Ideas for using a pair of ECL82 in a headphone amp.

Shoog

Member
2002-08-15 10:16 pm
Eire
Hi there,
I have a few nice ECL82 and don't really need a flea powered SE amp, so I was thinking of using them as a headphone amp. I have done some searching but can't find any schematics for them in this application, though I have seen them mentioned as suitable in this role.
My original idea was to just take the output off the cathode, but this just wouldn't provide enough current for a lot of headphones.
Has anyone tried this and what were there results?

Thanks
Shoog
 
Hi Shoog,

I have a few nice ECL82 and don't really need a flea powered SE amp, so I was thinking of using them as a headphone amp. I have done some searching but can't find any schematics for them in this application, though I have seen them mentioned as suitable in this role.

Well, even when triode strapping the output pentode of ECL82, it is not particulary linear, see this link.

My original idea was to just take the output off the cathode, but this just wouldn't provide enough current for a lot of headphones.

My best idea would be to build a "real" SE amp with it, output section triode strapped, Po about 1-1,5 watts. Use a stereo headphone jack with an integrated DTDP switch to cut off the "speaker output" and insert a 10 ohms dummy load across the secondary, to which the headphone is paralleled. You can drive any headphone even with extreme low impedance with this setup.

To get the picture, take a look at the schematic and description of my own headphone amp:

http://www.tubes.mynetcologne.de/roehren/pcl84pp/schematics_e.html

You can find an ECL82PSET on my homepage, too.

Tom
 
The ECL82 pentode triode-connected is perhaps not as linear as a 45 or whatever one might compare against. However, it's not bad. I've used it for a SE transformer-coupled amp with partial (plate) feedback and I get good sound for headphones too with a voltage divider off the 16-ohm tap (not ideal, but HPs need very litte power, so the sound is good and loud enough!). So I second the OPT suggestion.
The major issue, IME, with HP amps is noise more than distortion. Whether OPT or OTL, you'll get enough power/low enough distortion. With OTL, 300-ohm HPs could be direct connected, while 32-ohm ones might need a 270 ohm (or so) series resistor on the output to keep noise/sensitivity down. (Damping factor goes out the window, but with HPs is usually not a problem - if found to be, a 15-ohm resistor i parallel to the 32-ohm HPs raises DF to about 2, which solves that "problem".)
My OTL HP experiment was with EL86 rather than ECL82, but it's not so different. HT noise and heater hum are the major issues, plus avoiding ground loops (hum/buzz). Active regulation or heavy filtering solves the first two.
With ECL82 in triode mode and OTL, the criterion is linear current swing; you'll only need a few volts swing, so that's not a problem. 100Vpk and 30-40mA idle current on the ECL82 looks pretty linear to me (loadline is almost vertical), and should be sufficient for most HPs. You could use two pentode sections, upper triode connected, lower as pentode current sink. Or both in triode mode in White CF connection. Or just one tube with a silicon CCS. (Broskie's Aikido tricks might come in handy to get the noise down with WCF.) Gain may be sufficient without an input stage, just the CF. The ECL82 triode section has high mu, which you don't need/want, but it might be used as an input/feedback return stage, esp. for lo-Z HPs. (You prob. don't want a closed-loop gain of more than 5 or so.)
I can't upload a schematic right now, but if you're interested in a ECL82 long-tailed pair input/white CF output OTL HP amp schematic, send an email. I have both a direct-coupled version with +/- PS and a cap-coupled with single supply. Untried in reality as yet with ECL82, but same circuits work well with 5751/EL86, so shouldn't be a problem.

morgamp1 'at' yahoo 'dot' se

Morgan
 
ecl82 / 6bm8 headphone amp

Hello everyone,

I have built several SE and PP headphone amps with this tube and the one I like the most (and the one still in active service) is a design coming from Japan, you can see it here:
http://www.za.ztv.ne.jp/kygbncjy/tubeamp/tubeamp.html.
I have used the 6bm8pp design as a headphone amp, a very good design - in my humble opinion, and a very good sound. This is my main headphone amp - very recommended.

While you are in that site you can take a look at the other designs I think you will find many nice ideas.

Good Luck.

Glass_painter
 

Shoog

Member
2002-08-15 10:16 pm
Eire
Thanks for the detailed reply - they have touched on a few of my thoughts.

"Or both in triode mode in White CF connection. Or just one tube with a silicon CCS. (Broskie's Aikido tricks might come in handy to get the noise down with WCF.)"

I was looking at a white cathode follower using ECC88's on the headwize site. I was thinking of something similar but using some PC84's I have for one of the output valves, along with the ECL82's.. Seems as though this requires careful optimisation to get a satisfactory output power and minimum distortion. I'am not certain I'am up to the challenge.
I have built a MU stage SE amp with TT21's as the top current source, and 807's as the bottom output valve. I am happy with the result and don't see the limitations of not having a truly constant output impedance (though I do use simple to drive full range speakers). I burn a wopping 80V on current sensing resistors to make it approach a true CCS. I was thinking about using this topology as the basis of my headphone amp, and since these have fairly flat loads it might be satisfactory. The advantage of this circuit over the WCF is that it kills power supply noise dead with minimum power supply filtering. Any thoughts.

Another thing, I don't have any headphones at the moment so I can choose the headphones to suit the project.

Shoog
 
Hi Tom,:wave2:

Do you think your EL34SET with the Hammond 125ESE would be a good headphones amp?

Use a stereo headphone jack with an integrated DTDP switch to cut off the "speaker output" and insert a 10 ohms dummy load across the secondary, to which the headphone is paralleled. You can drive any headphone even with extreme low impedance with this setup.

I am going to try this with my JE Labs 300B. My headphones are Sennheiser HD650 (600Ohm).
What output of the OT (Hammond 1627SE) should I use? 4-8-or 16 Ohm?

Thanks :)

Manuel
 

PRR

Member
Paid Member
2003-06-12 7:04 pm
Maine USA
www.diyaudio.com
> just take the output off the cathode, but this just wouldn't provide enough current for a lot of headphones.

It can give current comparable to a common chip op-amp, and many people are happy with such things as headphone amps, especially for loads >>32Ω.

The cathode output resistance is ~140Ω, which means THD will be quite high when working hard in 32Ω, and the low-THD load should be well over 300Ω, higher than most phones.

Adding a transformer solves both problems. Use a 3K:8Ω or 20:1 tranny, output current is 20 times higher which is more than ample for any headphone; load is sufficiently high to get into low-THD zone.

A third feature is: you need less B+ power, so filtering is easier; and the ~20:1 transformer gives ~20:1 reduction in ripple at the load.

The basic problem is that vacuum conducts juice a million times worse than metal. Even if you flatten the vacuum tube structure (wide cathode, short plate spacing) and stretch the metal to long fine wire, a tube has 1K-10K resistance, a motor has 10Ω-100Ω resistance. Even a 6c33 is not the most efficient thing driving 32Ω. 300Ω phones have fine fragile wire, you don't want to go higher. Tubes need transformers to drive motors well.

The cute thing about headphones is we can do terribly wasteful things and not go broke. Instead of a good DC-rated transformer, resistance-load the tube and use a cheap core. Let's see.... 200V supply, 200V G2, 30mA, 100V on the plate, 100V/30mA= 3K 5W plate resistor. Transformer sees ~1K source, so it does not need a lot of Henries to go to deep bass. A 230VAC:12VAC power transformer and 10uFd coupling cap will rumble fine. A split-bobbin power transformer will have heaps of leakage inductance and poor treble, but a one-bobbin transformer may be OK, especially in the higher-Z loads. Or use some small core that would melt at 30mA DC but can handle milliWatts of DC-free audio.
 
Hi Manuel,

Do you think your EL34SET with the Hammond 125ESE would be a good headphones amp?

Frankly, I would consider it complete overkill and not suited very well. Even the 1 Watt headphone PP amp I built http://www.tubes.mynetcologne.de/roehren/pcl84pp/pcl84pp_e.html still is overkill, but it works excellently. If asked, I always would suggest and favour a "real" very small mini triode PP class A amp instead of an SE one. Too many objective advantages that cannot be ignored.

I am going to try this with my JE Labs 300B. My headphones are Sennheiser HD650 (600Ohm).
What output of the OT (Hammond 1627SE) should I use? 4-8-or 16 Ohm?

If you really want to do this, attach a nominal dummy load at the accordingly specified secondary tap. If in doubt, use 10 ohms dummy load at the 8 ohms tap and strap the headphone in parallel to the dummy resistor. Very low Z headphones like my 50 Ohms Sennheiser will lead to an effective load of about 8 ohms for the amp, while for high Z headphones (like yours) the mismatch is slightly less than 2 ohms in this situation and safely can be ignored. Any speaker is much worse as a load than this dummy-res/headphone combo.

