Here's the quote I was refering to in rec.audio.tubes from a Grover Gardner.
The "feedback" resistor is an interesting addition. I think it comes from the original Loftin-White design and was originally introduced to reduce hum, if I'm remembering correctly. But it has a very beneficial effect on the sound, tightening the bass and improving the sense of control, which can be a problem with DC designs. I don't know how to calculate the resistor value, but I've experimented with values ranging from 56K to 22K and it has a noticable effect on the sound. The lower the value the "tighter" the sound gets. Using this same idea in other DC topologies, I've found 33K is a good compromise.
The "feedback" resistor is an interesting addition. I think it comes from the original Loftin-White design and was originally introduced to reduce hum, if I'm remembering correctly. But it has a very beneficial effect on the sound, tightening the bass and improving the sense of control, which can be a problem with DC designs. I don't know how to calculate the resistor value, but I've experimented with values ranging from 56K to 22K and it has a noticable effect on the sound. The lower the value the "tighter" the sound gets. Using this same idea in other DC topologies, I've found 33K is a good compromise.
Giaime,
I found this circuit interesting, but it is not easy to analyze because all of the tubes are non-linear. (You could do some graphical load line stuff, but the feedback between stages makes that difficult.) So I decided to do an LTspice simulation using the tube models from Rydel_tubes.lib. As it turns out the DC operation point is very close to that shown on the schematic you referenced. The value of the 33K resistor has some effect on the DC current through the 2A3, but it has a greater effect on the biasing of the front end. For maximum signal swing you would want the front end biased about where it is with 33K.
As far as small signal AC performance goes, the 33K resistor is not invloved. However, it does effect the DC operating point and how much signal can be applied before saturation occurs. So it does indirectly effect real AC performance.
Rick
I found this circuit interesting, but it is not easy to analyze because all of the tubes are non-linear. (You could do some graphical load line stuff, but the feedback between stages makes that difficult.) So I decided to do an LTspice simulation using the tube models from Rydel_tubes.lib. As it turns out the DC operation point is very close to that shown on the schematic you referenced. The value of the 33K resistor has some effect on the DC current through the 2A3, but it has a greater effect on the biasing of the front end. For maximum signal swing you would want the front end biased about where it is with 33K.
As far as small signal AC performance goes, the 33K resistor is not invloved. However, it does effect the DC operating point and how much signal can be applied before saturation occurs. So it does indirectly effect real AC performance.
Rick
OK Giaime,
I worked it out, this analysis is for DC operating point ONLY.
Where:
Vc = Cathode voltage at the 2A3
Vg = Grid voltage at the 2A3
Assuming the current through the 390K grid resistor can be neglected.
Vg = (0.37 * Vc) + 47.0
So... with variations in 2A3's, this DC feedback wiil help to stablise the operating point. As Vc increases, for whatever reason, the Vg will increase by 1/3 that amount, thereby increasing the negative grid bias, thus working to lower the anode current, and Va.
I worked it out, this analysis is for DC operating point ONLY.
Where:
Vc = Cathode voltage at the 2A3
Vg = Grid voltage at the 2A3
Assuming the current through the 390K grid resistor can be neglected.
Vg = (0.37 * Vc) + 47.0
So... with variations in 2A3's, this DC feedback wiil help to stablise the operating point. As Vc increases, for whatever reason, the Vg will increase by 1/3 that amount, thereby increasing the negative grid bias, thus working to lower the anode current, and Va.
IMO, the 33K resistor provides a positive feedback for AC signal and a negative feedback for DC coupling.
As you can see, once the screen-grid current for the first stage is increased, the grid voltage for the 2A3 will be decreased, which open the tube more for amplification. This is the positive feedback (transient effect ).
In DC, assume the 33k resistor is not there, if the 2A3 is degrading, then the current flowing through it is decreasing, which causes the cathode current to decrease and the voltage drops across the cathode resistor will decrease and therefore the decrease with -ve grid voltage. Now with the 33K resistor, once the 2A3 is degrading, then the current flowing through it is decreasing, causing B+ to increase. Hence, the 33K resistor can feed a portion of the increased voltage to the grid, which increase back the supposedly designed grid bias voltage, which make sure that the output tube is operating at the correct bias point.
