Using less pair of higher power transistor can run into heat sink problem as heat concentrates at fewer spots. If you have 5 pairs like what I am doing, you spread the heat out better.
Yep, more pairs of cheap transistors can be cheaper rather to look for "the" transistor.
Something to consider is that On-Semi transistors almost always have much higher capacitances than similar Japanese ones, even the high Ft ones. This may have to do with the lower Ft (again, even "high Ft" On-Semis don't have such a high Ft).
I don't know why capacitance is always a few times higher, maybe they just use bigger dies, or they need bigger dies for the same SOA, or their fabrication process results in higher capacitances, or just they don't optimize capacitances at all(I think the last option).
Several times I had to increase miller capacitance when replacing dead On-Semi (Motorola) MJE340&MJE350 VAS and pre-drivers by Philips ones (no other brand available) in order to make the amplifiers stable again. If I remember properly, the ones from Philips also had higher hFE (and probably 5-10 times higher Ft... Philips seems like the opposite example, they always seem to do everything with as little silicon as possible, Japanese seem to be in the middle
I am glad there is so much information on this forum.
I have been reading through some old threads here. This is exactly what I see, the Motorola/On semi stuffs seems to have much higher capacitance. I am searching around on this forum on this. I am planning to design a power amp with between 5 to 7 pairs of output transistor. I am concern more popular ones from On Semi like MJW1302/3281 have very high capacitance. It would be hard of the driver transistor to drive 5 to 7 of these transistors.
I am searching on Digikey and find 2STA1962 and 2STC5242 to have collector base capacitance of 150pF compare to MJW3281 that output capacitance ( I assume is collector base capacitance) of 600pF. I am leaning towards the 2STA and 2STC for my case that I want to use 5 to 7 pairs to make it easier for the driver transistors.
Am I correct or is there other things about the MJW3281/1302 that make it better for power amps?
Well the MJW3281/1302 isn't obsolete from the vendor for one.
The high capacitance devices usually have SOA advantages. But you can only really tell by overlaying the expected load, and annoyingly not all manufacturer's graphs are well detailed. IIRC ON did some work and the MJL4281/4302 had superior SOA for a part of its time.
The ST equivalents have pretty neat specifications, but the capacitance is much more relevant on a CFP style output stage rather than the EF style.
The high capacitance devices usually have SOA advantages. But you can only really tell by overlaying the expected load, and annoyingly not all manufacturer's graphs are well detailed. IIRC ON did some work and the MJL4281/4302 had superior SOA for a part of its time.
The ST equivalents have pretty neat specifications, but the capacitance is much more relevant on a CFP style output stage rather than the EF style.
Thanks for your reply MT490
As I specified, I start out designing with minimum of 5 pairs and up to 7 pairs. SOA is not a concern at all. From my understanding, after conforming to Oliver's condition ( 26mV across the emitter resistor for EF output transistor), the only other way to lower crossover distortion is more pairs in parallel. SOA is not a concern, input capacitance becomes critical as it put very heavy load on the driver transistors.
I designed and test a 5 pairs stage with MJW, the drivers get hot fast when I sweep the frequency up.
Then again, these are text book knowledge, I want to hear from people that have actual experience. How do people choose transistors. This has been a mystery for me.
Thanks
As I specified, I start out designing with minimum of 5 pairs and up to 7 pairs. SOA is not a concern at all. From my understanding, after conforming to Oliver's condition ( 26mV across the emitter resistor for EF output transistor), the only other way to lower crossover distortion is more pairs in parallel. SOA is not a concern, input capacitance becomes critical as it put very heavy load on the driver transistors.
I designed and test a 5 pairs stage with MJW, the drivers get hot fast when I sweep the frequency up.
Then again, these are text book knowledge, I want to hear from people that have actual experience. How do people choose transistors. This has been a mystery for me.
Thanks
Hi Stephen,
The OnSemi 3281/1302 series of transistors is good. If you are using many output pairs, you should consider using a 3281/1302 pair as the driver and run significant bias through the driver, perhaps at least 100mA. Do not use a CFP output stage. Use a 3EF Triple or something similar, maybe like a diamond buffer triple (DBT) or diamond buffer quad as described in my book.
Cheers,
Bob
This experience I have make also.
By chance I have discovered the SM3159/SM3160:
http://uk.rs-online.com/webdocs/00bb/0900766b800bbc7e.pdf
Which kind of the in post #1 mentioned are these and what replacement shold I order?
Those MagnaTeck power transistors are not that great. They have a typical ft of only 10 MHz.
Bob
Did You recognize, that Toshiba gives "typical" values, and Onsemi specify "maximum" values?
Sajti
I realize the entire thread is pretty ancient, and many of the listed transistors are obsolete *and* hard to find. Perhaps a modern list of useful complementary output BJT's could be curated?
I'm also reminded by Winfred Hill's use of Cob/Pdiss as his figure of merit when designing a >1 MHz power bandwidth lab amplifier.
I'm also reminded by Winfred Hill's use of Cob/Pdiss as his figure of merit when designing a >1 MHz power bandwidth lab amplifier.
Are you do this observation in general or only in a certainly circuit topologies, where the output buffer operate in Class AB (idle current arround 20-50mA for each pair ?
From post #1 I have selected the versions with the highest value of ft - except the SANKEN versions most of them out of production unfortunately:
NEC
2SA1141/2SC2681 blue/bl 115V 10A 100W hfe: 60-200 80MHz similar TO247 (no orig. NEC datasheet)
2SA1227/2SC2987 gr/bl 140V 12A 120W hfe: 60-320 50MHz similar TO247
2SA1227A/2SC2987A gr/bl 140V 12A 120W hfe: 60-320 50MHz similar TO247
Toshiba
2SA1095/2SC2565 160V 15A 150W hfe: 40min 60MHz RM60 (34,3x16,3x5,3mm), old type, gr/bl
SANKEN
2SA1186/2SC2837 150V 10A 100W hfe: 50min 60MHz TO-3P (MT100)
2SA1303/2SC3284 150V 14A 125W hfe: 50min 50MHz TO-3P (MT100)
2SA1215/2SC2921 160V 15A 150W hfe: 50min 50MHz MT200
Fujitsu
2SA1043/2SC2433 120V 10A 150W hfe: 35-200 60MHz TO-3, old type
2SA1044/2SC2434 70V 10A 150W hfe: 35-200 60MHz TO-3, old type
2SA1075/2SC2525 gr/bl 120V 12A 120W hfe: 60-200 60MHz RM60 (34,3x16,3x5,3mm), old type
2SA1076/2SC2526 gr/bl 120V 12A 120W hfe: 60-200 60MHz RM60 (34,3x16,3x5,3mm), old type
Matsushita
2SA1064/2SC2488 150V 8A 100W hfe: 40-280 50MHz TO-3
2SA1065/2SC2489 150V 10A 120W hfe: 40-200 50MHz TO-3
Independent of the circuit topology the consequences are different by replace such types with such of low ft arround 4MHz.
What happens exactly by the various versions of front-ends and output buffers (like Harman's T-Circuit or Linn's CFP approach) - particularly concerning stability behaviour, THD-behaviour and audible sonic character ?
Are there special tasks in the individual circuit topology to avoid disadvantages by sound quality ?
Thank you for advices.
Last edited:
according the presently available line-up of audio power devices there are some new versions - OTOH the ones in MT-200 outline are not longer listet - go to
Audio ｜Sanken Electric
most interesting types for replace a lot of no longer available output devices seems to be this:
http://www.semicon.sanken-ele.co.jp/sk_content/2sa2151a_ds_en.pdf
http://www.semicon.sanken-ele.co.jp/sk_content/2sc6011a_ds_en.pdf
and
http://www.semicon.sanken-ele.co.jp/sk_content/2sa2223a_ds_en.pdf
http://www.semicon.sanken-ele.co.jp/sk_content/2sc6145a_ds_en.pdf
But the most imrortant difference I don't find in the datasheets.
Audio ｜Sanken Electric
most interesting types for replace a lot of no longer available output devices seems to be this:
http://www.semicon.sanken-ele.co.jp/sk_content/2sa2151a_ds_en.pdf
http://www.semicon.sanken-ele.co.jp/sk_content/2sc6011a_ds_en.pdf
and
http://www.semicon.sanken-ele.co.jp/sk_content/2sa2223a_ds_en.pdf
http://www.semicon.sanken-ele.co.jp/sk_content/2sc6145a_ds_en.pdf
But the most imrortant difference I don't find in the datasheets.
Last edited:
This is correct according this image under
Counterfeit Transistors
(fourth image)
and
http://i.ebayimg.com/images/i/201497692605-0-1/s-l1000.jpg
The die is attached to a TO3P type heatsink which is in turn bonded to a larger heatsink contact face (therefore a MT200 outline).
The abbreviation "LAPT" confuses me.
Under
2SA2223 - Power Transistor -260V -15A (Master Part)
I read LAPT (High frequency multi emitter transistor)
Under
http://www.semicon.sanken-ele.co.jp/sk_content/2sa2223a_ds_en.pdf
there is to read this:
LAPT (Linear Amplifier Power Transistor)
And under
Sanken audio components at Profusion
is mentioned this definition
LAPT (Large Area Parallel Transistor)
Only the last mentioned definition I know - go to
http://www.profusionplc.com/brands/sanken
and post 11 under
How many transistors inside 2SA1302/2SC3281 ?
so as post #2 under
The skinny on Sanken 2SA1216 Epitaxials
Last edited:
MT200 devices have been obsoleted with the advent of the newer generation thin die LAPT. The die geometry and electrical performance is the same but the thinner silicon makes it possible to dissipate MT200-like power in a MT100 case. Since the MT200 has become unique to Sanken, it is probably a large factor in the cost of a transistor, and also probably an increasing percentage of the price.