More Fun With Magnavox: The 9300 Series

Dave-

I've been listening to my 9304 amp for a few months now, after it sitting on a shelf for several years following the most basic redo (new caps and resistors and a matched quad of 6P14P-EV tubes) I must say the sound of this thing even in this form has something I like very much.

So I'd be willing to take it to the next level, including the 565 xformers. Can you (or anybody, really) direct me to the post in this thread that shows the cumulative EFB circuit mods and schematic, up until this point? The only one I see is in post #75, dated Nov 2015, and there's been quite a bit of posting since that point so it's not clear to me if that's the latest and greatest.

If that's the right one, I'll do those first and then the transformers as budget allows.
 
Last edited:
And yet another cookie is awarded. Good assumption Gadget! .........

Input Buffer Stage:

One of the things I disliked about the original and modified designs, is something that never seemed to find the right nitch (for me anyway): Input sensitivity.

The sensitivity level of the original design was such that an input signal of 1.0 vac rms produced full power output. This allowed it to work directly with the output a CD player, but could leave gain wanting with some classical music releases, and in any event, always compromised HF performance somewhat at mid level settings if a passive volume control was installed (which it invariably was) when a basic installation (no active preamp use) was desired. When used with a typical external active preamplifier employing a line stage gain of 20 db (X10), the gain was such that with sensitive speakers, the rushing sound of amplification was usually excessive under quiescent conditions.

With the modified design, input sensitivity was increased such that now only .60 vac rms would produce full power output -- a power level that itself was also significantly increased at that. For those needing more gain in a basic installation then, it helped that issue significantly -- but at the expense of hurting HF performance even more if an input level control was used, as now its setting would typically be at a position producing maximum HF attenuation (unless a very low value level control was used). For those wishing to use an active preamplifier, it made things worse, unless you also installed the compensated input sensitivity reduction network offered with the modified design. This eliminated any loss of HF performance, but still left the input sensitivity a little high for this type of use, returning it to the stock value of 1.0 vac.

Since these amplifiers see such a wide variety of installations then -- and because the results of the OPT upgrade were so promising, it was decided to fix this issue once and for all, by way of adding an input buffer to the amplifier. With this additional modification, now, either fixed resistors or a stereo pot of any practical value can be added to the input of the amplifier to set the gain (or sensitivity) level to whatever requirements might be desired, with no loss of HF response performance, regardless of the setting (if a pot is used), or of the values used (of either the pot or resistors if fixed values are used). With active external preamplifiers, the use of fixed input attenuation resistors is recommended, as then channel balance can be precisely maintained through the use of matched attenuation resistors between the channels. If it is desired to use a level control, any value can be used, although values over 500K serve no purpose, and are generally harder to find with balanced tracking between the sections to boot. In either case however, no compensation is required. In the development model, I have built it (and the schematic is shown) with two matched 470K resistors in each channel to provide 50% attenuation of the input signal, requiring the preamplifier to supply 1.30 vac to develop full power output. This is a power amplifier sensitivity figure I find most universally workable when external preamplifiers of various line stage gain levels are employed.

The beauty of the buffer stage then, is that any level of attenuation can be used, with no compensation required to maintain optimum HF response. With the modified amplifier that still retained the original output transformers, the use of a simple fixed -- compensated -- attenuation network was found as feasible to return the input sensitivity of the modified design back to that of the original design if required. However, because the HF response of the modified design with upgraded output transformers is so good, this approach was found to somewhat compromise the superb HF transient response that could otherwise be had, were no attenuation networks used at all -- compensated, or otherwise. The addition of the input buffer modification eliminates all of these concerns, allowing any level of signal attenuation to be had without any compromise of signal quality, and all without introducing any new concerns. It is a win-win scenario that should even be considered by those who do not wish to upgrade to the new OPTs, but wish to maintain optimum performance from the modified design, when any type of attenuation network is used at the input of the amplifier. A closeup pic of the buffer installation is included. To be consistent with the other small signal tubes of the design, it too is based on the 6EU7 tube.

Next time, I'll wrap things up with final comments, performance results, and pics.

