More Fun With Magnavox: The 9300 Series
- Thread starter dcgillespie
- Start date
crispycircuit
Super Member
I apologize for my "comments".... I understand getting the most possible out of your gear. Dave has taken the time/effort to help all the Magnavox owners get the most, without causing any harm. NICE!! This is a great thread! Since Magnavox is one of the last cheap amps available, " LONG LIVE MAGNAVOX"! I do not want to cause any harm . I'll get back in my cage....
Last edited:
Pio1980
AK Member
I dunno if anyone expects a giant killer from this project, but cleaning up the simplicity of these basic plentiful tube power amps will get the best out of their fluid transparent midrange potential and also improve measurable performance for chump change. That alone is worth the effort in order to spread the joy of good affordable sound.
Harlon
Active Member
Hopefully this won't offend, but when I read Dave's post about claiming his 9300 work up was hardly an effort to make it a Picasso I was instantly confused. My experience with Picasso brought images far from the simplistic beauty of a 9300.
My experience with amps is very basic. I haven't ever even seen or heard, much less owned or worked on many of the works of art I've read about here and elsewhere. But I generally wouldn't compare them with Picasso. Seems his lack of realism and distortion of perspective are contrary to the transparent clarity sought for in an amp.
As a fun sidebar, I will say that as far as 9300 power supplies go, I think my Maggie Console Power Supply is a bit of a Picasso.
Again, hope this didn't offend. Picasso is great.
Just gotta wonder what was going on in his head sometimes.
My experience with amps is very basic. I haven't ever even seen or heard, much less owned or worked on many of the works of art I've read about here and elsewhere. But I generally wouldn't compare them with Picasso. Seems his lack of realism and distortion of perspective are contrary to the transparent clarity sought for in an amp.
As a fun sidebar, I will say that as far as 9300 power supplies go, I think my Maggie Console Power Supply is a bit of a Picasso.
Again, hope this didn't offend. Picasso is great.
Just gotta wonder what was going on in his head sometimes.
Pio1980
AK Member
Attachments
Hummmm -- It seems that I surely need to bone up on just who the different artists were before using them as an example!! Thanks for the education!
Kid -- Frequency response is unchanged whether one or both channels of the original design are driven, as such measurements are always taken at low (1 watt) power levels.
Power and distortion are a different matter of course, which generated my deja vu comment in the post on the output stage, regarding how it suffered from the same ill effects as the stock small Dynaco amplifiers suffer from: When a single channel is driven with a bone stock 9300 into a 4 ohm load, power output at 1 kHz just reaches 15 watts, with distortion running about 2.3% at this frequency at 1 db below 15 watts. Power output and distortion improve in this scenario (over that of both channels driven) because the un-driven channel acts as an active regulator element to regulate both the B+ voltage, AND the bias voltage at the cathodes of the output tubes for the driven driven channel. But of course, that is not how the amplifier is used in a real world setting -- and also why specifying the power output of just one channel (and therefore simply assuming that when driving both everything would remain the same except that power would double) can be so deceptive. To cover for these faults, all manner of alternate power output rating systems were devised back in the day, such as, "music power", "instantaneous power" and the like -- all of which just added to the deception which ultimately caused the FTC to step in and standardize the rating system for stereo (2 channel) components that operate directly from the AC power line.
Of course, when the stock 9300 has both channels driven in this scenario, then the power output of each channel drops to just a hair over 10 watts as noted in the base line testing results (a 33% drop in rated per channel power), with the total unit power basically being 20 watts -- for only a 33 % increase in power. However, at this power level, cross over distortion is well intrenched at this point, making even this power level unusable, with THD soaring to over 6% in each channel. When power output is backed down so that the crossover distortion is contained, power output in each channel drops back to about 7.5 watts, for really a zero percent increase in usable power over that which a single channel can produce on it own, with THD dropping back into the mid 2% range again.
It is against this backdrop that the addition of EFB makes such a real world usable improvement in performance that can both be measured, and heard: Power output becomes 15.5 watts RMS in both channels with both channels driven, and THD at 1 kHz drops to well under 1% at 1 db below 15.5 watts. This effectively then represents slightly over a 100% increase in usable power output AND nearly a 10X reduction in distortion over that of the stock design at the new elevated power output level. The low drop 5AR4 helps to deliver this performance as well, since producing increased power output draws more power from the power supply as well. However, with less drop through the rectifier tube, there is less drop in B+ voltage as power increases. The drop that does occur, EFB automatically accounts for. Finally, since the power transformer is no longer supporting any tuner circuits, it has shown zero concerns in handling the increased performance of the modified design.
I hope this helps!
Dave
Kid -- Frequency response is unchanged whether one or both channels of the original design are driven, as such measurements are always taken at low (1 watt) power levels.
Power and distortion are a different matter of course, which generated my deja vu comment in the post on the output stage, regarding how it suffered from the same ill effects as the stock small Dynaco amplifiers suffer from: When a single channel is driven with a bone stock 9300 into a 4 ohm load, power output at 1 kHz just reaches 15 watts, with distortion running about 2.3% at this frequency at 1 db below 15 watts. Power output and distortion improve in this scenario (over that of both channels driven) because the un-driven channel acts as an active regulator element to regulate both the B+ voltage, AND the bias voltage at the cathodes of the output tubes for the driven driven channel. But of course, that is not how the amplifier is used in a real world setting -- and also why specifying the power output of just one channel (and therefore simply assuming that when driving both everything would remain the same except that power would double) can be so deceptive. To cover for these faults, all manner of alternate power output rating systems were devised back in the day, such as, "music power", "instantaneous power" and the like -- all of which just added to the deception which ultimately caused the FTC to step in and standardize the rating system for stereo (2 channel) components that operate directly from the AC power line.
Of course, when the stock 9300 has both channels driven in this scenario, then the power output of each channel drops to just a hair over 10 watts as noted in the base line testing results (a 33% drop in rated per channel power), with the total unit power basically being 20 watts -- for only a 33 % increase in power. However, at this power level, cross over distortion is well intrenched at this point, making even this power level unusable, with THD soaring to over 6% in each channel. When power output is backed down so that the crossover distortion is contained, power output in each channel drops back to about 7.5 watts, for really a zero percent increase in usable power over that which a single channel can produce on it own, with THD dropping back into the mid 2% range again.
It is against this backdrop that the addition of EFB makes such a real world usable improvement in performance that can both be measured, and heard: Power output becomes 15.5 watts RMS in both channels with both channels driven, and THD at 1 kHz drops to well under 1% at 1 db below 15.5 watts. This effectively then represents slightly over a 100% increase in usable power output AND nearly a 10X reduction in distortion over that of the stock design at the new elevated power output level. The low drop 5AR4 helps to deliver this performance as well, since producing increased power output draws more power from the power supply as well. However, with less drop through the rectifier tube, there is less drop in B+ voltage as power increases. The drop that does occur, EFB automatically accounts for. Finally, since the power transformer is no longer supporting any tuner circuits, it has shown zero concerns in handling the increased performance of the modified design.
I hope this helps!
Dave
Random note, I did some baseline testing on my own 9300 last night. It was essentially a stock circuit with the Edcor UL transformers. In UL mode with both channels driven to max output before clipping with a 5V4 rectifier, I was getting a teensy bit over 10 watts. Essentially no power change from the stock design, so the transformers are definitely not the limiting factor. Power bandwidth does improve considerably though, I was -1db at 13hz on the low end and -1 at 22.4khz on the high end, and flat between those points. No rise in response like the stock transformers seem to produce.
I should note those numbers are into an 8 ohm load, using the 8 ohm taps on the transformers. Feedback resistors were off the 16 ohm tap, and I was using a 5k feedback resistor.
I just finished re-working the driver stage, and I'll get some comparison numbers once I give it a second look to make sure I didn't miss anything. I have the EFB parts, so I'll do that next, but maybe tomorrow. I suspect turkey coma will come into effect tonight and I won't be up for it. I'm most curious to see what the inverter re-work does for my high frequency THD numbers. I was getting some stupid high figures, 18% THD at 22 khz. I'm curious if that is inverter imbalance coming into play or what.
and before anyone asks, I fixed the THD meter.
I should note those numbers are into an 8 ohm load, using the 8 ohm taps on the transformers. Feedback resistors were off the 16 ohm tap, and I was using a 5k feedback resistor.
