Use the 12AU7 as the splitter (1/2 per channel) and the *X7 as the voltage amplifier.
Can't do that unfortunately.
No extra tube socket and no room for another.
Now I could maybe use something between the passive preamp and amp to provide some gain if need be, however I think that I can get it right if I work at it.
Might see if I have a 9 pin socket and if so I may wire it up as a split load phase inverter and see what that does for me and use short wires to connect it up.
That way I don't have to remove the current inverter if it doesn't work as it's quite compact working where the 12AX7 tube is.
No extra tube socket and no room for another.
Now I could maybe use something between the passive preamp and amp to provide some gain if need be, however I think that I can get it right if I work at it.
Might see if I have a 9 pin socket and if so I may wire it up as a split load phase inverter and see what that does for me and use short wires to connect it up.
That way I don't have to remove the current inverter if it doesn't work as it's quite compact working where the 12AX7 tube is.
I see. A compromise (including solid state) would be using a mosfet for split load PI, DC-couple to the *X7.
My original thought was to use a DRV-134 board which uses an unbalanced to balanced/unbalanced chip that has a gain of 2 back when I didn't have enough gain in the amp initially.
It is good man!
In my opinion you can go to 200k on each grid of 6V6 without any issue
The schematic for BZ7 from Rongon is good, try to test it
The source is a Schiit Saga S preamp. It has a 75 ohm output impedance.
I use two 100 ohm resistors to combine it to mono.
I use two 100 ohm resistors to combine it to mono.
Since you have no intention of keeping the amp stock, which would solve the driver problem, and want to change it to fixed bias for more power, which creates these issues you're having with a LTP with no VA, it seems like a alternate option to get more power and be able to keep the existing PT and OPT would be to consider going to a 6L6GC common PP and not worry about 2 unfilled sockets... That would even give you a free socket to add a VA if you wanted GNFB or other input sensitivity adjustments.
Last edited:
One other solution can be the use of a input trafo 600:5k ohm or 10k with around 10 Vin max
Look on Jensen , Sowter, Lundhal,
I could but that would be going farther than I want.
I thought about a transformer and have been well pleased with the edcor matching transformers, however the source is about 12' from the amp and I need a lower impedance to keep hum pickup to a minimum.
I thought about a transformer and have been well pleased with the edcor matching transformers, however the source is about 12' from the amp and I need a lower impedance to keep hum pickup to a minimum.
With 600:5k installed close the amp you are safe
Per you can use the bal out ( if exist) to a bal- to- unbal trafo
You mentioned you're using a Schiit Saga S preamp. Doesn't that have balanced outputs on XLRs? Why not change the inputs to the power amp to XLR sockets (or 1/4" TRS jacks) and run a balanced (three wires: inverted, non-inverted, and ground) to the power amp input? You'd get slightly more gain from your power amp, and it would be perfectly balanced, which will reduce THD. Win, win, win. The only downside is that now you have to apply NFB differently, because you no longer have one triode's input grounded.
If you must stay with unbalanced connection to the amp, you might try an LR8 set to 3mA for the CCS in the tail of the LTP. However, I think it might be difficult to get it working properly in this circuit with this low amount of current and that second 10k ohm resistor in the tail. There's voltage dropped across that 10k resistor, which takes away from the plate to cathode voltage of the 12AX7/12BZ7. These are high impedance tubes that require at least 150Vp-k. Are you dead set on that adjustable NFB control with the 10k resistor and pot across it? There are better ways to accomplish that same thing.
Here's the best I could do on short notice with this circuit (using Adrian Immler's 12AT7 EH SPICE model). I found that the 12BZ7 just needs more than 275V of B+ to work in this circuit. 12AT7 works better, but yields only about 60% of the gain you can get out of a 12BZ7. Getting rid of the second 10k ohm tail resistor improves balance because of the increase in plate-to-cathode voltage. Gain is 15X from each triode.
What is the grid bias on the 6V6's? How much gain do you actually need from the LTP?
Note the rearrangement of the negative feedback. The above is the more usual way to do it, and it lets you get rid of that 10k resistor in the tail that was robbing the triodes of some plate-to-cathode voltage. With a B+ of only 275V, you need all the voltage you can get. Why is the B+ only 275V? Can that be raised a bit? That would help, a lot.
If you must stay with unbalanced connection to the amp, you might try an LR8 set to 3mA for the CCS in the tail of the LTP. However, I think it might be difficult to get it working properly in this circuit with this low amount of current and that second 10k ohm resistor in the tail. There's voltage dropped across that 10k resistor, which takes away from the plate to cathode voltage of the 12AX7/12BZ7. These are high impedance tubes that require at least 150Vp-k. Are you dead set on that adjustable NFB control with the 10k resistor and pot across it? There are better ways to accomplish that same thing.
Here's the best I could do on short notice with this circuit (using Adrian Immler's 12AT7 EH SPICE model). I found that the 12BZ7 just needs more than 275V of B+ to work in this circuit. 12AT7 works better, but yields only about 60% of the gain you can get out of a 12BZ7. Getting rid of the second 10k ohm tail resistor improves balance because of the increase in plate-to-cathode voltage. Gain is 15X from each triode.
What is the grid bias on the 6V6's? How much gain do you actually need from the LTP?
Note the rearrangement of the negative feedback. The above is the more usual way to do it, and it lets you get rid of that 10k resistor in the tail that was robbing the triodes of some plate-to-cathode voltage. With a B+ of only 275V, you need all the voltage you can get. Why is the B+ only 275V? Can that be raised a bit? That would help, a lot.
