Hi
Suppose we have an OPT with 0,4 and 8 ohm taps.
In the most common designs the NFB is taken from the 8 ohm tap.
But what happens when you are using a 4 ohm speaker and is connected to the 4 0hm tap?The ammount of the NFB remains constant?Or you are using an other ammount of NFB connected to the 4 ohm tap?
And what happens when the OPT has only a 6 ohm tap?
Suppose we have an OPT with 0,4 and 8 ohm taps.
In the most common designs the NFB is taken from the 8 ohm tap.
But what happens when you are using a 4 ohm speaker and is connected to the 4 0hm tap?The ammount of the NFB remains constant?Or you are using an other ammount of NFB connected to the 4 ohm tap?
And what happens when the OPT has only a 6 ohm tap?
The feedback doesn't change under the circumstances you're describing. In theory, you could take the feedback from the 4 ohm terminal, adjusting the size of the feedback resistor to keep the gain the same as it was when taking feedback from the 8 ohm tap. There might even be an advantage, albeit a pretty small one.
Hi SY
All these thoughts about FB started when I change tubes (the old ones where 5-6 years old) and speakers (the old ones where 4 ohm 3-way and the new 4 ohm 2-way,so it's a different load).
And now I have a problem with my amp.Especially with the left monoblock.
When I turn it too loud it starts to oscillates a little (some pops) and when I turn it lower it doesn't stop immediately even if I turn the pot at 0.
I don't now what's going wrong.I have built this amp about 6 years ago and I didn't have any problem before.
The only thing that I can remember is that at the beginning I had a problem with the FB.But I don't remember what I did exactly.
I think I'll calculate the FB again.
All these thoughts about FB started when I change tubes (the old ones where 5-6 years old) and speakers (the old ones where 4 ohm 3-way and the new 4 ohm 2-way,so it's a different load).
And now I have a problem with my amp.Especially with the left monoblock.
When I turn it too loud it starts to oscillates a little (some pops) and when I turn it lower it doesn't stop immediately even if I turn the pot at 0.
I don't now what's going wrong.I have built this amp about 6 years ago and I didn't have any problem before.
The only thing that I can remember is that at the beginning I had a problem with the FB.But I don't remember what I did exactly.
I think I'll calculate the FB again.
Yes, you may have an issue with coupling or decoupling caps as dhaen says. It can also be an instability problem if the tubes have a higher gain than the old ones- your description sure sounds like motorboating. This is caused by low frequency coupling through the power supply or instabilities from too high of an open loop gain interacting with LF poles which are not staggered enough. Fixing that will involve replacing and/or increasing the size of the decoupling caps and possibly tweaking coupling cap values. Or decoupling some of the stages separately. If you have a schematic you can post, maybe we can suggest some specifics.
I changed all the electrolytics and the problem doesn't exist any more.
But I'd like to calculate the FB again.
How much is the perfect value of FB?
I read in Menno van der Veen's book (Plitron) that with an ammount bigger than 15db of FB,the amp doesn't sound nice.
But which ammount is the perfect?
Second problem:The input sensitivity depends on the FB.So
when Icalculate the FB I had to know how much I'd like to be the input sensitivity.Morgan Jones suggests 2Vrms.But I think it's too big.In different books I see different values!I have seen amps with 500mVrms,380mVrms and 1Vrms.
But I'd like to calculate the FB again.
How much is the perfect value of FB?
I read in Menno van der Veen's book (Plitron) that with an ammount bigger than 15db of FB,the amp doesn't sound nice.
But which ammount is the perfect?
Second problem:The input sensitivity depends on the FB.So
when Icalculate the FB I had to know how much I'd like to be the input sensitivity.Morgan Jones suggests 2Vrms.But I think it's too big.In different books I see different values!I have seen amps with 500mVrms,380mVrms and 1Vrms.
The rule of thumb is to use 3-10dB for triode amps and 10-30dB for pentode amps.
The problem with using small amounts like 10dB on an amp which tends to produce a pretty good distortion band, is the resulting output is lower in overall distortion, but the harmonic spectrum is much worse. Triode amps produce primarily 2nd H, but it will be multiplied (3rd, 4th..) by using shallow amounts of NFB. The alternative, using none, leads to unacceptable levels of distortion (>1%) at loud levels. Using too much masks the triode sound.. (This of course means that if you must have triode, you need big fat expensive amps with like 15 or 30W output. Mmmm, I can just *FEEL* the global warming... )
Tim
The problem with using small amounts like 10dB on an amp which tends to produce a pretty good distortion band, is the resulting output is lower in overall distortion, but the harmonic spectrum is much worse. Triode amps produce primarily 2nd H, but it will be multiplied (3rd, 4th..) by using shallow amounts of NFB. The alternative, using none, leads to unacceptable levels of distortion (>1%) at loud levels. Using too much masks the triode sound.. (This of course means that if you must have triode, you need big fat expensive amps with like 15 or 30W output. Mmmm, I can just *FEEL* the global warming...
)Tim
Pjotr said:
What I came across in the literature from the last century is that NFB of more than 10 dB to 20 dB makes your amplifier dull and lifeless sounding. Regarding tube amps I can confirm that from own experience. Dunno what’s the magic behind this but it is what it is.
It's not hard to find all sorts of silly and unsubstantiated things if the "literature" you're searching is stuff like Absolute Sound and Stereophile. Put your hand over your wallet, smile, back slowly away...
I'm curious- have you ever heard a Futterman amp?
I'm not Steve. I'm much prettier.
I missed the JAES and JASA papers showing that feedback does all those horrible things you mentioned.
The Futterman amps, long reputed to have good sound, have huge levels of feedback.