I just finished re-working the driver stage, and I'll get some comparison numbers once I give it a second look to make sure I didn't miss anything. I have the EFB parts, so I'll do that next, but maybe tomorrow. I suspect turkey coma will come into effect tonight and I won't be up for it. I'm most curious to see what the inverter re-work does for my high frequency THD numbers. I was getting some stupid high figures, 18% THD at 22 khz. I'm curious if that is inverter imbalance coming into play or what.
and before anyone asks, I fixed the THD meter.
Last edited:
What is the primary impedance of your Edcors, and at what point is the screen tap placed (% of winding)?
Dave
Dave
I believe they are 8k, with 40% taps. I've got the part number in my notes somewhere to confirm. I added the EFB stuff and got numbers. I also pulled the UL wiring last night, and it didn't change very much. It got a smidge more power, but thanks to the scaling on the meter I can't tell precisely how much. It runs off the end of the 10w scale, but shows 11w on the 100w scale. The -1db point comes out 1.4 watts higher, even though it shows the same max power. Clearly something funny is going on there. I'm thinking the power meter is just bullshitting me. I'll have to measure it with the voltmeter and do the math to see what that has to say. With some luck I can use that to re-scale my other numbers to make it a more honest comparison.
Still getting a lot of distortion (around 10% THD) at high frequencies even with the EFB, but some other observation tell me those tubes are probably not well matched at all. Most of the distortion appears to be 2nd harmonic, and performance channel to channel varies quite a bit. It actually does better running much cooler. Best performance was running about .32v across the 10 ohm current monitor. I'll check through my box of tubes today to see if I can come up with 4 that behave themselves better.
Still getting a lot of distortion (around 10% THD) at high frequencies even with the EFB, but some other observation tell me those tubes are probably not well matched at all. Most of the distortion appears to be 2nd harmonic, and performance channel to channel varies quite a bit. It actually does better running much cooler. Best performance was running about .32v across the 10 ohm current monitor. I'll check through my box of tubes today to see if I can come up with 4 that behave themselves better.
That tapping point represents the intrenched belief by modern transformer manufacturers that 40% is THE UL tapping point for all tubes and transformers -- which of course is categorically untrue. The tapping point varies by tube class, and by loading and biasing conditions used as well, but modern transformers manufacturers do not do any development research to determine these things. To my knowledge, it's only been in the last few years that transformer manufacturers have finally realized that a 25% tapping point is optimum for 6BQ5 tubes when operated to require an 8K load, and therefore now offer that tapping point as a standard offering. Before that fact was understood, you had to specifically request that tapping point, which speaks volumes about the manufacturers who offer up specific transformers for specific tubes. For tubes of the 6BQ5 class however, the optimum conditions with fixed bias operation in fact includes an 8000 ohm load with taps at 25% of the winding. For cathode bias operation, a 10K plate-to-plate load is best, with the taps located at 50% of the winding. With an 8K transformer that has 40% taps, you would be better off using the transformer in pentode mode with some form of fixed bias operation.
Dave
Dave
I'm using my ST 1700A for this. Its a signal generator, THD meter, and has internal watt and voltmeter functions. It also happens to have a scale on there for wattage across an external 8 ohm load. Its not a wattmeter per se, its just a scale on a voltmeter that takes having to do the math out of the equation. Seems there is a bit of disagreement from the 10v/10w scale to the 30v/100w scale. Its 40 odd years old now, quite probably the divider network resistors are not dead-on anymore.
Its running pentode mode at the moment. Gave me less than 1w extra. I've got the EFB setup in there as well. Wouldn't that work effectively like fixed bias, or should I look into building a proper negative bias supply for this and grounding the cathodes? I'd think that the primary impedance is close enough to the stock 7628 to let this mostly work. I guess it might be worth confirming they are indeed an 8k primary, and that the screen and plate taps aren't reversed. I have it wired up properly per the schematic that came with them, but if they got the connections inside reversed that would certainly screw things up.
oldtexasdog
Well-Known Member
Edcor is known for doing that. My last set had one wired correct and the other reversed. I am not the only one either.
Ohm check confirms the wires are right. I didn't confirm the impedance, but its more ohms from center to what is supposed to be plate than what is supposed to be screen, roughly double, so the 40% seems about right as well. I checked both of them, and they come out the same.
EFB IS fixed bias, as in the bias is not affected by current flow through the output tubes like cathode bias is, but neither is it blindly fixed like traditional fixed bias is. Rather, it automatically adjusts accordingly based on the screen B+ voltage.
Have you measured the primary impedance to confirm their 8K specification? The original transformers were about 7600 ohms, but that simply won't make that much difference if the Edcors are in fact 8K units. How are you powering the screens?
If you are only getting about 11 watts of power output -- particularly if that is only with one channel driven, then something is not as it would appear to be. With a 121 vac line, the buck connection in place, using a 5AR4 rectifier tube, EFB, and good output tubes, you should have no problem developing at least 15 watts with a single channel driven (the unit here would produce 17 watts in that scenario). Have you got an alternate meter of known calibration you can use for verification of the output voltage delivered into a known 8 ohm load?
Dave
Have you measured the primary impedance to confirm their 8K specification? The original transformers were about 7600 ohms, but that simply won't make that much difference if the Edcors are in fact 8K units. How are you powering the screens?
If you are only getting about 11 watts of power output -- particularly if that is only with one channel driven, then something is not as it would appear to be. With a 121 vac line, the buck connection in place, using a 5AR4 rectifier tube, EFB, and good output tubes, you should have no problem developing at least 15 watts with a single channel driven (the unit here would produce 17 watts in that scenario). Have you got an alternate meter of known calibration you can use for verification of the output voltage delivered into a known 8 ohm load?
Dave
hm, trafo impedance comes out odd. Anyone mind checking my math and method here?
1vac into the 8 ohm tap gives me 40..4 vac plate to plate. That should be a turns ratio of 40.4:1, square of 40.4 is 1632, so into an 8 ohm load that should show a primary impedance of 13056 ohms ?
using the 16 ohm tap, I get 29.2v on the plates, so 6821 ohm primary impedance assuming an 8 ohm load.
using the 4 ohm tap, 55.1v on the plates, or 24288 ohms primary with an 8 ohm load.
measurements at 100hz
sanity check on the meter says 121.3v from the wall, and my must be accurate 1950s Weston panel meter on the same circuit says120v.
1vac into the 8 ohm tap gives me 40..4 vac plate to plate. That should be a turns ratio of 40.4:1, square of 40.4 is 1632, so into an 8 ohm load that should show a primary impedance of 13056 ohms ?
using the 16 ohm tap, I get 29.2v on the plates, so 6821 ohm primary impedance assuming an 8 ohm load.
using the 4 ohm tap, 55.1v on the plates, or 24288 ohms primary with an 8 ohm load.
measurements at 100hz
sanity check on the meter says 121.3v from the wall, and my must be accurate 1950s Weston panel meter on the same circuit says120v.
Last edited:
as for the other stuff:
That was my understanding as well. I was just confused when you said I'd be better off converting it to fixed bias. It does have the EFB circuit installed currently.
see above for trafo
Power supply is effectively identical to your schematic, except I'm using a 47uf first cap, 33uf second, and the two after that are 22uf, screens are fed after the 330 ohm resistor that the plate supply comes from. 5AR4 and the bucking winding are in place.
That was both channels driven. I didn't check it single channel. I would agree it seems low. There is a question of accuracy with my various meters, but I don't think they are that grossly off. When I was messing with the Bogens I got a max power range of 86 - 92, depending what meter I looked at. Load clocks in at 8.3 ohms, with 0.3 ohms of that being the test leads.
That was my understanding as well. I was just confused when you said I'd be better off converting it to fixed bias. It does have the EFB circuit installed currently.
see above for trafo
Power supply is effectively identical to your schematic, except I'm using a 47uf first cap, 33uf second, and the two after that are 22uf, screens are fed after the 330 ohm resistor that the plate supply comes from. 5AR4 and the bucking winding are in place.
That was both channels driven. I didn't check it single channel. I would agree it seems low. There is a question of accuracy with my various meters, but I don't think they are that grossly off. When I was messing with the Bogens I got a max power range of 86 - 92, depending what meter I looked at. Load clocks in at 8.3 ohms, with 0.3 ohms of that being the test leads.
didn't have a handy source of 10v, I was just using the audio oscillator to give me the 1v. I can see if I have a small transformer around here that will give me 10vac after the variac just to confirm though.
calculators make all math easy
