I'm going to build a ES9018 DAC/Headphone amplifier, with Parafeed 6n6p output stage, and i have been thinking about using Twisted Pear Legato 3.1 as i/v converter - but then i started thinking that why not use tubes to increse the gain instead.
Since the DAC output have to be loaded really low (<1R) i have to have tubes with quite much gain. Will run the hole thing balanced, so there i will at least have a Little bit help.
But i was first thinking to use ECC88 as i/v gain tubes, but i seem to get a Little bit to low gain with these - so i started thinking. is there any suitable TWIN TRIODE with a bit higher gain than ECC88, that is still quite linear? ECC83 and those will of course not be an alternative since they just never sounds good.
Since i will drive headphones i will have 18:1 OPT, which is why i will need the gain.
Any help would be appreciated!
Since the DAC output have to be loaded really low (<1R) i have to have tubes with quite much gain. Will run the hole thing balanced, so there i will at least have a Little bit help.
But i was first thinking to use ECC88 as i/v gain tubes, but i seem to get a Little bit to low gain with these - so i started thinking. is there any suitable TWIN TRIODE with a bit higher gain than ECC88, that is still quite linear? ECC83 and those will of course not be an alternative since they just never sounds good.
Since i will drive headphones i will have 18:1 OPT, which is why i will need the gain.
Any help would be appreciated!
SY is right about the 12AX7/ECC83. You will use transistors to get every last drop out of the type. 
CCS load each section at 900 μA. Don't bypass the cathode bias resistor. DC couple ZVN0545A source followers to the anodes, which decouples the triodes from the driven load. Betcha the composite block sounds damned good, while providing "gobs" of gain.
The 'X7 triode is a WIMP! It can't directly drive any sort of load. The technique outlined above deals with the type's deficiencies and maximizes its strengths.
If you remain unwilling to use the 'X7, the same sort of scheme can be employed with the 12AT7/ECC81. The 'T7 triode performs well with 200 to 220 V. on the plate and an IB of 3 mA.
CCS load each section at 900 μA. Don't bypass the cathode bias resistor. DC couple ZVN0545A source followers to the anodes, which decouples the triodes from the driven load. Betcha the composite block sounds damned good, while providing "gobs" of gain.The 'X7 triode is a WIMP! It can't directly drive any sort of load. The technique outlined above deals with the type's deficiencies and maximizes its strengths.
If you remain unwilling to use the 'X7, the same sort of scheme can be employed with the 12AT7/ECC81. The 'T7 triode performs well with 200 to 220 V. on the plate and an IB of 3 mA.
I will cascode ccs the circuit with two DN2540's per anode, and also it will be differential stage with common cathode bias resistor (without cap) to cancel a little bit of distortion.
didn't think of the 12AT7 - may be an alternative, is there any russian equivalent for that one? and wouldn't 6N2P-EV be better than 12AX7 if walking that path?
didn't think of the 12AT7 - may be an alternative, is there any russian equivalent for that one? and wouldn't 6N2P-EV be better than 12AX7 if walking that path?
Agree with everything else, but my measurements indicate that keeping the cathode firmly at AC ground drops out all harmonics but second and a minuscule amount of third, and the second is already ridiculously low. I used a LED with a "helper" resistor from B+ to boost the current through it to 6mA or so.
This topology is well suited for balanced I/V:
http://www.diyaudio.com/forums/digi...s-valve-output-stage-lundahl-transformer.html
http://www.diyaudio.com/forums/digi...s-valve-output-stage-lundahl-transformer.html
Remember, a differential gain block fed by a SE signal can produce only 1/2 μ of gain. Also, CCS loading of the LTP tail is "normal".
An unbypassed cathode resistor linearizes the triode, via local current NFB. Between local current NFB and a horizontal load line, SE 'X7 or equivalent triodes will be extremely linear. For all intents and purposes, the triodes in the 6H2П-EB (6n2p-eb) are equivalent to those in a 12AX7. The internal shield might be beneficial.