Part of the origin of the feedback Urban Legends is the problems of stabilizing voltage feedback loops taken around output transformers. Get the open loop gain of the amp too high and it will be a bear to stabilize. But keep the feedback on the primary side, or use nested loops, or compensate sensibly, and high levels of feedback can indeed work very well.
And in a direct hit on the low-feedback philosophy, Baxandall showed many years ago that using 6-10 dB can be worse than none at all- you increase high order distortion in an absolute (rather than relative) sense.
I missed the JAES and JASA papers showing that feedback does all those horrible things you mentioned.
The Futterman amps, long reputed to have good sound, have huge levels of feedback.
Part of the origin of the feedback Urban Legends is the problems of stabilizing voltage feedback loops taken around output transformers. Get the open loop gain of the amp too high and it will be a bear to stabilize. But keep the feedback on the primary side, or use nested loops, or compensate sensibly, and high levels of feedback can indeed work very well.
And in a direct hit on the low-feedback philosophy, Baxandall showed many years ago that using 6-10 dB can be worse than none at all- you increase high order distortion in an absolute (rather than relative) sense.
I think my question is a difficult question.
OK from where to start.
Right now my amp has about 12db FB.What to look for to see if it's OK.
I can do measurements only with a signal generator and an oscilloscope.
My problem for the input sensitivity is that I don't want to exist 1Vrms 'cause I'm planning to build Frank's buffer line stage.And I'm hearing only vinyl. The output of my phono with gain=70 is about 300mV.
Unless if I build a line stage with a little gain !
Honestly, I wouldn't worry about how much feedback you have, I'd look at what closed-loop gain you're targeting and get the amp stable at that gain. By "stable", I mean into a real load, not just a test resistor. Then you can go back, measure the open loop gain, and calculate how much feedback you ended up with.
SY,
I have some questions about the above post #13. In my Mark III copies we've talked about, with the Dynaco feedback circuit intact I have distortion, removed, the distortion decreases. I don't have a 16 ohm tap so I was taking the feedback from the 8 ohm, and then the 4 ohm. As you mentioned I have less voltage because of the taps I'm using. Could it be that I have too little feedback?
You also mention taking the feed back from the primary side? Would that not be in phase? And what are "nested loops?"
The Dynaco uses a 680 ohm resistor to ground. Then a 1k resistor with a 750 pico fared cap in parallel. Do these resistors act as a voltage divider? The feedback is then introduced between another 680 ohm cathode resistor and a 47 ohm cathode/shunt resistor to ground. The preamp design I'm using has the feedback introduced directly into the cathode of the preamp tube with a 280 ohm cathode resistor on the ground side.
If I wanted to increase the feedback to the level of the original, Would I increase the the value of the first 680 ohm resistor to ground in the feedback loop?
I have some questions about the above post #13. In my Mark III copies we've talked about, with the Dynaco feedback circuit intact I have distortion, removed, the distortion decreases. I don't have a 16 ohm tap so I was taking the feedback from the 8 ohm, and then the 4 ohm. As you mentioned I have less voltage because of the taps I'm using. Could it be that I have too little feedback?
You also mention taking the feed back from the primary side? Would that not be in phase? And what are "nested loops?"
The Dynaco uses a 680 ohm resistor to ground. Then a 1k resistor with a 750 pico fared cap in parallel. Do these resistors act as a voltage divider? The feedback is then introduced between another 680 ohm cathode resistor and a 47 ohm cathode/shunt resistor to ground. The preamp design I'm using has the feedback introduced directly into the cathode of the preamp tube with a 280 ohm cathode resistor on the ground side.
If I wanted to increase the feedback to the level of the original, Would I increase the the value of the first 680 ohm resistor to ground in the feedback loop?
If your feedback is configured like in the original Dyna amps, removing it will increase the distortion (and the gain!). Either there's something wrong in the amp (bad part, miswiring, something not connected properly) or you're doing something incorrect in the measurement. Can you give a bit more detail on what you did?
When it comes to what Pjotr said:
My experience confirms it, and I definitely didn't see it in any literature.
For the stereo project based on 807 PP finals, testing confirmed the production of lots of higher order harmonics, though the figure I came up with was quite close to the spec sheet THD= 1.8%.. after listening open loop, I decided to go with the recommendation of O. Schade and include parallel local NFB (10% of plate output back to the input). Connecting 20db of gNFB sounded horrible, very "solid statey" , and not good SS either. I finally settled for about 7.0db of gNFB for this project. I started out in SS design, and there, the more NFB the better.
The other project was a 40W monoblock based on the 6BQ6GA (PP Class AB1). This made mainly H3, with a trace of H5 and nothing else. Therefore, no local NFB, but variable gNFB. With no NFB, you have the usual problems: sloppy bass, excessive "brightness", some pentode harshness at near clipping.
Again, apply all of the gNFB (13db) and the sound is definitely beginning to sound "solid statey". Might be a good compromise value if you didn't want to include an extra twiddle knob. Backing off the gNFB to about 6 -- 10db sounds just about right: enough "edge" to bring Metal and Techno to life, to keep the woofers under control, and get rid of any remaining pentode nastiness.
"And in a direct hit on the low-feedback philosophy, Baxandall showed many years ago that using 6-10 dB can be worse than none at all..."
"Can be" doesn't mean every time, for every design, for all times either. I can't explain it either, but that's definitely what I'm hearing. Makes one wonder if this applies to poor open loop implementations?
As for the Futterman design, it probably needs more NFB than usual since its Zo is higher than usual, and needs the extra help to get that Zo down to a point that provides adequate damping.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.