In a class AB, How much output power can I get from 2 pair of these into 4 ohm? My simulations show about 4A and 18w per device dissipation for 125W output. Simulations was for TIP122/127. I don't have 142T/147T models.
https://www.mouser.com/datasheet/2/308/1/TIP142T_D-2320223.pdf
https://www.mouser.com/datasheet/2/308/1/TIP142T_D-2320223.pdf
TIP142/147
My advice is that these are the worst devices to use whether in To220 or TO218 package.
They are themally unstable unless you take special precautions.
Sounstream used these darlingtons exclusively with massive failure rates and eventually found out that one must use 1 ohm emitter resistors bypassed with Schottky diodes to guarantee thermal stability.
Why one would choose such a device for an amplifier is quite frankly beyond my understanding.
They are high THD devices, slow as dogs and again thermally unstable.
Zed Audio
My advice is that these are the worst devices to use whether in To220 or TO218 package.
They are themally unstable unless you take special precautions.
Sounstream used these darlingtons exclusively with massive failure rates and eventually found out that one must use 1 ohm emitter resistors bypassed with Schottky diodes to guarantee thermal stability.
Why one would choose such a device for an amplifier is quite frankly beyond my understanding.
They are high THD devices, slow as dogs and again thermally unstable.
Zed Audio
My output stage was originally 9540/540 but I'm having issues with thermal stability. I've been unable to determine the cause other then it being related to thermal tolerances because 2 out of 3 prototypes work. #3 powers up with no bias unless I warm the CM with a finger then it'll pop on. So I'm below the threshold of something.
So I thought I'd try darlingtons since my chassis limits me to 4x TO220 packages in the OPS.
So I thought I'd try darlingtons since my chassis limits me to 4x TO220 packages in the OPS.
These devices were used in the Allberry amplifiers a pair delivering about 40 watts into 8 ohms !These amplifiers were critically very well rated
I Personally dont like darlingtons but having said that they can be made to work very well
Trev
I Personally dont like darlingtons but having said that they can be made to work very well
Trev
Also their SOA doesn't allow much supply voltage, perhaps +/-25V to +/-30V would be prudent into reactive loads so 125W/4R would be pushing it.
Can´t talk about the TO220 versions and in any case you should NOT use much smaller TIP12* models,completely different, but I use the TO218/247 versions since forever, as in the 80's in the rounded top "TO3P" case, > TO218 > TO247, in the far more stressful Guitar and Bass amplifier market.
Switched to them between 2004 an 2006 when it became impossible to get non Fake 2N3055 (which had been my workhorse since the late 60´s, go figure) must have delivered some 4000-5000 amplifiers since, using them, with excellent track record, minimal Warranty problems, and in fact they are THE most used power transistor pair in the MI industry, the almost exclusive choice in Fender - Laney - Hughes & Kettner - Crate - many Marshall - Ampeg, and my own 😉
Always with +/- 40-42V rails and putting 60-80-100-120-150W RMS into 4 ohm using 1-2-3 pairs respectively.
Surprised about the repeated failures mentioned in comparatively mild (by comparison) Home Hi Fi use.
Again, not experience with the TO220 versions but in theory chip is exact same, so Hfe, voltage - current - SOAR specs "should" be about same, only stone in the shoe being small TO220 case and its much reduced case contact surface which heaviky impacts thermal trasfer.
But 100W into 4 ohm with +/-40V rails should be doable with 2 pairs per channel.
I´m not talking simulation but real World manufacturing experience.
Switched to them between 2004 an 2006 when it became impossible to get non Fake 2N3055 (which had been my workhorse since the late 60´s, go figure) must have delivered some 4000-5000 amplifiers since, using them, with excellent track record, minimal Warranty problems, and in fact they are THE most used power transistor pair in the MI industry, the almost exclusive choice in Fender - Laney - Hughes & Kettner - Crate - many Marshall - Ampeg, and my own 😉
Always with +/- 40-42V rails and putting 60-80-100-120-150W RMS into 4 ohm using 1-2-3 pairs respectively.
Surprised about the repeated failures mentioned in comparatively mild (by comparison) Home Hi Fi use.
Again, not experience with the TO220 versions but in theory chip is exact same, so Hfe, voltage - current - SOAR specs "should" be about same, only stone in the shoe being small TO220 case and its much reduced case contact surface which heaviky impacts thermal trasfer.
But 100W into 4 ohm with +/-40V rails should be doable with 2 pairs per channel.
I´m not talking simulation but real World manufacturing experience.
I'm stuck with TO220 packages in the chassis I have. I sort of painted myself into a corner in that respect. I might have to go back to the 540/9540 and solve the bias issue.
Yes, that´s another Darlington option.
But you should be able to solve termal stability without much problem.
A TO92 transistor (say BC546/556 or similar) used as a Vbe multiplier will have a factor of about 10-12 to 1 , because turn-on Vgs is about 3.5V each for those MosFets, so typical Vbe diode drop of 2mV/degC will be multiplied by same amount, which is ample compensation for them.
I use that in my own 300/600W Musical Instrument MosFet power amps built around IRFP250 or IRFP240/9240 and never have a problem.
Remember MI amps are used LOUD, in hot sweaty stages, under hot stage lights, for hours on end, a real endurance test.
I suggest you try around 10k:1k or 8k2:1k dividers, with 1 k made out of a 470 ohm resistor+470-500 ohm trimmer.
I Epoxy my TO92 transistors to heat sink but for better transfer you may drill a short 5mm hole , fill it with thermal grease and bury sensor transistor head there; you should NOT have any problem.
Try it, before comitting to final design breadboard such a power stage using the real heat sinks you will use later and try it.
Hope you are using source resistors, 0.22 or 0.33 ohm are classic values.
Again: physically build and test it, simulation is a wonderful tool but is not the Penultimate Truth as P K Dick would say ;o , "testing rules".
But you should be able to solve termal stability without much problem.
A TO92 transistor (say BC546/556 or similar) used as a Vbe multiplier will have a factor of about 10-12 to 1 , because turn-on Vgs is about 3.5V each for those MosFets, so typical Vbe diode drop of 2mV/degC will be multiplied by same amount, which is ample compensation for them.
I use that in my own 300/600W Musical Instrument MosFet power amps built around IRFP250 or IRFP240/9240 and never have a problem.
Remember MI amps are used LOUD, in hot sweaty stages, under hot stage lights, for hours on end, a real endurance test.
I suggest you try around 10k:1k or 8k2:1k dividers, with 1 k made out of a 470 ohm resistor+470-500 ohm trimmer.
I Epoxy my TO92 transistors to heat sink but for better transfer you may drill a short 5mm hole , fill it with thermal grease and bury sensor transistor head there; you should NOT have any problem.
Try it, before comitting to final design breadboard such a power stage using the real heat sinks you will use later and try it.
Hope you are using source resistors, 0.22 or 0.33 ohm are classic values.
Again: physically build and test it, simulation is a wonderful tool but is not the Penultimate Truth as P K Dick would say ;o , "testing rules".
I built two prototypes that worked fine. They've been in use for a year. I transferred the design to a new PCB intended to fit in a chassis I designed. This 3rd one powers up in class B. Sounds fine. DC offset is 1.5~2mV. I use 0.22r source resistors.
What dividers are you speaking of?
EDIT - Oh you must mean the Vbe multiplier divider. I can verify that in sim to get me in the ballpark. I calculate the total series value for about 1/10th the Vbe current then adjust the ratio to set bias.
What dividers are you speaking of?
EDIT - Oh you must mean the Vbe multiplier divider. I can verify that in sim to get me in the ballpark. I calculate the total series value for about 1/10th the Vbe current then adjust the ratio to set bias.
Last edited:
You should be proud 🙂 , best 2N3055 I ever used (besides mythical RCA I used in late 60s early 70s) were India made 2N3055H , labelled UR plus a diode symbol or USHA.
Myth says they were built by an Indian Military Fatory or specifically for Indian Armed Forces, truth is they were incredibly robust.
I bought a 3000 unit carton and used them for years, they stood everything (Musicians are brutal users), when I ran out of them started using ST, straight from ST Malaysia through official Importer so not Fake, but these started popping like fireworks or almost.
Importer confirmed they were not fake (showed me all the paperwork) BUT
1) these were Epitaxial instead of USA Hometaxial
2) in any case, since modern cutthroat competition is based on price, not quality, ST ones (and all other "serious" ones strictly met specs ... and nothing beyon, while USHA were well beyond published ratings, way overbuilt.
Oh well.
Myth says they were built by an Indian Military Fatory or specifically for Indian Armed Forces, truth is they were incredibly robust.
I bought a 3000 unit carton and used them for years, they stood everything (Musicians are brutal users), when I ran out of them started using ST, straight from ST Malaysia through official Importer so not Fake, but these started popping like fireworks or almost.
Importer confirmed they were not fake (showed me all the paperwork) BUT
1) these were Epitaxial instead of USA Hometaxial
2) in any case, since modern cutthroat competition is based on price, not quality, ST ones (and all other "serious" ones strictly met specs ... and nothing beyon, while USHA were well beyond published ratings, way overbuilt.
Oh well.
Bharat Electronics, or BE, (BEL on old components) is a Defense ministry factory, make transistors as a side business, really think tank sights, military radios, radars, ship suites...
One of the few such companies capable of making most of their own components. Even radar tubes.
Keltron has a license from Sprague for capacitors, again good quality. Also transistors well regarded.
CDIL has been discussed here, old school quality.
Usha Rectifier, UR, was OE to Fairchild, Marconi and Sumitomo, but now merged with Usha India, more or less defunct.
Fascinating how you got to try them, how they found their way to Argentina, and proved to be better than well known brands...
One of the few such companies capable of making most of their own components. Even radar tubes.
Keltron has a license from Sprague for capacitors, again good quality. Also transistors well regarded.
CDIL has been discussed here, old school quality.
Usha Rectifier, UR, was OE to Fairchild, Marconi and Sumitomo, but now merged with Usha India, more or less defunct.
Fascinating how you got to try them, how they found their way to Argentina, and proved to be better than well known brands...
Last edited:
BEL | