I will do the tuning of grid stoppers today to see if it helps... I tried reversing the OPT connections and it goes crazy so they were correct to begin with. It's not isolated to a channel it does the squeak with the volume to zero and even with only the EL84's installed. I will also try to add zobels on the OPT primaries as well to see if it helps... Strange is the noise doesn't come from the speakers but from the amp, seems like from the tubes or one of them at least.
So if any component in a zobel network heats even a little bit that indicates a clear oscillation even though there is no audible component in the speakers.
So if any component in a zobel network heats even a little bit that indicates a clear oscillation even though there is no audible component in the speakers.
Last edited:
So i am to understand it squeeled loudly or not? And all the time or was it only intermittent? I am trying to see if it applies to my situation. In theory at least, the way the OPT are wired now the feedback is negative as it should be. And since they squeal sometimes even on their own i take it it's not a case of accidental positive feedback. If i unhook the feedback it sounds dreadful, the Zout goes through the roof so the sound degrades. And of course the occasional squeal is still in place.
@ blu_glo yes, naturally they are soldered right on the tube sockets, in fact their lead there is only 1-2mm i was actually afraid the end connection would come off due to the soldering heat... Their other leads are about 20mm however as everything is "in the air" built.
Will get back after modifications. Also the switching transistor is badly chosen for a linear regulator. Is it possible it oscilaltes too? High frequency components tend to produce their own high frequency if they're not given any...
And that sounds great but not with the "classic" 8K p-p Afriend even recommended 10K in order to go beyond manufacturer specs.
More to the point, (even though they oscillate on their own it seems with only the bias voltage as input) would flip flopping the OPT connections bring a danger of destroying the amp? I'm willing to experiment with that further if you people think the feedback might be out of phase. Although i briefly tried that and it squealed audibly (not just the tubes).
And do the screen stoppers also have to be soldered very close to the socket?
More to the point, (even though they oscillate on their own it seems with only the bias voltage as input) would flip flopping the OPT connections bring a danger of destroying the amp? I'm willing to experiment with that further if you people think the feedback might be out of phase. Although i briefly tried that and it squealed audibly (not just the tubes).
And do the screen stoppers also have to be soldered very close to the socket?
Last edited:
The zobels go on the grid circuit, there is a difference. Input versus output.
This is going to be a beginner question but, i know how zobels are calculated for working in conjunction with the speaker inductance, impedance etc. but in this case i don't much know how i would size them.
In the meantime i am installing 5.6K grid stoppers, moving the screen stoppers on the socket itself and making them slightly larger and installing plate stoppers of 100R too. Also replacing cables from the output connectors to feedback with shielded type since they pass near other cables with audio signal. I think i will also replace the flimsy transistor with a simple resistor and two 150V avalanche diodes in series and see what happens. Should be stable enough.
Hopefully this exercise of frustration will soon be over.
In the meantime i am installing 5.6K grid stoppers, moving the screen stoppers on the socket itself and making them slightly larger and installing plate stoppers of 100R too. Also replacing cables from the output connectors to feedback with shielded type since they pass near other cables with audio signal. I think i will also replace the flimsy transistor with a simple resistor and two 150V avalanche diodes in series and see what happens. Should be stable enough.
Hopefully this exercise of frustration will soon be over.
The manufacturer's claim (this one guy from Italy) who builds professional Hi-Fi amplifiers claims these transformers can even reach into the RF domain (but no clear specification) so yes Zobels on the 440K grid resistors are a great idea. I will do just that and try not to make the amplifier sound bad with muffled highs.
Can hardly wait to receive this type of relic oscilloscope (dirt cheap but very well maintained) in about 3 days and see where the oscillation originates from (if there will still be an oscillation).
Oscilloscope BEM 009 Equipment MBLE, Manufacture Belge ... -
Also done more reading about tailoring the feedback compensation capacitor for after i am rid of oscillating power tubes, some articles i found right here in the forum
so once again, glad to be a member.
Oscilloscope BEM 009 Equipment MBLE, Manufacture Belge ... -
Also done more reading about tailoring the feedback compensation capacitor for after i am rid of oscillating power tubes, some articles i found right here in the forum
so once again, glad to be a member.Finally gave up as the scope shows only that i cannot stop the oscillation i can only lower it's frequency so i'll try something more typical designed by engineers not by me.
http://s27.postimg.org/4gkeurpvn/masco.jpg
The output stage here is straightforward and the simulation shows it does exactly what it claims. At least the simulation works... although somehow i think the real life version will not. My luck.
