissipation rating, which will ensure long life and excellent performance. However, it is high enough such that coupled with the smallish output and power transformer set, keeping the current draw tightly balanced between the tubes will ensure minimum magnetization of the OPT core and maximum LF performance, while keeping it dialed in at the correct level will ensure minimum distortion and full power transformer life as well. It was found that (as is usually quite typical) once a minimum distortion setting for the AC Balance control was determined, this setting held over a fairly wide range of output tubes and DC Balance controls settings. Therefore, eliminating the AC Balance control will have much less impact on actual performance, while eliminating the DC Balance and Bias controls will potentially have a much greater impact. Otherwise, the build is straight forward and not particularly complicated, although as I say, it is more involved than the previous to Magnavox projects. I do recommend replacing the can cap with one of a higher voltage rating, as the turn on voltage surge from the SS rectifier will push it up right to the limits of the stock can before the tubes warm up (365 vdc) , and you need a fourth cap section anyway. Therefore, replacing the original can with a four section 40-20-20-20 one rated at 525 vdc in each section is recommended. Antique Electronic's C-EC40-20X3-525 fills the bill, and allows maximum under chassis space for the new circuits. The choke I used is the standard reissue piece from Dynakit for all of their amplifiers, but any choke of about 50-70 Ohms and 1.5 Hy, rated for at least 200 mA will suffice. The same choke as used for the 8600 project (Hammond 156R) will work fine. The Bias and DC Balance controls can be of the wire-wound type (linear taper), and should be rated for 2 watts. The AC Balance control can be a standard linear taper 0.5 watt control, rated for 300 volts service. All resistors are .25 watt unless otherwise specified. Capacitors should have a 400 volt rating or better.

    Power Supply Changes

    1. Use an AC power switch and indicator of your choice if needed.

    2. Use a 1A Slo-Blo fuse for protection of the power transformer. No current limiter was found necessary.

    3. For bucking action, tie the Brown stripped Tuner Heater winding lead to the Black Primary lead. Apply AC power to the Brown Tuner Heater winding lead and the Reb/Blk Primary winding lead.

    4. Eliminate the hum balance control and use two 220Ω .25 watt resistors, connecting one lead of each resistor to each side of the amplifier tube heater winding. Connect the free end of each resistor to the positive lead of the output stage cathode bypass cap in either one of the two channels.

    5. Use two 1N4007 diodes to replace the rectifier tube.

    6. Use the previously specified filter can cap and choke, with the choke effectively replacing the original 100Ω 10W resistor. Connect the 40 uF section directly to the output of the rectifier diodes, and the remaining 20 uF sections for the original 3 positions.

    7. The original 470Ω 3W dropping resistor can be replaced with a 2W device if necessary, since the amplifier is not longer powering a tuner. I used a 500Ω 5W resistor, which is fine as well.

    Modified 9300 Amplifiers

    Since the original OPTs in the modified 8800 and 9300 amplifiers are the limiting factor for any further performance improvements, the effect of an appropriately balanced push-pull drive to the output stage is significant in achieving minimum distortion. If you would like to add an adjustable AC balance control to a modified 9300 amplifier and have access to the appropriate equipment to properly adjust it, it can easily be added as follows:

    1. Disconnect the 220K and 270K output tube grid return resistors from the bottom grid of the driver tube.

    2. Install a 50K linear taper control located convenient to these resistors.

    3. Connect the 220K resistor to one outside terminal of the control, and the 270K to the other outside terminal. Connect the wiper terminal of the control to the bottom grid of the driver tube.

    Adjust the control with conventional adjustment procedures. If more range is needed, the control can be increased to 100K, or preferably, the grid return resistors can be adjusted as necessary to bring the control into range.

    175/185 Series

    This series is extremely close to the 8800 series, with the 8800 series being a notably dumbed down version of the 175/185 series. The 175/185 series has a significant advantage in the design of the power supply and as a result, is one Magnavox design that actually biases the output tubes correctly from the get-go! Still however, the OPTs are no larger than those of the 8800, and so this series too would benefit from the addition of the Bias and DC Balance control modification offered in this project. As explained earlier however, this will also require the installation of the new driver circuit as well to accommodate the DC Balance circuit for the output stage.