Dave

SAM_1869.JPG
 
Last edited:
Thanks Gadget! Three things would be most helpful:

1. Full power bandwidth.

2. Frequency response.

3. 10 kHz square wave presentation.

Also, is your unit the modified design but simply using the Edcor transformers -- or where other modifications made? If so, what are they, and what are the specs of the transformers you used? Finally, is it still using pentode based operation?

Thanks!

Dave

Will try to get those soonish.

Current circuit is identical to what you posted earlier in the thread with the NFB resistor adjusted to keep the same feedback level off the 16 ohm tap. I'd have to look at the transformer tags to tell you absolutely for sure what model they are. It is running pentode mode currently, the screen taps are tied out of the way.
 
Dave, Does the 6EU7 buffer add any stress to the power transformer considering the 6.3V winding was not spec'd to handle this additional tube?
 
The 93xx amps were made to also run an outboard preamp, as well as an outboard tuner on some models, all off the power amp power transformer. These preamp/tuners had far more tube load for the PT, than a single 6EU7 would present. So, the PT on the amp is well capable of handling another 6EU7, without any distress.

Also, as I remember from a recent conversation with Dave, and as he alluded to above- the tube bias on the output tubes, at idle, is able to be considerably less current than stock and still provide optimal performance, due to the EFB conversion. This also allows for more power transformer "headroom"- in fact, the total B+ current for the amp at idle, even with the added 6EU7, should still be less than stock...

Regards,
Gordon.
 
All Grown Up

What a metamorphosis this Magnavox amplifier has gone through! Performance wise, there's simply no comparison to the original design -- and yet, the basic topology of the original design remains in tact, albeit with a few enhancements. But before I get into the performance results, let me stress that one of the most important things about the amplifier remains unchanged, which is what so many folks love about these amplifiers in the first place: The sound. Oh, its enhanced alright! But rather than considering it enhanced by way of a change in sonic character, consider it enhanced by a shot of steroids.

The upgraded version has significantly improved performance in virtually all areas but one, with that one area intentionally left at the value of the original design. Consider:

1. POWER OUTPUT: This is now 14.0 Watts RMS continuous power output per channel with both channels driven, which is slightly down from the 15.5 watts using the original OPTs of the non-upgraded modified design under the same conditions. This loss is insignificant, and due to the slightly higher primary impedance of the Z-565 transformers, versus that of the original devices. However, the upgraded Maggy will now supply this power level -- continuously in both channels -- from 20 Hz to 20 kHz! This is a performance level that the original OPTs simply cannot touch. And, it is high quality power output too:

2. DISTORTION: Within 1 db of 14.0 watts X2 (both channels driven), THD is < 1% from 25 Hz to 20 kHz. Even at 20 Hz under the same conditions, distortion is ~ 2.5%, which is still amazing performance for the upgraded design.

3. FREQUENCY RESPONSE: This is dead flat (+/- 0 db) from 20 Hz to 20 kHz, with a smooth roll-off on either side of this band. At 40 kHz, response is down just .5 db.

4. STABILITY: This is now Absolute, with no value of capacitance only loading causing any tendency towards oscillation. 10 kHz square waves are presented with a flat top and no ringing, while on the low end, pulse conditions result in very rapid settling -- even with no load of any kind attached.

This all represents very high performance, meaning that within its power class, the so modified and upgraded 9300 takes a back seat to very few comers, regardless of manufacture, and hangs in there with some of the very best 6BQ5 offerings out there, and surpasses most. In one way however, the upgraded version (intentionally) retains an important performance trait of the original design:

5. INTERNAL OUTPUT IMPEDANCE: With the original design, this was clocked at 1.59 Ohms, referenced to the original OPT having a 4 Ohm output winding. With the modified amplifier, this did not materially change, coming in at 1.52 Ohms, again referencing a 4 Ohm output. With the upgraded design, the internal output impedance again remains basically unchanged at 1.68 Ohms on the 4 Ohm output tap, and 3.27 Ohms referencing the 8 Ohm output tap. As a performance specification, this is hardly "high fidelity" like, with the best designs showing less than 1 Ohm internal output impedance, in reference to a 16 Ohm output tap. So what gives?

In an amplifier using the topology of the basic Magnavox design, the internal output impedance (that is, its ability to hold a steady output voltage in the face of a varying load impedance) is based almost entirely on the amount of NFB used. The more feedback used, the lower the internal impedance becomes. A low internal impedance means that an amplifier having that trait can exercise great control over the bass driver cone, to eliminate any "boominess" in the reproduction of bass frequencies. Another term for an amplifier's internal output impedance is its Damping Factor, with a low internal output impedance equating to a high Damping Factor. Sonically however, a very low internal impedance amplifier can tend to suck the life right out of the presentation, leaving the reproduced sound as having a "dry", un-engaging quality. On the other hand, a very high internal impedance amplifier will cause such boominess as to make the bass response sound artificial and bloated, without any definition at all. As with so many things in life then, the answer lies in the middle.

It is the somewhat elevated internal output impedance then that gives the Magnavox amplifiers their endearing listening quality that so many enjoy. With a little bit of internal output impedance, the bass will still be defined well enough, but there was also be a base that the sound builds on or a fullness that makes the sound for many so engaging to listen to. Because this IS a Magnavox amplifier, retaining this electrical quality is the best way to retain the sound that endears so many to it. With that inherent quality retained then, it is the improvement in the other specifications with the upgraded design that gives this version it's "on steroids" performance. On the other hand, if a high damping factor is what your speakers sound best with, then the original Dynaco SCA/ST-35 designs using the Z-565 might be better suited for you.


So, for those who think that the modified and upgraded Magnavox amplifier now has nothing left of its original build but its chassis, choke, and power transformer, understand that throughout all of the modification and upgrade efforts, they all still seek to retain the original essence of its sonic signature, allowing it to remain true to its heritage as a Magnavox console amplifier.

Finally, also note that the output tubes now idle at just 20 mA total cathode current each. This is largely due to the use of EFB for controlling the output stage operating point, but was also helped slightly by the new Z-565 OPTs as well. At this level of current draw, each output tube is dissipating a ridiculously low 6.4 watts at the plate, which amounts to just 53% of the conservative Design Center rating system. With such low power consumption, the power transformer remains just warm after several hours of operation. Tube and component life will be extended immeasurably.

Oh, two other housekeeping notes:

1. Using a good bottom plate will ensure the lowest hum and noise levels, and should be part of any quality rebuild. In the pics, I should a dirt cheap easy way to facilitate one.

2. It won't be long before one of the AK Magnavox fanatics wonders about hooking up the screen taps on the Z-565 to engage UL operation. Go for it, and knock yourself out. I didn't do it, and don't have any real plans to do it. If that is what you want, Dynaco already did that in their models with this transformer. This is a Magnavox thread, and in trying to stay as true to the original topology (and sonic signature) as possible, I have left the the output stage in it's original form, save the application of EFB. I mean no disrespect to those who wish to try it -- but there's only so many hours in the day, so I'll leave that to other inquisitive experimenters.

In the listening room, this amplifier has all the enjoyable listening qualities that define it's humble beginnings. But it can make itself known real quick, handling even the most challenging material with relative ease. I never ever say never, but it would be hard to imagine squeezing more from these little guys while still retaining the essence of what they are. For those who are so inclined to get the last ounce of performance from their Maggy, this one will make your friends stand up and take notice!

Dave

Pics include:

Below: Near textbook displays of 10 kHz square waves: Left on top, Right on the bottom.
SAM_1862.JPG

Below: 20 Hz at 1 db below 14.0 watts RMS in both channels. Distortion is starting to rise as discussed, but this is still excellent performance for this power level, that the vast majority of 6BQ5 offerings cannot touch.
SAM_1866.JPG

Below: At 25 Hz however, distortion at the same power level has dropped to well under 1% (about 0.65% avg).
SAM_1864.JPG

Below: Same power level at 20 kHz. This is exceptionally clean performance for an amplifier of this power class.
SAM_1865.JPG

Below: An easy bottom plate to minimize hum and noise: 3/8 inch plywood cut to size, with aluminum duct sealing tape lining the top, and four rubber feet underneath.
SAM_1870.JPG

Below: A completed underside shot.
SAM_1868.JPG

Below: Those bad boys behind the Magnavox don't scare it at all. Within its power capabilities, it can run with them just fine.
SAM_1874.JPG

Below: The schematic of the modified, upgraded version. The power supply page for the modified version remains the same for the upgraded version, except that the 340 volt supply point now becomes a 320 volt point.
Magnavox 9300 W:Z-565 OPTs.jpg
 
Last edited:
Setting the bias?
On the previous version the schematic had a single 10 ohm resistor for each channel, with instructions to adjust bias for .44VDC across the resistor. The new circuit has a 10 ohm resistor on each tube.

