Hi All
I know that the differential circuit that is low noise so i want to use vacuum tube to design an input circuit. This is only ideal and i haven't seen some one use type of circuit for input circuit. Is it not useful or not economical....?
Can anyone help me to answer this problem??
I know that the differential circuit that is low noise so i want to use vacuum tube to design an input circuit. This is only ideal and i haven't seen some one use type of circuit for input circuit. Is it not useful or not economical....?
Can anyone help me to answer this problem??
Its quite common to put a LTP up at the front end of many valve design's. A lot of these use an input transformer to drive both inputs of the LTP. There are a good few designs out there which are differential front to output.
http://www.nutshellhifi.com/triode1.html
http://www.vacuumstate.com/images_upload/pp_1c_s.gif
Shoog
http://www.nutshellhifi.com/triode1.html
http://www.vacuumstate.com/images_upload/pp_1c_s.gif
Shoog
Last edited:
Another differential amplifier...
An externally hosted image should be here but it was not working when we last tested it.
Much higher than most since there's lots of local feedback around the output valves.
Built a few versions of Garys designs and they all sound fantastic with solid bass.
Shoog
Yes, almost 100% looking at the zout of the phase splitter 😎
That's why I asked for a number 😉
That was a long time ago but IIRC, the output impedance was in the 2 to 3 ohm range.
The Tabor is not functional at this time. It got part way through a conversion to interstage coupling and got stalled.
The 47 P-P is working. Might be able to measure the output impedance this weekend.
I've had good luck using the LIMP module of ARTA to measure amplifier output impedance over frequency.
The Tabor is not functional at this time. It got part way through a conversion to interstage coupling and got stalled.
The 47 P-P is working. Might be able to measure the output impedance this weekend.
I've had good luck using the LIMP module of ARTA to measure amplifier output impedance over frequency.
Is there any advantage using LTP differential amp as input for power amp? I was thinking about this by I am leaning towards using 12AX7 with MOSFET mu-follower as load instead. Seems like using mu-follower gives lower distortion and provide a low impedance output to drive the miller capacitance effect of the next stage.
Generally, differential amps are less sensitive to noise supply, are more linear (H2 tends to cancel) and much more !
I have found my differential amps to have few issues with power supply noise. A lot depends on the specific design. That is to say if you create a wimpy front end passing just a mA or so of plate current through a high impedance triode - then its not going to be able to drive a final. Yvesm is correct in saying that unlike SE front ends which will distort more as the output signal increases, a good LTP will tend to intrinsically cancel out its own distortion because it is distorting fairly equally in antiphase at the outputs.
However you can push 20mA through a 5687 LTP and find it will have absolutely no issues driving the miller capacitance.
So if you design a LTP with enough grunt for the application I would say it has the edge on producing a lower overall distortion figure. However if you are looking for that warm valve sound (high 2nd harmonic) you would be best to avoid them.
Shoog
However you can push 20mA through a 5687 LTP and find it will have absolutely no issues driving the miller capacitance.
So if you design a LTP with enough grunt for the application I would say it has the edge on producing a lower overall distortion figure. However if you are looking for that warm valve sound (high 2nd harmonic) you would be best to avoid them.
Shoog
Not economical. Putting the LTP "up front" is quite common in solid state practice, and so that's what I do in hollow state designs. (Attached) You can also implement with small signal pents like the 6CB6, 6AU6, or 6JD6.
(FWIW: I would prefer these in the order of: 6CB6, 6JD6, 6AU6 based on projected harmonic profile. The 6CB6 tends to make more h2, but no h3 or h4. It should be clean in LTP operation where your even harmonics null out. The6JD6 comes closer to a harmonic "waterfall" pattern here, with the 6AU6 being close behind. Other pents that might work well include the 12BY7A, but that one really likes some bigvolts on the rail, as it was designed as a color TV video amp, and there's no shortage of rail voltage inside a color TV chassis.)
The other consideration is that when using vacuum tubes, the gm is orders in magnitude smaller than with transistors, and so you can have an AC, phase-to-phase imbalance problem. This requires a fix of unequal plate load resistors for AC balance, or an active tail load to increase Rtail and enforce plate-to-plate balance. This is much easier to do these days since you can SS the active tail load.
The main problem here is that this design requires two dual triodes, another hole in the chassis. A lot of commercial designs would rather use one dual as voltage amp + cathodyne, as this design has higher gain (you lose half your gain in an LTP). It's also a reason why the paraphase/floating paraphase are also favored: no need for a negative rail, the cathodes remain near ground potential, so no need for DC heating power or floating the heater supply lines.
Since this is a DiY design, I don't have to answer to any pencil-pushers in the accounting dept. Therefore, it doesn't matter if it needs an extra dual or a negative rail. That means I can include other seldom seen enhancements, such as active screen regulation.
Attachments
Last edited:
There's lots of reasons to have an LTP front;
1) Topology. Easiest and I would argue most linear way to balance the signal (assuming unbalanced input).
2) Distortion. Properly balanced LTP has lower distortion.
3) Noise. Properly balanced LTP will be completely impervious to B+ (or B-) noise. I would still advise regulating them.
4) Constant current draw. This is pretty much the same thing as the last point, but I feel merits mentioning. It really simplifies PSU design when the whole amp is at all times in constant current draw mode.
Constant current draw seems to improve transient response especially in my experience, when I compare very similar SE and balanced circuits. Both have similar PSUs, with B+ (and all other supplies) regulated, and with a 10µF paper in oil cap after the regulator, and all stages MOSFET plate loaded, with low impedance driving the OT. In these similar conditions the balanced amp seems to take a definite victory in the transient department ("sensation of moving air").
