Hi gents, I would like to learn about negative feedback in order to tweek my vintage magnavox tube amp. I appreciate your help, otherwise it will be long process.
I have had this magnavox vintage amp, which was part of radio console, for a long time, and when I got it, it was already old. I got it from ebay, do not know the model, I stripped the tuner section, and used it as SE amp. Always sounded lowly. I replaced the power supply caps, tubes, and so on...but over the years sockets and other stuff just got worse, so I decided to rebuilt it. For the purpose to use it as headphone amplifier and to power small desktop fullrange.
So I stripped all the parts I could use, boy they make it like a tank, some resistors are soldered directly to steel plate, there is no way I can unsolder this. No problem, as the most important part, output transformers and power trafo are easily removed. I am using new sockets, new filtration capacitors, and I will use solid state for rectifications. Its just easier.
I want to tweek it for best performance, while it is as breadboard, before I put it in the box. Here is where I need your help. Its playing nicely, but I do not have any distortion analyzer or spectrum analyzer. I have signal generator and osciloscope, so I can measure frequency response and visualy the sine wave.
I have found quite a few negative feedback values. From single 6k8, two 6k8 in series, one 20k ohm, and to comples two resistors with one cap parallel to one of those...
So I do I approach this?
Next post is few schematics as a start.
Thanks, ed
I have had this magnavox vintage amp, which was part of radio console, for a long time, and when I got it, it was already old. I got it from ebay, do not know the model, I stripped the tuner section, and used it as SE amp. Always sounded lowly. I replaced the power supply caps, tubes, and so on...but over the years sockets and other stuff just got worse, so I decided to rebuilt it. For the purpose to use it as headphone amplifier and to power small desktop fullrange.
So I stripped all the parts I could use, boy they make it like a tank, some resistors are soldered directly to steel plate, there is no way I can unsolder this. No problem, as the most important part, output transformers and power trafo are easily removed. I am using new sockets, new filtration capacitors, and I will use solid state for rectifications. Its just easier.
I want to tweek it for best performance, while it is as breadboard, before I put it in the box. Here is where I need your help. Its playing nicely, but I do not have any distortion analyzer or spectrum analyzer. I have signal generator and osciloscope, so I can measure frequency response and visualy the sine wave.
I have found quite a few negative feedback values. From single 6k8, two 6k8 in series, one 20k ohm, and to comples two resistors with one cap parallel to one of those...
So I do I approach this?
Next post is few schematics as a start.
Thanks, ed
Just few more things, I do not care much about its signal sensitivity, so if negative feedback reduces the output signal, fine, I have strong signal in...
I do not care much about flat frequency extension to 10Hz...30Hz is enough. This amp will never need to deliver deep low frequencies to the woofer. For headphones 30Hz is just enough. Btw, I used it before, with simple two resistor split, it works great, beats solid state headphone amps easily.
While I do care about single ended tube sound, I do not want to go overboard. I want clean smooth sweet sound.
So here are some schematics to start with. Its hard to say what original schematics was, as it was modified so many times. I am however sure it had this complex feedback with two resistors and one cap. It seems to me its there to shape the fr response, almost like in riaa preamp...
Anyway, righ now I have 40k resistor there and it sounds good. I even tried open, and it worked great as well. With one 6k8 resistor it motorboated. I though it would be other way around.
Any suggestions?
Thanks
I do not care much about flat frequency extension to 10Hz...30Hz is enough. This amp will never need to deliver deep low frequencies to the woofer. For headphones 30Hz is just enough. Btw, I used it before, with simple two resistor split, it works great, beats solid state headphone amps easily.
While I do care about single ended tube sound, I do not want to go overboard. I want clean smooth sweet sound.
So here are some schematics to start with. Its hard to say what original schematics was, as it was modified so many times. I am however sure it had this complex feedback with two resistors and one cap. It seems to me its there to shape the fr response, almost like in riaa preamp...
Anyway, righ now I have 40k resistor there and it sounds good. I even tried open, and it worked great as well. With one 6k8 resistor it motorboated. I though it would be other way around.
Any suggestions?
Thanks
Attachments
Hi adason,
Starting with a 6EU7 is extremely nice. That is an expensive tube these days. It is a 12AX7A improved for lower noise and hum. The circuit you are showing would be expected to have a 12AX7A instead. Anyway, just so you know what you have in that signal tube.
