Home |
Search |
Today's Posts |
#1
|
|||
|
|||
Long Tail Pair with SS CCS in the Cathodes
Anyone know of design info for this type of splitter.
I have one up and running on the bench using a 12AX7 @ 300V & 1.2mA each side but I can't seem to make it DC (150V / 168V) balance. Should I worry about the DC balance if the AC is balanced? In info apprecaited. Thanks RonL |
#2
|
|||
|
|||
wrote
Anyone know of design info for this type of splitter. I have one up and running on the bench using a 12AX7 @ 300V & 1.2mA each side but I can't seem to make it DC (150V / 168V) balance. Should I worry about the DC balance if the AC is balanced? In info apprecaited. Depends. It's pretty crucial if you direct couple to the next stage, so it must be possible. There are two ways. One is to apply a bias voltage to one of the LTP grids. Assumes previous stage is not direct coupled. Never seen it done. The other way is to use a pot, wiper to CCS and legs to cathodes. This should be of the least value necessary to allow the adjustment you need. You don't give your anode loads so you'll have to work it out yourself, or experiment with fixed resistors. Possibly start around 100R between one cathode and CCS. When you get it about right, sub for a pot, allowing some extra R for future adjustment. If you also vary one of the anode loads to achieve AC balance, then you will need to readjust the cathode pot to compensate. If there's 1.2mA each side how come it's not DC balanced? What you've written doesn't quite add up. cheers, Ian in message ... Thanks RonL |
#3
|
|||
|
|||
wrote: Anyone know of design info for this type of splitter. I have one up and running on the bench using a 12AX7 @ 300V & 1.2mA each side but I can't seem to make it DC (150V / 168V) balance. Should I worry about the DC balance if the AC is balanced? In info apprecaited. Thanks RonL Please find an example of an LTP with CCS in the amp at http://www.turneraudio.com.au/htmlwe...0ulabinteg.htm If the RLs to each side of the pair are equal, output signal also will be. But the dc anode voltage won't ever be exactly the same, since no tubes are exactly matched for bias conditions. The tubes could be quite different, one could be a 1/2 12AU7, the the other a 1/2 6SN7 triode and as long as the RLs are equal and the CCS is an extremely high resistance, such as in the schematic I have refered you to, ac balance will be ok even though dc balance is not good. But using dissimilar tubes to each 1/2 of an LTP results in 2H distortion that is normally cancelled away when each 1/2 is matched to the other in terms of Ra, gm, and µ. Patrick Turner. |
#4
|
|||
|
|||
Thats what I wanted to hear!. As long as the AC is balanced. (most
important) I wasn't sure if the DC should be balanced also. Thanks "Patrick Turner" wrote in message ... wrote: Anyone know of design info for this type of splitter. I have one up and running on the bench using a 12AX7 @ 300V & 1.2mA each side but I can't seem to make it DC (150V / 168V) balance. Should I worry about the DC balance if the AC is balanced? In info apprecaited. Thanks RonL Please find an example of an LTP with CCS in the amp at http://www.turneraudio.com.au/htmlwe...0ulabinteg.htm If the RLs to each side of the pair are equal, output signal also will be. But the dc anode voltage won't ever be exactly the same, since no tubes are exactly matched for bias conditions. The tubes could be quite different, one could be a 1/2 12AU7, the the other a 1/2 6SN7 triode and as long as the RLs are equal and the CCS is an extremely high resistance, such as in the schematic I have refered you to, ac balance will be ok even though dc balance is not good. But using dissimilar tubes to each 1/2 of an LTP results in 2H distortion that is normally cancelled away when each 1/2 is matched to the other in terms of Ra, gm, and µ. Patrick Turner. |
#5
|
|||
|
|||
Ian Iveson wrote: wrote Anyone know of design info for this type of splitter. I have one up and running on the bench using a 12AX7 @ 300V & 1.2mA each side but I can't seem to make it DC (150V / 168V) balance. Should I worry about the DC balance if the AC is balanced? In info apprecaited. Depends. It's pretty crucial if you direct couple to the next stage, so it must be possible. The guy does not require the DC balance to be exact. There are two ways. One is to apply a bias voltage to one of the LTP grids. Assumes previous stage is not direct coupled. Never seen it done. One way to get DC balance is to have *two* CCS so that one is used for each cathode of the LTP. Then one uses a cap between the cathodes for signal shunting. two CCS are effectively in parallel, but since their actual finite resistance is over 20Mohms, its ok to parallel them this way but yet make sure the DC in each anode RL is within 0.1% all the time. .. The other way is to use a pot, wiper to CCS and legs to cathodes. But then you have a pot in the signal path. This should be of the least value necessary to allow the adjustment you need. You don't give your anode loads so you'll have to work it out yourself, or experiment with fixed resistors. Possibly start around 100R between one cathode and CCS. When you get it about right, sub for a pot, allowing some extra R for future adjustment. This is ok until the next time the tubes are changed, and a reset for DC balance has to be attained. The twin CCS method avoids this. If you also vary one of the anode loads to achieve AC balance, then you will need to readjust the cathode pot to compensate. For an LTP with a single CCS with a an adjust pot for DC balance, the signal output will remain constant if the anode RL remain the same. The anode RL is the DC supply resistor and the cap coupled following grid bias resistor of the next stage in parallel. Once the use of an LTP is addopted, the only way to vary the output amplitudes is to vary the RLs. But why? The only reason would be to minimise thd for a given value of power; some slight imbalance of signals will cancel the 2H produced by an unmatched pair of output tubes. Trouble is that if you adjust for least 2H at 2 watts, the adjustment won't be right for 0.2 watts or 10 watts, or 50 watts. So use nice output tubes and you should find exactly balanced drive to give the lowest thd and the sweetest sound. If there's 1.2mA each side how come it's not DC balanced? What you've written doesn't quite add up. Well yes, but maybe if the RL is 100k, then to get an 18V difference in Ea, the current difference is only 0.18mA. I think he meant that he has a *nominal 1.2mA per side, so he must have a nominal 2.4mA in the CCS. Patrick Turner. cheers, Ian in message ... Thanks RonL |
#7
|
|||
|
|||
"Patrick Turner" wrote
The guy does not require the DC balance to be exact. None of us actually knows that. "The guy" wondered about it, you assumed it, and let's hope you are right in this case. Wouldn't have worked in my amp. One way to get DC balance is to have *two* CCS so that one is used for each cathode of the LTP. Then one uses a cap between the cathodes for signal shunting. Hadn't considered mentioning that idea. You would need a large value, non-polar cap. Never seen it done. Works in theory, but puts a cap where it otherwise wouldn't be necessary. Nice idea, but you need to give it more thought. two CCS are effectively in parallel, but since their actual finite resistance is over 20Mohms, its ok to parallel them this way Why shouldn't it be? You are confusing issues. CCS are OK in parallel, not in series. CVS is vice-versa. And they are not in parallel anyway because they are DC, and are isolated by a cap. Maybe there is another interpretation of what you mean to say? but yet make sure the DC in each anode RL is within 0.1% all the time. Whatever. If used to adjust bias in following stage, another criterion applies. The other way is to use a pot, wiper to CCS and legs to cathodes. But then you have a pot in the signal path. 100R or so between the cathodes isn't much, but the purist in me doesn't like it either even though it's only a spot of series-derived nfb equally applied to both sides. I cannot see a "purist" solution myself...between us we have outlined *every* logical possibility. It's a shame for the purist, who otherwise might be attracted to a chain of dc coupled ltp for a fully-balanced amp. Plot gets a bit more complicated still if you wish to use global nfb for such an amp. Rest of your contribution is quite correct. cheers, Ian in message ... This should be of the least value necessary to allow the adjustment you need. You don't give your anode loads so you'll have to work it out yourself, or experiment with fixed resistors. Possibly start around 100R between one cathode and CCS. When you get it about right, sub for a pot, allowing some extra R for future adjustment. This is ok until the next time the tubes are changed, and a reset for DC balance has to be attained. The twin CCS method avoids this. If you also vary one of the anode loads to achieve AC balance, then you will need to readjust the cathode pot to