High Anatech.
Guess that those ring emitter transistors were nothing else than japanese transistors that were manufactured by Motorola
among others (they started fabbing 2SA1302/2SC3281) after japanese firms were forced at gun point by the Reagan administration to give for free their process to US firms.
Guess that those ring emitter transistors were nothing else than japanese transistors that were manufactured by Motorola
among others (they started fabbing 2SA1302/2SC3281) after japanese firms were forced at gun point by the Reagan administration to give for free their process to US firms.
Even RCA started doing multi emitter types back when the Earth was cooling. But they were limited to HV switching types in the early days. It was Toshiba that picked it up and ran with it, making pretty much the best audio transistors in the business for 2 decades while poor Motorola played catch-up. But they DID make much better regular epitaxial types than the 3055. They made a very good 2N3773, and for a while, a PNP. Others didn’t have as wide an SOA but had a lot of current gain at 10A (2N5631). Fairchild made a convincing one too - and I bet that irked everyone at Motorola off. I’ve seen a LOT of the Fairchild units in vintage equipment. So along comes the MJ15001,2,3,4 which are actually still considered pretty good. It wasn’t until the 1502x, however, till Motorola stared doing multiple emitter. Finally, they come up with something that can best the 2SD424 - and that was because it would RUN on 200 Vce, where the 424 would not. It was still another two decades before the Jap technology was “acquired”.
According to my data book, Toshiba made a 2SD552/2SB552 pair, rated to 220V with spec a bit better than D424/B554. But I never got to see one in real life - they disappeared quickly. Maybe they couldn’t yield. I can’t believe it would be lack of demand - service techs would probably be beating down the factory doors trying to get them.
Both Toshiba and Sanken make multi emitter types. Toshiba’s are triple diffused, Sanken’s are epitaxial planar. Sanken makes triple diffused parts too but they are not as good (Cost less though). Some of the Sanyo outputs were triple diffused NPN with a planar type PNP, and still called “complementary”.
According to my data book, Toshiba made a 2SD552/2SB552 pair, rated to 220V with spec a bit better than D424/B554. But I never got to see one in real life - they disappeared quickly. Maybe they couldn’t yield. I can’t believe it would be lack of demand - service techs would probably be beating down the factory doors trying to get them.
Both Toshiba and Sanken make multi emitter types. Toshiba’s are triple diffused, Sanken’s are epitaxial planar. Sanken makes triple diffused parts too but they are not as good (Cost less though). Some of the Sanyo outputs were triple diffused NPN with a planar type PNP, and still called “complementary”.
Curiously well performing transistors had the paradoxal consequence to produce some kind of regression,
in the 80 some japanese manufacturers came back with basic schematics since a 2SA1302/2SC3281 pair driven
by a 2SA968/2SC2238 pair + a classical differential + non enhanced VAS + EF2 was good enough to yield decent
numbers, i ve seen 80s amps, from Pioneer among others, with +-45-50V rails using such a basic topology.
As for the unvailability of some japanese transistors this was due to strict business management where they wouldnt
produce anything if there was no orders, that was the era of the three Z, zero default, zero delay but also zero stock,
although in the 80s Hitachi breached this rule and made available their much sought 2SJ48-50/2SK133-135 lateral fets,
at the time, in 1988, i picked 6 pairs of each for a 2 x 100W/8R amplifier using 3 complementary/channel and wich is
perfectly functional to this day despite a treatment as PA amplifier that would have killed any BJT based amp.
in the 80 some japanese manufacturers came back with basic schematics since a 2SA1302/2SC3281 pair driven
by a 2SA968/2SC2238 pair + a classical differential + non enhanced VAS + EF2 was good enough to yield decent
numbers, i ve seen 80s amps, from Pioneer among others, with +-45-50V rails using such a basic topology.
As for the unvailability of some japanese transistors this was due to strict business management where they wouldnt
produce anything if there was no orders, that was the era of the three Z, zero default, zero delay but also zero stock,
although in the 80s Hitachi breached this rule and made available their much sought 2SJ48-50/2SK133-135 lateral fets,
at the time, in 1988, i picked 6 pairs of each for a 2 x 100W/8R amplifier using 3 complementary/channel and wich is
perfectly functional to this day despite a treatment as PA amplifier that would have killed any BJT based amp.
Hi wahab,
Yes, On Semi markets them using the same serial number.
