syn08,
You haven`t seen 2SC3600/2SA1406, truly incredible properties and exceptional linearity, as a result of Sanyo`s FBET process. 400MHz@10mA.
lineup said:
Oh no, I don't mean nicotine. I myself trapped by this nasty substance.
Lumba Ogir said:
Lumba;
I prefer to leave non-linearity privileges to hearing created by God, and build my devices as humble creatures, not pretending to be created by Himself. So, I have to measure, how deeply I involve myself on God's territory, and try to be there as less as possible.
Lumba Ogir said:
Well, I have seen and used 2SC3600/2SA1406. They are simply a scaled down version of 2SC3601/2SA1407: lower Cob, same Ft, lower Icmax but, unfortunately, a much lower Early voltage. As such, 2SC3600/2SA1406 are much worse in terms of linearity than 2SC3601/2SA1407. But you don't in particular seem to like linearity, so no wonder you like that pair.
The single advantage that I can think of is cost; 2SC3600/2SA1406 pairs are significantly cheaper.
Syn,
Interesting results, Im very much surprised by the sanyo small signals result, it doesnt make too much sense looking at the datasheet specs. Could you perhaps post a pic of them, Id like to see whether they arent fakes, Ive got a few of them. Is it possible for you to try the fairchild small signals as well, ksa1845.
Interesting results, Im very much surprised by the sanyo small signals result, it doesnt make too much sense looking at the datasheet specs. Could you perhaps post a pic of them, Id like to see whether they arent fakes, Ive got a few of them. Is it possible for you to try the fairchild small signals as well, ksa1845.
homemodder said:
I can assure you they are genuine, I got all my Sanyo stuff from a very reputable dealer. This would be for the first time to see a fake from their hands, and for 10 pennies devices? And I know very well how genuine Sanyos are looking like.
OTOH, you certainly know about the NF curves in the datasheets as telling little to nothing about rbb and noise for the use of audio. Look at those curves: if the low noise areas are extending to low Rg, then there are chances those devices are good for low noise audio applications.
Hi Syn08
Thanks for sharing; it’s always good to see somebody sharing some interesting and useful information.
I’m looking forward to the results with different Ic,
The transistors will off course have their lowest noise at different Ic and other conditions.
A different story is the BS gang from Sweden.
Lumba: I see no point in answering your post.
lineup - halojoy - Carl Groman gromanswe or wathever you like to call yourself:
It’s the same answer.
PS: O I know you are busy, but don’t forget the VSOP.
Cheers
Stinius
Thanks for sharing; it’s always good to see somebody sharing some interesting and useful information.
I’m looking forward to the results with different Ic,
The transistors will off course have their lowest noise at different Ic and other conditions.
A different story is the BS gang from Sweden.
Lumba: I see no point in answering your post.
lineup - halojoy - Carl Groman gromanswe or wathever you like to call yourself:
It’s the same answer.
PS: O I know you are busy, but don’t forget the VSOP.
Cheers
Stinius
I want to point out something to everyone. Back in 1978, I attended the Tokyo Audio Fair, in order to give a paper. At the exhibition park they also had an ENGINEERING SHOW, just like WESCON, or EASTCON in the USA. They let me in to the engineering show when I showed them my credentials. It was amazing! Everybody and his brother made ultra low noise, high quality jfets and bipolars. Why so many manufacturers? Why such quality? It would seem, and someone please correct me if I am wrong in this, that each major company was tied to certain audio manufacturers, and they might make the request for a special series of parts to make their design best. This is what created so much competition between semiconductor manufacturers, and why so much diversity. Alas, most of these devices can't be found anymore, but some were amazing! I love this spirit of true competition, it is a lot like the competition between Charles Hansen (Ayre), Nelson Pass (Pass) and me. Competition will always bring out the best.
At least, that is what I think, at the moment.
Now, be careful what you fall for. You might find a GREAT part, but after that bin is empty, there may be no more.
At least, that is what I think, at the moment.
Now, be careful what you fall for. You might find a GREAT part, but after that bin is empty, there may be no more.
@10mA and @2mA
Here are the results at 10mA for the top three contenders:
Model Scope Beta Vn0 [uV/rtHz] Vni [nV/rtHz]
2SC3503 MP 107 821.6 0.46
2SC3601 MP 135 379.5 0.21
2SC2547 LN 488 434 0.24
Results are pretty much as expected. The Sanyo wonder slightly beats the Hitachi TO-92 device. However, in my stash of 150 2SC2547 at least 60% have beta over 400 (measured here). 0.2nV/rtHz at beta 400 is really almost incredible.
Unfortunately, also as expected, the corner frequency for the flicker noise also increased significantly. It's about 7KHz at 10mA, certainly way to high, and noise increases with exactly 10dB/decade as theory predicts. Therefore, I would not recommend 10mA bias. 1-2mA seems to be a decent compromise between low noise and low corner frequency, at well under 1KHz.
Here's the data at 2mA:
Model Scope Beta Vn0 [uV/rtHz] Vni [nV/rtHz]
2SC3503 MP 107 396.6 0.63
2SC3601 MP 135 236.31 0.37
2SC2547 LN 488 270 0.42
2SC3601 is still a little better, but has a corner frequency of 700Hz, while 2SC2547 is way lower at about 250Hz.
Currently, while supply lasts, 2SC2547/2SA1084 is the best option. Then, 2SC3601/2SA1407 currently in mass production at Sanyo. Ultimately, there are Fairchild japanese clones that will do an still excellent low noise job. These are very cheap and the availability in NA is very good as well. In all cases, Ic=2mA and Vce=6V appears to be a good compromise between plateau noise and low flicker noise corner frequency. For the 2SC2547, consider the beta of 400 and you'll get a formidable competitor for the extinct Toshiba JFETs. At 2mA, the base current is 5uA and if you consider 90% of this cancelled by the complementary part (matched for beta) you got only max. 500nA through a MC cartridge. That's already very good and could even get much better, if beta matching is tighter. Drift has also to be considered, but one to another I think nobody should be concerned for the low noise JFET extinction (John Curl is excepted, his age and experience entitles him to complain about missing parts).
I'll continue testing other models, unfortunately I don't currently have any more candidates.
Edit: a picture of the test jig, now with LME49710 opamps, testing a Hitachi device.
Here are the results at 10mA for the top three contenders:
Model Scope Beta Vn0 [uV/rtHz] Vni [nV/rtHz]
2SC3503 MP 107 821.6 0.46
2SC3601 MP 135 379.5 0.21
2SC2547 LN 488 434 0.24
Results are pretty much as expected. The Sanyo wonder slightly beats the Hitachi TO-92 device. However, in my stash of 150 2SC2547 at least 60% have beta over 400 (measured here). 0.2nV/rtHz at beta 400 is really almost incredible.
Unfortunately, also as expected, the corner frequency for the flicker noise also increased significantly. It's about 7KHz at 10mA, certainly way to high, and noise increases with exactly 10dB/decade as theory predicts. Therefore, I would not recommend 10mA bias. 1-2mA seems to be a decent compromise between low noise and low corner frequency, at well under 1KHz.
Here's the data at 2mA:
Model Scope Beta Vn0 [uV/rtHz] Vni [nV/rtHz]
2SC3503 MP 107 396.6 0.63
2SC3601 MP 135 236.31 0.37
2SC2547 LN 488 270 0.42
2SC3601 is still a little better, but has a corner frequency of 700Hz, while 2SC2547 is way lower at about 250Hz.
Currently, while supply lasts, 2SC2547/2SA1084 is the best option. Then, 2SC3601/2SA1407 currently in mass production at Sanyo. Ultimately, there are Fairchild japanese clones that will do an still excellent low noise job. These are very cheap and the availability in NA is very good as well. In all cases, Ic=2mA and Vce=6V appears to be a good compromise between plateau noise and low flicker noise corner frequency. For the 2SC2547, consider the beta of 400 and you'll get a formidable competitor for the extinct Toshiba JFETs. At 2mA, the base current is 5uA and if you consider 90% of this cancelled by the complementary part (matched for beta) you got only max. 500nA through a MC cartridge. That's already very good and could even get much better, if beta matching is tighter. Drift has also to be considered, but one to another I think nobody should be concerned for the low noise JFET extinction (John Curl is excepted, his age and experience entitles him to complain about missing parts).
I'll continue testing other models, unfortunately I don't currently have any more candidates.
Edit: a picture of the test jig, now with LME49710 opamps, testing a Hitachi device.
Attachments
syn08,
However, I made a mistake again, this should be more correct:
2SC3600/2SA1406 300MHz@10mA, 400MHz@30mA
2SC3601/2SA1407 200MHz@10mA, 400MHz@50mA
I would not use these devices at high currents.
I don`t agree at all.
However, I made a mistake again, this should be more correct:
2SC3600/2SA1406 300MHz@10mA, 400MHz@30mA
2SC3601/2SA1407 200MHz@10mA, 400MHz@50mA
I would not use these devices at high currents.
Lumba Ogir said:
Then obviously, you don't have a clue about what the Early voltage is and what is the impact on linearity.
Soon, we'll find we have different definitions and understandings of sleep, eat, drink, ****.
Wavebourn said:
No; I'm sending my boards to PCB houses for years; my time is to precious to spend it in painting and drilling PCB boards. Plus that I have a genuine concern about disposing all that chemical crap.
syn08,
these devices were highly successfully optimized to their purpose. Just employ them suitably and don`t worry about the Early voltage.
these devices were highly successfully optimized to their purpose. Just employ them suitably and don`t worry about the Early voltage.
Wavebourn said:
Never used it. I have a perpetual license for the Cadence Orcad suite, currently at v. 16.2
Re: @10mA and @2mA
Great work Syn, thanks for this contribution!
The 3601 is quite an intriguing device, low Rbb and ultra low Cob
would appear to me to be conflicting attributes - am I right here?
Here is a paper which has some info on low noise higher Ic BJT's.
www.janascard.cz/PDF/Ultra low noise amplifiers.pdf
cheers
Terry
syn08 said:
Great work Syn, thanks for this contribution!
The 3601 is quite an intriguing device, low Rbb and ultra low Cob
would appear to me to be conflicting attributes - am I right here?
Here is a paper which has some info on low noise higher Ic BJT's.
www.janascard.cz/PDF/Ultra low noise amplifiers.pdf
cheers
Terry
- Status
- Not open for further replies.