The Sanken 2SA1186/2SC2837 are my favorite pair. They may have the lowest current and voltage rating of all Sanken LAPT, but they have 60 MHz f_T, good gain linearity and above all, a low cob capacitance of 110 pF (at 80 V). As they are increasingly difficult to source, I have begun looking for alternatives.
I stumbled on the Toshiba 2SA1093/2SC2563 (by the way, does anybody have a data sheet for the SC part?). They feature slighlty worse hFE linearity, but sport 90 MHz fT which is also incredibly flat with collector current). Cob of the PNP is given as 110 pF at 10 V, so it is probably even better than for the Sanken. Unfortunately, they have been discontinued and are even harder to source.
Toshiba go on in their selector guide about their improved high-MET process.
As a replacement, they give the A1941/C5198 pair which happens to be a 10 A part.
Even the A1940 which has exactly the same current rating as the SA1093 has higher Cob (260 instead of 150 pF). hFE linearity is equivalent on the high current end and maybe a tad better on the low current end (if you believe that it was not the data sheet artist at work). fT is given with 30 MHz but no plot is provided. IC vs. VCE looks just the same.
So what are the advantages of the new triple diffused process over the old PCT process?? I notice all the new Toshiba power parts only have 30 MHz fT. Are they maybe better than the data sheets suggest?
I stumbled on the Toshiba 2SA1093/2SC2563 (by the way, does anybody have a data sheet for the SC part?). They feature slighlty worse hFE linearity, but sport 90 MHz fT which is also incredibly flat with collector current). Cob of the PNP is given as 110 pF at 10 V, so it is probably even better than for the Sanken. Unfortunately, they have been discontinued and are even harder to source.
Toshiba go on in their selector guide about their improved high-MET process.
As a replacement, they give the A1941/C5198 pair which happens to be a 10 A part.
Even the A1940 which has exactly the same current rating as the SA1093 has higher Cob (260 instead of 150 pF). hFE linearity is equivalent on the high current end and maybe a tad better on the low current end (if you believe that it was not the data sheet artist at work). fT is given with 30 MHz but no plot is provided. IC vs. VCE looks just the same.
So what are the advantages of the new triple diffused process over the old PCT process?? I notice all the new Toshiba power parts only have 30 MHz fT. Are they maybe better than the data sheets suggest?
Looking at 2SA1302, it is easier to see that Toshiba did implement an improvement, even within the "triple diffused process":
The 2SA1943 has 30 MHz, 360 pF vs. 25 MHz, 470 pF in the 2SA1302. IC vs. VBE looks a little more linear in the newer part, and hFE linearity is a little flatter and flatness extends out a little farther now.
Comparing the 1943 to OnSemi MJL1302A, these may well be very similar parts. MJL has a max 600 pF vs. 470 pF typ in the 1943 which is probably as identical as one can get. hFE flatness looks very much the same. OnSemi do not give IC vs. VBE. They state a typical fT of 30 MHz at 5V, 1A for both PNP and NPN, but in contrast to Toshiba, they give also typical plots that contradict the tabular information. fT at 5V, 1A looks more like 40 MHz (1302) and 48 (3281), and the curves peak at 45 and 55 MHz.
As pointed out, there is no fT plot for the 1943, but there is one for the original 1302 which peaks at 28.
=> is there some hope that the new Toshiba parts are faster?
The 2SA1943 has 30 MHz, 360 pF vs. 25 MHz, 470 pF in the 2SA1302. IC vs. VBE looks a little more linear in the newer part, and hFE linearity is a little flatter and flatness extends out a little farther now.
Comparing the 1943 to OnSemi MJL1302A, these may well be very similar parts. MJL has a max 600 pF vs. 470 pF typ in the 1943 which is probably as identical as one can get. hFE flatness looks very much the same. OnSemi do not give IC vs. VBE. They state a typical fT of 30 MHz at 5V, 1A for both PNP and NPN, but in contrast to Toshiba, they give also typical plots that contradict the tabular information. fT at 5V, 1A looks more like 40 MHz (1302) and 48 (3281), and the curves peak at 45 and 55 MHz.
As pointed out, there is no fT plot for the 1943, but there is one for the original 1302 which peaks at 28.
=> is there some hope that the new Toshiba parts are faster?
2SA1943...
Hi,
I've been using the 1934 for a while. It seems to be a good part, but I do wish it was a little faster. I'm using it as a vas transistor, and have found it to be the slowest part in the amp.
