Most of them don't but 99% they can handle about 0.1-0.5 mA unbalance with plenty of headroom left. Grid chokes in general are not a solution for positive grid drive anyway....in this cad wit's A1 or AB1.
push pull removed(cancelled) most of the 2nd harmonics unlike a single ended amp...
Only when the splitter is symmetrical. There is always an option to use an asymmetric splitter to retain some H2.
analog_sa,
I have seen some people's implementation of Parallel SE amplifiers. Most of them ignore one or another of my recommendations in Post #13. The best topologies in the world do not sound good if they are not done properly. And first impressions always stick in peoples minds. For any engineering problem, there are 100 solutions, with at least 3 solutions that will work, but only when they are implemented properly.
indra1,
You can use a symmetrical splitter, and still get 2nd harmonic distortion.
Just do both of these together:
Do not use global negative feedback
And . .
Use unmatched output tubes.
ColinA123,
Can you provide a schematic of the parallel 2A3 SE amp?
Did it use different correctly scaled impedance, current, and power rated output transformers; different higher current, lower impedance drivers; Individual bias for each 2A3, higher current power supply with 2X the current; etc.?
We wrote the article of "Paralleling Tubes Effects" in Glass Audio Volume 12, Number 5, 2000.
With 20 6SN7 dual triodes (40 sections), and over 600 sets of data, testing results on an HP 3585 spectrum analyzer.
Then we did the Taylor series math, and proved the results of all the data that was collated and analyzed.
Then a dual SE amp with triode drivers, and 300B outputs was built on a single chassis. The tube gains were very closely matched. The two channels both received the same Left channel signal from a CD player, with several different CDs to test different types of music. The two channels drove separate speakers.
There were two switches operating relays. The settings were:
Driver plates connected together and 300B plates connected together
Driver plates connected together and 300B plates Not connected together
Driver plates Not connected together and 300B plates connected together
Driver plates Not connected together and 300B plates Not connected together
Then, listening tests were done at 3 venues, on different speakers, with nobody knowing what was being tested, just listening for differences. One participant was so aggravated at not hearing any difference, he threw the test switches across the room. Differences were either not perceived, or not able to be described.
I know, other amps, other speakers, may give different results to the paralleled versus unparalleled debate.
But think about this:
A 2A3 arrived from the factory. It sounded real good. But . . .
A real 2A3 has two separate triodes in parallel.
It did not have exactly equal spaces from the filaments to one side of the grid versus the filament to the other side of the grid. And there are two filaments and two grid structures and they both may have imperfect spacings relative to themselves, and relative to the other triode (that is in the same glass envelope).
Or, it did not have exactly equal spaces from one side of the grid to its closest plate side, versus to the other side of the grid to its closest plate side. And there are two grid structures and two plate structures, and they both may have imperfect spacing relative to themselves, and may be unequal from one triode to the other (in the same glass envelope).
Now, how many of you are suddenly going to throw your real 2A3s out, in favor of the modern mono-plate "2A3s"
And of course, the modern mono-plate 2A3s can have unequal spacings just like each half of the dual plate 2A3.
Parallel tubes indeed. You need to use multiple true spaced triodes like the 416A tube. Just parallel enough of them so that you can get the power you need (don't take this last suggestion seriously, I am just trying to get the point across).
I have seen some people's implementation of Parallel SE amplifiers. Most of them ignore one or another of my recommendations in Post #13. The best topologies in the world do not sound good if they are not done properly. And first impressions always stick in peoples minds. For any engineering problem, there are 100 solutions, with at least 3 solutions that will work, but only when they are implemented properly.
indra1,
You can use a symmetrical splitter, and still get 2nd harmonic distortion.
Just do both of these together:
Do not use global negative feedback
And . .
Use unmatched output tubes.
ColinA123,
Can you provide a schematic of the parallel 2A3 SE amp?
Did it use different correctly scaled impedance, current, and power rated output transformers; different higher current, lower impedance drivers; Individual bias for each 2A3, higher current power supply with 2X the current; etc.?
We wrote the article of "Paralleling Tubes Effects" in Glass Audio Volume 12, Number 5, 2000.
With 20 6SN7 dual triodes (40 sections), and over 600 sets of data, testing results on an HP 3585 spectrum analyzer.
Then we did the Taylor series math, and proved the results of all the data that was collated and analyzed.
Then a dual SE amp with triode drivers, and 300B outputs was built on a single chassis. The tube gains were very closely matched. The two channels both received the same Left channel signal from a CD player, with several different CDs to test different types of music. The two channels drove separate speakers.
There were two switches operating relays. The settings were:
Driver plates connected together and 300B plates connected together
Driver plates connected together and 300B plates Not connected together
Driver plates Not connected together and 300B plates connected together
Driver plates Not connected together and 300B plates Not connected together
Then, listening tests were done at 3 venues, on different speakers, with nobody knowing what was being tested, just listening for differences. One participant was so aggravated at not hearing any difference, he threw the test switches across the room. Differences were either not perceived, or not able to be described.
I know, other amps, other speakers, may give different results to the paralleled versus unparalleled debate.
But think about this:
A 2A3 arrived from the factory. It sounded real good. But . . .
A real 2A3 has two separate triodes in parallel.
It did not have exactly equal spaces from the filaments to one side of the grid versus the filament to the other side of the grid. And there are two filaments and two grid structures and they both may have imperfect spacings relative to themselves, and relative to the other triode (that is in the same glass envelope).
Or, it did not have exactly equal spaces from one side of the grid to its closest plate side, versus to the other side of the grid to its closest plate side. And there are two grid structures and two plate structures, and they both may have imperfect spacing relative to themselves, and may be unequal from one triode to the other (in the same glass envelope).
Now, how many of you are suddenly going to throw your real 2A3s out, in favor of the modern mono-plate "2A3s"
And of course, the modern mono-plate 2A3s can have unequal spacings just like each half of the dual plate 2A3.
Parallel tubes indeed. You need to use multiple true spaced triodes like the 416A tube. Just parallel enough of them so that you can get the power you need (don't take this last suggestion seriously, I am just trying to get the point across).
Miller of two 300B is similar to one 12AX7; about 100pFd.
About 100K reactance at 16KHz. (Note that Rg-kMax for 300B is 250K/tube, so 125K a pair.)
I'll speculate that any driver too lame to pull a 12AX7, or ~~100K, shouldn't be under consideration.
I expect that the signal driving a 12AX7 grid is usually less than +/- 3 Volts. 3 Volts/100k Ohms only requires 30uA drive at 16kHz. But driving a 300B is a little harder than that.
A 300B might be biased at -74V. A pair of 300B tubes with a total 100pF of miller capacitance at 20kHz is 80k Ohms of capacitive reactance. The 125k grid resistor is in parallel with that. We may have 5 pF more in the wiring that is also multiplied by the stage gain of about 3 (15 more pF of miller C) Lets say the total impedance is about 50k Ohms. 74V/50k Ohms = about 1.5mA of drive current required at 20kHz. That is not too bad, but I don't normally calculate the individual circuits of an amplifier at a top frequency of 20kHz.
A 300B might be biased at -74V. A pair of 300B tubes with a total 100pF of miller capacitance at 20kHz is 80k Ohms of capacitive reactance. The 125k grid resistor is in parallel with that. We may have 5 pF more in the wiring that is also multiplied by the stage gain of about 3 (15 more pF of miller C) Lets say the total impedance is about 50k Ohms. 74V/50k Ohms = about 1.5mA of drive current required at 20kHz. That is not too bad, but I don't normally calculate the individual circuits of an amplifier at a top frequency of 20kHz.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.