Magnavox AMP 142 rebuild / troubleshooting

[gadget73]

The pic above the one with all the color is electrically fine. Its a differential amplifier, it amplifies the difference between the two grids. Input signal to both, you get the difference as output. Its electrically equivalent to the one I scribbled out, just with the positions of the 10k and the 0.47uf flipped but that makes no difference. If it got poor balance, thats usually because the tail resistor is too small. Only 10K here, should probably be in the 47k-68k range. Or use a CCS, that also works.

Basically the larger the tail resistor, the better the balance. You can slightly shift the plate resistors to improve the balance but it shouldn't need to be 50% larger to make it happy. After I get around to eating breakfast I have an amp on the bench with an LTP that I need to make some final measurements on. 47k tail, 82k plate resistors on both sides. I'll see how well balanced that is. No feedback on that, its in the middle of a more conventional setup with a voltage amp stage before the inverter.

 

[Tube Radio]

The circuit in color is wired how mine currently is, but apparently it's not right for this particular amplifier given how the square waves look.

Or maybe it only works good with smaller plate resistors than I have.

 

[gadget73]

I suspect its going to provide unequal feedback depending on which side, negative to one triode and positive to the other. Might explain why the drive output looks so strange and why the plate resistors have to be so mismatched to equalize the output.

 

[Tube Radio]

So should I wire it with the 10k going to ground then make the plate resistors 200k and increase the 10k until the outputs are balanced?

 

[gadget73]

I'd probably start with a 47k on the tail and 100k on the plate resistors and see what that gets you.

 

[Tube Radio]

Would 200k on the plates be ok?

I need the gain.

So basically keep it wired as is with the .47uF cap connected where it is and just change the tail resistor, right?

 

[gadget73]

Probably depends on supply voltage, but see what you get.

The other option is a gain stage in front and bring feedback in there. 6c4 would do it, thats half a 12au7.

 

[Tube Radio]

So it's now currently wired like this. Tried removing the compensation cap to see what it would do.





Here's the result.




I then wired it how it was with the 10k tail resistor connected to the .47uF cap with the compensation cap disconnected and here's the result.





Which one looks closest to the textbook example?





Also why did I have to use no compensation cap at all?

 

[gadget73]

The top and bottom part of the wave don't look like they are at the same frequency. Is your generator doing weird things with duty cycle?

ringing looks like it needs some compensation. Is this with or without feedback connected?

also what sort of voltages are you getting?

 

[Tube Radio]

That's with feedback connected.

The generator looks like it puts out a proper square wave.

Testing with a sinewave at full undistorted output at 400Hz it takes 2Vrms input. So the gain has been reduced. Maybe that's why I do not need the compensation cap and could be the issue with the waveforms.

I may use a 10k pot so I can tweak the gain to what I need it to be then select the closest value fixed resistor. I'll then need the compensation cap.

I'll set the input to 1Vrms and adjust the 10k pot until I get full undistorted output. That way the amp at least has its original input sensitivity.

The schematic then becomes this.




Now to adjust the compensation cap I set my dual channel scope up with both probes in 1X mode and the signal generator goes to channel 2 while the amp goes to channel 1. I set the scope up so both waveforms are the exact same amplitude at 1kHz. I then set the generator to 20kHz and adjust the cap until channel 1 (connected across the 4 ohm non-inductive load) is once again equal in amplitude to channel 2. The problem is I now need a slightly smaller compensation cap as the trimmer just is out of its range. Gonna see what trimmers I have and maybe I can ditch the series cap if I have the right trimmer.

I don't know if I'll ever get the squarewave to be textbook.

So I decided to try the more traditional method of compensation which is a cap across the feedback resistor. Using a 1.5-20pF trimmer cap got the sinewaves equal in amplitude at 20kHz.

The schematic then becomes this.





The amp seems to sound a little better in the upper treble now.

I do have a slight problem though.

With the input shorted or open I hear a very slight bit of hissing which I didn't think a tube amp could do and I also hear a slight bit of noise that seems like what one might hear if they move their cellphone real close to some older amplifiers.

So the gain will need to come down some. The tube buffer is really not needed anymore as the preamp itself with one channel seems to drive the amp just fine, however with needing to lower the gain I'll likely need the tube buffer or the DRV134 board. At that time I may replace the 47k resistor between the input and ground with a 50k volume control or a 50k trimpot so that I can adjust the level as the DRV134 board has no level controls.

The issue with the squarewave response is that I need to adjust my generator if it is possible given the squarewave is not 100% symmetrical.

 

[Tube Radio]

Moderator, is it possible to move this topic to the end of the following topic?

Magnavox AMP-142 rebuild

I have a Magnavox AMP-142 that I am going to rebuild. Here's the stock schematic. I'm going to modify it several ways. 1. I will use the following phase inverter, will put an octal socket where the almost punched hole is between the filter caps and use a 6SN7 tube. I've used that phase...

audiokarma.org

 

[gadget73]

thats just going to trim the tail resistor smaller than 10k.