If you locate a similar part that is faster, let me know... ...I'd be very interested.
-Dan
Hi,
I've been using the 1934 for a while. It seems to be a good part, but I do wish it was a little faster. I'm using it as a vas transistor, and have found it to be the slowest part in the amp.
If you locate a similar part that is faster, let me know... ...I'd be very interested.
-Dan
Hi Al,
actually, bipolar output stages are the most common thing. As an exhaustive primer, have a look at: http://www.dself.demon.co.uk/index.htm. You will also find quite a few schematics here, e.g. search for Krell.
Regards,
Eric
actually, bipolar output stages are the most common thing. As an exhaustive primer, have a look at: http://www.dself.demon.co.uk/index.htm. You will also find quite a few schematics here, e.g. search for Krell.
Regards,
Eric
capslock said:
Thanks Eric, I am familiar with the idea of separate NPN and PNP output trannies, ( I can't say that I totally understand them yet!), but I was querying the use of bipolar transistors in one package, such as the OnSemi MJL1302A. Or am I getting the wrong end of the stick again!
Dan,
have a look at the VAS thread. There are various video transistors in TO-126 package that can dissipate 8W with proper cooling, have a 120 or 180 V rating and still > 200 MHz f_T.
If you need higher power, just use the 1943 as a cascode transistor and use a fast small signal type for the steered current source.
Or buffer the VAS with a class-A complementary emitter follower.
Eric
have a look at the VAS thread. There are various video transistors in TO-126 package that can dissipate 8W with proper cooling, have a 120 or 180 V rating and still > 200 MHz f_T.
If you need higher power, just use the 1943 as a cascode transistor and use a fast small signal type for the steered current source.
Or buffer the VAS with a class-A complementary emitter follower.
Eric
Hi Dkempai,
Eric is 100% right. You need a superior transistor to the 1943 in this role.
Your application is 100V rails, so you need a voltage amplifier which will withstand around 50% more than 2 x 100V; viz a Vceo of 300V. It is clearly driving a large output stage, so it would be reasonable to run it at around 18mA. This means it will be dissipating 1.8W, so the transistor must be able to cope easily with at least three times this, namely 8W, npv.
Ft for the VAS is crucial. You make it as fast as possible. 30MHz of the 1943 is a doddle, and will sound slow and leaden. Go to at least 100MHz and preferably 200MHz, or more.
Now, the Sanyo Ultrahigh Definition CRT Display Video Output device, the 2SA1476, will cope with 200V, 15W, 200mA and offers ft of 400MHz. However, it is a little low on voltage, but there are higher voltage ones there; merely go to the Sanyo site and check it out. The other manufacturer of note is Toshiba. But it will mean some fairly serious websurfing, and sourcing. But they are around, and the TO-220AB case is good for 15W.
In your search, try to find Cre of 2.5pF or less.
Cheers,
Hugh
Eric is 100% right. You need a superior transistor to the 1943 in this role.
Your application is 100V rails, so you need a voltage amplifier which will withstand around 50% more than 2 x 100V; viz a Vceo of 300V. It is clearly driving a large output stage, so it would be reasonable to run it at around 18mA. This means it will be dissipating 1.8W, so the transistor must be able to cope easily with at least three times this, namely 8W, npv.
Ft for the VAS is crucial. You make it as fast as possible. 30MHz of the 1943 is a doddle, and will sound slow and leaden. Go to at least 100MHz and preferably 200MHz, or more.
Now, the Sanyo Ultrahigh Definition CRT Display Video Output device, the 2SA1476, will cope with 200V, 15W, 200mA and offers ft of 400MHz. However, it is a little low on voltage, but there are higher voltage ones there; merely go to the Sanyo site and check it out. The other manufacturer of note is Toshiba. But it will mean some fairly serious websurfing, and sourcing. But they are around, and the TO-220AB case is good for 15W.
In your search, try to find Cre of 2.5pF or less.
Cheers,
Hugh
Dan, in addition the 1943 has excessive parasitic Miller capacitance.
Hugh, what is your reasoning behind saying that the VAS transistor should have 1.5x the total (postive + negative) rail voltage? High voltage transistors usually have all their electrodes doped to have higher resistance which ruins both speed and hFE linearity.
Anyway, I was hoping anybody knew some high f_T, low C_ob output transistors other than the Sanken 1186/2837 parts...
Hugh, what is your reasoning behind saying that the VAS transistor should have 1.5x the total (postive + negative) rail voltage? High voltage transistors usually have all their electrodes doped to have higher resistance which ruins both speed and hFE linearity.
Anyway, I was hoping anybody knew some high f_T, low C_ob output transistors other than the Sanken 1186/2837 parts...
switching transistors for audio?
I have found the Sanyo 2SA1292/2SC3256 which are 60 V (only), 15 A switching transistors. Typical f_T is 100 MHz and pretty flat. C_ob is not specified. In spite of these devices being labeled switchting transistors, h_fe is pretty flat out to 10 A. Sourcing is kind of difficult, though. Has anybody used these?
Next interesting find are the NEC 2SA1232/2SC3012. I only have a Japanese data sheet without any graphs. f_T is 60 MHz, C_ob 250/150 MHz. Does anybody have a better data sheet?
I have found the Sanyo 2SA1292/2SC3256 which are 60 V (only), 15 A switching transistors. Typical f_T is 100 MHz and pretty flat. C_ob is not specified. In spite of these devices being labeled switchting transistors, h_fe is pretty flat out to 10 A. Sourcing is kind of difficult, though. Has anybody used these?
Next interesting find are the NEC 2SA1232/2SC3012. I only have a Japanese data sheet without any graphs. f_T is 60 MHz, C_ob 250/150 MHz. Does anybody have a better data sheet?
capslock said:
Well, in reality, I am running the 1943 in constant current mode, being driven by an upstream current regulator, on the positive supply rail. I'm using an 2SA1837 as the current handling device of a complementary pair on the negative rail. In both cases the larger slower transistors are seeing the voltage changes on their emitters, not collectors. So, I would probably need to look at the base to emitter capacitance.
In any case, the amp works well and is stable. (I'm only using 5pf in series with 10K to compensate the complementary pair used as the VAS transistor. With good board layout, I may not even need that. (currently it is on a breadboard, with lots of coupling and inductance
) -Dan
Sanken transistors, LAPT-3
Does anybody know what kind of replacements (or source) are suited for Sanken audio output transistors marked as LAPT-3?
On the Sanken web site I found only as info that these are custom made for Sansui, now discontinued, and nothing more! No suggested replacement, no equivalents... nothing at all! Even no e-mail contact addresse for asking them...
I sent an e-mail to AlegroMicrosystems (distributers of Sanken), but there's no any answer from them.
Here are the collected infromaton I found:
Sanken catalogue lists several transistors to belong to LAPT group, and I can recognize 2SC3519/2SA1386, 2SC3284/2SA1303, 2SC3264/2SA1295 which are incorporated in various Sansui amplifiers like: AU-X901, AU-X701, and AU-X911DG, respectively.
From the 1990's to the end of the company, they were used transistorsmarked as LAPT-2, LAPT-3 and LAPT 23 (these can be recognized as 200W devices, probably 2SC3264/2SA1295, but I dont know to what equivalents belong LAPT-3)
LAPT-3 looks lake standard TO-3P trannies, but collector base looks copper-alike (I guess it is a non-magnetic/antivibrating issue)
Thank you in advance
Does anybody know what kind of replacements (or source) are suited for Sanken audio output transistors marked as LAPT-3?
On the Sanken web site I found only as info that these are custom made for Sansui, now discontinued, and nothing more! No suggested replacement, no equivalents... nothing at all! Even no e-mail contact addresse for asking them...
I sent an e-mail to AlegroMicrosystems (distributers of Sanken), but there's no any answer from them.
Here are the collected infromaton I found:
Sanken catalogue lists several transistors to belong to LAPT group, and I can recognize 2SC3519/2SA1386, 2SC3284/2SA1303, 2SC3264/2SA1295 which are incorporated in various Sansui amplifiers like: AU-X901, AU-X701, and AU-X911DG, respectively.
From the 1990's to the end of the company, they were used transistorsmarked as LAPT-2, LAPT-3 and LAPT 23 (these can be recognized as 200W devices, probably 2SC3264/2SA1295, but I dont know to what equivalents belong LAPT-3)
LAPT-3 looks lake standard TO-3P trannies, but collector base looks copper-alike (I guess it is a non-magnetic/antivibrating issue)
Thank you in advance
2SA1943 is a high current device and therefore it's intended to be used in low impedance circuits
When such a big device is used in high impedance circuits [like VAS, input stage, etc...] it turns very slow since the magnitude of the currents available to charge and discharge B-E and B-C capacitances is very small. A 2SA1943 used as a VAS works like a TO-92 small signal device with a 5nF capacitor placed between B and E and a 360pF miller capacitor placed between C and B.
In the other hand, when you put such a device in an output stage and drive it from a much lower impedance it turns to be pretty fast
If a single small-signal device has not enough power handling for a certain VAS application, just use two, three, etc... in paralell. Toshiba has nice high voltage low current transistors
When such a big device is used in high impedance circuits [like VAS, input stage, etc...] it turns very slow since the magnitude of the currents available to charge and discharge B-E and B-C capacitances is very small. A 2SA1943 used as a VAS works like a TO-92 small signal device with a 5nF capacitor placed between B and E and a 360pF miller capacitor placed between C and B.
In the other hand, when you put such a device in an output stage and drive it from a much lower impedance it turns to be pretty fast
If a single small-signal device has not enough power handling for a certain VAS application, just use two, three, etc... in paralell. Toshiba has nice high voltage low current transistors
Better alternatives than 2SA1186 & 2SC2837?
Dear All,
like to respond to the original start of this thread by suggesting:
2SB817E & 2SD1047E from Sanyo - very linear transfer characteristics, but 'beta droops' a bit beyond 2-3A...
2SA1146 & 2SC2706 from Toshiba - these look to be quite remarkable devises...
The new 0302G & 0281G from OnSemi - They are available as both MJW in TO-247 case and NJW in TO-3P. There seem to be 'quite a following building' for these devises...
I have done some spread sheet and other severe desktop manoeuvres in order to have similar Vbe vs. Ic graphs for these OnSemi devises, like the ones published by most far eastern manufactures, to aid the benefit of possible direct comparison.
Please find the graphs attached below in standard JPG format. I have a better detailed version of the file available in open office drawing format as well if you like to request it by mail.
All the best
a1greatdane aka 'Dr' O
Chilli Sound Works
London, UK
Dear All,
like to respond to the original start of this thread by suggesting:
2SB817E & 2SD1047E from Sanyo - very linear transfer characteristics, but 'beta droops' a bit beyond 2-3A...
2SA1146 & 2SC2706 from Toshiba - these look to be quite remarkable devises...
The new 0302G & 0281G from OnSemi - They are available as both MJW in TO-247 case and NJW in TO-3P. There seem to be 'quite a following building' for these devises...
I have done some spread sheet and other severe desktop manoeuvres in order to have similar Vbe vs. Ic graphs for these OnSemi devises, like the ones published by most far eastern manufactures, to aid the benefit of possible direct comparison.
Please find the graphs attached below in standard JPG format. I have a better detailed version of the file available in open office drawing format as well if you like to request it by mail.
All the best
a1greatdane aka 'Dr' O
Chilli Sound Works
London, UK
Attachments
a1greatdane I couldn't find the link to your graphs. I'd be interested to see them?
My vote for BJTs are the workhorse (and cheap) 2SA1943/2SC5200 and the On Semi ThermalTrak parts especially the NJL4281D and NJL4302D for Class B designs where real-time bias can matter quite a bit. In my experience the genuine Toshiba 1943/5200's measure (and work) better than the Fairchild or other versions.
For those of you concerned about speed and fT I'm curious what you think of MOSFETs? They solve the speed problem nicely and don't suffer from the turn off problems that BJTs have. Of course you have to deal with the gate capacitance, but unless you have 4 or more pairs, even the LME49830 can do a respectable job driving them directly (i.e. with slew rates well in excess of what the amp will ever have to deal with).
My vote for BJTs are the workhorse (and cheap) 2SA1943/2SC5200 and the On Semi ThermalTrak parts especially the NJL4281D and NJL4302D for Class B designs where real-time bias can matter quite a bit. In my experience the genuine Toshiba 1943/5200's measure (and work) better than the Fairchild or other versions.
For those of you concerned about speed and fT I'm curious what you think of MOSFETs? They solve the speed problem nicely and don't suffer from the turn off problems that BJTs have. Of course you have to deal with the gate capacitance, but unless you have 4 or more pairs, even the LME49830 can do a respectable job driving them directly (i.e. with slew rates well in excess of what the amp will ever have to deal with).
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.