Quick question I'd love to have a conclusive answer on. I'm sure that it's splitting hairs but nonetheless...
My amp uses 970/2240 transistors for the input stage and 1145/2705 for the VAS stage.
There is a Marantz service bulletin that actually recommends replacing some of the input stage with their bigger counterparts.
My question is whether it's actually a good idea in the sound quality department? I have new-old stock of the 1145/2705s - about 20 of each so I can hFE match.
View attachment NPNB 2sc2240.pdf
View attachment NPNC 2SC2705.pdf
The higher powered transistors actually have a lower output capacitance, but no noise figure is stated in the data sheet. That tells me they should stay where they were intended for - the VAS stage.
Any insight would be much appreciated.
My amp uses 970/2240 transistors for the input stage and 1145/2705 for the VAS stage.
There is a Marantz service bulletin that actually recommends replacing some of the input stage with their bigger counterparts.
My question is whether it's actually a good idea in the sound quality department? I have new-old stock of the 1145/2705s - about 20 of each so I can hFE match.
View attachment NPNB 2sc2240.pdf
View attachment NPNC 2SC2705.pdf
The higher powered transistors actually have a lower output capacitance, but no noise figure is stated in the data sheet. That tells me they should stay where they were intended for - the VAS stage.
Any insight would be much appreciated.
Oh good question.
The A970/C2240 are ranked GR, and measuring between 200 and 240
The A1145/C2705 stock I have is ranked Y and measuring consistently around 140.
The A970/C2240 are ranked GR, and measuring between 200 and 240
The A1145/C2705 stock I have is ranked Y and measuring consistently around 140.
Can you show what the service bulletin statement is - sometimes the noise figure may not be the criterion for the function that transistor is used.
Because if Cob is the main concern, then KSA992/KSC1845 have even lower at 1.6pF and you'll find higher hFE.
Probably the bulletin is aiming for higher hFE devices or rank. What model is the Marantz?
The PSB is actually addressing potential failure due to temperature issues. Not worried about that. I want what sounds best. Noise is my main concern and Cob is probably second.
PSB and schematic attached.
View attachment Marantz Sm6100 Mz05-013 Service Manual.pdf
I was hoping I could get more insight as to the true nature of the two transistors however. Were the TO-92L transistors just a more expensive version of the smaller ones with the same golden sound?
I know the 992/1845s are great replacements. I only want the best for my baby so what is it?
PSB and schematic attached.
View attachment Marantz Sm6100 Mz05-013 Service Manual.pdf
I was hoping I could get more insight as to the true nature of the two transistors however. Were the TO-92L transistors just a more expensive version of the smaller ones with the same golden sound?
I know the 992/1845s are great replacements. I only want the best for my baby so what is it?
1. Are you aware that the Service Manual is from 2007?
2. Where do you intend to purchase 2SA1145/2SC2705 in 2021?
Now you have KSA1381/KSC3503 but the problem is to get transistor pairs with the same hFE classification.
Otherwise, as someone else mentioned KSA992/KSC1845 are the closest eqivalents of 2SA970/2SC2240, but you are looking for larger transistors as suggested by your Service Manual.
And hurry up, all these transistors are increasely difficult to obtain from reputable sources.
Otherwise, there are ISC generic products 2SC3423/2SA1360 which are equivalent to 2SA1145/2SC2705.
Good luck in searching the right replacements.
2. Where do you intend to purchase 2SA1145/2SC2705 in 2021?
Now you have KSA1381/KSC3503 but the problem is to get transistor pairs with the same hFE classification.
Otherwise, as someone else mentioned KSA992/KSC1845 are the closest eqivalents of 2SA970/2SC2240, but you are looking for larger transistors as suggested by your Service Manual.
And hurry up, all these transistors are increasely difficult to obtain from reputable sources.
Otherwise, there are ISC generic products 2SC3423/2SA1360 which are equivalent to 2SA1145/2SC2705.
Good luck in searching the right replacements.
Last edited:
Thanks Berlusconi but as stated before I have brand new stock sitting in my tickle trunk waiting for my next amp rebuild:
The next iteration of this build I will use 2SA1163/2SC2713 which are the SMD versions of the A970/C2240 and hopefully just as good.
For now I just want to know if it's a good idea to update the old trannies in the existing rig with the larger/newer ones as per the PSB, and if so can I replace all 6 in the main input stage.
The next iteration of this build I will use 2SA1163/2SC2713 which are the SMD versions of the A970/C2240 and hopefully just as good.
For now I just want to know if it's a good idea to update the old trannies in the existing rig with the larger/newer ones as per the PSB, and if so can I replace all 6 in the main input stage.
How much power are the A970/C2240 dissipating? Is the PCB showing a dark area around them?
These are very good 100mA TO-92 transistors, made for audio. If you have the A1145/C2705 then follow the bulletin. These are also good transistors, Accuphase used them as well.
But bulletin also says to replace 2SA1048 and 2SC2458 with KTA1268 and KTC3200, which have similar parameters to the A970/C2240 a stage earlier, so maybe you want to swap them.
But if your amp is not having any issues, I wouldn't worry about it.
This is already a rebuild on double-sided and drilled perf-board with passive cooling channels in the bottom of the case. Not to mention even component spacing.
Not worried about heat. The next design will thermally couple the differential pairs and provide better electrical symmetry.
But I digress. Is the 2SA1145/2SC2705 ideal for the input stage, albeit with slightly lower gain? The existing transistors have an hFE variance of 20% whereas the new ones would only be about 5%. That is what I really want to know.
I've got these Pirelli tires on my car that are unevenly worn, should I replace them with a new set of Pirellis that are .5" larger diameter than the original spec?
Not worried about heat. The next design will thermally couple the differential pairs and provide better electrical symmetry.
But I digress. Is the 2SA1145/2SC2705 ideal for the input stage, albeit with slightly lower gain? The existing transistors have an hFE variance of 20% whereas the new ones would only be about 5%. That is what I really want to know.
I've got these Pirelli tires on my car that are unevenly worn, should I replace them with a new set of Pirellis that are .5" larger diameter than the original spec?
Last edited:
Maybe you don't read the app notes on the datasheets for the original Toshiba parts? The bigger package size of the 2705/1145 should hint that its intended for more power dissipation (1W), like in the VAS stage, where low capacitance is also advantageous to a linear, high signal voltage swing between the power rails. Perhaps the pair could also double as a pre-driver pair for the output stage, a cascode transistor or general purpose drivers for even a relay or indicator etc. Such other applications may not need any special qualities but if there is a lower price for a larger quantity of particular, essential type of transistor, a smaller inventory will also mean an even cheaper overall cost to the manufacturer.
On the other hand, 2240/970 are definitely intended specifically for the input pair - very few low cost transistors beat them on that critical low noise level but for economic purchasing and inventory treasons, they may also wind up in general purpose apps too.
I should add that non-original parts will be a different kettle of fish. Whilst they likely work and give acceptable gain, noise figures and hFE won't be stellar. I found in that across 4 allegedly good types of input pairs from a Chinese platform seller, they proved statistically, to be all the same components, probably labelled according to the supplier's marketing strategy, I'd guess. So the best, genuine new stock part substitutes for 2240/970 are undoubtedly KSC1845/KSA992 from Onsemi resellers. Don't even think of platform seller's junk.
As you are aware, SMD is fast becoming the only reliable way to obtain original quality substitute semis. Coming to grips with making or buying stock PCBs to suit them is a big problem for DIYs and I certainly understand the wish to stay with thru-hole tech. Unfortunately, the fakes issues just keep increasing with thru-hole parts purchasing.
On the other hand, 2240/970 are definitely intended specifically for the input pair - very few low cost transistors beat them on that critical low noise level but for economic purchasing and inventory treasons, they may also wind up in general purpose apps too.
I should add that non-original parts will be a different kettle of fish. Whilst they likely work and give acceptable gain, noise figures and hFE won't be stellar. I found in that across 4 allegedly good types of input pairs from a Chinese platform seller, they proved statistically, to be all the same components, probably labelled according to the supplier's marketing strategy, I'd guess. So the best, genuine new stock part substitutes for 2240/970 are undoubtedly KSC1845/KSA992 from Onsemi resellers. Don't even think of platform seller's junk.
As you are aware, SMD is fast becoming the only reliable way to obtain original quality substitute semis. Coming to grips with making or buying stock PCBs to suit them is a big problem for DIYs and I certainly understand the wish to stay with thru-hole tech. Unfortunately, the fakes issues just keep increasing with thru-hole parts purchasing.
@qwadrofonik
I would stick with 970/2240 because, obviously, the initial problem was poor heat transfer. The initial design as a whole has not provided enough air flow to remove the heat. Instead of telling the customers the ugly truth they suggested replacement of transistors.
As you obviously are rebuilding the amplifier you have the opportunity to provide more air flow to cool your transistors and use advantages of adequate transistors at the same time.
Right now I am making almost the same set of decisions in cloning project of Accuphase A60. I have approached some interesting conclusions, based on measurements exclusively. All what I wanted was to know how transistors perform instead of learning their rating on discussion platforms (echo-chambers).
Based on my measurements I think it is the right time for Marty McFly to go back to 1955 and tell Doc that Chinese transistors aren't that bad at all.
Look at the snapshots below (Uce-Ic). One is of KSC3503 (ON Semi) and the other is of SA1360 (ISC). Both snapshots have been made with the same settings and were arbitrarily chosen. The both are currently available close equivalents of your 2SC2705.
Which appears to be better? I have not disclosed which is which - you decide.
Good luck
I would stick with 970/2240 because, obviously, the initial problem was poor heat transfer. The initial design as a whole has not provided enough air flow to remove the heat. Instead of telling the customers the ugly truth they suggested replacement of transistors.
As you obviously are rebuilding the amplifier you have the opportunity to provide more air flow to cool your transistors and use advantages of adequate transistors at the same time.
Right now I am making almost the same set of decisions in cloning project of Accuphase A60. I have approached some interesting conclusions, based on measurements exclusively. All what I wanted was to know how transistors perform instead of learning their rating on discussion platforms (echo-chambers).
Based on my measurements I think it is the right time for Marty McFly to go back to 1955 and tell Doc that Chinese transistors aren't that bad at all.
Look at the snapshots below (Uce-Ic). One is of KSC3503 (ON Semi) and the other is of SA1360 (ISC). Both snapshots have been made with the same settings and were arbitrarily chosen. The both are currently available close equivalents of your 2SC2705.
Which appears to be better? I have not disclosed which is which - you decide.
Good luck
Attachments
Last edited:
+10000 points to Berlusconi. That is EXACTLY what I wanted to know. Thank you. I'm hoping the KSC3503 is the one on the left.
Also thank you to Ian Finch for your helpful insight.
Ironically I've already upgraded the VAS stage from A1145/C2705 and A1360/C3423 TO A1381/C3503 and Toshiba's new generation TTA004B/TTC004B respectively. I guess I have no use for the 1145/2705 at this time.
By any chance have you compared the Sanyo 2SC3503 to the ONSEMI KSC3503? I have a small inventory of both pairs but the older Sanyo ones are more closely hFE matched. I'm not skilled enough to current-compensate my pairs yet.
Also thank you to Ian Finch for your helpful insight.
Ironically I've already upgraded the VAS stage from A1145/C2705 and A1360/C3423 TO A1381/C3503 and Toshiba's new generation TTA004B/TTC004B respectively. I guess I have no use for the 1145/2705 at this time.
By any chance have you compared the Sanyo 2SC3503 to the ONSEMI KSC3503? I have a small inventory of both pairs but the older Sanyo ones are more closely hFE matched. I'm not skilled enough to current-compensate my pairs yet.
I recently bought 25 KSC3503, and the HFe was between 70 and 80. Quite close.
Still can't source the compliment though... 🙁
Same story with the new Toshiba tt series. Can't get the NPN anywhere, but oodles of PNP about.
With the relative ease in obtaining PNPs at this time, I almost started to think about redesigning to use PNP
Will 2SC2655 and 2SA1020 not fit?
Still can't source the compliment though... 🙁
Same story with the new Toshiba tt series. Can't get the NPN anywhere, but oodles of PNP about.
With the relative ease in obtaining PNPs at this time, I almost started to think about redesigning to use PNP
Will 2SC2655 and 2SA1020 not fit?
Last edited:
Hey Mondo I think this has been discussed in another thread but you will have a hard time matching the KSC3503. The most common stock is the KSC3503DS and KSA1381ESTU.
My guess is that they downrated their entire inventory of NPNs at some point because they couldn't truthfully market them in the same rank. They differ in hFE by as much as 10-20% between the complementary pairs. Still the conclusion is that they are fine to use together if some resistance is added to the collector of the PNP based on output current. I can't even hazard to guess what that value would be.
My guess is that they downrated their entire inventory of NPNs at some point because they couldn't truthfully market them in the same rank. They differ in hFE by as much as 10-20% between the complementary pairs. Still the conclusion is that they are fine to use together if some resistance is added to the collector of the PNP based on output current. I can't even hazard to guess what that value would be.
KSC1845/KSA992 is more than enough for the input stage. I used it in my Marantz based amplifier, even with +/-55V power supply.
They are available, reliable, and easy to match.
They are available, reliable, and easy to match.
Indeed I have heard many people extolling the virtues of the 992/1845. However if I'm substituting then I'd rather go with the Toshiba SMDs.
Lets be clear though. I have a working set of 970/2240s in the amp. They all measure over 200 but the hFE variance is enough to suggest that they're a little tired.
When new, the 970/2240 are as good or better than the 992/1845. Are the 2705/1145 still as good as the 970/2240? I think the answer here is yes but only in the VAS stage.
In fact all Korean manufactured transistors are a minor compromise. They represent the best On Semi can do in a Korean environment - and there is a reason why they are produced in Korea. On Semi has systematically bought out smaller Japanese transistor lines and moved operations to Korea. They're not alone. Toshiba's TTC series represent a design that can be made pretty decent in Korea, unlike their designs made for Hitachi machines. Thankfully newer fab process makes them consistent and reliable.
Maybe On Semi's new generation fab process has finally eclipsed Japan's meticulous attention to quality, and maybe China has come up from the rears. Who really knows?
What is my point? All of this is conjecture until you can measure real world sound reproduction quality like Berlusconi has. The results can be surprising.
Lets be clear though. I have a working set of 970/2240s in the amp. They all measure over 200 but the hFE variance is enough to suggest that they're a little tired.
When new, the 970/2240 are as good or better than the 992/1845. Are the 2705/1145 still as good as the 970/2240? I think the answer here is yes but only in the VAS stage.
In fact all Korean manufactured transistors are a minor compromise. They represent the best On Semi can do in a Korean environment - and there is a reason why they are produced in Korea. On Semi has systematically bought out smaller Japanese transistor lines and moved operations to Korea. They're not alone. Toshiba's TTC series represent a design that can be made pretty decent in Korea, unlike their designs made for Hitachi machines. Thankfully newer fab process makes them consistent and reliable.
Maybe On Semi's new generation fab process has finally eclipsed Japan's meticulous attention to quality, and maybe China has come up from the rears. Who really knows?
What is my point? All of this is conjecture until you can measure real world sound reproduction quality like Berlusconi has. The results can be surprising.
2240/970 are made for different purposes, than 2705/1145. High hfe is more important at the input stage, and not as much in the VAS stage.
Manufacturing in Korea is not compromise, I would say. I have 100-100pcs of KCS3503D/KSA1381E. They are remarkably consistent. All 3503 has the hfe between 93-95. 1381s arrived from 2 different batch, first has hfe between 111-113, and 2nd has 124-126.
I had same experience with other Onsemi devices.
Manufacturing in Korea is not compromise, I would say. I have 100-100pcs of KCS3503D/KSA1381E. They are remarkably consistent. All 3503 has the hfe between 93-95. 1381s arrived from 2 different batch, first has hfe between 111-113, and 2nd has 124-126.
I had same experience with other Onsemi devices.
I have to correct my earlier post, seeing the one above, and rechecking these on Semi ksc3503 with hfe meter.
I agree, hfe very consistent at 94 to 98.
Whether it can be matched to its counterpart, is another question I can't answer (I can't source any at the moment)
I agree, hfe very consistent at 94 to 98.
Whether it can be matched to its counterpart, is another question I can't answer (I can't source any at the moment)
From Farnell I was able do get KSC3503-DS (hFE about 80) and KSA1381ESTU (hFE about 120). Pretty useless.
However, at Reichelt I have purchased SC3423Y (hFE 175-180) and SA1360Y (hFE175-180). They've matched pretty good.
Now, let me first reveal that in my previous post the left snapshot corresponds to SC2705, on the right is plot of KSC3503. Which one do you prefer?
Now, let me present you snapshots of SC3423 (ISC, China) and KSC3503 (ON Semi).
I think quadrofonik should keep his original 970/2240 and purchase 1845/992 if and only if they were damaged.
Figure 1. KSC3503 (I have decided to show it with Y=50mA to enable direct comparisons with SC3423
Figure 2. KSC3503 (now with Y=25mA)
Figure 3. SC3423Y (with raw measurements shown along with XY presentation, compare it to its complementary device from my previous post, left image. They match just fine)
Attachments
Last edited: