Hi everyone.
I would like to build an SE amplifier with two parallel KT88 output tubes.
There is a schematic of what I've done so far.
I assembled one channel on test bench and it seams that it works fine. I am currently running 120 mA through transformer and eventually I will increase current to 150 mA which is the max bias for my transformers.
My concern is the driver stage. I just copied it from a similar design but I don't see that anyone else is using that driver stage. Actually I couldn't find any more designs with just two paralleled output tubes in SE amp.
Also there is no separate RC filter for driver stage. Maybe it isn't needed if CCS is used?
I am really new in building amplifiers and have no any experience with tube amplifiers but as I can understand paralleling output tubes cuts impedance on half so driver stage impedance should be lower as well? And CCS is there to inshore low distortion for driver tubes.
I understand that B+ voltage of 420 V I have with 2.5 K output transformer would suit better for UL operation but at this point I just want to make it work.
Once I learn more about tube amps I will want to reconfigure this amp into SE UL amplifier so I also want to be able to implement negative feedback.
I would like to hear opinions of more experienced people about this design before I start drilling holes on amp cabinet.
I should note that I am using high efficiency speakers and amp is more than laud enough.
Tanks.
I would like to build an SE amplifier with two parallel KT88 output tubes.
There is a schematic of what I've done so far.
I assembled one channel on test bench and it seams that it works fine. I am currently running 120 mA through transformer and eventually I will increase current to 150 mA which is the max bias for my transformers.
My concern is the driver stage. I just copied it from a similar design but I don't see that anyone else is using that driver stage. Actually I couldn't find any more designs with just two paralleled output tubes in SE amp.
Also there is no separate RC filter for driver stage. Maybe it isn't needed if CCS is used?
I am really new in building amplifiers and have no any experience with tube amplifiers but as I can understand paralleling output tubes cuts impedance on half so driver stage impedance should be lower as well? And CCS is there to inshore low distortion for driver tubes.
I understand that B+ voltage of 420 V I have with 2.5 K output transformer would suit better for UL operation but at this point I just want to make it work.
Once I learn more about tube amps I will want to reconfigure this amp into SE UL amplifier so I also want to be able to implement negative feedback.
I would like to hear opinions of more experienced people about this design before I start drilling holes on amp cabinet.
I should note that I am using high efficiency speakers and amp is more than laud enough.
Tanks.
Attachments
try it without the cathode bypass caps on the input cathode and see if you like it....
your driver is like the input stage of the broskie aikido line amps...
your driver is like the input stage of the broskie aikido line amps...
Why You will use prestage with gain more 100? Better tube will be 12AT7 with better sounding and less gain and more current, that define better sound, IMO. Parallel ECC83 is not good choice. It's enough one 12AT7 /ECC81/ in SRPP. And gain is enough for Your sensitivity speakers.
Why You need PSE, if You will use speakers with High sensitivity? IMO, SE sounds better PSE....For SE You can use the same OT with connecting speakers to 16 0hms.
And You can add switch in grid 2 of KT88 to define UL and Triod mode.
Why You need PSE, if You will use speakers with High sensitivity? IMO, SE sounds better PSE....For SE You can use the same OT with connecting speakers to 16 0hms.
And You can add switch in grid 2 of KT88 to define UL and Triod mode.
Last edited:
I never heard before a tube amp and didn't know how laud I should expect it to be so I decided to go with a bit more power.
Now as I have 20 W OT and 4 x KT88 I would like to get the best of it.
Driver stage with parallel ECC83 I just copied from a similar design and I don't understand completely what would best suit there.
Are you suggesting that I should stay with SRPP and just use different tubes or there is a better circuit I should use if I want to stay with PSE?
(I actually didn't know circuit I am using is SRPP)
Many thanks.
Now as I have 20 W OT and 4 x KT88 I would like to get the best of it.
Driver stage with parallel ECC83 I just copied from a similar design and I don't understand completely what would best suit there.
Are you suggesting that I should stay with SRPP and just use different tubes or there is a better circuit I should use if I want to stay with PSE?
(I actually didn't know circuit I am using is SRPP)
Many thanks.
Note that UL is not normally done with SE, only with push-pull.
The gain of a tube for an input on the screen is more non-linear than for input on the control grid. This means that UL with SE reduces odd order distortion, just as it does for push-pull, but it adds considerable even order distortion.
Whe UL is used on push-pull circuits, the extra even order distortion doesn't matter, as it cancels out in the transformer.
The gain of a tube for an input on the screen is more non-linear than for input on the control grid. This means that UL with SE reduces odd order distortion, just as it does for push-pull, but it adds considerable even order distortion.
Whe UL is used on push-pull circuits, the extra even order distortion doesn't matter, as it cancels out in the transformer.
Hi,
Burning lots of power(watts)(*) for 2 tubes in parallel, you may want to just use 1 tube(KT120 or 150) apply high tension of 450-500Vdc, both can withstand this voltage, you should be able to get at least 15W/channel. That's what you need really for spkrs >= 90dB.
*I just remember globe warming
Burning lots of power(watts)(*) for 2 tubes in parallel, you may want to just use 1 tube(KT120 or 150) apply high tension of 450-500Vdc, both can withstand this voltage, you should be able to get at least 15W/channel. That's what you need really for spkrs >= 90dB.
*I just remember globe warming
You can try: DIY Single-Ended (SE) KT88 / 6L6 / EL34 / 6CA7 Tube Amplifier
Good working project.....
Good working project.....
As I heard how laud can few watts be with 93 dB speakers I realize I would be happy with smaller amp and if I was starting this project now I would just make SET with 10W 3K OT and single KT88.
