I got an amplifier from a friend yesterday , it use a packed
complementary darlington STK050 output stage , including
the VBEM , out of curiosity i checked the schematic and
did a few sims whose results where expected since it s
what D Self call a blameless but with a variant in VAS
enhancement , results being about the same , amplifier
is JVC JAS-44 circa 1977 or so , schematic is below ,
the transistors are of course different set apart for the IPS.
complementary darlington STK050 output stage , including
the VBEM , out of curiosity i checked the schematic and
did a few sims whose results where expected since it s
what D Self call a blameless but with a variant in VAS
enhancement , results being about the same , amplifier
is JVC JAS-44 circa 1977 or so , schematic is below ,
the transistors are of course different set apart for the IPS.
Attachments
Q12 bootstrapping R2 has not the same radical effect as an emitter follower to feed the base of Q5 (prefered method) or a common base loading the collecter of Q5 (in a cascode scheme) to linearise the VAS.
So this JVC amp has not blameless performance.
An interesting feature is C8 in parallel with the emitter load of Q5, R5.
I wonder why a cap in this position is so rarely used.
So this JVC amp has not blameless performance.
An interesting feature is C8 in parallel with the emitter load of Q5, R5.
I wonder why a cap in this position is so rarely used.
Perfs are very good , not only THD but particularly IMD
is very low , at 1KHZ it is ppm level , IMD can be lowered
to 120dB when doing a 19 + 20khz test , OLG is 120dB....
Actualy the only caveat is the huge compensation cap
but that is due to the construction where the front
end is connzected to the OS using wires as long
as 30cm....
I ll do some sims using a double pair and a lighter compensation,
as well as a few passive components that were neglected at the time
but for sure it perform very well at first tries.
is very low , at 1KHZ it is ppm level , IMD can be lowered
to 120dB when doing a 19 + 20khz test , OLG is 120dB....
Actualy the only caveat is the huge compensation cap
but that is due to the construction where the front
end is connzected to the OS using wires as long
as 30cm....
I ll do some sims using a double pair and a lighter compensation,
as well as a few passive components that were neglected at the time
but for sure it perform very well at first tries.
A note on purchasing here, Nige. I bought some recently and discovered to my dismay that it's extremely hard to buy both the NPN and the PNP versions from the same manufacturer. No idea why.
On MPSA42/92. You might consider paralleling them, 2 or 3 in place of just one. If you get Motorola/ON Semi's originals, pairing them should be no problem from a population as small as 30 or 40 pcs, although they are dirt cheap and buying 100+100 shouldn't financially burden anyone.
You may be interested to learn that MPS and MPSA trannies, 118 of them, consititute some of Krell's front ends, several stages in parallel threes, some in also parallel cascode connection. Marantz also liked paralleled VAS trannies. Using just 2 you can do as is, but if using 3 or more, you would need to add 10 Ohm resistors for improved matching. Not a must, but very handy indeed.
As an FYI/BTW --- Paralleling devices for VAS works very well in lowering distortion.... I used in the late 1970's and early 1980's (published in TAA).
Thx-RNMarsh
Thx-RNMarsh
Yes, indeed it does.
Although I'm not surprised. It allows one to use low power transistors at lower currents, but such devices typically have lower distortion and higher Ft than more powerful ones. Once you parallel, a normally weedy device, oridinarily safe to use at merely 2.5 mA or so in that spot suddenly delivers three times that when using parallelling, and with no sacrifice in quality of operation.
The only caveat is that if one wants to play it safe, one should carefully match those parallel devices, or as best as one can.
The almost ideal device that comes to mind here are Motorola/ON Semi's MPSA 42/92. At 300 V, they are hard to misuse, they can dissipate more than is usual for TO-92 package, although I would cauton agains their use at more than 200 mW without snap-on heat sinks. I'd love to see them in another version with their Ft improved from 60 MHz to something over 100. Just for the hell of it.
Here I thought I was the only person who did this. You can parallel a bunch of SOT23, SOT223, or other SMD package transistors in a small space and end up with an impressive aggregate device.
I will be buying KSP 44/94 also before it is too late . I don't find MPSA 44/94 now .
I parallel transistors a lot in my work . It often means we can avoid a heat sink . As you say a big free lunch often as the small devices are better . For example a power transistor of 400 V might have a gain of 12 . 3 x MPSA 44 might have a gain of 50/3 = 17 ish . The power transistor Ft 4 Mhz ( on a good day ) , MPSA ( KSP ) 50 MHz . Even at 200 V average we might get 3 to do a VAS . Put a cascode at the bottom and it's looking good . That's about 330 R per device if CCS . Make it 4 and 430R and it is looking great . If my recipe that is 8 x KSP44 , 8 x KSP94 , 2 x BC550C , +/- 160 V and double VAS ( with HF adjustment like Hitachi ) . Tell me that isn't linear and will not move a mountain . Only problem is output devices . It would have to be All NPN in the style of Quad or Bryston . That is get the MPSA94/44 to do the first bit . My boss has asked me to build a very high voltage amp . I think a Bryston clone is my only option . It will be Quad really . I told a lie before . For want of one KSP 44 I wouldn't make it 2 transistors one side and 3 the other . For the application it would make no difference . You might ask yourself how I ever get anything done and ramble on at DIYa ? I try to multitask and in a written sense fail . Dvv the people in Somerset certainly taught you English very well , my school failed me and I am a native .