Here I was never able to get Japanese data sheets for individual parts. The closest I was able to get were data books in list form with basic parameters written in Japanese. That was a huge step up from people that cross referenced in and out of ECG or NTE replacement books (the book of lies as I call them). You can't do that. The only details on semiconductors I could get was from American firms like Motorola / On Semi and others. "Back in the day" I used a lot of 2N5631 and 2N6031, then various MJ series (but never 15015/15016 garbage).
The MJ1502x transistors were the "new type" of faster and more linear power devices. I used tons of 2SD555 and 2SB600, plus 2SC424 and 2SA554. The 2SD424/B554 were my metal stocked parts in addition to the On Semi parts. In plastic power, most of what was listed above. The lateral FETs I saw mostly in Luxman and a couple others. My supplier for Japanese parts bought them direct from Japan, he went about twice - three times a year and got timed shipments. He was a little more expensive, but I bought solely from him and only ever had three defective devices over a decade, including ICs.
Freedom from BJT SOA curves allow FETs to survive short term overloads, but they still die from overheating. So if you abuse devices without current protection (SOA for BJTs), FETs will survive longer. But, the still need drivers that can handle current. Folks think they don't require gate current. Not at DC, but they do at higher frequencies.
Yes, On Semi markets them using the same serial number.
Here I was never able to get Japanese data sheets for individual parts. The closest I was able to get were data books in list form with basic parameters written in Japanese. That was a huge step up from people that cross referenced in and out of ECG or NTE replacement books (the book of lies as I call them). You can't do that. The only details on semiconductors I could get was from American firms like Motorola / On Semi and others. "Back in the day" I used a lot of 2N5631 and 2N6031, then various MJ series (but never 15015/15016 garbage).
The MJ1502x transistors were the "new type" of faster and more linear power devices. I used tons of 2SD555 and 2SB600, plus 2SC424 and 2SA554. The 2SD424/B554 were my metal stocked parts in addition to the On Semi parts. In plastic power, most of what was listed above. The lateral FETs I saw mostly in Luxman and a couple others. My supplier for Japanese parts bought them direct from Japan, he went about twice - three times a year and got timed shipments. He was a little more expensive, but I bought solely from him and only ever had three defective devices over a decade, including ICs.
Freedom from BJT SOA curves allow FETs to survive short term overloads, but they still die from overheating. So if you abuse devices without current protection (SOA for BJTs), FETs will survive longer. But, the still need drivers that can handle current. Folks think they don't require gate current. Not at DC, but they do at higher frequencies.
Hi wg_ski,
The RCA devices were faster for the same part number. Remember we had to adjust the HF compensation cap on older amplifiers depending on the make of output transistor?
The MJ15003,4 are decent transistors. Especially compared to the earlier device we had to choose from.
I really wish I had access to the actual Japanese data sheets back then. Big secret in my world.
The RCA devices were faster for the same part number. Remember we had to adjust the HF compensation cap on older amplifiers depending on the make of output transistor?
The MJ15003,4 are decent transistors. Especially compared to the earlier device we had to choose from.
I really wish I had access to the actual Japanese data sheets back then. Big secret in my world.
Back when you get D424’s there was no “On Semiconductor”. It was Motorola.
I had the “Japanese Transistor Manual’s” you speak of, from 1985-89. It’s just amazing how much Good Stuff disappeared between the 4 of them. I also had a handful of original Toshiba dats sheets. Including 2SD424, 2SD552, 2SD873, 2SC3182, 2SC3281, and 2SC2565. As well as a few horizontal output transistors. I had a D555 data sheet too, and it was darn near the same as the elusive MJ15011/12. Never could find those for sale anywhere, but could get D555/B600 from MCM.
I had the “Japanese Transistor Manual’s” you speak of, from 1985-89. It’s just amazing how much Good Stuff disappeared between the 4 of them. I also had a handful of original Toshiba dats sheets. Including 2SD424, 2SD552, 2SD873, 2SC3182, 2SC3281, and 2SC2565. As well as a few horizontal output transistors. I had a D555 data sheet too, and it was darn near the same as the elusive MJ15011/12. Never could find those for sale anywhere, but could get D555/B600 from MCM.
Hi wg_ski,
MCM is a place I didn't buy from very often. That involved shipping and customs clearance to Canada. I bought semis locally and from a place called "Future Electronics" in Montreal. Early on I bought from EBC, or Electronics Buyers Club in CA. I would buy large bulk from them when I could wait.
I have data books sets from 1986 up to the early 2000's I think. These are highly sought after here and were invaluable. Cross reference information is typically incorrect. It really paid to go from the real specs. I have partial Sanyo data for the STK packs, not one single complete or detailed datasheet on a Japanese semiconductor. It was always a sore point, and Japanese semi representatives here wouldn't part with any.
Probably, could be. It didn't matter enough to remember this stuff as far as dates go. As I said, I would have to look at my data books and catalogs to determine this stuff, mind garbage otherwise. I bought direct from both incarnations.
MCM is a place I didn't buy from very often. That involved shipping and customs clearance to Canada. I bought semis locally and from a place called "Future Electronics" in Montreal. Early on I bought from EBC, or Electronics Buyers Club in CA. I would buy large bulk from them when I could wait.
I have data books sets from 1986 up to the early 2000's I think. These are highly sought after here and were invaluable. Cross reference information is typically incorrect. It really paid to go from the real specs. I have partial Sanyo data for the STK packs, not one single complete or detailed datasheet on a Japanese semiconductor. It was always a sore point, and Japanese semi representatives here wouldn't part with any.
Hi Anatech.
For Hitachi laterals the required drive current is very low, using a single pair for say 100W the necessary VGS imply
no more than 3.24 mW at 100kHz to charge/discharge the gate/source capacitance, with 3 such parraleled devices
the capacitance will be tripled but the necessary VGS will be reduced by a 1.7 factor, so that s still the same required
power drive, that s why those transistors can be driven directly by the VAS, in my old lateral fets amp the VAS current
is no more than 2.5mA with a +-60V idle supply voltage and driving 3 laterals pairs.
For Hitachi laterals the required drive current is very low, using a single pair for say 100W the necessary VGS imply
no more than 3.24 mW at 100kHz to charge/discharge the gate/source capacitance, with 3 such parraleled devices
the capacitance will be tripled but the necessary VGS will be reduced by a 1.7 factor, so that s still the same required
power drive, that s why those transistors can be driven directly by the VAS, in my old lateral fets amp the VAS current
is no more than 2.5mA with a +-60V idle supply voltage and driving 3 laterals pairs.
Hi wahab,
Hmm, interesting. I never did take the time to figure it all out. I did note that amplifiers with a buffered VAS driving mosfets seemed to perform better. With my instrumentation today I could determine much more.
One interesting design was the Revox B242 amplifier. It used a Mosfet stage to isolate the output stage from the VAS circuitry.
Hmm, interesting. I never did take the time to figure it all out. I did note that amplifiers with a buffered VAS driving mosfets seemed to perform better. With my instrumentation today I could determine much more.
One interesting design was the Revox B242 amplifier. It used a Mosfet stage to isolate the output stage from the VAS circuitry.
High Anatech, indeed even Hitachi s laterals perform better with a buffered VAS, that s due to the capacitive loading
being completely neutered, hence the OLG at high frequencies is better sustained at high frequencies and there s
more NFB available at said frequencies, if the VAS has enough gain the result will be about the same as with a buffer.
On a side note you are right for the much better SOA of those devices, the 2SJ48-50/2SK133-135 are only 7A/100W
and yet they can be used at 5A peak current without any trouble, that s the big difference with BJTs where a 10A/100W
device would never sustain such a current, in my exemple of power amplifier with 3 pairs, that s 21A and 300W per side
and the amp was loaded with 4R fed by a 2 x 43V AC 560VA toroid and used as PA amp working at full power.
being completely neutered, hence the OLG at high frequencies is better sustained at high frequencies and there s
more NFB available at said frequencies, if the VAS has enough gain the result will be about the same as with a buffer.
On a side note you are right for the much better SOA of those devices, the 2SJ48-50/2SK133-135 are only 7A/100W
and yet they can be used at 5A peak current without any trouble, that s the big difference with BJTs where a 10A/100W
device would never sustain such a current, in my exemple of power amplifier with 3 pairs, that s 21A and 300W per side
and the amp was loaded with 4R fed by a 2 x 43V AC 560VA toroid and used as PA amp working at full power.
Yes!
With bipolars, once you get over 50 V C-E, SOA often restricts you to as little as 2 amperes (or even less) averaging over many types. Over 150V you're into the 100 mA or so scale depending on part. You can get away with short term peaks, but anything sustained will get you.
I just find I like bipolar outputs when you aren't trying to deliver massive power. In other words, to listen to. I think maybe the FET amps I have heard were too simplified and therefore didn't sound as good as they could have.
With bipolars, once you get over 50 V C-E, SOA often restricts you to as little as 2 amperes (or even less) averaging over many types. Over 150V you're into the 100 mA or so scale depending on part. You can get away with short term peaks, but anything sustained will get you.