But I already bout transformers 20W 2.5K and 4 x KT88.
I just need someone to point me out what kind of driver stage I should look for so I can use parts I already have.
I can't get 450-500 V from my power supply as after filtration I get 420 V with 300 mA load.
But I already bout transformers 20W 2.5K and 4 x KT88.
I just need someone to point me out what kind of driver stage I should look for so I can use parts I already have.
I can't get 450-500 V from my power supply as after filtration I get 420 V with 300 mA load.
Where this "information" is from ?
I know based on my own tests that the case is fully contrary.
UL (used in SE) reduces the THD. Both 2nd and 3rd harmonics are smaller with UL than with triode at the same power level.
Also the available power level with UL is essentially bigger than with the triode connection. The difference of output impedance is not huge.
The cancellation of even order distortion with push-pull is one of the two major advantages of it (the other being the dramarically smaller amount of iron and copper for the same bass response) and is explained in a vast multitude of textbooks on amplifiers. It is easily shown by drawing the fundamental and 2nd harmonic for each tube on graph paper or with the 1st year high school trigonometry identity
sin(2x) = 1 - 2sin(x)^2.
The curved gain transfer of tubes when the input is the screen is documented in a multitude of books on tubes. It is easily seen that applying feedback into a curved transfer input is equivalent to feeding back distortion into a linear input instead of feedding back just the whole output.
It follows from these two facts that using UL with SE will increase even order distortion and with push pull it is in any case cancelled as stated.
It is possible to tweak things with resistance in series with the screen, but only very slightly, and you will always get greater distortion reduction in an SE amp with normal negative feedback than with UL feedback of the same amount.
However your 2nd statement is correct, the power output in UL is indeed greater than for triode operation, as the screen excursion is less. The output impedance is of similar order to triode output as you say.
How exactly did you do your test? With what equipment?
Last edited:
To be more precise,
Sin(2X) = 2SinxCosx, and Cos(2x)= 2(Cos(x))^2 - 1 = Cosx^2 - Sinx^2 = 1 - 2(Sin(x))^2
Pick the one you need.
Do you think I can use ECC83 as an input tube and ECC81 as driver same way as Patrick did in that amp in 1997 with 12SL7 and ECC32?
Should I parallel them?
I wouldn't use UL and NFB at this time.
Should I parallel them?
I wouldn't use UL and NFB at this time.
I don't think I want to go that far with this thread. It is more about driver stage for this amp
Thanks
I have tested SE output stages with various TV sweep tubes and also with big 813.
In all cases optimum UL gives lower THD than the triode and much higher output power.
For this discussion I made a LTspice simulation with KT120 SE triode and 50 % UL.
In both cases the conditions were as follows: +Ub = 500V, Ia = 100 mA, Ug1 = -58 V, RL = 3k5, Pout = 6,8 W.
Triode: 2nd = -20 dB, 3rd = -41 dB
UL (50%): 2nd = -22 dB, 3rd = -54 dB.
These simulation results are similar as my test results with actual circuits.
There is no sign of increasing 2nd harmonic with UL.
Another simulation. 813 SET and SE UL.
Conditions: +Ub = 700 V, Ug1 = -52 V, Ik = 135 mA, RL = 5k, Pout = 12 W.
Triode: 2nd = -30 dB, 3rd = -42 dB, THD = 3,4 %
UL (40 %): 2nd = -34 dB, 3rd = -38 dB, THD = 2,6 %
This is in well agreement with my measurements with actual 813.
UL reduces 2nd harmonic and THD compared to triode.
Ofcourse there me be some tube types that this is not the case, but I do not know such tubes so far.
Conditions: +Ub = 700 V, Ug1 = -52 V, Ik = 135 mA, RL = 5k, Pout = 12 W.
Triode: 2nd = -30 dB, 3rd = -42 dB, THD = 3,4 %
UL (40 %): 2nd = -34 dB, 3rd = -38 dB, THD = 2,6 %
This is in well agreement with my measurements with actual 813.
UL reduces 2nd harmonic and THD compared to triode.
Ofcourse there me be some tube types that this is not the case, but I do not know such tubes so far.
You still haven't explained how you did the test. Without being abile to review that, your test result means nothing.
Being misled by SPICE on distortion is a common error. The models do not necessarily model the curvature of the grid characteristics accurately, particularly as regards the screen grid.
I suspect any correlation between SPICE results and your test is due to luck with some sort of test error or other factor you haven't mentioned.
However, the preceding stage will also create even order distortion. The phase of this distortion is such that it will partly compensate for the distortion in the SE output stage. However with typical operating conditions the distortion created by the preceding stage is low enough that the compensation is ineffective. If you are operating the preceding stage under very unusual conditions, it just might be enough to confuse the situation by compensating the extra even order distortion caused by UL operation, but I very much doubt it.
UL connection with SE was never used commercially when tubes were king, and there's good reasons why:-
1) To improve performance over SE, it is cheaper to go to push pull. Sue there's an extra tube, but the transfoerm is very much smaller and the design wrt leakage inductance and stray capacitance is less.
2) It doesn't offer the marked reduction in distortion that it does with PP, as I explained.
Last edited:
Commercial considerations hardly apply to DIY, don't they?
If distortion is important to you, you're not going for SE, triode or other.
If cost is important to you, go for sand amps.
Many people have gotten good results with SE-UL. George of Tubelab gives the option to switch between triode and UL on his Simple SE design. I think he uses the UL setting himself when playing loud rock. Triode for more subtle music.
With a regular SE OPT you can use the secondary for cathode feedback, which can work very well in some cases.
- Status
- Not open for further replies.