Watch out with the Pd the dummy res must be capable to dissipate.

Watch out for your ears and headphone, since some milli-Watts will give extreme soundpreassure levels already.

Tom
 

Shoog

Member
2002-08-15 10:16 pm
Eire
"The cute thing about headphones is we can do terribly wasteful things and not go broke. Instead of a good DC-rated transformer, resistance-load the tube and use a cheap core. Let's see.... 200V supply, 200V G2, 30mA, 100V on the plate, 100V/30mA= 3K 5W plate resistor. Transformer sees ~1K source, so it does not need a lot of Henries to go to deep bass. A 230VAC:12VAC power transformer and 10uFd coupling cap will rumble fine. A split-bobbin power transformer will have heaps of leakage inductance and poor treble, but a one-bobbin transformer may be OK, especially in the higher-Z loads. Or use some small core that would melt at 30mA DC but can handle milliWatts of DC-free audio."

Hi there PRR,
Thats just the sort of thing I have been thinking of. I like the idea of only needing two tubes and a rectifier. Using mains transformers has worked for me on my Parafeed 807 amp so I don't need to be convinced that this can work satisfactorily. I have access to some nice small potted 12V toroidals which should do an excellent job.
The only thing I am not entirely certain about is using just a 3K load resistor. A constant current source would allow the ECL82 to perform to its best (as poor as that is).
The only problem is finding a suitable pentode/triode for the CCS. It would need a DC resistance of about 3K so that it consumed about 100V. I was looking at the PL84 in this role, but it doesn't look to have a happy operating point at about 30mA and 100V. I was also looking at the 6aq5 (a mini 6V6) but again it wants to see more voltage on its anode. I really want to use some of the valves I already have, so exotics are not ideal.

Do you think I would see a significant increase in performance using the CCS rather than the 3K resistor?

Thanks for the great input.

Shoog
 
I second PRR's idea.

Shoog wrote:
'The only thing I am not entirely certain about is using just a 3K
load resistor. A constant current source would allow the ECL82 to
perform to its best (as poor as that is).'

Triode-connected, it won't be bad at all. And a little bit of shunt NFB
(think Pass' basic Zen) might improve it further. Considering how
little power a headphone needs, I think the somewhat wasteful
3k resistor will be fine. But I think a larger-ratio transformer might
be better for HPs (230:15 for 30-ohm and 230:22+22 for 300 ohm
HPs perhaps?).

'The only problem is finding a suitable pentode/triode for the CCS.
It would need a DC resistance of about 3K so that it consumed
about 100V. I was looking at the PL84 in this role, but it doesn't
look to have a happy operating point at about 30mA and 100V. I
was also looking at the 6aq5 (a mini 6V6) but again it wants to see
more voltage on its anode. I really want to use some of the valves
I already have, so exotics are not ideal.

Wouldn't the obvious choice for the CCS be another ECL82?:)
Running it in pentode mode might require an extra power supply
for the screen, though (it must be bootstrapped from its cathode).
In triode op. (200 ohm cathode resistor), the resultant CCS would
only have the same dynamic resistance as a 3k resistor - bummer.
We were approaching the tube version of the Zen with CCS load!

It looks like a more workable design, if you want CCS, would be to
stack a driven CF triode-ct ECL82 pentode on top of another one
pentode-ct as current sink. Unbypassed cathode resistor, or better,
a depletion MOSFET (or maybe even an LM337 or 317?) in cascode
with the CCS pentode. Screen current via dropping resistor from B+.
You would need a gain stage, but no problem, just stack the
unused triode sections SRPP-wise and direct-couple to the CF.
Too much gain? Make a common-cathode input stage and use
the inverting input for a touch of NFB. But I'm rambling...

Just my SEK 0.02

Morgan
 

Shoog

Member
2002-08-15 10:16 pm
Eire
I'am now thinking of lashing it up with just the 3k resistor and a single ECL82 per channel. I really like the simplicity and elegance. If the thing need a bit of feedback I will probably mess about with a bit of "Partial Feedback".
Your right that using another pair of ECL82's would allow a PRSS driver stage and a CCS loaded output stage, but I only have two tubes at the moment, and this whole project is an exercise in seeing "what if" without spending much/any cash.
I am hoping that at such low power outputs the distortion wont ramp up to much, and what distortion there will be will present itself as 2nd order. The idea of mellow lush sounding headphones appeals to me much more than harsh and intrusive.
I have the sujested circuit draw up ion scrap paper - but its so simple its hardly worth drawing up for general consumption.