2700 Ω cathode bias resistors look good to me. The TO92 case version of the DN2540 will be fine at 900 μA.
An unbypassed cathode resistor linearizes the triode, via local current NFB. Between local current NFB and a horizontal load line, SE 'X7 or equivalent triodes will be extremely linear. For all intents and purposes, the triodes in the 6H2П-EB (6n2p-eb) are equivalent to those in a 12AX7. The internal shield might be beneficial.
2700 Ω cathode bias resistors look good to me. The TO92 case version of the DN2540 will be fine at 900 μA.
oshifis: i will run the circuit differential all the way into the headphones with LL1689AM-PPZ output transformer.
The circuit will look like this - Figure 18
except that it will have an extra tube gain-stage before the one in picture - the gain stage will look the same - except that it will use caps between the first and second stage instead of a transformer. output will be possible to run in 18:1 and 18:2
Eli: the ES9018 does have balanced output, so that won't be any trouble
LPT = ???
The circuit will look like this - Figure 18
except that it will have an extra tube gain-stage before the one in picture - the gain stage will look the same - except that it will use caps between the first and second stage instead of a transformer. output will be possible to run in 18:1 and 18:2
Eli: the ES9018 does have balanced output, so that won't be any trouble
LPT = ???
Two problems here: first, the mu is high, which means you massively increase the plate resistance, which tends to increase distortion. Second, the relatively small amount of feedback transforms a very benign native distortion spectrum (at 8.5VRMS out, -85dB second, and no higher in my -120dB sight) into something with measurable higher order distortion and only 6-10dB less second.
Yes, not surprising, I've written about them many times. HLMP6000.
What about 5751 tube running >3mA with DN2540 on anode - looks like a lil bit better/lower gain than 12ax7?
As said i will run + and - output from dac, straight into each of the tubes grids. DN2540 on anode, and the anode signal output through cap into the output stage (look at schematics above).
This tube looks to be a "better" / lower gain version than 12ax7.
As said i will run + and - output from dac, straight into each of the tubes grids. DN2540 on anode, and the anode signal output through cap into the output stage (look at schematics above).
This tube looks to be a "better" / lower gain version than 12ax7.
If you are going to use a differential setup, go the 12AT7 route. Most of the harmonic distortion the 'T7 triode produces is 2nd order and that's cancelled by the differential topology.
THE place for a CCS in a differential setup is in the tail, where it forces symmetry between the 2 sides.Resistively load the 'T7 anodes with 100 KOhm/5 W. parts and use DC coupled MOSFET source followers. You'll get plenty of gain, along with very little distortion. A negative rail of about 50 V. is needed under the tail CCS. Set the CCS for 6 mA. The B+ rail should be about 520 V. A 510 Ω carbon composition stopper should go on each FET's gate.
BTW, the 12AT7/ECC81 is not a wimp and it can drive loads. Therefore, the IRFBC20 is a good candidate for the buffering FET. Regardless of the MOSFET you select, be certain to install a Zener diode between gate and source to protect against start up transients.
I haven't built stuff with the AT7, but wit AU7, 6N6P, 6N30P-DR and such tubes, and i have always liked the anode CCS. I can get hold of matched DN2540's close (within 1%) so current matching is no problem that i even use resistors normally, not pot for setting. Also, i wouldn't want to have as high B+ as i already will be useing 6n6p in the output stage. I'm aiming B+ at around 250V max for that stage (they will be anode loaded CCS). I think anode loading will be the way to go for the input stage also - that'll give me lower output impedance, which makes me not needing the buffer between the stages(?)
Why would it be better loading the cathode? less PSRR/isolation also (since next stage will have same HV-supply) - the only thing i can think of is easier to balance the current between the channels with a Centertap output of a pot?
Why would it be better loading the cathode? less PSRR/isolation also (since next stage will have same HV-supply) - the only thing i can think of is easier to balance the current between the channels with a Centertap output of a pot?
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.