compensate. For an LTP with a single CCS with a an adjust pot for DC balance, the signal output will remain constant if the anode RL remain the same. The anode RL is the DC supply resistor and the cap coupled following grid bias resistor of the next stage in parallel. Once the use of an LTP is addopted, the only way to vary the output amplitudes is to vary the RLs. But why? The only reason would be to minimise thd for a given value of power; some slight imbalance of signals will cancel the 2H produced by an unmatched pair of output tubes. Trouble is that if you adjust for least 2H at 2 watts, the adjustment won't be right for 0.2 watts or 10 watts, or 50 watts. So use nice output tubes and you should find exactly balanced drive to give the lowest thd and the sweetest sound. If there's 1.2mA each side how come it's not DC balanced? What you've written doesn't quite add up. Well yes, but maybe if the RL is 100k, then to get an 18V difference in Ea, the current difference is only 0.18mA. I think he meant that he has a *nominal 1.2mA per side, so he must have a nominal 2.4mA in the CCS. Patrick Turner. |
#8
|
|||
|
|||
"Patrick Turner" wrote in message
... wrote: Thats what I wanted to hear!. As long as the AC is balanced. (most important) I wasn't sure if the DC should be balanced also. If the LTP is a twin triode, or a low µ triode like the 12AU7, 6CG7, 6SN7, then you should find the dc balance very close with most tubes that test well. Maybe you have a 6SL7, or 12AX7, 12AY7, 12AT7, and the gain is higher, so expect more difference in Ea for the two triodes. Usually is never more than 10V difference. The biasing of each LTP triode shouldn't use greater than 470k grid R lest the tiny grid current that flows develops a voltage which can effect the quiescent anode volts, ( Ea ). Direct coupled LTPs like in a Mullard 520 etc have the first triode direct coupled to the anode of the input tube, then a 1M or more is taken to the second triode grid which is grounded via a 0.1uF or suchlike. Sometimes there is a slight dc voltage across the 1M, so this dc unbalances the LTP. BTW, for the Mullard circuit to be properly ac balanced while a common cathode R of say 22k is used from cathodes to 0V, the usual way to achieve balance is to adjust the value of the DC carrying RL to each triode until balance is achieved. This usually upsets the dc balance which does not matter much in most amps because the LTP will clip well after the output stage clips. But the LTP with a CCS allows much easier setting up with equal RL values which are cheaply available at 1% tolerances or better. Its the RL values which automatically balance the LTP with a CCS that makes it such an attractive gain block. Just make sure you have a high impedance meter for measuring the outputs and dcV values; it should be better than 9M input Z to the meter. Most digital meters are like this and read ok between 20Hz and 1 kHz. Above 1kHz, digital meter signal volt readings roll off due to the capacitance of the meter. Patrick Turner. I've nearly completed a "spares box" 10 watt amplifier design (EF86, 6SN7, 2 x 6V6) that uses a "Mullard 5-10" EF86 circuit (120K, 2,2K, 100 ohms, etc), directly coupled (108 VDC) to g1 of the first 1/2 6SN7, with a CC circuit in cathode. The latter is an MJE340 transistor with a 5.6 Kohm emitter resistor and a +25 volt (approx) DC reference to the base for 4.3 mA CC (2.15 mA for each 1/2 6SN7 with 33Kohms on each plate.) DC balance is close to perfect and drive to 6V6's very balanced. Getting 10.4 watts RMS max but still tweeking things. I plan to post a schematic on the binaries later for comments/criticism. Cheers, Roger Thanks "Patrick Turner" wrote in message ... wrote: Anyone know of design info for this type of splitter. I have one up and running on the bench using a 12AX7 @ 300V & 1.2mA each side but I can't seem to make it DC (150V / 168V) balance. Should I worry about the DC balance if the AC is balanced? In info apprecaited. Thanks RonL Please find an example of an LTP with CCS in the amp at http://www.turneraudio.com.au/htmlwe...0ulabinteg.htm If the RLs to each side of the pair are equal, output signal also will be. But the dc anode voltage won't ever be exactly the same, since no tubes are exactly matched for bias conditions. The tubes could be quite different, one could be a 1/2 12AU7, the the other a 1/2 6SN7 triode and as long as the RLs are equal and the CCS is an extremely high resistance, such as in the schematic I have refered you to, ac balance will be ok even though dc balance is not good. But using dissimilar tubes to each 1/2 of an LTP results in 2H distortion that is normally cancelled away when each 1/2 is matched to the other in terms of Ra, gm, and µ. Patrick Turner. |
