Well guys, the presence of this man, is something that may turn Michael so happy as
As i am....
Dimitri has enormous International experience in electronics, his degrées papers, University Certificates will be shocked compared to many of us.
This man, some of the Russian Brains that are turning more rich the US know how, are someone very special, he is the one that is called when no one can find solutions for hard problems.
His presence here, loosing his precious time reading the thread and watching the schematics, is by itself, an evidence of Mike design quality.
thank you Dimitri, your presence here have enormous weigth to this thread value.
In my point of view, as i constructed this amplifier, is that everyone will be very surprised and satisfied with it.
If result was not very impressive.... check your circuit....i guarantee that something was made wrong related the construction or parts selection.
Since the first moment you switch this amplifier on, you will perceive how good it is... and prepare yourself to fall down from your chair when some dinamic peak appear, punching your stomach hard and shaking your home glasses.
Dimitri, i will prepare, rigth now, a Vodka, with lemmon and suggar....a Caipiroska!... in your honor.
Cheers!
Carlos
As i am....
Dimitri has enormous International experience in electronics, his degrées papers, University Certificates will be shocked compared to many of us.
This man, some of the Russian Brains that are turning more rich the US know how, are someone very special, he is the one that is called when no one can find solutions for hard problems.
His presence here, loosing his precious time reading the thread and watching the schematics, is by itself, an evidence of Mike design quality.
thank you Dimitri, your presence here have enormous weigth to this thread value.
In my point of view, as i constructed this amplifier, is that everyone will be very surprised and satisfied with it.
If result was not very impressive.... check your circuit....i guarantee that something was made wrong related the construction or parts selection.
Since the first moment you switch this amplifier on, you will perceive how good it is... and prepare yourself to fall down from your chair when some dinamic peak appear, punching your stomach hard and shaking your home glasses.
Dimitri, i will prepare, rigth now, a Vodka, with lemmon and suggar....a Caipiroska!... in your honor.
Cheers!
Carlos
dimitri said:
Sorry for this...
This is symasym4:
http://www.diyaudio.com/forums/attachment.php?s=&postid=683532&stamp=1121552051
This is symasym5:
http://www.diyaudio.com/forums/attachment.php?s=&postid=684856&stamp=1121720926
These are the 2 interesting ones, symasym4 is the one carlos have built,
symasym5 has sligthly changed feedback, i think it sounds better and
is the one i actually listen to...
Mike
Hi Saijti,
I had the same feeling, but simulations showed that it is enough
(With this topology), so i tried in realworld, and it really works !
It sounds very powerful, not weak or compressed in any form.
I just wanted to keep the smallsignalbjts.
When hardly driven, current goes to 5ma, and when saturating
it even goes up to 15ma, but this happens only when reaching
slewratelimit.
But, this low current makes it difficult to tune this amp to more power.
(For me, more than enough for now)
Of course, high beta devices are necessary for the outputstage,
was no problem, these onsemi devices measured hfe > 100,
this gives a currentgain above 1:10.000 for the output.
(actually, mje15030 = 130, mjl3281 = ~110)
Mike
I had the same feeling, but simulations showed that it is enough
(With this topology), so i tried in realworld, and it really works !
It sounds very powerful, not weak or compressed in any form.
I just wanted to keep the smallsignalbjts.
When hardly driven, current goes to 5ma, and when saturating
it even goes up to 15ma, but this happens only when reaching
slewratelimit.
But, this low current makes it difficult to tune this amp to more power.
(For me, more than enough for now)
Of course, high beta devices are necessary for the outputstage,
was no problem, these onsemi devices measured hfe > 100,
this gives a currentgain above 1:10.000 for the output.
(actually, mje15030 = 130, mjl3281 = ~110)
Mike
I choosed the ~2.5ma to avoid heatup the smallsignal-bjts, with
5ma they will get very hot. Now they operate at ~40° (or less),
just feeling very warm, not hot yet.
I tried a tripledarlington with this amp, but already gave up in sims,
it get's too difficult to get good stability. With a stable version i ended
up in an amp having much higher phasehifts and higher thd.
It's too much gain in this case...
I use this amp to drive 4ohm speakers, and it plays effortless up to
very high volumes, so i think it's currentcapabilities should be high enough.
Mike
5ma they will get very hot. Now they operate at ~40° (or less),
just feeling very warm, not hot yet.
I tried a tripledarlington with this amp, but already gave up in sims,
it get's too difficult to get good stability. With a stable version i ended
up in an amp having much higher phasehifts and higher thd.
It's too much gain in this case...
I use this amp to drive 4ohm speakers, and it plays effortless up to
very high volumes, so i think it's currentcapabilities should be high enough.
Mike
I would use some emitter resistors is the VAS as well, to get less gain, and better linearity. It means some more voltage drop, but it's not too much in my point of view.
Another issue I thinking to use ccs in the VAS too, instead of 68ohm resistor.
BD139/140 works well in the VAS up to 40V rails. Bit slower than 2Ns (50MHz), but I think this is no problem...
sajti
Another issue I thinking to use ccs in the VAS too, instead of 68ohm resistor.
BD139/140 works well in the VAS up to 40V rails. Bit slower than 2Ns (50MHz), but I think this is no problem...
sajti
Using a ccs insted of the 680hms changes the whole nature of
the topology, rendering it "uninteresting"... Don't ask me why,
i did not fully understand it.
I will play with the gain and feedback more, there is still something
to optimize, maybe giving a symasym6, without RC and without
millercap, only the caps in the vas left...
bd139/140 works, carlos used the to92 versions...
Mike
the topology, rendering it "uninteresting"... Don't ask me why,
i did not fully understand it.
I will play with the gain and feedback more, there is still something
to optimize, maybe giving a symasym6, without RC and without
millercap, only the caps in the vas left...
bd139/140 works, carlos used the to92 versions...
Mike
MikeB said:
Yeah, but a couple of questions. Background first. I've built a couple of chip amps and a pp valve amp - that's it. Haven't tackled a discreet ss amp. This one looks easy to make and I have the output transistors and the others are easy to get. I'm making a speaker system that will have SET's for the top end horn (Unity Horn), maybe this amp for the midbass (60-350hz) and class d for the low bass. Kinda multicultural (ala Brazil). The questions:
You mentioned transistor matching a couple of times. Is that required for any of the transistors in this amp? If so, do you have a link to a procedure.
What does the Q signify in the transistor names? In the catalogs I see mje15031, but not qmme15031.
What are you using for a supply, or more to the point, what would you recommend? Transformer rating, cap size, etc..
Thanks
Well Dimitri, i can capture some meaning of your silence, as the same way we can see
defects in one image, observing it througth strong lenses...showing enormous empty spaces and defects, i imagine that you can see a lot of fails in every schematic you observe.
This may be not so happy to you....as some Super Vision always show errors, mistakes, and you cannot open your mouth everytime to avoid to be undelicated.
Very lonely those that have enormous know how...things may be boering, have to read a lot of foolishes silently, not to be excluded by the group.
In your case, have to use sub-liminary instructions...saying that made this other way and result good...and this way, go showing the correct way to do...a good way to introduce your know how in this amateur world, without the scandalous sittuation related to point errors or start discussions, when you will be speaking greek and others speaking Chinese.
We had, ...and we still have people like you, big brains, some could reduce to the average conversation level...others go just reading very boered, others lost patience and gone away.
Some others, so nervous they are, tired to read many foolishes, as many guys, and i have to include myself in this package, had not theorical bases, and we may wrote enormous non sense things...those, very nervous, when find someone that can discuss in a reasonable leveled stage, explode in a dog figth way, as a result of tensions acumulated and enormous lonelyness.
Observe, by example, Mikeks...the one is An electronic alive Enciclopedia...but very slowly now a days.... entering here or there...may be lonely too.
Well, world not perfect.... for me it is all rigth...i am low know how level too..so..i am the same as the big majority.
regards,
Carlos
defects in one image, observing it througth strong lenses...showing enormous empty spaces and defects, i imagine that you can see a lot of fails in every schematic you observe.
This may be not so happy to you....as some Super Vision always show errors, mistakes, and you cannot open your mouth everytime to avoid to be undelicated.
Very lonely those that have enormous know how...things may be boering, have to read a lot of foolishes silently, not to be excluded by the group.
In your case, have to use sub-liminary instructions...saying that made this other way and result good...and this way, go showing the correct way to do...a good way to introduce your know how in this amateur world, without the scandalous sittuation related to point errors or start discussions, when you will be speaking greek and others speaking Chinese.
We had, ...and we still have people like you, big brains, some could reduce to the average conversation level...others go just reading very boered, others lost patience and gone away.
Some others, so nervous they are, tired to read many foolishes, as many guys, and i have to include myself in this package, had not theorical bases, and we may wrote enormous non sense things...those, very nervous, when find someone that can discuss in a reasonable leveled stage, explode in a dog figth way, as a result of tensions acumulated and enormous lonelyness.
Observe, by example, Mikeks...the one is An electronic alive Enciclopedia...but very slowly now a days.... entering here or there...may be lonely too.
Well, world not perfect.... for me it is all rigth...i am low know how level too..so..i am the same as the big majority.
regards,
Carlos
Hi Sheldon !
Ok, the PSU: I use a RKT toroid 250watts,2*25volts,5amperes, a "high end"
version, means that is completely enclosed.
This toroid is ok for stereo use, but too small for full power into 4ohms,
but i did not recognize any weakness.
As rectifier i use a standard 250v25a metal-bridge, as caps i have 2
BC 22000uf,40v. As fuses i choose 2.5T (slow).
I have no snubbercaps, a very simple PSU, but with the onboard caps
this seems to be no problem.
Transistor matching !
You need to match single pairs only, the 2 mpsa18, the 2 2n5401 in 2nd diffamp,
and the 2 2n5551. The single 2n5401 with its base to gnd can't be
matched to anything. To be honest, i did not match the outputdevices...
That makes 3 pairs to match per channel, not too much.
For a stereo amp it might be a good idea to keep the transistors
in range between the 2 channels, not exact match, but not too different.
especially the lower right bc546 in the ccs could be matched between
channels. I can tell you later how to match transistors.
The simplest but not best way is to measure ohms between base and emitter,
it seems that matched transitors of same type show the same measuring.
This method is still better than no matching. With good matching the
DC-offset is ~2mv.
This "Q" in the naming of transistors is for the sim-models, the realworld
names are of course "mje15030". Be sure to get originals, best would be
original onsemi for outputdevices. The drivers (mje15030/31) need to be
mounted on heatsink, they are heavily biased and would overheat without
heatsink. The best method is to mount them on the same heatsink like
the outputdevices, this gives "perfect" thermal stability.
The changes in quiscentcurrent (set to 55ma) where below my ability
to measure. (~12mv across a 0.22ohms)
I also used onsemi devices for mpsa18,2n5401 and 2n5551. The bc546
is philips.
The resistors are of course all metalfilm, except the 0.22ohms.
The smaller caps (22pf,100pf,330pf) are all micas, for 100nf,47nf,4.7uf
i use wima mks2, for the electrolyts i used what i got. (63v)
The 220uf in feedback is on my board a 470uf because of the distance
of the legs, the 220uf was too small in dimensions. For this cap
the polarity is "plus" to gnd as inputoffset is negative.
I use a 16v version for this one.
The heatsinks i use on my prototype are too small (see photo),
for fullpower the amp might overheat. For "normal" listeninglevel
they are adequate. Heatsinks with double the size would do the job,
that would be ~0.9k/W.
I think the midrange of this amp is excellent, you might use it for a
wider bandwidth, just for topend trebles a single ended classa is of
course unbeatable...
Mike
Ok, the PSU: I use a RKT toroid 250watts,2*25volts,5amperes, a "high end"
version, means that is completely enclosed.
This toroid is ok for stereo use, but too small for full power into 4ohms,
but i did not recognize any weakness.
As rectifier i use a standard 250v25a metal-bridge, as caps i have 2
BC 22000uf,40v. As fuses i choose 2.5T (slow).
I have no snubbercaps, a very simple PSU, but with the onboard caps
this seems to be no problem.
Transistor matching !
You need to match single pairs only, the 2 mpsa18, the 2 2n5401 in 2nd diffamp,
and the 2 2n5551. The single 2n5401 with its base to gnd can't be
matched to anything. To be honest, i did not match the outputdevices...
That makes 3 pairs to match per channel, not too much.
For a stereo amp it might be a good idea to keep the transistors
in range between the 2 channels, not exact match, but not too different.
especially the lower right bc546 in the ccs could be matched between
channels. I can tell you later how to match transistors.
The simplest but not best way is to measure ohms between base and emitter,
it seems that matched transitors of same type show the same measuring.
This method is still better than no matching. With good matching the
DC-offset is ~2mv.
This "Q" in the naming of transistors is for the sim-models, the realworld
names are of course "mje15030". Be sure to get originals, best would be
original onsemi for outputdevices. The drivers (mje15030/31) need to be
mounted on heatsink, they are heavily biased and would overheat without
heatsink. The best method is to mount them on the same heatsink like
the outputdevices, this gives "perfect" thermal stability.
The changes in quiscentcurrent (set to 55ma) where below my ability
to measure. (~12mv across a 0.22ohms)
I also used onsemi devices for mpsa18,2n5401 and 2n5551. The bc546
is philips.
The resistors are of course all metalfilm, except the 0.22ohms.
The smaller caps (22pf,100pf,330pf) are all micas, for 100nf,47nf,4.7uf
i use wima mks2, for the electrolyts i used what i got. (63v)
The 220uf in feedback is on my board a 470uf because of the distance
of the legs, the 220uf was too small in dimensions. For this cap
the polarity is "plus" to gnd as inputoffset is negative.
I use a 16v version for this one.
The heatsinks i use on my prototype are too small (see photo),
for fullpower the amp might overheat. For "normal" listeninglevel
they are adequate. Heatsinks with double the size would do the job,
that would be ~0.9k/W.
I think the midrange of this amp is excellent, you might use it for a
wider bandwidth, just for topend trebles a single ended classa is of
course unbeatable...
Mike