Mystery between an Old Magnavox 9301 and a Grommes of Similar Architecture

[gadget73]

The stock Magnavox circuit uses a paraphase inverter. The first triode is a bog standard voltage amp section. No special magic to it, the only thing close to odd is the need to complete the feedback loop with either a resistor or the balance pot.

The second triode is also a voltage gain stage, but fed through a voltage divider so that the output level is basically the same as the first triode. Also nothing special here, just the two resistors forming the divider need to be reasonably close to right for decent inverter balance. Voltage divider is R113/R114. If that were failed, you'd have much more signal appearing at the grid of one EL84 than the other.

Try lifting C104 out and see what you get for AC voltage output at that end. I'd expect a good 12+ volts AC of audio signal at that point. If that is good, put it back in circuit and measure it while connected to the EL84 grid. It should not show much change in ability to deliver voltage, though it may need a bit more signal input to get there.

 

[Dioretsa]

All, I cannot believe that I did this but per your suggestions I found that the capacitor from pin 7 of the 6EU7 (not the coupling cap) was not running to ground but instead run to pin 9 on the 6EU7. I disconnected this capacitor and "voila" I am now getting 6 VRMS at the output with 8 ohm load. That still is low but much improved. I believe I copied this mistake from the first Maggie I received which had been modified some already when I received it or I just messed up when I rebuilt them both. Either way....my mistake or copy mistake, this appears to be the problem. I suspect that now I can go back over the NFB making the upgrades/tweaks to increase the gain. I will keep you posted but thank you so all so much. I am headed back to my lab now to "tweak a little"!! 6DZ7 was right on with the 100mmf capacitor comment. Also my confidence is a little higher that I am not incurably dumb and might actually be able to help someone else if one of these Maggies come up in discussion. I AM NOT DECLARING VICTORY JUST PROGRESS! YEAAAA!!!

 

[gadget73]

I'd expect that to be frequency dependant feedback though. 100pf into whatever that cathode resistor is would have to be a very high pass network.

 

[Dioretsa]

No I left out something in my report! This is all making sense now! The capacitors I put in during the rebuild that I accidentally ran to ground were 100 NANO Farad not Pico Farad. I am guessing that that may have passed more NFB into the cathode. Another math error!!! Now I am getting about 8 VRMS at the output before my scope shows clipping (the waveform is squaring at the peak of the wave). This is still not up to where I expected because this is only about 8 watts. But it is way better. I looked at the output of the 6EU7 and the sin wave at 1KHz gets some weird distortion as the input approaches 1.5 VRMS from my wave generator. I assume that is why the output is clipping and squaring off. Does that make sense?

 

[gadget73]

ah, that would explain it a whole lot better.

What sort of a load are you running this into? 8 volts into 8 ohms is 8 watts, which is about right. These only do a little over 10 when properly loaded to 4 ohms.

 

[Dioretsa]

Yes I'm using 8 ohm speaker cabinets that I made from some quite high efficiency speakers in anticipation of using with tube amps. Guess I'll have to make a 4 ohm version! I've been tweaking the NFB loop and I have it so that it is just starting to clip at 8.5 when I supply a 1.3 VRMS signal at 1KHz. I can move the sensitivity around by monkeying with the NFB but I don't know what the affect is on the frequency response. Tonight or tomorrow I am going to test 10 Hz to 50 KHz and see if the response is still OK. I'm just happy to have it working and to have learned so much during this little debacle! It was fun!

 

[Dioretsa]

That is clips at 8.5 VRMS output

 

[Dioretsa]

Does anyone know any tricks for improving the frequency response of these amplifiers?

 

[gadget73]

Yes.

http://audiokarma.org/forums/index.php?threads/more-fun-with-magnavox-the-9300-series.687735/

Lots of background and discussion but if you'd like to skip to the schematic

More Fun With Magnavox: The 9300 Series

 

[Dioretsa]

Another question. I have got the Maggies working fine. But I have some questions. The phase inverter coupling cap that runs to the second 6BQ5 is 0.047 mF cap while the audio amplifier side couples via a 0.0047 mF cap (or 0.0015 mF) depending on the schematic and chassis model. Does anyone know why the phase inverter couples with a cap that is 10 times the capacitance as the other side? I have been experimenting with a resistance decade box and have found that the NFB loop can be modified within reason to provide above 12 VRMS or above output (even closes in on the theoretical 15 VRMS) but the amp starts to lose linearity with regard to frequency. The center frequencies are high and the low end is low. Changing caps helps this but at some point I introduce some distortion when below 35 Hz. Also the amp shows high end distortion beginning above 10KHz. I read the stuff that was sent to me in this thread and this all makes since. The higher the gain, the more pronounced the non-linearity in the center frequencies. Any suggestions or experiments I can run to optimize?

 

[gadget73]

It has to do with the way the phase inverter works, and maintaining stability. Something about the time constants for the two RC networks not lining up in order to keep it from going bonkers.

Much of this is fixed in Dave's re-design. it sorts out many of the shortcomings of the phase inverter, the stability, and the output stage biasing. Its still of course limited by the stock transformers but its definitely an improvement.

 

[Dioretsa]

Continuing questions on the Maggies if anyone is still listening!
Been experimenting with the NFB loop to see range of power possible. I can actually get to the sorta theoretical 15 Watts per channel BUT...anything over about 7 Watts and I'm seeing distortion in the high frequencies. I will put photo of the oscilloscope at about 10KHz. I see what I think may be "parasitic oscillation" that can actually red plate the 6BQ5 tube if allowed to go on long. If I set the NFB loop to a point where the frequency response is flat and the weird oscillation goes away the power output drops. Any experience with this, suggestions or comments? I think I can hear the higher frequency distortion when I play actual program material into the amp when I reduce the NFB to get the power.

 

[gadget73]

Its probably at least partly the phase inverter going unbalanced. Its a problem with the paraphase design, which is part of why Dave converted to a floating paraphase setup. It has better balance. I believe he also had some high frequency instability problems that were resolved with some tuning of the feedback and adding a step network after the voltage amp stage.

 

[Dioretsa]

Its probably at least partly the phase inverter going unbalanced. Its a problem with the paraphase design, which is part of why Dave converted to a floating paraphase setup. It has better balance. I believe he also had some high frequency instability problems that were resolved with some tuning of the feedback and adding a step network after the voltage amp stage.

 

[Dioretsa]

Thank You, Gadget! What is a step network? I am using the 9304-10 NFB setup which is a 3900 ohm resistor followed by a 2200 ohm resistor after the circuit splits off to where the balance control once was (if it ever had one). I dink with the 3900 ohm resistor only. I have had it as high as 6800 ohm. At 6800 ohm the amplifier is quite loud and I can get >12 Watts across a 4.3 ohm load but then the oscilloscope starts to show the HF distortion. Here's a rough plot.

 

[gadget73]

Its the 390 pf cap in series with the 1500 ohm resistor. Basically its just a low pass filter arrangement, high frequency stuff is shunted to ground. In stock form, some had a 100pf cap straight to ground in the same spot. Its doing the same job.

 

[Dioretsa]

Wow! That looks more like the Grommes design (at least the parts I've digested so far). This is VERY interesting. Thanks again!

I think, I learn therefore I am....

 

[Dioretsa]

In the original design the NFB loop comes only from output transformer. In Davids design, the NFB comes also directly from pin 7 (output of 6BQ5) through a 47 (pF?) capacitor. Do you know why? Then the whole biasing system of the 6BQ5s is changed to a voltage controller network with a 5K pot controlling the bias. Not sure I understand all the detail in that. I've never used a variable voltage controller before. I have used the fixed voltage types.The specs look great on David's design. Very slick and smart!. I may try this on the second unit. But I will probably need some more guidance. Maybe I'll try some of the changes. I need to think about it !

 

[gadget73]

Most of it comes from the output transformer, the 47pf cap is for some high frequency stability stuff. If you read through the thread I linked earlier, there is a very in depth explanation of all the whats and whys of it.

The variable regulator is part of the EFB system. The extremely short and sweet of it is that it tracks bias level to screen voltage in order to maintain the voltages relative to one another. Dave could explain this much better I'm sure, but what it ends up doing is decreasing distortion and improving power output all in one go.