JLH 10 Watt class A amplifier

Q3 Substitute?

I'm collecting parts to build an updated- JLH. Both the updated and high-power ESL versions call for a 2SC3421 for Q3. Mouser.com shows a TIP47 as a cross-reference match for the 2sc3421. Are there any opinions out there as to the suitability of this substitute?


Cheers.

Ryan
 
hello,

after building my amp a jlh variation , the DoZ

using as output devices a pair of tip3055, i decided to improve it by puting another pair of tip3055 (so now we have 2 pairs of tip3055)

as i found the improvement was so great i even decided to put another paraleled pair, wich has improved the "precision" of the sound even more.

today i will probably try another pair, but this is getting expensive...

do you thik that if i try with 10 paraleled pairs of ..lets say bd139
the sound will be even more precise, since those are faster and there will be more of them so they will work "lighter"??

thanks for the answer, see you later!:confused: :confused: :confused:
 
I built a headphone amplifier in class A with BD139/BD140.

Based on an Elektor class A headphone amplifier, but with 8 output devices.
After the first design Elektor came up with a second edition of the headphone amplifier a number of years later with 4 output devices instead of the original 2.

Mine delivered the entire output in class A, i used the amplifier on my DIY loudspeakers for some time.
I used the BD versions with the highest Ft for the amp.
It was my first class A amplifier and the sound was astonishing, even with only a 24va transformer and double LM regulators.

Seemed to me that part of the sonic quality was due to the extreme bandwidth of the BD devices.
Graham has the 3281 in his design. Very hard to come by, pretty expensive, and BD bandwidth is 4 times higher.
Sounds like a great idea to me.
 
wel the current in the DoZ version of jlh is controlled by a transistor so the shift in quiescent current was really very little if there is any, furthermore im doing no matching at all with transistors.
may be if you use a lot of transistors in the output , statistically the mean Hfe are very alike because of the large number used it tends more to a gaussian dsitribution.
may be there is a "chorus effect" , after all its more nice to hear a chorus of 20 singers tan only one (well i know it depends of....),
anyway i like the chorus of 8 transistors i have created :D

i'm using 30v at 1,5A , a little more than the recommended , because i think there is a small difference in sound definition.

the high ft of the bd139 is good for the resolution of the sound, but i'm afraid that the stability of the amp will be severely damaged...

i have heard of some expensive "faked" a class that claims to have 16 output transistors, but those are mosfets, and i didn't like that sound at all.

reading more i find that there is a builder that has constructed the jlh for esl but that used it actually for non electrostatic loudpseakers, common 8ohm speakers, and he has compared that with his friend version that uses 4 mj15003 instead of the 6 he actually uses , he has reported an improved sound too.

i wonder if geoff still reads this forum :) mey be he could bring to us some light .

:)
 
Hola,

There is company here in Holland that builds amplifiers under the name Array.

The amplifiers have an encridible number of paralleled small power devices, they claim sonic benefit.

For my headphone amp i selected BD's from one batch, turned out there were only a couple that had a large deviation in Hfe.
With 0.50 ohm emitter resistors( 2*1 Ohm parallel) i thought it good enough to use the rest, with even average Hfe for the 140 and 139.
Mosfets in full class A are ok, class AB Mosfet stages are not.
(only my opinion)
BD's are so small that many can be placed close to eachother and signal traces are still small.
Threshold made solo amplifiers with 64 output Mosfets.

Hasta.
 
Good News!

i've finished the first test with 11 pairs of bd 139,after the whole day cuting wires and soldering...

The result is incredible!!! i have not listen to something like that long time ago...
much more precision in the voice, the bass are perfectly defined,an incredible improvement, no sign of unstability, increased sensibility...
this amp is a dream... mi final goal are 24 pairs (yes 48!!)
bd139, to give a good safety margin, because i think that the sound i so smooth and well defined that i don't think that the increased number qill give a sound improvement.

and this is cheaper , since an tip3055 equals more less 10 bd139

and the bd139 are more tan 10 times cheaper ....well i cant see any shortcommgs.

the sensitivity of the amp is greatly improved, this has to be treated carefully, for example if i put my hand 1 cm near the heatsink or the pcb some hum pases the loudspeakear, but i know how tu cut this pb out... so for now i am extremelly happy....

so i recommend you to give a try to this mod ( replacing hiagher power transistors by a lot of small ones)

i repaet the improvement is real , and very noticeable, i whish you could hear the diff on my amp!!!

if someone is interested in a more precise description of how this mod has improved the voice playback, just ask for it , i will be very happy to explain so.

bye , and i hope you can try it some day ....
 
The one and only
Joined 2001
Paid Member
jacco vermeulen said:
There is company here in Holland that builds amplifiers under the name Array.

The amplifiers have an encridible number of paralleled small power devices, they claim sonic benefit.

I built a nice little amp some years ago (last seen in the
possession of Bob Lord) which consisted of arrays of TO-92
devices (a hundred or so) each operating at about 10 mA,
Class A of course. He loved them, but not enough to buy a
second pair.

:cool:
 
Nelson i was thinking about putting a hundred or two of bc559 working togheter, since there is so many the heathsink is not an issue, in fact not needed at all right?


about my listening impressions, and level of detail, i have a very nice MD recording of cindy lauper, in the song altrought the night
you can hear the soft sound of her tongue hiting the lips and tooth while she is singing and pronouncig the "th" sound, i was able to hear that only in a class A GOOD amp.

my schematic is very simple, the doz of rod eliot, but instead of paraleling 2 tip3055 i have paralleled 22 bd139.

just conecting them togheter, of course i put a resistor in each emiter 1 ohm resistor, this could seem large , but there is so many devices that i doesnt matter.
i can describe in another way the sound:

when you are adjustind the quiescent current , and the quiescent current is too smal the amp sounds awfull, like a trash can, as you increase the iq sound is more detailled an more authoritative,
well that whats happened, when i shifted the 3ºpairs of tip3055 for the bunch of bd139 with the same current, the impresion was the same than i have explained before, and "saddly" when i think of what my amp used to sound before, i think the sound was like a "trashcan" .

i wonder if i replace the bunch of bd139, by 10 times more bc559, will the amp sound even better?

:smash: :smash: :smash: :smash: :smash:

i like to know wich was the kind of a classA circuite you used Nelson single ended i presume but no mosfets in it right?

and i also like to know the more important reason of why the amp
sonds actually better, is there any 2d order harmonics effect?

personnaly i don't think so because the saound is more warm but also more "fine"...if i close my eyes i can even feel the lips of carly simon vibrating...
i can hear as i have said before clearly the sound of te tongue softly hiting the lips an then the tooth,that is for real!



Maybe the reason is more likely to be the statistical reason i have outlined before the , way that when the size of the sample of a given population is somewhat near 30 then the distribution tends rapidly to be gaussian distribution , with a clear maximum near the center, not like the t-student that is used for smaller sampling sizes.
that is one of the reason why in the school even if each student has a bad voice, the whole class singing toghether sonds almost always fine ( but there i admit there is a loss of resolution so this theory lacks something important)
Maybe someone can show us some light on the concept itself?

thanks to all that are interested by this and will give their opinion
(i apologize for my poor english , i can speak correctly only french and spanish)

see you soon friends!

:D
 
why do something that has already be well done?

http://diyaudio.8m.com/Bebet/bebet_sch.html

this page is ok
but do NOT put paralleled transistor as this author said
to paralel them use only one resistor in the emiter conection.
(the author uses somewhere else a resistor in the colector too, this is not a good thing to do)


there is one pair of transistor to paralel another pair just conect base to base, colector to colector , and emitor to a resistor then to the correspondig pcb conector.

very easy project, hope you can complete it soon to exchange opinions :D
 
hello Graham,

here i am posting some benefits of replacing ONE power device by several , as much as possible , much smaller devices,
that is all about this...

until now i have used only 3 paralelled power devices, but now i can see that benefits of paralleling a lot of "small power devices"

11 pairs are just enough but i like to have a good reserve of power in my outpute devices so i calculated in a total of 24 pairs of bd139.

and a as future project , i am considering to put 200 bd560 working in paralel, for a class a this will be a sure benefit as for clss ab. but we have to take care of stability i know....

some exeprimenting is still necessary.

further more i hve bught the toshiba you are talking about in 4 differents shops, all were fake . so i wil never use again those.

any new idea or explanation is welcome.





;)