wg_ski said:Whew! Glen got back to this before I did.
Let's not get too carried away here - we don't want you to end up in the 'bin again!
My personal experience with all this is that once sufficiently-rated and suitable components for the job are selected the things that influence the sound the most are all the things that are NOT typically put into a Spice simulation. (You could, but it would be a complete RF model including electromagnetic effects.)
Sure, but we were specifically talking about frequency compensation, which is a complex subject (more complex than most realise). I don’t want to spark off another generalised SPICE-wars thread.
WRT to frequency compensation, unless copying or elaborating without too much deviation on an existing, proven design, SPICE can be invaluable (and surprisingly accurate) tool for quickly evaluating and optimising a concept – especially if more complex forms of frequency compensation are used.
Cheers,
Glen
Why use one with 5 cob if their are are ones with less than half of that. But for vas duty the early voltage is just as important.
DrG, Im a ex south south african, lived there most of my short life. What you say has some reason, I have found people that couldnt tell the diffrence between a delibertly bad designed amp of mine and a electrocompaniet which cost me a enough to purchase a new small car and is highly acclaimed. They acknowledged that the two amps sounded different but equally good, it was scary because I underbiased this amp and it was enough for me to pick up some crossover distortion. Like some people that are not good at sports some are not good at judging differences in audio sound. I did the same test with my sister who is a music teacher and plays various instruments, she told me that something was wrong after 10 seconds of listening to my bad amp. This was a blind test.
Ive also built some amps which for me and some people that listened was very good, in a lot of cases she disagreed with me saying yes they do sound good and pleasant but not like what the real instruments sound like and that this one or that one sounded more like to true life. This is how I do listening tests now, with input from someone that actually knows what the real thing sounds like.
So I thought of testing her hearing and invited some of her fellow members of an orchestra to the same listening tests and looked like a fool when they also prefered the same amps as my sis, also saying they sounded more like the real thing. Clearly I also have a deficiency compared to someone that dedicates their life to playing musical instruments, at least my deficiency doesnt seem as bad as i can clearly hear differences in transistors, capacitors, resistors and minor ones in cables. At least with my scope I can then verify that their are indeed performance differences in active components, much harder to find with passives.
I dont think that anyone will ever do such a controlled experiment looking at costs and time involved but if they do I hope they use people that are musicians and know what they listening at. Im afraid even most so called audiophiles are gonna be embarressed when told their equipment do not sound like the real thing not to mention the designers of these audiophile products.
DrG I will also suggest you build 2 diamond buffers,one with bc or 2n parts and the other with toshiba transitors. All the rest the same passives, hook it direct to your cdplayer and listen to one then the other several times through good quality phones, I suggest something like hd600s and come back here to this thread and tell us honestly if they sounded the same, not necessarily which one sounded better as i too sometimes fail on this and prefer to use an experienced musicians opinion.
DrG, Im a ex south south african, lived there most of my short life. What you say has some reason, I have found people that couldnt tell the diffrence between a delibertly bad designed amp of mine and a electrocompaniet which cost me a enough to purchase a new small car and is highly acclaimed. They acknowledged that the two amps sounded different but equally good, it was scary because I underbiased this amp and it was enough for me to pick up some crossover distortion. Like some people that are not good at sports some are not good at judging differences in audio sound. I did the same test with my sister who is a music teacher and plays various instruments, she told me that something was wrong after 10 seconds of listening to my bad amp. This was a blind test.
Ive also built some amps which for me and some people that listened was very good, in a lot of cases she disagreed with me saying yes they do sound good and pleasant but not like what the real instruments sound like and that this one or that one sounded more like to true life. This is how I do listening tests now, with input from someone that actually knows what the real thing sounds like.
So I thought of testing her hearing and invited some of her fellow members of an orchestra to the same listening tests and looked like a fool when they also prefered the same amps as my sis, also saying they sounded more like the real thing. Clearly I also have a deficiency compared to someone that dedicates their life to playing musical instruments, at least my deficiency doesnt seem as bad as i can clearly hear differences in transistors, capacitors, resistors and minor ones in cables. At least with my scope I can then verify that their are indeed performance differences in active components, much harder to find with passives.
I dont think that anyone will ever do such a controlled experiment looking at costs and time involved but if they do I hope they use people that are musicians and know what they listening at. Im afraid even most so called audiophiles are gonna be embarressed when told their equipment do not sound like the real thing not to mention the designers of these audiophile products.
DrG I will also suggest you build 2 diamond buffers,one with bc or 2n parts and the other with toshiba transitors. All the rest the same passives, hook it direct to your cdplayer and listen to one then the other several times through good quality phones, I suggest something like hd600s and come back here to this thread and tell us honestly if they sounded the same, not necessarily which one sounded better as i too sometimes fail on this and prefer to use an experienced musicians opinion.
homemodder said:
If the VAS uses a cascode, and the transistor in question is the common-emitter portion of the cascode, then the Early voltage of that transistor isn't important. For a given BJT process, as beta increases, the Early voltage decreases. You could actually improve performance by choosing a transistor (such as the "C" version of a BC550) with a higher beta and a lower Early voltage for the common-emitter portion of the VAS cascode. The improvement would come from increased current gain of the VAS.
I started reading this thread for enlightenment on output
choices but now it seems to have been transformed
into a war zone...
Actually , this thread is a lesson in "high order mudslinging"
I won't comment , as this thread is a personal battleground
best viewed for entertainment only.
To the real topic , My favorites are MJL 4302/4281..very linear
,cheap,readily available and mj15024/5 ... durable, good for
abuse(subs) Sanken- I got made fun of for using them
,but they sound more "musical' than any of the MJ-X
devices..OS
choices but now it seems to have been transformed
into a war zone...
Actually , this thread is a lesson in "high order mudslinging"
I won't comment , as this thread is a personal battleground
best viewed for entertainment only.
To the real topic , My favorites are MJL 4302/4281..very linear
,cheap,readily available and mj15024/5 ... durable, good for
abuse(subs) Sanken- I got made fun of for using them
,but they sound more "musical' than any of the MJ-X
devices..OS
Thanks ostripper, I agree with your evaluation, If I wanted an amp for blasting high sound levels in a disco or for pa and price is in consideration I would happily use Onsemis. For music playing for my listening pleasure I ll rather use sankens or Toshibas.
My prefered ones 2sa1186/2sc2837 and for higher power 2sa1303/2sc3284.
Andyc, what you say is absolutley true, one could go the route of extra complexity and cascode and so forth or you could just use a high quality transistor to get better results knowing what to look at for. I prefer not to cascode BJTs, something to me just doesnt sound right subjectively when doing so, and is odd as I have no complaints cascoding Jfets. D.Self seems to prefer going the very high beta way using darlingtons and cascoding in his vas judging by his more recent design inputs at cambridge. Ill have to try this sometime.
In all Hugh is correct and I agree with him, one can get stunning results with a typical Lin design with proper, fastidious parts selection and very careful selection of operating points.
I have nothing against complexity and my fav diyamp that ive built is the stochino fast amp. Well its not that complex but is a stunning sounding amp. I have used different trannies and made slight modifications to the compensation though. A lot of people that have listened to it prefer it to my megabuck electrocompaniet and as Ive commented before my listeners are mostly musicians.
My prefered ones 2sa1186/2sc2837 and for higher power 2sa1303/2sc3284.
Andyc, what you say is absolutley true, one could go the route of extra complexity and cascode and so forth or you could just use a high quality transistor to get better results knowing what to look at for. I prefer not to cascode BJTs, something to me just doesnt sound right subjectively when doing so, and is odd as I have no complaints cascoding Jfets. D.Self seems to prefer going the very high beta way using darlingtons and cascoding in his vas judging by his more recent design inputs at cambridge. Ill have to try this sometime.
In all Hugh is correct and I agree with him, one can get stunning results with a typical Lin design with proper, fastidious parts selection and very careful selection of operating points.
I have nothing against complexity and my fav diyamp that ive built is the stochino fast amp. Well its not that complex but is a stunning sounding amp. I have used different trannies and made slight modifications to the compensation though. A lot of people that have listened to it prefer it to my megabuck electrocompaniet and as Ive commented before my listeners are mostly musicians.
homemodder said:
Douglas Self doesn't cascode his VAS because the emitter follower buffer (don't call it a Darlington please!) combined with the LTP current mirror maximises VAS gain. This greatly increases the local NFB applied to the VAS by the miller compensation cap to such a degree that VAS distortion is reduced to a negligible (ie. not dominant) level. Cascoding would therefore make little (if any) measurable improvement to the performance of the design as a whole.
However, this is only an argument for not cascoding the VAS of the Blameless amp, not amps in general. It is of course design specific!
Also I’m glad that you like the 300V/us Stochino amp as that mean fast amps aren’t necessarily “bad”, however I’m a bit confused. You say that you don’t subjectively like a cascoded VAS, but the Stochino amp VAS is indeed cascoded:
http://www.soton.ac.uk/~apm3/diyaudio/Stochino_circuit.jpg
????
Cheers,
Glen
Homemodder - those BJTs surely belong to the creme de la creme among power BJTs. I will try to get hold of 100pcs or so of the Y-grade.
Thanks for sharing!
For the 25-50W class A power amp I am working on now, I have sucessfully used the Onsemi
NJW0302/0281
with an idling current between 1 - 2A.
At these currents Ft is about 60MHz which I find is outstanding.
Hfe is between 75 - 150 and Cob = 400pF.
I am curious: what is it that you do not like about the NJW0302/0281?
Sigurd
Thanks for sharing!
For the 25-50W class A power amp I am working on now, I have sucessfully used the Onsemi
NJW0302/0281
with an idling current between 1 - 2A.
At these currents Ft is about 60MHz which I find is outstanding.
Hfe is between 75 - 150 and Cob = 400pF.
I am curious: what is it that you do not like about the NJW0302/0281?
Sigurd
homemodder said:
Glen, you right thats not a darlington as used in blameless, but the scheme I mean is completly different and is not in his books. Lately he has used a single transistor darlington vas , cascoded and followed by a buffer. I dont think its ethical and would not do without D.self or Cambridge permission put schematics here as I think these units are still in production, its from azur range. A friend back engineered one some months back to see what the good doc was upto while modifying a unit for the mayor of my town. I was very surprised too at first but never had a second thought on it till now when andy mentioned the benifits of very high beta. I dont have as much time to spend with electronics as I would like.
Mmmm seems the good doc had in fact a new idea here, since most here dont regard XD as new. For D.Self to use this scheme it must have some positive attributes, i doubt someone would design something worse performing than previous attempts.
I dont like cascoding any bjt be it vas or ltp, from most amps Ive designed in Lin topology cascoding didnt do it for me subjectively. I tried some tips given in this forum but still not entirely happy. The stochino is not my design I built it exactly to the schematics, but not with those horrible 2n devices but sanyo small signals bjts, in place of bd139 i have 2sc2910, mine has one extra copensation cap and smaller miller caps as i found the sound just a tad bit clinical, otherwise its identical to schematics. Glen I have no problem with the claimed high slew rate of this amp, even if it was less than a third of it but still sounded as good Id be very happy. I forgot to mention I included a small resister in the bases of the ltp current source trannies as well as resistors in between current source and ltp, this was for purely subjective reasons. Some here say this amp easily blows its outputs, I guess Im lucky then so far, they drive my BW804s effortly to high levels. I prefer my comercial amp, maybe its just in my head to justify the small fortune i spent on it.
Jacco excuse my ignorance, what is SOTA, sorry to dissapoint you but I didnt use those 2n devices. Nice pic of Nec drivers you showed, near impossible to obtain , i use 1507/3902, 2sb1144/2sd1684, 1249/3117 or 1930/5171 and in that particular order of preference. High frequency crt display transistors work very well here too but its better to have some safety margin.
Mmmm seems the good doc had in fact a new idea here, since most here dont regard XD as new. For D.Self to use this scheme it must have some positive attributes, i doubt someone would design something worse performing than previous attempts.
I dont like cascoding any bjt be it vas or ltp, from most amps Ive designed in Lin topology cascoding didnt do it for me subjectively. I tried some tips given in this forum but still not entirely happy. The stochino is not my design I built it exactly to the schematics, but not with those horrible 2n devices but sanyo small signals bjts, in place of bd139 i have 2sc2910, mine has one extra copensation cap and smaller miller caps as i found the sound just a tad bit clinical, otherwise its identical to schematics. Glen I have no problem with the claimed high slew rate of this amp, even if it was less than a third of it but still sounded as good Id be very happy. I forgot to mention I included a small resister in the bases of the ltp current source trannies as well as resistors in between current source and ltp, this was for purely subjective reasons. Some here say this amp easily blows its outputs, I guess Im lucky then so far, they drive my BW804s effortly to high levels. I prefer my comercial amp, maybe its just in my head to justify the small fortune i spent on it.
Jacco excuse my ignorance, what is SOTA, sorry to dissapoint you but I didnt use those 2n devices. Nice pic of Nec drivers you showed, near impossible to obtain , i use 1507/3902, 2sb1144/2sd1684, 1249/3117 or 1930/5171 and in that particular order of preference. High frequency crt display transistors work very well here too but its better to have some safety margin.