#9
|
|||
|
|||
"Patrick Turner" wrote in message
... wrote: Thats what I wanted to hear!. As long as the AC is balanced. (most important) I wasn't sure if the DC should be balanced also. If the LTP is a twin triode, or a low µ triode like the 12AU7, 6CG7, 6SN7, then you should find the dc balance very close with most tubes that test well. Maybe you have a 6SL7, or 12AX7, 12AY7, 12AT7, and the gain is higher, so expect more difference in Ea for the two triodes. Usually is never more than 10V difference. The biasing of each LTP triode shouldn't use greater than 470k grid R lest the tiny grid current that flows develops a voltage which can effect the quiescent anode volts, ( Ea ). Direct coupled LTPs like in a Mullard 520 etc have the first triode direct coupled to the anode of the input tube, then a 1M or more is taken to the second triode grid which is grounded via a 0.1uF or suchlike. Sometimes there is a slight dc voltage across the 1M, so this dc unbalances the LTP. BTW, for the Mullard circuit to be properly ac balanced while a common cathode R of say 22k is used from cathodes to 0V, the usual way to achieve balance is to adjust the value of the DC carrying RL to each triode until balance is achieved. This usually upsets the dc balance which does not matter much in most amps because the LTP will clip well after the output stage clips. But the LTP with a CCS allows much easier setting up with equal RL values which are cheaply available at 1% tolerances or better. Its the RL values which automatically balance the LTP with a CCS that makes it such an attractive gain block. Just make sure you have a high impedance meter for measuring the outputs and dcV values; it should be better than 9M input Z to the meter. Most digital meters are like this and read ok between 20Hz and 1 kHz. Above 1kHz, digital meter signal volt readings roll off due to the capacitance of the meter. Patrick Turner. I've nearly completed a "spares box" 10 watt amplifier design (EF86, 6SN7, 2 x 6V6) that uses a "Mullard 5-10" EF86 circuit (120K, 1.8K + 100 ohms, etc), directly coupled (108 VDC) to g1 of the first 1/2 6SN7, with a CC circuit in cathode. The latter is an MJE340 transistor with a 5.6 Kohm emitter resistor and a +25 volt (approx) DC reference to the base for 4.3 mA CC (2.15 mA for each 1/2 6SN7 with 33Kohms on each plate.) DC balance is close to perfect and drive to 6V6's very balanced. Getting 10.4 watts RMS max but still tweeking things. I plan to post a schematic on the binaries later for comments/criticism. Cheers, Roger (snip) |
#10
|
|||
|
|||
Patrick and group:
I've tried the MJE340 setup and it works great. I'm going to try the SuperTex LR8 http://www.supertex.com/pdf/datasheets/LR8.pdf in a CCS mode to see if I can get a little better PSRR. The MJE340 that Patrick uses works good, but thr LR8 in a nice TO92 package would be a little neater. Anybody use any of these.? RonL |
#11
|
|||
|
|||
On Wed, 21 Sep 2005 06:04:56 GMT, "Ian Iveson"
wrote: There are two ways. One is to apply a bias voltage to one of the LTP grids. Assumes previous stage is not direct coupled. Never seen it done. The other way is to use a pot, wiper to CCS and legs to cathodes. This should be of the least value necessary to allow the adjustment you need. You don't give your anode loads so you'll have to work it out yourself, or experiment with fixed resistors. Possibly start around 100R between one cathode and CCS. When you get it about right, sub for a pot, allowing some extra R for future adjustment. Another way, sometimes done in mic preamp inputs in the solid-state world is to use two CCS's, and couple the cathodes together with either a resistor or a series resistor and (large) capacitor. Can't really see the need in this case. But fun to speculate. Thanks, as always, Chris Hornbeck |
#12
|
|||
|
|||
