Nothing wrong with the design at all, i gonna try to build PMP some time. Try these, 2sa1016k|2sa2362k as input transistors. Use these in place of all the 2sa970| 2sc2240. The toshibas arent that bad but ......., have a look at datasheets. Also use them in current mirrors, makes big difference especially comparing to 2n5551|2n5401. I like to use cascoded current mirrors, that is when im in complexity and ultimate mood, they simply sound better.
I have tested a lot of transistors a couple of years back, mostly for lineariaty, cob and early affect using a method similar to Renardson and actual distotion measurements on a basic amp and found that using certain active devices performance can be greatly enhanced especially when one also pays attention at operating them in their most linear regions.
Thanks for posting some of these great amp ideas, i hope i can find some time to build them.
Regards
Alex
I have tested a lot of transistors a couple of years back, mostly for lineariaty, cob and early affect using a method similar to Renardson and actual distotion measurements on a basic amp and found that using certain active devices performance can be greatly enhanced especially when one also pays attention at operating them in their most linear regions.
Thanks for posting some of these great amp ideas, i hope i can find some time to build them.
Regards
Alex
Hi Alex,
These 2SA1016k/2SC2362k trannies really look nice. Thank for the tip. I hope I can get them in the Netherlands.
As for building the PMP amp, there is more to come: the MCP and PCP amp. The latter has an auto bias circuit and four pairs of MOSFETs in the output stage, probably FQA12P20/FQA19N20 instead of 2SJ201/2SK1530, because the FQA's have lower capacitances.
You already noticed that these amps are only design ideas, so be warned. The devil hides in the detail.
As for the input stage, what do you think about the THAT300 series?
Regards,
Edmond.
These 2SA1016k/2SC2362k trannies really look nice. Thank for the tip. I hope I can get them in the Netherlands.
As for building the PMP amp, there is more to come: the MCP and PCP amp. The latter has an auto bias circuit and four pairs of MOSFETs in the output stage, probably FQA12P20/FQA19N20 instead of 2SJ201/2SK1530, because the FQA's have lower capacitances.
You already noticed that these amps are only design ideas, so be warned. The devil hides in the detail.
As for the input stage, what do you think about the THAT300 series?
Regards,
Edmond.
Hi Edmond
That 300 trannies do look ok. They have very good noise performance and fast, pity about the max vce.
I couldnt bring all my equipment with me as i emigrated here, otherwise i would test them to see how they faired but i think they would also come short compared to 2sa1016K, the next best i tested was 2sa872. The noise figure might be just slightly better, the 2sa also having very low rbb. But regarding early effect i suspect they will lose out, lineariaty hard to say as the datasheets dont provide much info, bandwith info is also scarse and for all we know after IC 1ma it may drod substancially, drops are usually in bigger increments. You run your ltps with higher currents and here youll find 2sa outperforming it too as its ft is fairly constant and then reaches max of 270mhz at about 10ma. The cob of the 2sa also sustancially lower.
The only real advantage i see is from the matching. I especially like sanyo as when trannies purchased from the same batch you will notice that they are already very closely matched and unlike other manufacturers where you have to purchase big quantities to be able to match hfe.
The biggest obsticle i have building one of your amps is the pcboard layout. Headaches, big headaches designing a board for such a complex design. Has anyone done boards yet??, with my limited time it will take me ages to do this.
I havent been to your site lately, got some reading to do.....
That 300 trannies do look ok. They have very good noise performance and fast, pity about the max vce.
I couldnt bring all my equipment with me as i emigrated here, otherwise i would test them to see how they faired but i think they would also come short compared to 2sa1016K, the next best i tested was 2sa872. The noise figure might be just slightly better, the 2sa also having very low rbb. But regarding early effect i suspect they will lose out, lineariaty hard to say as the datasheets dont provide much info, bandwith info is also scarse and for all we know after IC 1ma it may drod substancially, drops are usually in bigger increments. You run your ltps with higher currents and here youll find 2sa outperforming it too as its ft is fairly constant and then reaches max of 270mhz at about 10ma. The cob of the 2sa also sustancially lower.
The only real advantage i see is from the matching. I especially like sanyo as when trannies purchased from the same batch you will notice that they are already very closely matched and unlike other manufacturers where you have to purchase big quantities to be able to match hfe.
The biggest obsticle i have building one of your amps is the pcboard layout. Headaches, big headaches designing a board for such a complex design. Has anyone done boards yet??, with my limited time it will take me ages to do this.
I havent been to your site lately, got some reading to do.....
Hi Alex,
I'm not an expert in selecting the right trannies, So, when it comes to drawing the PCB art work, I will try to make provisions for both the discrete devices and the THAT3xx chips. So, everybody can put there what they like most.
Drawing PCB layouts isn't my fort either, actually, I hate it. What I really like is playing (read: simulating) with all kind of topologies to get the lowest distortion, a nice step response and good recovery from overload.
When I've finished the design of the PCP amp, not yet published on my website, as I'm still wrestling with some minor details, I will start with the PCB, four layers! Not only to ease laying out the traces, but also to maximize the trace width to the output devices in order to lower inductances (less chance of HF oscillations). BTW, Halcro uses six layers!
Please, do have a look at my website from time to time, to follow my progress (but don't feel disappointed when I'm a bit slow, please
).
Cheers,
Edmond.
I'm not an expert in selecting the right trannies, So, when it comes to drawing the PCB art work, I will try to make provisions for both the discrete devices and the THAT3xx chips. So, everybody can put there what they like most.
Drawing PCB layouts isn't my fort either, actually, I hate it. What I really like is playing (read: simulating) with all kind of topologies to get the lowest distortion, a nice step response and good recovery from overload.
When I've finished the design of the PCP amp, not yet published on my website, as I'm still wrestling with some minor details, I will start with the PCB, four layers! Not only to ease laying out the traces, but also to maximize the trace width to the output devices in order to lower inductances (less chance of HF oscillations). BTW, Halcro uses six layers!
Please, do have a look at my website from time to time, to follow my progress (but don't feel disappointed when I'm a bit slow, please
).Cheers,
Edmond.
Dear Alex -
very interesting info about the 2SA1016/SC2362! I have used 2SA2240/SA970 many times as input stage transistors, and if there any better ones out there, I am very much interested in knowing.
It would be interesting to know what transistor parameters that you think would make a difference (and why) between the SC2240/sa970 pair vs the SC2362/sa1016 pair in for ex a lne level differential input stage running at 2-5mA and 10 mA?
Regards,
Sigurd
very interesting info about the 2SA1016/SC2362! I have used 2SA2240/SA970 many times as input stage transistors, and if there any better ones out there, I am very much interested in knowing.
It would be interesting to know what transistor parameters that you think would make a difference (and why) between the SC2240/sa970 pair vs the SC2362/sa1016 pair in for ex a lne level differential input stage running at 2-5mA and 10 mA?
Regards,
Sigurd
homemodder said:
homemodder said:
Sigurd Ruschkow said:
Quattor said:
Hi
I am very interested, too.
I have found & downloaded Datasheets of them:
http://www.datasheetarchive.com/
Two things:
1. Where can we buy these SANYO: 2SC2362 / 2SA1016
... if they are not already obsolete ...
2. Any of you guys have SPICE Models for these ??
... with good & fair models, we may be able to start makin' comparing
... in different circuit configurations
Thanks. Lineup
----------------------------------------------
PS. I already knew about 2SC1815 / 2SA1015 from Toshiba.
They are probably better than 2SC2240 / 2SA970, in some parameters.
And 2SC2240 / 2SA970 are certainly not bad! DS.
An externally hosted image should be here but it was not working when we last tested it.
TOSHIBA - 2SA1015 datasheet PDF
Quattor,
I have the datasheets and know them.
I am still interestid in YOUR oppinons to "It would be interesting to know what transistor parameters that you think would make a difference (and why) between the SC2240/sa970 pair vs the SC2362/sa1016 pair in for ex a lne level differential input stage running at 2-5mA and 10 mA?"
Sigurd
I have the datasheets and know them.
I am still interestid in YOUR oppinons to "It would be interesting to know what transistor parameters that you think would make a difference (and why) between the SC2240/sa970 pair vs the SC2362/sa1016 pair in for ex a lne level differential input stage running at 2-5mA and 10 mA?"
Sigurd
Quattor said:
Hi everyone
Sigurd youre right 2sa970|2sc2240 are good no doubt, but there is better....... Lets say its the difference between a rolls royce and a mercedes. Look at the datasheet guys, it speaks for it self. The 2sa1016 has lower early voltage, lower cob, higher ft and lower noise. These are parameters i regard as important.
I tested several hundred transistors and found these to be the rolls royce of small signal transistors. There are simple tests that can be done with osciliscope and opamp, i used somethig similar to renardson. Did the same for vas and driver transistors, one learns quite a bit about transistor behaviour. Theres no doudbt that the early effect and low cob is a very important factor, just look at all the cascodes being used. Also 2sa872 is better. There are better performers regarding early voltage and cob, but none with low noise.
Sanyo transistors outperform toshiba transistors when it comes to lower Ic devices, just a pity they not much into high output transistors. Have a look at jfets as well, they are excellent too.