All that ringing says it wants some compensation, usually across the next stage's grid to ground resistor. It shouldn't bounce like that. Quite possibly thats what you are hearing if its at the edge of oscillation.

how much feedback is this getting? You may just need more open loop gain to make this sort of inverter work correctly. Converting it back to a paraphase might also be an option. Those tend to have more gain, and if its floating paraphase they don't have the balance problems that the stock design has.

 

[dcgillespie]

The 200K/300K plate load resistors for the inverter is an unusually large differential. If the top (inverting) tube uses a 200K, then the bottom tube would more likely use a 240K resistor — not 300K, particularly with a High Mu tube like a 12AX7.

Otherwise, I smell an OPT issue……

Dave

 

[Tube Radio]

thats just going to trim the tail resistor smaller than 10k.

The tail resistor is still 10k, I just reduced the value of the feedback divider resistor which increased the gain to where a 1Vrms input produces full output.

All that ringing says it wants some compensation, usually across the next stage's grid to ground resistor. It shouldn't bounce like that. Quite possibly thats what you are hearing if its at the edge of oscillation.

I cannot trust the square wave testing until I get the generator adjusted to where the square wave has a 50% duty cycle provided it has an adjustment for that.

It could indeed be at the edge of oscillation, however I can usually see that when doing testing with a sinewave as I'll normally see something on some portion of the waveform, but a sinewave at any frequency fed through the amp looks 100% clean.

So it looks like I have a self biased long tailed pair inverter.

how much feedback is this getting? You may just need more open loop gain to make this sort of inverter work correctly.

Not sure on the feedback, but an input of 1 volt produces full output.

The 200K/300K plate load resistors for the inverter is an unusually large differential. If the top (inverting) tube uses a 200K, then the bottom tube would more likely use a 240K resistor — not 300K, particularly with a High Mu tube like a 12AX7.

Unless I have an issue with the scope and probes at work and my scope and probes at home, that's the value that got both 12AX7 outputs exactly equal at 400Hz.

I measured the signals at the 6V6 control grids.

I moved the compensation cap to where it was originally and the only hiss now is what's coming from the tube buffer.




The main problem I have is that the Schiit Saga S preamp does not go to full 0 volume by design.

So consequently in order for the audio to be barely audible I need the tube buffer set to a lower output such that I cannot get the full 22 watts out of the amp.

For a 9.8mVrms input to the amp a 400Hz tone is moderately loud and music is at a decent listening level.

The gain of the amp at 400Hz is 11.17.

To hear such a small input signal be at a good listening level the open back speaker I have must be very efficient. That could be why I had to take a few extra steps to ensure the amp is as quiet as possible.

Decided to do a test and a MM phono cartridge connected to the amp will drive it to a decent output level.

I connected one channel of the preamp straight to the amp and I do not need the tube buffer. Now if I feed a stereo signal to the amp I'll need to use summing resistors unless it's possible to short the outputs of the preamp together in passive mode and have no ill effects. Now if I do have to use the summing resistors I will need some gain given the resistors will reduce the level some, however if I keep the preamp amp in active mode I could use 100 ohm summing resistors and only lose 1mV of signal.

What I don't know is if in active mode the preamp uses coupling caps on the output.

In passive mode I don't think it does.

So I need a gain of 1 buffer. I can go ahead and use the DRV134 buffer for that. Will need to either only use the + outputs and leave the - outputs floating as that provides a gain of 1 or I can short the - outputs to ground which provides a gain of 2 and use a voltage divider after the board to get the gain back to 1.

That said I could keep the gain of 2 and use a dual gang 50k volume control on the input so that I can have some way to tweak the signal level.

I fixed the symmetry of the squarewave out of my audio generator. All I had to do was set the output to 400Hz or 4kHz and adjust the time/div on my scope to where the top portion of the waveform fills 1/2 horizontal and the bottom half fills the other half horizontal and adjust the pot to where the end of the top portion of the waveform and the beginning of the bottom portion is right dead center horizontally of the screen.

Here's the result of the squarewave testing at 10kHz. It looks much better. I'm sure the squarewave being non-symmetrical caused some issue as the width would have been different for the top and bottom parts of the waveform. I'd be willing to bet the mismatched 6V6 tubes are causing the issue especially if the 6V6GT and 6V6 GTA have slightly different specs. At some point I may get a matched quad of 6V6 tubes. Who would y'all recommend I use to purchase the quad from?





The maximum undistorted output power at 400Hz is 25 watts.

I also noticed that at full output the milliammeters read the following.

6V6GT 95.9mA
6V6GTA 100.4mA

At idle
6V6GT 69.9mA
6V6GTA 69.2mA

So that tells me that the tubes are just different enough to where there's imbalance in the cathode current at full output. A matched quad of 6V6 will likely make the amp even better as the cathode currents will be near equal at full output.

With a matched quad of 6V6 I wonder if I'll gain any power output.

I also noticed that since I got the compensation cap right it seems like that I have slightly more upper treble.

So for now until I get the matched quad of 6V6 I'll not do anything else to the amp.

Also the DRV134 board will go in the case the tube driver is in now and will replace it. The toroidal power transformer will be used as it has two 12.6V AC windings which will connect to the DRV134 board to create the +/- supplies needed. I'll use a DPST switch to connect both output jacks together for a mono output. That way if I ever find another set of drivers and another amp I can go stereo just by flipping a switch. Also the board is X2 gain from input to balanced output and if the + or - out is shorted to ground the chip is then X2 gain on whichever output is not shorted to ground so I can plug in a balanced or unbalanced amp and have the same X2 gain.

The only thing is I don't like the way the meters are mounted in the round holes left from the analog meters I used. Once I get another Dremel I may try to cut the hole square then cut a piece of metal and cut out the meter holes and use bolts and nuts to hold the panel to the chassis.

 

[gadget73]

Still strange to me that the top and bottom of the square wave aren't the same.

 

[Tube Radio]

The only thing I can figure is it's the 6V6GT and 6V6GTA tubes used. I can swap one 6V6GT with one 6V6GTA so that there's one of each paired up and see if that makes a difference.

I paired the 6V6GT tubes together and the 6V6GTA tubes together as the 6V6GT tubes were closely matched to each other in individual cathode current measured. Same for the 6V6GTA tubes.

If I find it works better with a 6V6GT and 6V6GTA tube paired I might leave it like that until I get a matched quad of 6V6 tubes.

 

[gadget73]

They aren't meaningfully different. The GTA has higher plate voltage and dissipation ratings, though I believe some of that may actually be from different rating systems than actually higher ratings. Design center v design max. The rest of the parameters match and a GTA can sub for a GT.

 

[Tube Radio]

Could them having to be biased slightly differently for the same plate current affect anything?

For the 300k resistor it is made up of a 200k and a 1200k in series. To make 270k I can parallel a 1 meg resistor with the 200k resistor.

Will try that and see if the squarewave improves.

I checked my probes and both channels of my scope by shorting the probe tips together fed from my audio generator. Both waveforms are exactly the same amplitude. So unless the 300k resistor didn't change the signal enough for me to see it, I don't know why 300k was what I needed.

Then again this whole amp has been weird from day 1.

I do have the two chokes that were in two 6V6 plates.

Would there be any benefit to putting those back in?

The 12AX7 plate resistor is 266.3K now. Should be close enough.

Unless my scope and/or probes are only off when connected to the amp, 270k is not the correct resistor and I must use 300k. Partly confirmed by the lower portion of the waveform measured across a 4 ohm load starring to distort the - portion before the + portion whereas with the 300k resistor both + and - portions distorted evenly.

What could cause a 270k resistor to not be enough?

Scope set to add mode.





Channel 2 inverted.




I removed the 1 meg resistor so that the resistance would be 300k and all is well.

I took my Schiit Sys two input volume control and made a summing box out of it.

1. Shorted input 2 right and left channels together.
2. Connected a 100 ohm resistor from input 1 right channel to input 2 and input 1 left channel to input 2.

I now have a mono output on input 2 and a variable mono output on the output. I turned on the active mode of the Schiit Saga S since it has a 75 ohm output impedance.

With the Schiit Saga S set to minimum volume the audio is barely audible from the speaker which is good.

Also given I no longer need the tube buffer I had to reverse the + and - input leads to the speaker crossover given the buffer inverted the signal.

Concerning the DRV134 board I'll install it in the box the tube buffer is in and use it for when I need to convert an unbalanced signal to a balanced signal. Most likely I will use it in my main stereo as I have a USB audio interface which has balanced inputs and my unbalanced sources are near the rear of my 12' X 12' building while the sound card is near my PC at the front of the building. That way I can run a balanced cable and not worry about any hum being picked up by the cable.

 

[Tube Radio]

Been listening to the amp and open baffle speaker for a few days.

Am well pleased so far.

Decided to do a low frequency response test on the Saga S and stereo to mono combiner and the Saga S is flat to 10Hz with 0 phase shift at all.

 

[Tube Radio]

Given how I have the cathodes wired with the top two 6V6 connected to one milliammeter and the bottom two 6V6 connected to one milliammeter, how would I go about reducing IMD by using a common cathode resistor as mentioned in this post of the More Fun With Magnavox: The 9300 Series thread?

More Fun With Magnavox: The 9300 Series

I have a feeling I'm about to feel really stupid... I don't know. The 4.4 Ohm set I was using just look like regular big resistors, but the new ones I just bought are those aluminum ones for plate mounting. I bet they're inductive huh? Could that be the whole issue here with the wonky patterns?

audiokarma.org

Given I have four 6V6 would I use one 10 ohm resistor for all four cathodes connected between both milliammeter negatives and ground or would a 5 ohm resistor be appropriate given it's four cathodes not two?