I'm not sure how low that single-ended transformer will go. They have to be large for LF response. At low levels I imagine you'll be fine for low frequency cut-off.
Lower feedback generally is more stable. Don't forget that the phase shift from the output transformer adds to all the rest, so you have less margin than you think you do. That's why tube circuits often had low values of feedback.
-Chris
Starting with a 6EU7 is extremely nice. That is an expensive tube these days. It is a 12AX7A improved for lower noise and hum. The circuit you are showing would be expected to have a 12AX7A instead. Anyway, just so you know what you have in that signal tube.
I'm not sure how low that single-ended transformer will go. They have to be large for LF response. At low levels I imagine you'll be fine for low frequency cut-off.
Lower feedback generally is more stable. Don't forget that the phase shift from the output transformer adds to all the rest, so you have less margin than you think you do. That's why tube circuits often had low values of feedback.
-Chris
The lower resistor value allows more fb . Excessive fb will cause instability or motorboating. If you are copying a circuit it is best to use the same or similar fb value resistor that is in the circuit. You should not go with a lower value resistor as that can cause problems. If you are breadboarding you can put a pot in place of the resistor and have adjustable feedback. Playing with the control can help you dial in the level of fb that is acceptable to your ears. cheers.
Thanks both for reply. I learn something every day.
First, i am using testing speakers, read 'crappy'. I am not going to put it in the main system and experiment on it, especially when it can motorboat occasionally. So judging by the sound itself is difficult. I hooked up headphones, and noticed slight humm. Not noticeable on small bookshelf. Once i address hum, i will be good to experiment with feedback.
My plan is to install that feedback with two resistors/one cap and use pot for one resistor. Dial highest to lowest resistance and measure fr response and sound...that's all can do for now.
I do have about 3-4 of those 6eu7s, did not know their value. Good to know. One is such that when you turn the power on, it lights up brightly, then dims to normal heater glow. I was told its normal.
First, i am using testing speakers, read 'crappy'. I am not going to put it in the main system and experiment on it, especially when it can motorboat occasionally. So judging by the sound itself is difficult. I hooked up headphones, and noticed slight humm. Not noticeable on small bookshelf. Once i address hum, i will be good to experiment with feedback.
My plan is to install that feedback with two resistors/one cap and use pot for one resistor. Dial highest to lowest resistance and measure fr response and sound...that's all can do for now.
I do have about 3-4 of those 6eu7s, did not know their value. Good to know. One is such that when you turn the power on, it lights up brightly, then dims to normal heater glow. I was told its normal.
Hi adason,
Yes, the heater will glow brightly and then dim. Those 6EU7 tubes were designed specifically for mic and phono amplifier front end amplification. They even moved the heater lead position to reduce hum pickup. So this tube type is a no-holds barred type of low noise tube. Bogen used to use them a lot.
Drag out some 12AX7A's for this amplifier, or maybe even a 6AV6 and ignore the diode connections (tie them to the cathode). This is 1/2 of a 12AX7A, so if you only need one section, or want better stereo separation (okay, it's extreme), use the 6AV6. Note that these are usually dirt cheap as well.
-Chris
Yes, the heater will glow brightly and then dim. Those 6EU7 tubes were designed specifically for mic and phono amplifier front end amplification. They even moved the heater lead position to reduce hum pickup. So this tube type is a no-holds barred type of low noise tube. Bogen used to use them a lot.
Drag out some 12AX7A's for this amplifier, or maybe even a 6AV6 and ignore the diode connections (tie them to the cathode). This is 1/2 of a 12AX7A, so if you only need one section, or want better stereo separation (okay, it's extreme), use the 6AV6. Note that these are usually dirt cheap as well.
-Chris
Hi Chris, so what would be the rational to replace this nice 6eu7 will 12AX7A? I already have about four of those 6eu7s, I might as well use them. This hobby is not about saving money. If it was, we would not be doing it.
Can you explain what the capacitor role is in that feedback, and how its value changes the resulting fr response? Thanks.
Can you explain what the capacitor role is in that feedback, and how its value changes the resulting fr response? Thanks.