People please go read up on the net on the origin of the newer onsemi devices, and for that matter the fairchild ones too. These are basicaly japanese copies of the well known 1943/5200 combo except made properly with uncompromised soar and quality control and not in the way Chinese companies like Mospec and Winsling, you know the white paste ones do. One could use those white paste transistors too with some risk but compare the sound to a original. Now this is what I would call trash.
Onsemi has also introduced a copy of a very good old japanese driver , looking at fairchild they are producing plenty of copies of well known and excellent japanese transistors, ksc/ksa. As in human nature onsemi and fairchild have done improvements once they had their hands on the original concept, like we do here with amp schematics. Once i saw the blameless I built it and shortly afterwards started going about doing improvements, well its in my nature. What does all this tell you, simple, they saw a japanese advantages and have decided to follow their lead. Most of these transistors have been replaced by new types by the japanese. Toshiba have introduced new drivers and replacement for 1943/5200. Do you think its the americans or the EU that come up with ever decreasing die making in the processor of your computer, think again. The japanese are perfectionists. One of my subjects at varsity was industrial electronics and I remember having lectures about the japanese culture, how it affected their products and how the west could keep up but this is offtopic sorry..
All those transistors mentioned will work, but to evalute one has to buy some other pairs,I place them in the same design and listen. I cannot do this for someone, and can only comment on what Ive found, those Sankens I mentioned are miles ahead of the older mj series in sonic terms, more specifically greater and faster impact on bass frequencies, excellent highs. In the case of the 2sa1186 its a tranny that has been a fav of mine since my varsity days. Found it in a state of he art then pioneer car amp that i used to blast music in my car.
Sigurd I didnt have a proper listen to those yet, I like the compensation diode idea though. The fairchild pair are good, Id say on even sonic terms as original 1943, better soar. By the way I recieved 10 jfet order of 2sk222 and 10 2sk223 to try. I think ill build nelsonvandals headphone amp to compare with the toshibas. The problem is if I want to purchase these it has min order and Im still bargaining with a dealer.
Onsemi has also introduced a copy of a very good old japanese driver , looking at fairchild they are producing plenty of copies of well known and excellent japanese transistors, ksc/ksa. As in human nature onsemi and fairchild have done improvements once they had their hands on the original concept, like we do here with amp schematics. Once i saw the blameless I built it and shortly afterwards started going about doing improvements, well its in my nature. What does all this tell you, simple, they saw a japanese advantages and have decided to follow their lead. Most of these transistors have been replaced by new types by the japanese. Toshiba have introduced new drivers and replacement for 1943/5200. Do you think its the americans or the EU that come up with ever decreasing die making in the processor of your computer, think again. The japanese are perfectionists. One of my subjects at varsity was industrial electronics and I remember having lectures about the japanese culture, how it affected their products and how the west could keep up but this is offtopic sorry..
All those transistors mentioned will work, but to evalute one has to buy some other pairs,I place them in the same design and listen. I cannot do this for someone, and can only comment on what Ive found, those Sankens I mentioned are miles ahead of the older mj series in sonic terms, more specifically greater and faster impact on bass frequencies, excellent highs. In the case of the 2sa1186 its a tranny that has been a fav of mine since my varsity days. Found it in a state of he art then pioneer car amp that i used to blast music in my car.
Sigurd I didnt have a proper listen to those yet, I like the compensation diode idea though. The fairchild pair are good, Id say on even sonic terms as original 1943, better soar. By the way I recieved 10 jfet order of 2sk222 and 10 2sk223 to try. I think ill build nelsonvandals headphone amp to compare with the toshibas. The problem is if I want to purchase these it has min order and Im still bargaining with a dealer.
obscurion said:
Without going into what to look for on a datasheet, I think the question illustrates a point.
Apart from the datasheet, you've got very little to go on other than measurements that you make yourself, or that have been made by people you trust.
You can't depend on your own ears. You have to use them, but distrust them too.
There's no meaningful data on which specific devices people prefer to listen to.
There's reputation, but it's hard to know to what extent it's reliable.
When we're designing, for example, a radio stage, we don't run round our colleagues asking them which transistors they 'like' in that application, or rather we do, but we don't mean 'like'. We mean 'based on the datasheets and other known facts, which will meet the specification cheapest?' 'Will the rep. buy us lunch?'
The only choice is to bone up a bit on what is known about human perception, and aim for a design with faults lower than those known to be perceptible. This way you have a specification to meet.
Otherwise you're just at the mercy of the storm.
w
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.