Ian Iveson wrote: "Patrick Turner" wrote The guy does not require the DC balance to be exact. None of us actually knows that. "The guy" wondered about it, you assumed it, and let's hope you are right in this case. Wouldn't have worked in my amp. One way to get DC balance is to have *two* CCS so that one is used for each cathode of the LTP. Then one uses a cap between the cathodes for signal shunting. Hadn't considered mentioning that idea. You would need a large value, non-polar cap. Never seen it done. Works in theory, but puts a cap where it otherwise wouldn't be necessary. Nice idea, but you need to give it more thought. Well it works well. At DC the separate CCS act like a high value cathode resistors which makes the anode voltage stay very stable. two CCS are effectively in parallel, but since their actual finite resistance is over 20Mohms, its ok to parallel them this way Why shouldn't it be? You are confusing issues. CCS are OK in parallel, not in series. CVS is vice-versa. And they are not in parallel anyway because they are DC, and are isolated by a cap. Maybe there is another interpretation of what you mean to say? The two CCS are effectively in parallel as far as signal working is concerned. Nothing else was meant or implied. but yet make sure the DC in each anode RL is within 0.1% all the time. Whatever. If used to adjust bias in following stage, another criterion applies. The other way is to use a pot, wiper to CCS and legs to cathodes. But then you have a pot in the signal path. 100R or so between the cathodes isn't much, but the purist in me doesn't like it either even though it's only a spot of series-derived nfb equally applied to both sides. I cannot see a "purist" solution myself...between us we have outlined *every* logical possibility. It's a shame for the purist, who otherwise might be attracted to a chain of dc coupled ltp for a fully-balanced amp. Plot gets a bit more complicated still if you wish to use global nfb for such an amp. with a dual CCS in an LTP and direct coupling to an output stage, things sure do get complex with the need to have the anode dc voltage at about the output stage grid bias voltages, so a negative supply is needed. I cannot see any reason for such complexity, since couping output tube grids to cathodes is a pita if anything goes wrong. Anyway, with a pair of paralleled CCS, the gain of the LTP falls to near zero at DC because the cap coupling the cathodes becomes an open circuit at VLF, but yet there is no ultimate 90 degree phase shift, so if anything the cap across cathodes may not jepodize the LF stability with NFB. Just how much NFB is also important. I have never felt a need or wanton desire to build such an amp with an LTP with dual CCS and a cap coupling to tie the cathodes together. Rest of your contribution is quite correct. Good to see you not bitching about something this time. Patrick Turner. cheers, Ian in message ... This should be of the least value necessary to allow the adjustment you need. You don't give your anode loads so you'll have to work it out yourself, or experiment with fixed resistors. Possibly start around 100R between one cathode and CCS. When you get it about right, sub for a pot, allowing some extra R for future adjustment. This is ok until the next time the tubes are changed, and a reset for DC balance has to be attained. The twin CCS method avoids this. If you also vary one of the anode loads to achieve AC balance, then you will need to readjust the cathode pot to compensate. For an LTP with a single CCS with a an adjust pot for DC balance, the signal output will remain constant if the anode RL remain the same. The anode RL is the DC supply resistor and the cap coupled following grid bias resistor of the next stage in parallel. Once the use of an LTP is addopted, the only way to vary the output amplitudes is to vary the RLs. But why? The only reason would be to minimise thd for a given value of power; some slight imbalance of signals will cancel the 2H produced by an unmatched pair of output tubes. Trouble is that if you adjust for least 2H at 2 watts, the adjustment won't be right for 0.2 watts or 10 watts, or 50 watts. So use nice output tubes and you should find exactly balanced drive to give the lowest thd and the sweetest sound. If there's 1.2mA each side how come it's not DC balanced? What you've written doesn't quite add up. Well yes, but maybe if the RL is 100k, then to get an 18V difference in Ea, the current difference is only 0.18mA. I think he meant that he has a *nominal 1.2mA per side, so he must have a nominal 2.4mA in the CCS. Patrick Turner. |
Reply |
Thread Tools | |
Display Modes | |
|
|
Similar Threads | ||||
Thread | Forum | |||
FS: Pair of CDT Audio HD-63 Braxials AND a Pair of CDT Audio HD-62's | Car Audio | |||
Quad ESLs - 57 or 63? | Vacuum Tubes | |||
FS: Audio Cables & Adapter Cables | Pro Audio | |||
2nd Classical mic pair for location work | Pro Audio | |||
wrap test | Pro Audio |