negative feedback vs. attenuation

[black_labb]

after reading through daz'a thread (any reason this wont work) i started to reconsider a control i was going to put into a circuit for the monster of a bass amp. i was going to put in a control to limit the bass. i was planning to either use a high pass RC network where i could control the f3 so i could adjust it from ~5hz to ~150hz. i was also going to do a low pass where i could adjust the f3 from ~20khz to 500hz or wherever. was also considering similar designs. i was going to use these along with a james style tone stack to get quite a versatile eq as long as you didnt want to boost the low lows without the low mids, or the inverse with the highs.

i understand that negative feedback is inserting an out of phase signal to a circuit, which essentially cancels the signal to a certain extent. negative feedback is described as tightening up the sound and making things more controlable.

other than attenuation, how does it make things sound "tighter and under control". it seems to me that it is more useful than just turning down the volume, or turning down the treble in the case of a presense control.

looking at what it does technically (cancelling out part of the waveform) doesnt seem to me as anything more than attenuation, but it is very important to hifi systems to make the amp linear from what i've heard. this seems more than just attenuation.

i'd like to understand a bit more about how negative feedback works so i can possibly use it as low cut or high cut either around a gain stage or from the output tap to the PI. on http://www.freewebs.com/valvewizard/localfeedback.html it seems that they use -ve feedback to greatly decrease gain (gain of 65 to 10 or less). i'm thinking i may use -ve feedback around a pentode stage (6bl8) through an inductor or possibly a RC filter to tighten up/minimise the lows. they seem to suggest that not minimising the gain alot can make the input impedance too high? this looked to me to be due to the grid resistor. would it be reasonable to go without the grid resistor and just use a resistor from the signal out to the grid (+ a filter in my case). i would replace the value of the grid resistor with the output impedance of the previous stage?
the other thing is a 6bl8 is not designed for being used as a gain stage, so the datasheets dont seem to help much in calculating it theoretically. i'm thinking i may use a low pass rc network in the feedback circuit and use a pot to adjust the f3. this way i can make the circuit without the feedback and measure the gain, and then tweak and test to suit.

thanks

luke

[Steve Conner]

Hey Luke

I just made a post to Daz's thread rambling about NFB and bass, so I guess I might as well carry on rambling here.

I guess what you're asking is, what is the difference between frequency shaping by NFB, and frequency shaping by a passive tone stack? Well, in a preamp stage, the difference is that NFB reduces the distortion generated by the tube, but the passive network doesn't. So, turning down a bass control that works with NFB will not only reduce the amount of bass, it will reduce the amount of distortion on the bass, which could make it sound cleaner, or thinner, depending on taste.

I've tried both kinds with bass guitar, the passive Fender style stack, and the active Baxandall. I used the Bax around a 12AX7 gain stage, reducing the gain to 10 with both knobs in the flat position. I could just be imagining it, but I thought the Bax was cleaner, up until it starts to clip, then the distortion is a bit harsher and comes on more suddenly.

The real fun starts when you play with the NFB loop in the power amp. Pentode and tetrode power tubes have no ability to damp the speaker by themselves, they rely completely on the NFB to do it. So frequency shaping by messing with this NFB interacts with your speaker in subtle ways. If you bled off NFB at low frequencies to create a bass boost, it would also undamp your speaker and emphasize its resonances. So a vented cabinet with two resonances would react differently than a sealed cabinet with its single resonance. (Marshall's VBA400 has a resonance control in the power amp, and I wonder if it's just coincidence that all the cabinets they sell with it are sealed.)

A similar argument goes for the presence control, cone resonances and voice coil inductance.

[black_labb]

thanks steve. thats very helpful. i think that the feedback around the pentode will work perfectly for the design. the preamp is a hiwatt signal flow, so there is a 12ax7 where each triode with individual input feeds into the next gain stage through a volume pot each. the next stage is a 6bl8 pentode, which will have the lowpass feedback around it to minimuise/tighten up the low end. the 12ax7's should be able to overdrive the pentode when the gain/first volume is all the way up, and the feedback will keep the low end clean and in control while the mids overdrive when the gain is up.
there will be a james style tone stack with volume and then the triode from the 6bl8 as a gain stage before the phase inverter and cathode followers.
i will probably use diodes on the cathodes of the first triodes to get more gain and a flat as possible response so the low end feedback can still operate without making the sound too thin. i will try to set them up as differently as possible so the channels sound different.

sorry to be rambling a bit, but i've been getting very excited about this amp recently. i love how just reading people's queries can give you good ideas for your own amps. its why i love this forum.

[Chuck H]

Another reason NFB sounds tighter than padding the signal is that shunt NFB circuits reduce impedance. If you want tighter then sending a low impedance into a comparably high impedance is a good idea because it keeps the input from loading the signal. The damping effect is non linear and the closer the impedances are the more non linear it becomes. I use shunt NFB circuits often for mixing stages and such. Or whenever I need a stage to be more transpearant and not distort. Which, come to think of it, is't all that often really . But it does com up.

Chuck

[Bill Guthke]

new user

goodmorning or good evening where ever you may be located.my name is billg.i build and repair amps and have been since i was 13 yrs old.if any one has info on a legend las 27oo amp it would be helpfull.need a schematic.if you have one e-mail kid286@juno.com thanks billg

[black_labb]

Another reason NFB sounds tighter than padding the signal is that shunt NFB circuits reduce impedance. If you want tighter then sending a low impedance into a comparably high impedance is a good idea because it keeps the input from loading the signal. The damping effect is non linear and the closer the impedances are the more non linear it becomes. I use shunt NFB circuits often for mixing stages and such. Or whenever I need a stage to be more transpearant and not distort. Which, come to think of it, is't all that often really . But it does com up.

Chuck

thanks, good to know.

[lowell]

goodmorning or good evening where ever you may be located.my name is billg.i build and repair amps and have been since i was 13 yrs old.if any one has info on a legend las 27oo amp it would be helpfull.need a schematic.if you have one e-mail kid286@juno.com thanks billg