It looks like segor has them, but i have some doubts whether they not fakes or manufactured in india or china as the pic they show dont conform to the the markings that sanyo uses. Gain group and batch markings different. Maybe sanyo has changed this, ill find out. All original sanyos are very hard to find. Im lucky, i have contact which works at rotel factory and those are assured originals.
Transistors from different manufacturers sound quite different and the ones from sanyo and toshiba and hitachi for me tend to sound great but then it could be just the superior specs where the difference lies. Like i explained to edmond, for those that only want simple designs superior performance could be had by using better trannies which might just cost a few pennies more.
Quattor, yes rotel uses these, they use the finest components available, thats how they manage to extract some decent performance from such a simple and outdated design.
Alex
Sigurd youre right 2sa970|2sc2240 are good no doubt, but there is better....... Lets say its the difference between a rolls royce and a mercedes. Look at the datasheet guys, it speaks for it self. The 2sa1016 has lower early voltage, lower cob, higher ft and lower noise. These are parameters i regard as important.
I tested several hundred transistors and found these to be the rolls royce of small signal transistors. There are simple tests that can be done with osciliscope and opamp, i used somethig similar to renardson. Did the same for vas and driver transistors, one learns quite a bit about transistor behaviour. Theres no doudbt that the early effect and low cob is a very important factor, just look at all the cascodes being used. Also 2sa872 is better. There are better performers regarding early voltage and cob, but none with low noise.
Sanyo transistors outperform toshiba transistors when it comes to lower Ic devices, just a pity they not much into high output transistors. Have a look at jfets as well, they are excellent too.
It looks like segor has them, but i have some doubts whether they not fakes or manufactured in india or china as the pic they show dont conform to the the markings that sanyo uses. Gain group and batch markings different. Maybe sanyo has changed this, ill find out. All original sanyos are very hard to find. Im lucky, i have contact which works at rotel factory and those are assured originals.
Transistors from different manufacturers sound quite different and the ones from sanyo and toshiba and hitachi for me tend to sound great but then it could be just the superior specs where the difference lies. Like i explained to edmond, for those that only want simple designs superior performance could be had by using better trannies which might just cost a few pennies more.
Quattor, yes rotel uses these, they use the finest components available, thats how they manage to extract some decent performance from such a simple and outdated design.
Alex
Sigurd
Do a simple distortion test, not simulated, of say just a ltp or a simple diamond buffer. The results will tell you why. At anything from 1 ma to around 8 the 2sa1016 is far superior and although i didnt test at larger currents, im pretty sure theyll be better at larger currents too, but here you start reaching the end of the most linear curve and then also other parameters set in depending on working voltages. The 2sa2240 can hande more Ic but who runs ltp at these currents.
Even Walt Jung likes these sanyos and has mentioned using them in current sources to obtain better performance in a published paper paper.
The other thing is that, its not on any datasheets but these trannies also have better t on and t off times. In amplifiers you will find objectively that transistor with faster times tend to sound more pricise and cleaner. Im not talking of harmonic spectrum here, vis a vis, same spectrum with same amp,these provide lower noise, lower distortion more precise, larger soundstage. These are the reasons i use them.
Alex
Do a simple distortion test, not simulated, of say just a ltp or a simple diamond buffer. The results will tell you why. At anything from 1 ma to around 8 the 2sa1016 is far superior and although i didnt test at larger currents, im pretty sure theyll be better at larger currents too, but here you start reaching the end of the most linear curve and then also other parameters set in depending on working voltages. The 2sa2240 can hande more Ic but who runs ltp at these currents.
Even Walt Jung likes these sanyos and has mentioned using them in current sources to obtain better performance in a published paper paper.
The other thing is that, its not on any datasheets but these trannies also have better t on and t off times. In amplifiers you will find objectively that transistor with faster times tend to sound more pricise and cleaner. Im not talking of harmonic spectrum here, vis a vis, same spectrum with same amp,these provide lower noise, lower distortion more precise, larger soundstage. These are the reasons i use them.
Alex
homemodder said:
Hi Alex,
Why do Sanyo's have a higher Early voltage?
Is it because they are manufactured by means of a better process or is it because hFE is just lower?
As you know, the product of these two parameters is fairly constant, although the bad guys reveal a product about 10k and the good guys about 20k. So, can you check these figures against your measurements on real devices, please?
Regards,
Edmond.
@Jacco, thanks for the tip.
homemodder said:
Alex,
I was a 2SC2240/2SA970 user and switched recently, for my new designs, to 2SC2362/2SA1016. Indeed, the Sanyo pair is significantly better in terms of noise (1dB vs. 6dB at 10Hz, 6V/100uA and Rg=10k and 0.7dB vs. 2dB at 1KHz, 6V/100uA and Rg=10k) better in terms of Cob, slightly better in terms of Early voltage (but then of course slightly worse in terms of beta) and about the same in terms of Ft.