    The stock 175/185 series also benefits from an effort to address the otherwise poor supersonic frequency and transient response shown in the stock 8800 series, when components to control those characteristics are omitted. In looking at the approach used however, the results from this project are likely superior to those of the stock 175/185 design. Therefore, even if you elect not to add the Bias and DC Balance controls to the output stage, installing the new driver circuit of the modified 8800 amplifier into the 175/185 series will be of benefit. The characteristics of the OPTs used between these two models are likely so close as to use the new driver circuit with no modifications.

    ALSO: In developing the modifications for the 8800 amplifier, a glitch developed when Channel 1 was being converted to the new design, alongside Channel 2, which already had been converted. The glitch manifested itself as a significant parasitic oscillation in both channels above about 2 watts of power output in both channels. This was traced to the need for separating the screen grid circuits of the two amplifiers because the cathode circuits have been separated, in combination with the higher Gm operating point of the modified design. This resulted in the addition of the two 68Ω resistors that separate the screen grid circuits of the two amplifiers. The point of the comment is that the output stages of the 175/185 operate at the same Gm level as that of the modified 8800 amplifier does, but the design of the 175/185 amplifiers simply connects all four screen grids together. Therefore, if a 175/185 amplifier is modified to include the Bias and DC Balance Controls of this project, then the 68Ω resistors separating the screen grid circuits need to be installed as well.

    No Controls

    If desired, the Bias, DC Balance, and AC Balance controls can be eliminated from this project as follows:

    1. The resistors making up the Bias and DC Balance circuits can all be eliminated, and a single 220Ω 2W resistor be connected between the paralleled output tube cathodes and ground (along with the cathode bypass cap).

    2. The AC Balance control can be eliminated by replacing the 430K resistor with one of two matched 470K resistors, and the existing 470K resistor with the other matched 470K resistor, that also has a 10Meg resistor connected in parallel with it. Tie the free end of the two matched 470K resistors together, and where they connect, also connect the free lead of the .047uF cap that goes to the bottom grid of the driver tube.

    Listening Impressions

    From a sonic presentation standpoint, on my Cornwalls (8Ω), the modified amplifier still has the warmth that endears so many to these amplifiers to their users, but is smoother and definitely more natural sounding -- this a product of taming the supersonic transient and frequency response issues in the modified design. The clarity and dynamic presentation that comes from properly biasing the output stage and ensuring well balanced push-pull operation is unmistakable. I have always found this to be the case when these performance characteristics are optimized in a design that did not originally do so. When driving a 4Ω load (two Cornwalls per channel for my tests), the same improvement in sonic qualities are noted, but it definitely takes on a more powerful sound -- and not because there is simply more speaker board to deliver it. Properly load, the amplifier really comes into its own, and really shows its newfound punch and ability to deliver. The listening experience is very engaging, full and well balanced, and simply fun to listen to.

    Closing

    That brings this project to a close other than some final pics and a schematic which follows.

    Happy listening!

    Dave

    Magnavox 8800 Modification.jpg

    Below: The final build.
    SAM_2424.JPG

    Below: Power Supply closeup.
    SAM_2426.JPG

    Below: Output Stage closeup. The NFB resistors are located beneath the 750 pF caps.
    SAM_2427.JPG

    Below: Driver/Inverter Stage closeup.
    SAM_2428.JPG

    Below: Top view of the finished 8800 (for now), with the pile of old parts removed, and a hint of what can turn it into a dragon slayer!
    SAM_2425.JPG
     
    Last edited: May 1, 2018
  16. kward

    kward AK Subscriber Subscriber

    Messages:
    3,343
    Location:
    Utah
    Hi Dave. Another neat project!

    A few questions.
    1. Still not sure I get the reasoning for the 68Ω resistors. I assume these resistors connect to a common point on the power supply for both channels. But the purpose is not the same as what you have previously called the "screen stability" resistors with separate resistor for each screen? I'm not quite seeing how these resistors separate the screen circuits of the two channels.

    2. It didn't dawn on me until I reviewed the schematic that there is no 8Ω tap on the output transformer secondaries. So your specs quoted for 8Ω loads are running an 8Ω load on the 4Ω tap. But under those conditions, I would have thought that full power mid band THD would have been less than a 4Ω load on the 4Ω tap since the reflected impedance would be ~20KΩ under those conditions.
     

     

    Please register to disable this ad.

  17. dcgillespie

    dcgillespie Fisher SA-100 Clone Subscriber

    Messages:
    9,622
    Location:
    Ball Ground, GA
    Hi Kevin --

    1. Your correct in that the 68Ω resistors are not Screen Stability resistors in the sense that they are not providing protection from output tube arcing. The 6V6 family of tubes is not a particularly high Gm tube, and operating them with cathode bias and within all published ratings nearly guarantees that any arc events simply wouldn't happen (not from circuit conditions anyway). But that's not to say that parasitic oscillations can't happen as a result of the two separate amplifiers operating from a common power supply. Normally, with a common cathode connect between all four tubes, as well as a common connection between all four screen grids, and with the use of cathode bias, tubes of the 6V6 class will easily remain stable in spite of the four tubes representing two separate channels. But in this case, with the cathode circuits separated, and the plate circuits separated, but the screen grids not separated, the two output stages didn't like that when operating the screens from a common supply point, and would become unstable rather easily. The addition of the two resistors (one feeding one pair of screens, the other feeding the other pair) completely stamped out that non-social behavior, with the amplifiers remaining stable regardless of drive or loading conditions.

    In the beginning, I had one new channel, one old channel, and a big bleeder to represent the additional power that the second new channel would draw, over what the remaining old channel was drawing -- this to let the power supply see the full load of what it would see with both new channels installed, even though the second channel had not been installed yet. In this scenario, both the old and new channel were perfectly stable at all times. Only when I converted the remaining channel to the new design did the instability crop up -- which would be understandable. The old channel and bleeder drew the equivalent current of on one new channel, but the bleeder is a passive device, while the tube element is a dynamic device. In that regard then, it is easy to see how the two screen circuits -- operating at a higher quiescent current, would in fact start fighting each other unless the power supply impedance feeding the screens were in fact 0.0 Ohms. The resistors completely resolve the issue, and are basically invisible to the operation of the circuit.

    2. Interestingly, that is exactly how the original design acted: lower distortion at a higher load impedance. Ultimately, at the bias level the original amplifier operates with, that makes perfect sense as the tubes are so over-biased to begin with. Feeding a numerically higher load would minimize any greater increase in bias voltage, and let the stage operate with comparatively lower distortion. In the modified amplifier, the quiescent bias level, type of bias used, and load conditions all come together as an optimized piece, so that it produces the lowest distortion into the designated load for those operating conditions. Deviation from that load then reduces power, and increases distortion, as theory says it will.

    I hope this helps!

    Dave
     
  18. Kidmoe

    Kidmoe AK Subscriber Subscriber

    Messages:
    321
    Location:
    Southeast Mass.
    Thanks for the hard work Dave, it looks like a lot of thought went into the design..I look forward to modding mine in the near future.
     
  19. frankiebull

    frankiebull Active Member

    Messages:
    211
    Location:
    Oklahoma City, Oklahoma
    This has been interesting. I have an 8802 that I like real well. I am going to redo to this thread. Where do you get the 50 ohm and 50k ohm pots?
     
  20. Kidmoe

    Kidmoe AK Subscriber Subscriber

    Messages:
    321
    Location:
    Southeast Mass.
    Dave, The schematic shows a 1000uf/10v cap and 1m resistor coming off the lower 6eu7 meeting up with a 470k resistor coming from the input jack but when I look at the closeup image of the driver area it looks to me like the cap and 1m resistor are going to a ground connection and not connecting with the 470k at all. Am I seeing this wrong?
     

Share This Page