I had one of these 6EU7 6BQ5 stereo SE amps when I was a teen - as I remember it sounded quite good. Since you have the tubes I would run them. (Schematic was 3rd from left - solid design)
I used very efficient 6" Lesa FR drivers in small open back boxes, was pretty happy with that set up for several years, played lots of Pink Floyd, Deep Purple, etc. Sounded good to me. (Garrard SP-25 MKIV TT with Goldring cart, and homebrew RIAA amp)
I used very efficient 6" Lesa FR drivers in small open back boxes, was pretty happy with that set up for several years, played lots of Pink Floyd, Deep Purple, etc. Sounded good to me. (Garrard SP-25 MKIV TT with Goldring cart, and homebrew RIAA amp)
Improving the Magnavox 8600
Dave Gillespie did an excellent thread at Audiokarma on rebulding/improving a magnavox 8600 with instructions that average diyers can follow.
The improvements are astounding. He starts with an amp that is only producing 1.35 watts per channel before the onset of heavy distortion and finishes with an amp that puts out 4 watts per channel with much lower distortion.
This is the thread title "Magnavox Flea Power: Getting More Out Of The 8600 Series - A Lot More!"
Steve
Dave Gillespie did an excellent thread at Audiokarma on rebulding/improving a magnavox 8600 with instructions that average diyers can follow.
The improvements are astounding. He starts with an amp that is only producing 1.35 watts per channel before the onset of heavy distortion and finishes with an amp that puts out 4 watts per channel with much lower distortion.
This is the thread title "Magnavox Flea Power: Getting More Out Of The 8600 Series - A Lot More!"
Steve
Hi ArcticBrew, thanks for the tip, however, as I am not a member, I can only read the text, not see the pictures. I may join to see what is he doing.
In the mean time, I measured frequency response. With no feedback, with 110k ohm, 43k ohm, and 20k ohm feedback resistor. It measures (and sounds) best at no feedback, or just 110k ohm. It distorts at 20k ohm feedback. And motorboats at 6k8. First of all, it measures pretty badly, that is...its not flat at all. How can something measuring so poorly sound so good. I should measure it in dB, I know, I just took the average voltage from the oscilloscope.
In the mean time, I measured frequency response. With no feedback, with 110k ohm, 43k ohm, and 20k ohm feedback resistor. It measures (and sounds) best at no feedback, or just 110k ohm. It distorts at 20k ohm feedback. And motorboats at 6k8. First of all, it measures pretty badly, that is...its not flat at all. How can something measuring so poorly sound so good. I should measure it in dB, I know, I just took the average voltage from the oscilloscope.
Attachments
Hi adason,
Well, if you can get 4 wpc out of this, leave the 6EU7 tubes in there. The reason I suggested the 12AX7A is because it has the same characteristics and can be run from a 6.3 V heater supply. Additionally, these tubes are easy to find just about everywhere tubes can be found. The 6EU7 is a specialized tube for low noise, lower than the other 12AX7 clan. It is more expensive to replace and harder to find out in the wild. It would sound identical to the 12AX7a. Save the 6EU7 for a phono amp project. It just seems like a massive waste to throw those tubes into a power amplifier where it's very low noise isn't being used. I was very surprised to see it was being used in that amp to begin with. They belong in a nice preamplifier.
Feedback. Your curves don't look right at all. Feedback will progressively flatten your frequency response at it is increased. The main component that determines how much feedback you are working with is the resistor. The capacitor in parallel with that resistor sets the high frequency roll-off point where the response should begin dropping. The larger value of that capacitor will lower the frequency and reduce the phase margin. What you really want to do is have the gain fall to less than unity before the frequency that the phase shift makes the negative feedback into positive feedback (from the value of the resistor). The capacitor works with the resistor in setting this frequency, so you can't just change one without calculating the proper new value for the capacitor in the feedback network. The other very useful trait of feedback is that it lowers the apparent output impedance of the amplifier as well. That increases the damping factor and is one reason why the response curve becomes flatter and extends flat to higher and lower frequencies. As the feedback is increased beyond the ideal range you will begin to see a peak forming at the high frequency end of your frequency response plot. Eventually it will increase to a point where the amplifier oscillates at that frequency.
-Chris
Well, if you can get 4 wpc out of this, leave the 6EU7 tubes in there. The reason I suggested the 12AX7A is because it has the same characteristics and can be run from a 6.3 V heater supply. Additionally, these tubes are easy to find just about everywhere tubes can be found. The 6EU7 is a specialized tube for low noise, lower than the other 12AX7 clan. It is more expensive to replace and harder to find out in the wild. It would sound identical to the 12AX7a. Save the 6EU7 for a phono amp project. It just seems like a massive waste to throw those tubes into a power amplifier where it's very low noise isn't being used. I was very surprised to see it was being used in that amp to begin with. They belong in a nice preamplifier.
Feedback. Your curves don't look right at all. Feedback will progressively flatten your frequency response at it is increased. The main component that determines how much feedback you are working with is the resistor. The capacitor in parallel with that resistor sets the high frequency roll-off point where the response should begin dropping. The larger value of that capacitor will lower the frequency and reduce the phase margin. What you really want to do is have the gain fall to less than unity before the frequency that the phase shift makes the negative feedback into positive feedback (from the value of the resistor). The capacitor works with the resistor in setting this frequency, so you can't just change one without calculating the proper new value for the capacitor in the feedback network. The other very useful trait of feedback is that it lowers the apparent output impedance of the amplifier as well. That increases the damping factor and is one reason why the response curve becomes flatter and extends flat to higher and lower frequencies. As the feedback is increased beyond the ideal range you will begin to see a peak forming at the high frequency end of your frequency response plot. Eventually it will increase to a point where the amplifier oscillates at that frequency.
-Chris
Thanks Chris, yup, something is not right, that's why I said I do not understand that feedback, it should be the other way around. I better check if everything is connected right. But I connected everything as it was in the original amp...hmm.
Right now I am quite busy at work (started new job), perhaps over the weekend.
Thanks everyone!
Right now I am quite busy at work (started new job), perhaps over the weekend.
Thanks everyone!
Gentleman! As I came home and was checking it twice...kidding aside, I found no mistake in circuit, besides, I am listening to it and it sounds fantastic...but, but I realized those fr responses I posted were measured using small 6 ohm fullrange speaker and not using true resistor! So that could most likely explain uneven fr response, right? When I will find some time, I will measure again using resistor, 8 ohm, 20 watts or so. Sorry it did not occured to me earlier.
(still it does not explain why more feedback made things worse...)
Another thing occured to me which could perhaps explain the behaviour of feedback. What if i am actually applying positive feedback? Perhaps i switched secondary wires. Is it possible?
Output Transformer Impedance
Did you measure the impedance of the output transformers?
I have seen specs that say they are 5,000 ohms into a 3.2 ohm load. Dave replaced the transformers with ones designed for an 8 ohm load so they would be more compatible with modern speakers.
There are variations among these amps based on the outputs that Magnavox put into the consoles.
I have the 6V6 stereo bi-amped version and high frequenct OTs are for 4 ohm output. I put a pair of Heil AMTs (3.7 ohm measured resistance) on the outputs and get ~4 watts at clipping. Response is -3db at 45khz.
Steve
Did you measure the impedance of the output transformers?
I have seen specs that say they are 5,000 ohms into a 3.2 ohm load. Dave replaced the transformers with ones designed for an 8 ohm load so they would be more compatible with modern speakers.
There are variations among these amps based on the outputs that Magnavox put into the consoles.
I have the 6V6 stereo bi-amped version and high frequenct OTs are for 4 ohm output. I put a pair of Heil AMTs (3.7 ohm measured resistance) on the outputs and get ~4 watts at clipping. Response is -3db at 45khz.
Steve
The last post in this thread is the impedance measurement I referenced. 40.3 turns ratio on the 6BQ5 SE amp output transformers.
http://www.diyaudio.com/forums/tubes-valves/169676-magnavox-console-speaker-impedances.html
Steve
http://www.diyaudio.com/forums/tubes-valves/169676-magnavox-console-speaker-impedances.html
Steve
hi friends, I am not ignoring you, I just have no time right now, but weekend is slowly coming...
I have just measured the amp with 4 ohm 20 watts resistor, at 0.5 volt output set first at 1 kHz for all the feedback types...this is what I got.
Off course this is after I switched the output wires 🙂 My bad. Now the feedback behaves exactly as it should. Frenkly no feedback, or strong feedback, the amp sounds great to me. Yes, some feedback will be used, I will further experiment, but I am slowly starting thinking about putting it into permanent box.
Now, how about that low frequency hump?
Off course this is after I switched the output wires 🙂 My bad. Now the feedback behaves exactly as it should. Frenkly no feedback, or strong feedback, the amp sounds great to me. Yes, some feedback will be used, I will further experiment, but I am slowly starting thinking about putting it into permanent box.
Now, how about that low frequency hump?
Attachments
- Status
- Not open for further replies.