What are the instructions for setting bias on the new circuit? Is it .2V across each 10 ohm resistor? Are the 4 test points instead of 2 to prevent using mismatched pairs?
 
Last edited:
Morrism -- Correct. On the original modified design, a bias value of .44 vdc was specified -- but this was the result of two tubes drawing current through a single common resistor to each channel, and equates to 22 mA per tube. The target figure has dropped slightly to 20 mA per tube, with the four resistors simply allowing each tube to be monitored individually. The bias is set the same way (by adjusting it for an average of 20 mA across all four tubes), with a matched quad always the best tube scenario to have.

I meant to put a note on the schematic that the bias is measured between point "X" as shown on the schematic, and each test point terminal. The X made it on the schematic, but the note didn't. :(

Kirk -- I did not develop one, but some enterprising AKer probably might. Besides the two Z-565 transformers and the LM-337 regulator, everything else is pretty standard stuff that would be available at Antique Electronic Supply, with a full list easy enough to put together.

Thanks for the interest!

Dave
 
I seem to have missed it somewhere, but what was the original PI vs the PI of the new output transformers?
 
The schematic is in fact hand drawn. I'm so hopelessly old-school..... It's all I can do just to get the silly thing posted. I should also say however that I've always favored old school (had drafted) schematics, if only for the human quality they project.

The original OPTs were 7600 Ohm units, while the Z-565s measure 7800+ Ohms. Not a big deal, but enough to change the output slightly.

Dave
 
The schematic is in fact hand drawn. I'm so hopelessly old-school..... It's all I can do just to get the silly thing posted. I should also say however that I've always favored old school (had drafted) schematics, if only for the human quality they project.

I like to draw, so when I see the straight lines, attention to details, cleanliness of a hand drafted (with emphasis on drafted) schematics you post, I'm not only favored to it, I respect and appreciate it as well.
I remember drafting class back in eighth grade.

I better start using a ruler...
 
Hey Dave- would it put anything out of whack, to try either a 5V4 or 5AR4 rectifier? With slightly increased B+ offsetting the increased transformer primary impedance, the amp might just make the same power as with the original output transformers...

Regards,
Gordon.
 
Thread is awesome, educational, and I'm going to put my 9304 under the knife! For cheap guys who don't want to buy new OPTs what to do? My guess would be to follow the later Jan 15 2017 schematic from the new buffer up to the .047 caps. Then follow the earlier Nov 13 2015 schematic from the .047 caps to the Maggy OPTs, For feedback I'd guess the later schematic would be the one to use because of the lower 1.29V at the 6EU7 cathode. Cookie / No Cookie?

scan0002.jpg Magnavox 9300 W-Z-565 OPTs.jpg
 
You'd need the feedback from the earlier schematic since that has been tweaked to match up with the habits of the stock transformer. Different output impedance needs a different value resistor to maintain the same amount of feedback.

Basically marry the 6EU7 buffer stage to the early schematic if you want to run stock iron.
 
You'd need the feedback from the earlier schematic since that has been tweaked to match up with the habits of the stock transformer. Different output impedance needs a different value resistor to maintain the same amount of feedback.

Basically marry the 6EU7 buffer stage to the early schematic if you want to run stock iron.

And truly, there's nothing wrong with that plan. IME, the 93xx has the best and heaviest iron of any of the non-PPP Magnavox units (the ones that only use two output tubes per channel). Many people have greatly enjoyed listening to those with the stock OPTs. With Dave's enhancements, they'd be even better than the stock ones people have been enjoying...

The replacement OPTs are markedly better- but, there's nothing fundamentally wrong with leaving the originals.

Regards,
Gordon.
 
Back
Top Bottom