Lower overall distortion could be the reason, but I suspect it's the constant current draw aspect that is inherent in fully balanced Class A circuits.
Edit: Both circuits with no gNFB.
1) Topology. Easiest and I would argue most linear way to balance the signal (assuming unbalanced input).
2) Distortion. Properly balanced LTP has lower distortion.
3) Noise. Properly balanced LTP will be completely impervious to B+ (or B-) noise. I would still advise regulating them.
4) Constant current draw. This is pretty much the same thing as the last point, but I feel merits mentioning. It really simplifies PSU design when the whole amp is at all times in constant current draw mode.
Constant current draw seems to improve transient response especially in my experience, when I compare very similar SE and balanced circuits. Both have similar PSUs, with B+ (and all other supplies) regulated, and with a 10µF paper in oil cap after the regulator, and all stages MOSFET plate loaded, with low impedance driving the OT. In these similar conditions the balanced amp seems to take a definite victory in the transient department ("sensation of moving air").
Lower overall distortion could be the reason, but I suspect it's the constant current draw aspect that is inherent in fully balanced Class A circuits.
Edit: Both circuits with no gNFB.
Last edited:
Hi Jaap, not really wanting to display the schematic of the transformer coupled Tabor as I consider it as an experiment that did not work. It will most likely be converted back to the DC coupled state.
LTPs can also make for a great sounding output stage. Granted 38W/Chn. from four KT-120s is not very efficient, but running them Class A triode differential with a high impedance current source "tail" makes for very engaging sound.
All of the HiFi P-P amps that I have built in the past 10 years have been fully differential from the input to the output (except a couple that were someone else's designs). I think they just sound better.
I made a P-P amp with three cascaded LTP stages all class A with a CCS in the tail. The first two stages were 6SN7's and the outputs were KT88's. About 12 watts in triode, and 25 in UL, but they were VERY NICE watts! I never assembled it into a chassis, but the breadboard is still in a box somewhere. I wanted to try it with 300B's before deciding on which one to build.
I made a P-P amp with three cascaded LTP stages all class A with a CCS in the tail. The first two stages were 6SN7's and the outputs were KT88's. About 12 watts in triode, and 25 in UL, but they were VERY NICE watts! I never assembled it into a chassis, but the breadboard is still in a box somewhere. I wanted to try it with 300B's before deciding on which one to build.
I read a raving review of a very high end amplifier Vacuum State. They won award in 2003 with their amp using differential PP output stage. It is a triode PP LTP. I am thinking about designing my amp using this configuration. But seems like a lot of people don't think it's a good idea. Attached is my conceptual design
Attachments
Last edited:
Alan,
Had a quick look at your conceptual design.
Did'nt see any "show stoppers".
Running a single CCS in the output tube cathodes means strictly Class A operation. You may want to lower the B+ a fair bit so that the EL34 can run at higher idle current to better suit Class A. As shown (200mA CCS) the EL34s are way over dissipating.
2mA per side in the front end diff amp is too high for a 12AX7, drop that to 1mA per side.
Not mad keen on the "in-line" zener dropper in the MOSFET Follower drains but it would probably work.
Allan Wright's (RIP) DPA300 amp used 300B filamentary triodes with a single CCS in the tail like you have shown with differential cascode 6DJ8 front end, that is, a differential amp where each side is a cascode rather than a single triode. I would expect that to behave VERY similarly to the 6AU6 diff. amp above.
Cheers (Merry Xmas All),
Ian
Had a quick look at your conceptual design.
Did'nt see any "show stoppers".
Running a single CCS in the output tube cathodes means strictly Class A operation. You may want to lower the B+ a fair bit so that the EL34 can run at higher idle current to better suit Class A. As shown (200mA CCS) the EL34s are way over dissipating.
2mA per side in the front end diff amp is too high for a 12AX7, drop that to 1mA per side.
Not mad keen on the "in-line" zener dropper in the MOSFET Follower drains but it would probably work.
Allan Wright's (RIP) DPA300 amp used 300B filamentary triodes with a single CCS in the tail like you have shown with differential cascode 6DJ8 front end, that is, a differential amp where each side is a cascode rather than a single triode. I would expect that to behave VERY similarly to the 6AU6 diff. amp above.
Cheers (Merry Xmas All),
Ian
Last edited:
Thanks Ian.
Do you know what is the advantage of using differential power tubes over the convention grid bias with ground cathode?
I was thinking about cascode LTP for the first stage, but just don't feel like floating another tube at over 100V. That kills the headroom. I did cascode on a 12AX7 before using a BJT, still I have to hold the base at at least 100V.
I am open to a mu-follower first stage to drive the LTP to get more loop gain. The mu-follower is going to use a N MOSFET for the top follower part instead of a tube floating on top. Also the low output impedance of the mu-follower makes lowering the miller cap of the LTP unnecessary.
Thanks
Do you know what is the advantage of using differential power tubes over the convention grid bias with ground cathode?
I was thinking about cascode LTP for the first stage, but just don't feel like floating another tube at over 100V. That kills the headroom. I did cascode on a 12AX7 before using a BJT, still I have to hold the base at at least 100V.
I am open to a mu-follower first stage to drive the LTP to get more loop gain. The mu-follower is going to use a N MOSFET for the top follower part instead of a tube floating on top. Also the low output impedance of the mu-follower makes lowering the miller cap of the LTP unnecessary.
Thanks
- Status
- Not open for further replies.