Thank you.
As you can see, once the screen-grid current for the first stage is increased, the grid voltage for the 2A3 will be decreased, which open the tube more for amplification. This is the positive feedback (transient effect ).
In DC, assume the 33k resistor is not there, if the 2A3 is degrading, then the current flowing through it is decreasing, which causes the cathode current to decrease and the voltage drops across the cathode resistor will decrease and therefore the decrease with -ve grid voltage. Now with the 33K resistor, once the 2A3 is degrading, then the current flowing through it is decreasing, causing B+ to increase. Hence, the 33K resistor can feed a portion of the increased voltage to the grid, which increase back the supposedly designed grid bias voltage, which make sure that the output tube is operating at the correct bias point.
Thank you.
An insult to call Loftin & White. Remiss in not utilizing any L&W advances.
Large cathode bypass caps are sure sign this author had no clue how and
why the original Loftin & White circuit functioned. As much misinformation,
blind leading blind, oversimplification almost to the point of lies exits. Very
few aside from Darius champion the far less easily understood whole truth.
Loftin & White topology cancels cathode feedback, cancels power hum,
and provides an ultrapath for the output. This one does none the above.
Instead, capacitors far larger than practical in L&W times abused simply
to bypass all such references to GND. Misses the whole point.
The original amplifier called for no capacitor larger than 2uF, and none
larger than 1uF in the audio path. Did not need big bypass because it
worked smarter. Go read Darius about seventeen times. Sorry, learning
curve on this one was VERY difficult for me. But well worth it.
Large cathode bypass caps are sure sign this author had no clue how and
why the original Loftin & White circuit functioned. As much misinformation,
blind leading blind, oversimplification almost to the point of lies exits. Very
few aside from Darius champion the far less easily understood whole truth.
Loftin & White topology cancels cathode feedback, cancels power hum,
and provides an ultrapath for the output. This one does none the above.
Instead, capacitors far larger than practical in L&W times abused simply
to bypass all such references to GND. Misses the whole point.
The original amplifier called for no capacitor larger than 2uF, and none
larger than 1uF in the audio path. Did not need big bypass because it
worked smarter. Go read Darius about seventeen times. Sorry, learning
curve on this one was VERY difficult for me. But well worth it.
Last edited:
I concur. Darius deserves full credit for explaining the engineering elegance contained within this design. It takes a while to get your head around it; but the original Loftin-White was a true stroke of genius. The legion of pretenders who have offered ‘improved’ versions only demonstrate their lack of understanding of the operation of the original.
Thanks for the flame. Now if you just look at what I said; I used the word 'inspiration'. To inspire: stimulation or arousal of the mind, to special or unusual activity or creativity.
Again, thanks for the pedantic response.
I sympathize with Kenpeter on this. Inspiration normally implies taking a kernel of an idea and expanding on it. Most of what is described as Loftin/White topology today, is simply direct coupling. That's actually shrinking the concept they developed, and misses their main point entirely.
I learned a lot from Darius in the short time he was here. I was inspired to remake one of my amps into a true LF. Described here: http://www.diyaudio.com/forums/tubes-valves/125488-loftin-white-801-amp.html
My caps are a little larger than the original, but I had them available, and they are all film caps. I did apply one modern touch, and use DC heating on the 801 filaments. LF works. This amp sounds very nice, and is now my headphone amp - yes it's that quiet. If you weren't looking at it, you would know that it's on when it's quiet. It actually has significant hum when it's off, because the unbiased OPT's pick up inductive fields from surrounding gear.
One general concept that was driven home to me from this exercise, and from thinking about grounding issues (or, more precisely, reference issues), is that I have to think in current. My mantra now, when trying to understand circuits, is follow the current, close the loops. Voltage will follow. Might seem like semantic playtime, but it's not.
Sheldon
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.