The only impedament to getting started is that my friend who has the suitable 12V toroidals has them somewhere in a huge pile of boxes and hes not willing to look until November. I will test with whatever I can lay my hands on until then.
 
Sounds like a good idea to me. And I think you should triode-
connect the pentode section. Not sure what PRR meant by
200V on the screen, 100V anode...? Surely, 100V on the
screen is enough (and this of course is what you get in
triode mode with the anode sitting at 100V). If you go
for pentode mode, a dropping resistor from B+ that sits
the screen at about 100V and a bypass cap to ground
are called for. The gain might then be enough to do with-
out an input stage. But if you apply partial (shunt) NFB,
gain may be too low. If triode-connected, it may also be
a bit low. I'm too lazy to calculate, but anyway you'll see
for yourself.
Carefully balanced and biased AC heaters should be
quiet enough, though (current-) regulated DC might be
even better. Triode mode and/or shunt feedback will
worsen PSRR, so better filtering is required than in
pure pentode mode. If you want lush/dark, you'll get
that in spades in pentode mode w/o feedback. But,
I'm guessing, a bit too much...
Hammond BSE125 output transformers are suitable
and not too expensive if the 230:12V idea fails.

Good luck with your project!

Morgan
 

Shoog

Member
2002-08-15 10:16 pm
Eire
I was thinking triode connected with a 100R resistor from g2 to the anode.
I will also try plate to plate feeback with a 100K resistor from the anode of the triode to the anode of the pentode section. I will initially lash the thing up with other transfomers until the toroidals become availble.
I was originally going to use a EZ80 for rectification, but I will only have 240VAC so it doesn't give me enough headroom to do the required filtering. I will now use a BYV44 400 which is an ultra fast soft recovery diode pair. Might help to keep the edge on the sound.

Shoog
 
planet10 wrote:
"Or the IXYS CCS chip?"

Yes, but maybe I was unclear. I meant tucking an LM317 between
cathode and ground of an ECL82 pentode section, to improve the
constancy of the current provided. There's only about 5-6V there,
so an IXYS wouldn't be happy, if I read the right data sheet in all
haste. It's a high-voltage part, and needs about 10V to regulate
properly.
(I bet it's much more expensive too, but that was not my point.:) )

OTOH, used alone as the cathode current sink of an ECL82 "on top",
the IXYS would be ideal. (Oh, as a plate load too, certainly - as an
alternative to that 3k resistor in PRR's design example, for instance.)

Morgan
 
Shoog,

100k anode-to-anode *may* be OK, but note that the output
stage effectively "sucks" (I mean sinks - and sources) quite a bit of
current from the input stage due to the shunt feedback action.
No free lunch...
I guess you will need more than 3 mA quiescent current in the input
triode (quite a lot for that valve section), which it turn will require
quite a bit of anode-cathode voltage to bias the triode properly.
I'd back off to 150k-180k to make sure the input doesn't clip before
the output. A 'scope or good ears will tell you how far you can go,
but I thought I'd better alert you to this possible problem. Like me,
you may have read, but not quite grasped, JRB's article on partial FB
at the TubeCAD site. I wish he'd put a bit more effort in writing that
particular one - there are some vague bits -, but maybe there are
other references to PF calcs. (BTW, I'm not criticising John - I know
he has too little time, and still manages some fabulous articles about
topologies hitherto unknown to man...:D )

Morgan
 
Morgan L said:
Lo Scherzo looks just fine per se, but uses global FB, so "results may
vary" unless you use the specified OPT. Starting with partial FB is
safer. (Not too fond of the starved input stage - no doubt to maxi-
mise gain, in turn to make FB more effective. I'd rather pop up the
current and use less FB. E/PCL82 are cheap.:) ) Just my SEK0.02.

Morgan

Lo Scherzo has many configurations.
You can use it with or without FB, in triode mode or ultralinear. On the pc-board there are 2 jumpers for make these changes in simple way.
I made one many years ago(with junkbox components, I only ordered output transformers, because I know the story and the quality of transformers of these kits...all these kits are from the magazine "costruire hi-fi", and transformers come from an ambicious project by Gruppo Euterpe...).

I used it in triode mode without feedback with grado RD-60 headphone and a receiver Meridian 504 for a long time. I remember with nostalgy these nights listening a wonderful music to radio...