Then again it might since the original problem was that with that much voltage difference between plate and screen a EL84 tube would probably have too much gain, hence the auto-oscillation would readily happen.
As i've read here in a thread don't know which: problem 1-excessive gain; problem 2-phase error is too great in the global NFB. Since it would not globally oscillate just the EL84 then the problem must be gain.
http://s27.postimg.org/4gkeurpvn/masco.jpg
The output stage here is straightforward and the simulation shows it does exactly what it claims. At least the simulation works... although somehow i think the real life version will not. My luck.
Then again it might since the original problem was that with that much voltage difference between plate and screen a EL84 tube would probably have too much gain, hence the auto-oscillation would readily happen.
As i've read here in a thread don't know which: problem 1-excessive gain; problem 2-phase error is too great in the global NFB. Since it would not globally oscillate just the EL84 then the problem must be gain.
Thank you very much! and not a moment too soon, already had started dismantling it for modification. I did not say giving up completely, just that i should try a more classic schematic and try to somehow lower the voltage of the transformer with a choke. But that would be a bit of a waste since the output trannys can take 20W with a flat response from 20Hz and more if the F-3db is acceptable higher for example 60Hz (which is acceptable for what i usually listen). What i know about them is only this:F-3db at 20W = 20Hz-30KHz, the N=32 , Z p-p 8Kohm, Zout is 8ohm no ultralinear taps in primaries.
But as it is i will for the moment only make connections with solid 1.5mm wire and more orderly in case you or some of the other tube enthusiasts here come up with any idea other than what i've already tried and failed. I am eliminating the star grounding partially since there were too many wires and hard to work inside the amplifier.
But as it is i will for the moment only make connections with solid 1.5mm wire and more orderly in case you or some of the other tube enthusiasts here come up with any idea other than what i've already tried and failed. I am eliminating the star grounding partially since there were too many wires and hard to work inside the amplifier.
Last edited:
This is a very good first step. Star ground is more difficult to do right. For now, better do "earth follows signal" with ONE connection to chassis from each channel (at the inputs) .
cjkpkg sorry for being a bad student here but it's all dismantled including power supply. Only EL84 sockets left right now for testing purposes but modified power supply far from ready.
There are many things on my mind right now i'm not sure which is more important. Ok so regarding phase issues, for testing that i changed the grid resistors on EL84 to just 10K and disconnected from every other stage, still oscillating. At this point i also shaved off socket terminals that are NC and awe and behold - not all are NC in fact it's a lie. pins 2 AND 1 are control grid and a control grid pin left open is bad mojo can pick up many nasty signals. But that didn't help either. If it was a phase issue it would be simple to tailor it's FB or just add capacitance on top of the Miller capacitance of the first tube in the inverter until it killed the offending frequency (like that Masco schematic i posted earlier).
And i must admit i am a bit jealous of SY here for his achievement with standard transformers, but at this point given the amazing "preview" i've had from this amp i'm willing to settle for the very standard setup and give up the quest for endless power.
There are many things on my mind right now i'm not sure which is more important. Ok so regarding phase issues, for testing that i changed the grid resistors on EL84 to just 10K and disconnected from every other stage, still oscillating. At this point i also shaved off socket terminals that are NC and awe and behold - not all are NC in fact it's a lie. pins 2 AND 1 are control grid and a control grid pin left open is bad mojo can pick up many nasty signals. But that didn't help either. If it was a phase issue it would be simple to tailor it's FB or just add capacitance on top of the Miller capacitance of the first tube in the inverter until it killed the offending frequency (like that Masco schematic i posted earlier).
And i must admit i am a bit jealous of SY here for his achievement with standard transformers, but at this point given the amazing "preview" i've had from this amp i'm willing to settle for the very standard setup and give up the quest for endless power.
An externally hosted image should be here but it was not working when we last tested it.
Photo a bit low quality. Still not much to look at but will be much tidier than before i hope. Trying the typical schematic then if it will be stable only then i'll try to get the full 17W of EL84 at 300V
http://s23.postimg.org/dbsc6fo2j/ampl.jpg
edit: actually i increased the C16 to 39pF the audio effect will be inaudible but it should make the amp more stable with lower bandwidth.
http://s23.postimg.org/dbsc6fo2j/ampl.jpg
edit: actually i increased the C16 to 39pF the audio effect will be inaudible but it should make the amp more stable with lower bandwidth.
Last edited:
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.