I just find I like bipolar outputs when you aren't trying to deliver massive power. In other words, to listen to. I think maybe the FET amps I have heard were too simplified and therefore didn't sound as good as they could have.
That s something i cant answer technically speaking for sure, i ll just point Hitachi s words on the subject.
They state that bipolars have about 38S transconductance while their lateral have only 1S, and hence they require
more NFB to achieve as low THD as bipolars, so in this respect bipolars are more forgiving when low amount of NFB
are used, and eventually also when the designs are not much refined.
That being said the higher Ft of laterals allow for higher NFB at higher frequencies, so it s possible to design good amps,
FTR there was a Hitachi commercial amp circa 1978 and it had very good numbers THD and IMD wise, all while delivering
2 X 70W/8R and 2 x 109W/4R out of a single 2SJ48/2SK133 pair, and the french review did load it at full power, including
at 4R, for one hour before doing the measurements.
They state that bipolars have about 38S transconductance while their lateral have only 1S, and hence they require
more NFB to achieve as low THD as bipolars, so in this respect bipolars are more forgiving when low amount of NFB
are used, and eventually also when the designs are not much refined.
That being said the higher Ft of laterals allow for higher NFB at higher frequencies, so it s possible to design good amps,
FTR there was a Hitachi commercial amp circa 1978 and it had very good numbers THD and IMD wise, all while delivering
2 X 70W/8R and 2 x 109W/4R out of a single 2SJ48/2SK133 pair, and the french review did load it at full power, including
at 4R, for one hour before doing the measurements.
Last edited:
That's what I saw in the market coming across my service bench.
Really good bipolar amps tended to be more complicated and unsurprisingly, most performed better. In the literature I read for Mosfets (lateral types), they pushed easier drive and less complicated circuitry. That's what we got. It's too bad designers didn't apply themselves to create equal designs, but I guess it was for more profit. The other mosfets types were not recommended for audio (but some still used them). Counterpoint in particular.
I think the average "okay" amplifier sounded better with Mosfets - maybe. That points to poor design with bipolars more than anything. The first I saw was the Nikko Alpha III. Numerous others became popular. The Luxman LV-105u was an "okay" amplifier for example, but we had things like the Bryston 4B, which wasn't a good sounding amplifier at all. I'm not picking on Mosfets at all, but I don't think designers gave them a chance to perform as well as they could have. They were much more expensive devices, that's for sure.
Really good bipolar amps tended to be more complicated and unsurprisingly, most performed better. In the literature I read for Mosfets (lateral types), they pushed easier drive and less complicated circuitry. That's what we got. It's too bad designers didn't apply themselves to create equal designs, but I guess it was for more profit. The other mosfets types were not recommended for audio (but some still used them). Counterpoint in particular.
I think the average "okay" amplifier sounded better with Mosfets - maybe. That points to poor design with bipolars more than anything. The first I saw was the Nikko Alpha III. Numerous others became popular. The Luxman LV-105u was an "okay" amplifier for example, but we had things like the Bryston 4B, which wasn't a good sounding amplifier at all. I'm not picking on Mosfets at all, but I don't think designers gave them a chance to perform as well as they could have. They were much more expensive devices, that's for sure.
So a made a rapid test on 5 2N3055 from Motorola, and excepted one wich seems a little deaf and has a 41 gain at 3.6A
all others are in a range of 55-59 gain at 4.2-4.5A, but guess that those 4 transistors are from a same batch since there s
only 10% gain variability.
At a little lower current the reluctant one has about 63 gain at 2.3A while the four others are in a 74-82 range for about 2.75A.
all others are in a range of 55-59 gain at 4.2-4.5A, but guess that those 4 transistors are from a same batch since there s
only 10% gain variability.
At a little lower current the reluctant one has about 63 gain at 2.3A while the four others are in a 74-82 range for about 2.75A.
Yep, my experience was beta highly variable, but generally low. I was not encouraged to use them from my testing back in the 1970's. You'd get some with reasonable beta.
Hi wahab,
I highly suspect those were selected samples somehow. I would have been shocked to see a group of four that good. 10% variations was excellent for any number back then. A 2N3055 wasn't a premium part by any stretch, and their industrial applications were not demanding.
Hi wahab,
I highly suspect those were selected samples somehow. I would have been shocked to see a group of four that good. 10% variations was excellent for any number back then. A 2N3055 wasn't a premium part by any stretch, and their industrial applications were not demanding.