Or you can do what Nelson Pass did in the early 90ies, which is to use series/parallel cascode stages, with 2 pass and 2 gain transistors, incidentally, MPSA 42/92.
INTERMEDIATE LIN
Ben Duncan calls the amplifiers often refereed to here as Intermediate Lin . Reference , High Performance Audio Power Amplifiers page 103 Fig 4.24 . Also " Slew limiting in the Lin Topology " page 186 fig 5.10 .
Ben is well read and respected . He might have caused the idea that it is a H C Lin circuit ? My hunch is , it is H C Lin if anyone . Some of the valve amps were not so different , Radford for example . It is special in that it's differential input is pentode and triode in something like LTP ( it has a short tail ) . I recently found out how brilliant that is . It is a TV valve like 7199 . My brother always talked of extending the tail in similar amps so as to get proper symmetry . Quad I think uses 100 K and 91 K to get it about right . Quad valve amp to me is an op amp .
Ben Duncan calls the amplifiers often refereed to here as Intermediate Lin . Reference , High Performance Audio Power Amplifiers page 103 Fig 4.24 . Also " Slew limiting in the Lin Topology " page 186 fig 5.10 .
Ben is well read and respected . He might have caused the idea that it is a H C Lin circuit ? My hunch is , it is H C Lin if anyone . Some of the valve amps were not so different , Radford for example . It is special in that it's differential input is pentode and triode in something like LTP ( it has a short tail ) . I recently found out how brilliant that is . It is a TV valve like 7199 . My brother always talked of extending the tail in similar amps so as to get proper symmetry . Quad I think uses 100 K and 91 K to get it about right . Quad valve amp to me is an op amp .
Dvv and I like BF720/721 .
BC847BS ( dual ) also and BCV61/62 . MPSA 92/42 are in SMD .
No , it's how long from now can we get them . The list is getting shorter every day if lets say TO 92 .
Dvv . Shall we start a thread of nice transistors ? Preferably from Farnell or RS where possible or other easy to deal with suppliers .
I would love it is someone told me of a TV transistor with remarkable Rbb for example . I am told that Philips had one for TV's . They made it as a service spare at a subsidized price . Fed up with audio people buying them they demanded a serial number of the TV . Not sure it's true , possible .
Dvv . Shall we start a thread of nice transistors ? Preferably from Farnell or RS where possible or other easy to deal with suppliers .
I would love it is someone told me of a TV transistor with remarkable Rbb for example . I am told that Philips had one for TV's . They made it as a service spare at a subsidized price . Fed up with audio people buying them they demanded a serial number of the TV . Not sure it's true , possible .
Certainly, why not, unless there's such a topic already?
Although, I'm not really sure how useful can I be, given that I do not have access to that many transistors, and that I already have my own list of preferred devices. My own local market is exceptionally poor with Japanese devices.
As for MPSA 42/92 trannies, I believe we'll be seeing them for quite a while yet. Motorola is not quick to delete products, although they stabbed me in the back when they stopped making BD 529/530 and the ultra, hyper low noise NPN BC 651. Both were outstanding products, truly outstanding.
Last edited:
This sounds like a very good idea. Maybe one of the mods would be so kind as to make it sticky since this is a big issue facing builders as we move forward. The KSA/KSC VAS devices everyone likes to use are now only available in lower grades, TO-92 choices getting smaller etc. Tough times ahead for bip leaded devices and even worse for JFET's
Heres a proposal for how to structure the thread
Circuit
function Device Comments
____________________________________________________________________
LTP BC547/557 45V high hFE, high Ft SS TO-92; needs cascode
LTP 2N5501/5401 150V med hFe high Ft TO-92; good for cascode also
VAS KSC1301 etc
Heres a proposal for how to structure the thread
Circuit
function Device Comments
____________________________________________________________________
LTP BC547/557 45V high hFE, high Ft SS TO-92; needs cascode
LTP 2N5501/5401 150V med hFe high Ft TO-92; good for cascode also
VAS KSC1301 etc
I think the LTP to VAS connection is the critical one . Advertising men call it slew rate in my opinion . I call it stage matching . The much derided Rod Elliot says the increase in LTP current is almost always lost in the Cdom . He concedes that one does win in the end . I say make the VAS as high gain as possible . A cascode seems ideal . Dvv told me sometime ago MPSA 42 was considered highly linear . Seeing as we wrap the amp in feedback this seldom seems important . One could say with current gain of about 50 it should be ( TV transistor I assume for gun drive ) . Adding a BC 550 C as cascode , did anyone ever analyze this ? What surprises me is people speak as if the loosing of a volt or two of drive is a big crime . It can be won back for pennies . BTW , if you deliberately have less drive voltage the clipping is nicer ( - 5 V VAS compared with outputs ) . Have a switch for maximum power also ( + 5 V ) .