And this is the CCS I used under the ECC83. Now I run them at about 2.5mA for the whole LTP, I regulated the trimmer in the CCS until I got equal voltage swings on the two plates (but still retaining gain).
I also tried another CCS (simple jfet with drain resistor), results were the same.
Also, obviously, I re-measured all passive components and tried many different 12AX7/ECC83.
I also tried another CCS (simple jfet with drain resistor), results were the same.
Also, obviously, I re-measured all passive components and tried many different 12AX7/ECC83.
Attachments
Ciao Gianluca,
thank you for your reply.
Aside from my problem of lack of gain and output voltage swing (because if I understand well, your suggestion doesn't improve that), are you saying of moving the AC coupling to the first and second stage? So I'd have to use a BIG negative rail, to allow the cathode of the 5687 to be at the needed bias voltage for the EL36. It could be done, yes, it's definitively an improvement and I could even try some A2 drive.
If not, I find extremely hard to DC-couple the whole amp, but I'm sure this wasn't what you meant.
I'll keep open this possibility, but I'd like first to solve the gain problem of the first stage.
Thank you very much!
thank you for your reply.
Aside from my problem of lack of gain and output voltage swing (because if I understand well, your suggestion doesn't improve that), are you saying of moving the AC coupling to the first and second stage? So I'd have to use a BIG negative rail, to allow the cathode of the 5687 to be at the needed bias voltage for the EL36. It could be done, yes, it's definitively an improvement and I could even try some A2 drive.
If not, I find extremely hard to DC-couple the whole amp, but I'm sure this wasn't what you meant.
I'll keep open this possibility, but I'd like first to solve the gain problem of the first stage.
Thank you very much!
I did not have a closer look to the numbers so I am just throwing ideas on the desk now ... but what I meant is to DC couple the CFs and output stage and AC couple the input to the driver stage. Plus a CCS would give quite a great result under that CF.
Ok ... thats not a minor modification and I am not saying anything about your gain issue.
Do you mean your simulation (with actual measured voltages) is suggesting you should have a higher gain? Where? At CFs or ECC83s' plates?
Gianluca
Ok ... thats not a minor modification and I am not saying anything about your gain issue.
Do you mean your simulation (with actual measured voltages) is suggesting you should have a higher gain? Where? At CFs or ECC83s' plates?
Gianluca
Ciao Gianluca,
it's very helpful your suggestion and surely I'll implement that, it looks promising. But it shouldn't interfere with gain, if done correctly.
What I'm saying, is that both TubeCAD (the "LTP with current source" circuit) and SPICE (actual circuit simulated) give about double output voltage and gain for the LTP stage. Of course I measured this without any load, neither the CF stage nor the output tubes.
So, even without load (instead TubeCAD and SPICE simulations were with load) I have low gain and I can't understand why, since I haven't found severe limitations in the possibility of swing of the LTP.
Spice says that I should have 50x gain and about 100Vpk output... instead I have about 26x gain, and 29Vrms maximum undistorted output. And I'll add that I have distortion on the positive peaks of the sine wave first, so it's the tube that reaches cutoff too soon. Increasing current in the LTP didn't help, I even tried putting in an ECC81, same results of half gain than simulation.
The problem is that I can accept some volts lost from the simulation, because SPICE is a semplified world than reality, but loosing half the gain and half the voltage swing is too much, so I think there's something wrong.
Can someone help me?
it's very helpful your suggestion and surely I'll implement that, it looks promising. But it shouldn't interfere with gain, if done correctly.
What I'm saying, is that both TubeCAD (the "LTP with current source" circuit) and SPICE (actual circuit simulated) give about double output voltage and gain for the LTP stage. Of course I measured this without any load, neither the CF stage nor the output tubes.
So, even without load (instead TubeCAD and SPICE simulations were with load) I have low gain and I can't understand why, since I haven't found severe limitations in the possibility of swing of the LTP.
Spice says that I should have 50x gain and about 100Vpk output... instead I have about 26x gain, and 29Vrms maximum undistorted output. And I'll add that I have distortion on the positive peaks of the sine wave first, so it's the tube that reaches cutoff too soon. Increasing current in the LTP didn't help, I even tried putting in an ECC81, same results of half gain than simulation.
The problem is that I can accept some volts lost from the simulation, because SPICE is a semplified world than reality, but loosing half the gain and half the voltage swing is too much, so I think there's something wrong.
Can someone help me?
poobah said:Giaime,
You should test your OPT... perhaps you are not modeling with the correct impedance?
Hello poobah,
the opt doesn't matter here, I'm talking about the first stage in the amp, the long tailed pair that has much less gain than what Spice says.
Neither the load does matter: however, all my tests are done with a dummy load on the secondary of the OPT. But... that still doesn't matter.
I've tried disconnecting the LTP from the cathode follower, so I took measurements of the LTP ALONE with NO load, it gave only a couple of volts more, maybe 32Vrms.
Also I should mention that a trusted friend of mine is simulating the amp in Spice (since I'm not too good with it) and the LTP gives 48Vrms with 2Vrms input, with 5% distortion.
Thank you all for the interest.
However the problem wasn't there! I did measure the output between one anode and ground.
My fault was that I regulated the current in the CCS (for max output and minimal imbalance) looking at the waveforms on my scope.
But my scope is so old and rusty that I shall remember to AVOID trusting him!
In fact, the balance wasn't right. I measured the output with a digital RMS multimeter, and matched it with a 0.1Vrms difference.
Now the gain of the whole driver stage (EL36 grid signal / input of the amp) is over 26, and I finally got 46.7Vrms at the EL36 grid, so I have ample margin.
It put out 8W now, I should look at the waveforms because I smell there's some AB class going on there.
So, for now, problem solved, I should start thinking to how to implement what Gianluca said some posts ago about DC coupling.
Thank you all! I need a new oscilloscope...
However the problem wasn't there! I did measure the output between one anode and ground.
My fault was that I regulated the current in the CCS (for max output and minimal imbalance) looking at the waveforms on my scope.
But my scope is so old and rusty that I shall remember to AVOID trusting him!
In fact, the balance wasn't right. I measured the output with a digital RMS multimeter, and matched it with a 0.1Vrms difference.
Now the gain of the whole driver stage (EL36 grid signal / input of the amp) is over 26, and I finally got 46.7Vrms at the EL36 grid, so I have ample margin.
It put out 8W now, I should look at the waveforms because I smell there's some AB class going on there.
So, for now, problem solved, I should start thinking to how to implement what Gianluca said some posts ago about DC coupling.
Thank you all! I need a new oscilloscope...
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.