KSC3503 has been discontinued and is no longer in stock at Mouser or Digikey. But there are many BJTs and MOSFETs one could use instead. I have a stash of ZVN5045A I use for this kind of thing.
Last edited:
I just knocked out a simulation of the original circuit, tweaked with MOSFET source followers after the 12AX7 plates, as suggested by stocktrader200. The source followers reduce the Zout to the point where you could use 47k ohm grid leaks on your 6V6s and that would work perfectly fine. No more worries about driving the 6V6 grids, or grid current in the fixed bias 6V6s.
Balance is not quite as good as the 12AT7 w/ LR8 version in post 34. But THD is very low, and gain is about 29X per triode, which is almost double what you'd get from the circuit in #34.
Each MOSFET dissipates 330mW with 4.5mA Ids. The only resistors that need to be bigger than 1/2W are R14 and R15, the cathode load resistors for the MOSFETs. Those should be 2W minimum (3W would be better).
The grid bias on the 12AX7s is about as low as it can go, at -1.2V.
When you take a 12AX7 and let it drive a very light (high impedance) load, you'll get great results. The 12AX7 is a very linear triode, but only if it's very lightly loaded. If you try to drive a 100k ohm load with a 12AX7, it will load down pretty badly; its gain will go down and THD will go up as it struggles to drive the heavy load. The source followers after the 12AX7 plates solve that problem.
Balance is not quite as good as the 12AT7 w/ LR8 version in post 34. But THD is very low, and gain is about 29X per triode, which is almost double what you'd get from the circuit in #34.
Each MOSFET dissipates 330mW with 4.5mA Ids. The only resistors that need to be bigger than 1/2W are R14 and R15, the cathode load resistors for the MOSFETs. Those should be 2W minimum (3W would be better).
The grid bias on the 12AX7s is about as low as it can go, at -1.2V.
When you take a 12AX7 and let it drive a very light (high impedance) load, you'll get great results. The 12AX7 is a very linear triode, but only if it's very lightly loaded. If you try to drive a 100k ohm load with a 12AX7, it will load down pretty badly; its gain will go down and THD will go up as it struggles to drive the heavy load. The source followers after the 12AX7 plates solve that problem.
It's unbalanced.
It was mentioned either here or Audiokarma that I could ground the 4 ohm tap and use the common and 16 ohm taps for the balanced feedback signal.
I do have that DRV-134 board though which I can use to convert the unbalanced signal to balanced.
Only thing is I'd need two 1/4" jacks that are isolated from the chassis. Think I do have two of those. The 4 ohm jack I'd leave as is given the 4 ohm tap goes to ground. The 4 8 and 16 ohm taps would go to the sleeve connection of the three jacks and the ground would go to the tip.
I can use 1k and 10k, right?
That would drop less voltage.
I like how you show the negative feedback connected, however does it need to be cap coupled and do I need some sort of voltage divider to tweak the gain?
If you decide to use the secondary for cathode feedback, test the transformer first to see if the signal is truly balanced for each half winding from the grounded 4ohm. If not very well balanced you might get an oscillator.
you can use this
https://www.jensen-transformers.com/transformers/line-output/
you got 12 dB of gain, can be enough with your driver
and use 200K for each g1 of power tube
Walter
What I don't understand is by all rights this phase splitter should be better than the phase splitter I had
however that one produces a worse 10kHz squarewave than this original does.
Results with first schematic.
Results with second schematic. By all rights based on what several have told me about the second schematic not being right far as the load I'm driving is concerned and other things the squarewave shouldn't look as good as it does, yet here we are.
Not sure why, but the version with the 82k and 100k plate resistors either had slightly lower output at 10kHz or the output at 10kHz is right and it increased closer to 20kHz. If the latter is true that would explain the square wave not being right as the amp would have been overcompensated at 10kHz due to 20kHz being higher than it should and having to adjust the compensation cap to make it flat at 20kHz.
At some point maybe today or tomorrow I might try the split load phase inverter.
I did a test and indeed the transformer does not put out equal voltage from the 16 ohm tap to the 4 ohm tap and the 4 ohm tap to ground. So it would be no good for feedback.
How would I then do the feedback? Cap couple the 6V6 plate to a resistor feeding one 12AX7 cathode and do the same for the other 6V6 plate?
I may also play around with moving the feedback to the spot indicated. Is the feedback cap coupled to the grid and do I need a voltage divider?
Given how the squarewave looks is this a case of I need to leave well enough alone?
however that one produces a worse 10kHz squarewave than this original does.
Results with first schematic.
Results with second schematic. By all rights based on what several have told me about the second schematic not being right far as the load I'm driving is concerned and other things the squarewave shouldn't look as good as it does, yet here we are.
Not sure why, but the version with the 82k and 100k plate resistors either had slightly lower output at 10kHz or the output at 10kHz is right and it increased closer to 20kHz. If the latter is true that would explain the square wave not being right as the amp would have been overcompensated at 10kHz due to 20kHz being higher than it should and having to adjust the compensation cap to make it flat at 20kHz.
At some point maybe today or tomorrow I might try the split load phase inverter.
I did a test and indeed the transformer does not put out equal voltage from the 16 ohm tap to the 4 ohm tap and the 4 ohm tap to ground. So it would be no good for feedback.
How would I then do the feedback? Cap couple the 6V6 plate to a resistor feeding one 12AX7 cathode and do the same for the other 6V6 plate?
I may also play around with moving the feedback to the spot indicated. Is the feedback cap coupled to the grid and do I need a voltage divider?
Given how the squarewave looks is this a case of I need to leave well enough alone?
Last edited: