Working on rebuilding an NVA AP50 amplifier from scratch, I have a question about the output transistors. NVA is known to have used BDV64/65 and TIP142/147 darlington output transistors in these amps, driven by TIP31/32 transistors. Mine actually has a mix of these BDV / TIP output transistors. Whatever parts one has laying around I guess...
Would a pair of Sanken 2SB1560 / 2SD2390 darlingtons be drop-in replacements? This here thread from 2013 suggests they may be. I looked for spice models but couldn't find any. It looks like the maximum collector current of the BDV is 15A vs the TIP's 10A and the Sankens are 10A also. The Sanken's gain looks to be considerably higher. Internal construction too. The Sanken omits an internal resistor (and a protection diode?).
Here is the schematic:
Would a pair of Sanken 2SB1560 / 2SD2390 darlingtons be drop-in replacements? This here thread from 2013 suggests they may be. I looked for spice models but couldn't find any. It looks like the maximum collector current of the BDV is 15A vs the TIP's 10A and the Sankens are 10A also. The Sanken's gain looks to be considerably higher. Internal construction too. The Sanken omits an internal resistor (and a protection diode?).
Here is the schematic:
Depending on which manufacterers datasheet you look at, they will either show none, 1 or 2 resistors inside. These BDV's and TIP's can still be had, for instance Reichelt (DE) still lists them. It is my feeling though, that with these old parts - which were manufacturer specific when they were new - what you can buy now from ONSemi or ISCsemi, is something more generic that will meet the minimum spec's of the original part.
This would require changing the TIP31A/32A divers as they are too slow for the Sanken and replace them
with a pair of BD139/140, actualy those TIP are inadequate even for the BDV64/65.
Beware that the Sanken have vastly higher gain than the BDV and a biais control using a pot is necessary before
doing any change, something that allow to start with a low enough bias current.
with a pair of BD139/140, actualy those TIP are inadequate even for the BDV64/65.
Beware that the Sanken have vastly higher gain than the BDV and a biais control using a pot is necessary before
doing any change, something that allow to start with a low enough bias current.
The TIP drivers would be the first thing to go. ANY time you see them. EVERY time you see them.
MJH11017/8 or MJH6284/7 are the current comparable output types. TIP142/7 are among the more rugged types, but they just don’t have any gain left at 10A. Also, do not mix brands or families - The internal Rbe values have a nasty habit of being different, even between old and new vintage of the same manufacturer. Pick one and stay with it.
MJH11017/8 or MJH6284/7 are the current comparable output types. TIP142/7 are among the more rugged types, but they just don’t have any gain left at 10A. Also, do not mix brands or families - The internal Rbe values have a nasty habit of being different, even between old and new vintage of the same manufacturer. Pick one and stay with it.
Pretty elaborate frontend to lead up to tip's and Darlington's.
Not sure if scratch means repairing a factory unit or new design with diy board.
Diy design the whole output section could be changed to discrete Darlington's.
Repair, compensation should be tested, since higher speed outputs would require changes
to amplifier compensation in the fronted.
Additional comp was added to the Darlington's as well C9 and C11
high frequency performance and slew rate, turn on time be pretty close to horrible.
Such elaborate frontend for such horrible output power section
No sanken models for darlington, test in sim with discrete high Ft transistors
and absolutely remove tip's. Aside from the fact 30 to 40 watt low gain transistors are not
needed to drive Darlington's.
Not sure if scratch means repairing a factory unit or new design with diy board.
Diy design the whole output section could be changed to discrete Darlington's.
Repair, compensation should be tested, since higher speed outputs would require changes
to amplifier compensation in the fronted.
Additional comp was added to the Darlington's as well C9 and C11
high frequency performance and slew rate, turn on time be pretty close to horrible.
Such elaborate frontend for such horrible output power section
No sanken models for darlington, test in sim with discrete high Ft transistors
and absolutely remove tip's. Aside from the fact 30 to 40 watt low gain transistors are not
needed to drive Darlington's.
BDW93C/94C are quite better than BDV64/65 and TIP142/147 but they are in a TO220 package that doesnt
allow for a lot of dissipation despite their 80W rating.
FI their junction/case thermal resistance is 1.86°C/Watt, they are good for amps whose suply voltage is +-24V,
moreover on 4R where it should be +-20V at most, those voltages being in idle conditions.
allow for a lot of dissipation despite their 80W rating.
FI their junction/case thermal resistance is 1.86°C/Watt, they are good for amps whose suply voltage is +-24V,
moreover on 4R where it should be +-20V at most, those voltages being in idle conditions.
ST is only making those in TO-220 nowadays. They would work SOA-wise, but a bigger package is advised to get the heat out. Comparable to TIP142/7 otherwise, gain is pretty low at 10A. Using a driver with them is the only thing that saves the low beta issue. Linearity is still going to be better with a 15 or 20 amp device.
BDW93/94 They are TO-220 packages so power output is limited.
So +/- 38 volt rails shown not advised
Unless your real world duty cycle is 20% power at 8 ohms.
Aside being slow with poor gain driving real world loads.
MJH6284/7 are only " slower" devices worth even considering.
Aside from mentioned high speed Sanken.
93/94C and 142/147 are Barely .1 to .4 MHz
Are poorly matched pairs and most the brute force compensation seen like
C9 / C11 over the power device. Is typical crutch to fix that the PNP or NPN
will be considerable slower and gains dont match well.
When 6284/7 get up to 2 MHz and pnp and npn actually somewhat matching is tolerable
compared to 142/147 which is non tolerable.
Would have to skim through datasheet or even torture myself in sim.
Far as I remember the PNP is basically garbage.
And the advice to match specific brands somewhat makes sense.
But regardless the devices are not well matched from the start.
So +/- 38 volt rails shown not advised
Unless your real world duty cycle is 20% power at 8 ohms.
Aside being slow with poor gain driving real world loads.
MJH6284/7 are only " slower" devices worth even considering.
Aside from mentioned high speed Sanken.
93/94C and 142/147 are Barely .1 to .4 MHz
Are poorly matched pairs and most the brute force compensation seen like
C9 / C11 over the power device. Is typical crutch to fix that the PNP or NPN
will be considerable slower and gains dont match well.
When 6284/7 get up to 2 MHz and pnp and npn actually somewhat matching is tolerable
compared to 142/147 which is non tolerable.
Would have to skim through datasheet or even torture myself in sim.
Far as I remember the PNP is basically garbage.
And the advice to match specific brands somewhat makes sense.
But regardless the devices are not well matched from the start.
At the time ST had some TO3 variants, sort of, with the BDX87C/88C, these shouldnt be confused witr the TIP and BDV
as they had 20MHz Ft and the gain was, like the also 20MHz BDW, much better and well above the minimal 750 rating, their max
gain was specified as 18-20k, comparable to some Sanken devices.
All OnSemi “complementary” darlingtons are on the 3 MHz process. TIPs of all varieties have always been generic - pretty much the bottom of the barrel. Poor matching is par for the course. I has a circuit using the 142/7 that should have worked oscillate like crazy and nothing would tame it. Put in Sankens and it just quieted right up and didn’t give a moments trouble. Unfortunately, the Sanken B1560 and it’s NPN are not generally available anymore. MOQ of like 1000 each with a year lead time. Those 6284/7 used to be TO-3’s, and available from several sources. ON recently started putting them in TO-247, and they discontinued the TO-3 version.
ST isn’t making ANY TO-3 anymore. They also had a complementary epitaxial planar power process ON didn’t have (until the Fairchild merger, and then they got it in spades). Motorolas original fast darlingtons we’re only NPNs.
Nobody will get 10 amps from a TO3P and especially a T0-220
at 50 to 75c
Temperature derated dont expect more than 1 or 2 amps.
With 80 volts try 1 amp.
Gain is poor. Driving 4 ohm loads distortion would be very high.
And a single device would sag and be far far far from the rail voltage
when starting to clip.
Minimum would be 2 output devices per rail with 93/94
4 ohms is more like 4 devices.
Even with 142/147 or 6284/7
Soa is ok for real world with single pair. Even then driving 4 ohm loads
they sag. 2 pairs wakes thing up or improves needed gain for low impedance.
at 50 to 75c
Temperature derated dont expect more than 1 or 2 amps.
With 80 volts try 1 amp.
Gain is poor. Driving 4 ohm loads distortion would be very high.
And a single device would sag and be far far far from the rail voltage
when starting to clip.
Minimum would be 2 output devices per rail with 93/94
4 ohms is more like 4 devices.
Even with 142/147 or 6284/7
Soa is ok for real world with single pair. Even then driving 4 ohm loads
they sag. 2 pairs wakes thing up or improves needed gain for low impedance.
Indeed. I have looked around and the old parts can be found at least somewhat reliably from some NOS parts sellers. My question was to make an inventory of my options, including these Sanken transistors.
So definitely not a drop in replacement, thanks.
That makes sense and I was wondering about this before you offered this. Thank you.
Will take a look at these, thanks. May stick with the old Philips (BDV) transistors at the output then.
Absolutely. Had no intention of keeping it like that, all mixed up.
New design with DIY board. I like this idea to use discrete darlingtons. Transistors, even as they will be of the same type, do not have to be matched to form a darlington do they? What could be suitable transistors for modeling purposes?
There s eventualy the BDW83C/84C also from ST and in a TOP3, they dont specify the Ft on their datasheet
but they are eventualy plastic case variants of the BDX87C/88C and are also specified as minimal 750 gain and 18k max.
That being said anything targeting very good perfs should go Sanken, as far as we re talking darlingtons, and if they can
be found as a lot of very such good device are obsolete.
Can check the data sheets for turn on time.
No mystery here. If Ft isnt listed it is = garbage.
Only modern Darlington with listed Ft is Sanken and MJH6284/7
83/84 and 93/94 if you doubt the gain and turn on time.
Run them into 4 ohms loads at high frequency.
Then your answers will be incredible visible.
Not trying to argue or shun peoples suggestions.
These are poor transistors for low impedance/ Audio Frequency applications.
They belong in ancient vending machines.
No mystery here. If Ft isnt listed it is = garbage.
Only modern Darlington with listed Ft is Sanken and MJH6284/7
83/84 and 93/94 if you doubt the gain and turn on time.
Run them into 4 ohms loads at high frequency.
Then your answers will be incredible visible.
Not trying to argue or shun peoples suggestions.
These are poor transistors for low impedance/ Audio Frequency applications.
They belong in ancient vending machines.
Agree for all those old devices excepted BDW93/94, for low powers like 15W they are more than good enough,
at 3A typical gain is 5k for both N and P, Ton are 0.4-0.2us, Tf are 2-1.5us and Ft is 20Mhz, wich say that
their internal drivers are not as slow as the ones of the TIPs and BDVs.
at 3A typical gain is 5k for both N and P, Ton are 0.4-0.2us, Tf are 2-1.5us and Ft is 20Mhz, wich say that
their internal drivers are not as slow as the ones of the TIPs and BDVs.
For 15 miserable watts I’d rather have a discrete pair of run of the mill 10A 3 MHz epi-base non darlingtons driven by MPSA06/56. All generic, can find them anywhere. TIP142/7’s are for 50 watt guitar amps.
Classic T0-126 driver transistor in lower voltage range would be BD139/40
Higher Voltage some have used KSC3503 and KSA1381 which will soon be discontinued.
Audio Applications will likely resort to Toshiba Current production TTC004B and TTA004B
Wide range of power transistors Assuming Toshiba 2SC/2SA or TTC TTA 5200/1943
I tend to use equivalent ON -SEMI models for 5200/1943
or 3281 / 1302 or MJL 3281 / 1302
Most modern search engines for parts likely need a G suffix added.
All power devices likely limited to T03P packages use to get lucky some
still made larger 263 or 264 packages. Been awhile think ON-SEMI will list 5200/1943 variants
Believe not much different than FJL4315 and FJA4313
High Speed T0-220 most use MJ15034 / 35
and numerous variables for higher voltage.
So the last 2 numbers go up and down for voltage tolerance.
Basically find models directly at ON-SEMI website.
Or often Mouser might include links that work to models.
I go straight to ON- SEMI website
Last edited:
Say up to 20W/8R as they are good up to roughly 3A, i prefer these to any TIP, darlingtons or EF2.
For guitar amps i used only BDV64/65, actualy i dont remember anything good from any TIP apart as switches and also AM modulator for some CB and amateurs radio transceiver, although Fender used them in some cheap guitar amp along with
a slow high voltage MC1436 opamp, see schematic below as well as the limited BDW datasheet but still with some curves.
Attachments
Last edited: