Maybe the LTP solution is simpler than normally supposed. The classic solutions are
However musing over data sheets the thought hit me. The 6922/6DJ8/ECC88 really likes low Vp-k so why not just use that tube with a really large tail. It has low rp so doesn't need huge plate load and it likes running with Vp-k under 100V leaving plenty of voltage for a goodly tail.
- Special power supply (e.g. negative leg)
- Unequal load resistors
- Active tail
- Inductive tail
However musing over data sheets the thought hit me. The 6922/6DJ8/ECC88 really likes low Vp-k so why not just use that tube with a really large tail. It has low rp so doesn't need huge plate load and it likes running with Vp-k under 100V leaving plenty of voltage for a goodly tail.
Plenty of those already, especially dc-coupled ones, 100~140v cathode voltage, 20K tail, for example a 6SN7, 33K plate resistors, 350V B+
Would you mind comparing that to a gain stage/concertina combo? for the given supply voltage I really like the concertina, as you can get twice the gain compared to a single-side driven differential. Then you get to actually use all the voltage you have available, and can still direct couple the 2nd stage grid to the 1st stage plate when using something like the 6dj8.
That's in fact the classical solution: it's why a long-tailed pair is called long-tailed.
Just ran a really quick (not optimized) run and sensitivity is noticeably higher imbalance is obviously zero and can swing 45V per phase. Distortion profile is obviously different.
To be fair to the LTP in reality I would probably shorten the tail somewhat and go ahead and use unequal loads which would allow for more swing. That said the old split load looks really nice with that tube.
Good point. I would probably make some adjustments to get a more reasonable current.
So increasing the current to 1.6mA per tube and changing the tail to 50K increased imbalance slightly to just over 0.2dB but distortion improved and swing jumps to almost 60V. Of course to get that swing requires over 5V peak input.
Mentioning the need for high peak voltage to drive the differential- another way to go would be to drive the differential with a gain stage/concertina combo- low mu tube up front should bring a decent gain margin- and then we get perfect balance from the beginning, equal load resistors, and less need for a big tail. You can even use a decent mosfet for the concertina to save half a dual triode per channel. I've used 6FQ7/6CG7 into 6DJ8 this way, and it works great for driving sweep tubes. Cheap option would be 6n1p up front.
All 4 methods in Post # 1 work very well . . . if they are properly designed and implemented with quality parts.
Each topology has various advantages and disadvantages.
Typically one of the lowest burden voltages of an LTP uses a choke.
I once used a Hammond 5H 200mA choke with a pair of 6C45pi triodes, and series 3V batteries from Rg, to the grid stoppers and through the grid stoppers to the grids.
That worked very well, and was very simple. That did Not reflect a DC voltage back to the RCA input connector.
The circuit in Post # 3, Will reflect a positive DC voltage back to the RCA input connector, until the input coupling capacitor is charged to the grid's DC voltage.
I had a commercial amplifier using a concertina phase splitter. It had hum that had lots of harmonics of the hum frequency (higher harmonics are more easily heard and very distracting). The cause was the filament to cathode leakage.
Whichever circuit you choose, Have Fun!
Each topology has various advantages and disadvantages.
Typically one of the lowest burden voltages of an LTP uses a choke.
I once used a Hammond 5H 200mA choke with a pair of 6C45pi triodes, and series 3V batteries from Rg, to the grid stoppers and through the grid stoppers to the grids.
That worked very well, and was very simple. That did Not reflect a DC voltage back to the RCA input connector.
The circuit in Post # 3, Will reflect a positive DC voltage back to the RCA input connector, until the input coupling capacitor is charged to the grid's DC voltage.
I had a commercial amplifier using a concertina phase splitter. It had hum that had lots of harmonics of the hum frequency (higher harmonics are more easily heard and very distracting). The cause was the filament to cathode leakage.
Whichever circuit you choose, Have Fun!
Last edited:
Edit wrong reply: I was thinking of doing something similar but different for a 6EL7 amp (same could be done for 6LU8). The idea was VAS->concertina and then using each sweep tube as a driver and output with local FB from the big triode back to the small one. The idea being to normalize the gain of each phase and lower effective Zout.
Last edited:
Do you know if that amp had elevated heaters? Would seem like a good idea.
I always thought the choke tail was an elegant solution. Hope to try it someday.
You can use another ECC88 as CCS, one triode per channel. With the right cathode resistor, it will result in an equivalent tail resistor in the region of 80-100K but will allow you to use larger anode current for better linearity, lower rp and higher gm for the LTP.....
Have a look at M. Jones's book for examples.
45;
Yes of course but my original thought was to simplify. That said there are a lot of lonely 7 pin small signal pentodes out there. 🙂
Yes of course but my original thought was to simplify. That said there are a lot of lonely 7 pin small signal pentodes out there. 🙂
Yes, I have some of those and some others like 6CB6 all pulls.
Would you be willing to share the schematic you used? Valve Wizard shows an example using EF86.
Would you be willing to share the schematic you used? Valve Wizard shows an example using EF86.