However, if you look carefully at the datasheets, you would of course notice that the much better noise performance is reached at an Ic of about 100uA or below. For good reasons, 100uA is not the current you would like to run your input stage at. At 1mA and above, the noise performances are almost at par. This is the reason why I was not really able to measure any significant noise improvement over the 2SA970/2SC2240.
Which leaves the Cob as the other significant advantage. If you look again at the datasheets, you would of course note that the Cob's are about in the same ratio as ICmax. Which simply means that the Sanyo pair is a about half the (base/emitter) area of the Toshiba device.
If one is happy with 50mA ICmax, then the Sanyo devices are better. Only one last note: while the Toshiba's can take some abuse in terms of Ic and Pd, the Sanyo's are not; they die at currents as low 60-70mA, either catastrophically or have a very large beta drop and a huge increase in the noise figures. This is another price to pay for a very small area bipolar device.
Hi Edmond
Im not well versed in transistor physics, but reading from the datasheets the main reason seems to be the Ic versus Vce linearity which implies a more linear transistor although i found cob plays a factor here too. The more constant the IC versus Vce changes the higher the early voltage.current the Trannies with lower cob also have higher early voltage. It seems the early effect, affects the higher frequencies most and the high frequency is effected by the cob. Pity theres no graph that gives info how cob affects the early voltage.
Hfe didnt get into the equation the ones i tested *2sa1016*were from group g and had hfe around 500. Manufacturing process might be better yes, but the doping ect play a factor too. Guys like halcro and opamp designers use ultra high frequency trannies in ltps which brings to mind they were chosen for low cob and high early voltage altough im pretty sure they found low noise too.
Im sure theres some info on this subject on the net somewhere, i had some in my varsity books but like my equipment and audio related stuff stayed at home. The girlfriend was complaining that i seemed to pay more attention to my plastic and metal babies *transistors and cars* than to her so they stayed at home, for now that is.
Jacco is spot on. The trannies pictured are more like the sanyos i know. Although one cannot see the markings the rounded triangular edges on the front is the sign. Segor is said to be a reputable supplier so maybe sanyo are just using different casing now.
Regards
Alex
Im not well versed in transistor physics, but reading from the datasheets the main reason seems to be the Ic versus Vce linearity which implies a more linear transistor although i found cob plays a factor here too. The more constant the IC versus Vce changes the higher the early voltage.current the Trannies with lower cob also have higher early voltage. It seems the early effect, affects the higher frequencies most and the high frequency is effected by the cob. Pity theres no graph that gives info how cob affects the early voltage.
Hfe didnt get into the equation the ones i tested *2sa1016*were from group g and had hfe around 500. Manufacturing process might be better yes, but the doping ect play a factor too. Guys like halcro and opamp designers use ultra high frequency trannies in ltps which brings to mind they were chosen for low cob and high early voltage altough im pretty sure they found low noise too.
Im sure theres some info on this subject on the net somewhere, i had some in my varsity books but like my equipment and audio related stuff stayed at home. The girlfriend was complaining that i seemed to pay more attention to my plastic and metal babies *transistors and cars* than to her so they stayed at home, for now that is.
Jacco is spot on. The trannies pictured are more like the sanyos i know. Although one cannot see the markings the rounded triangular edges on the front is the sign. Segor is said to be a reputable supplier so maybe sanyo are just using different casing now.
Regards
Alex
Thanks Syn08 for the explanation, if i had to go read up again on manufacturing processes and the physics involved, would give me headaches all over again. Im busy with too many things at same time. Yes its axactly where i use them, anything below Ic 10ma operating. There are always tradeoffs yes and the best device is the one operating at its intended parameters and these operate great at input stages where these currents prevail. I dont know about beta though i have some gain group g s which measure close to 500 and 2 or 3 group h which measure around 780 on my fluke meter. The h are near impossible to find though. I have a batch of some hundred 2sa970 the highest was group bl at 460. The noise figure is very close to the same as i posted before, i did notice the 100uA. I ran some tests a couple of years back to better understand the which parameters were good for audio. Noise i just looked at datasheets. I thought it strange when i first joined this forum, saw excellent designs but being built with not so great parts. Building great amps involve not only good design but also using great parts and the layout and construction important too.
I also see my favourite driver is sold at profusion, its another sanyo. . Pity about vas transistors. Thanks again jacco. They have some transistors manufactured by unisonic, anybody know if these are as good as original japanese ????.
Alex
I also see my favourite driver is sold at profusion, its another sanyo.
. Pity about vas transistors. Thanks again jacco. They have some transistors manufactured by unisonic, anybody know if these are as good as original japanese ????.Alex
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.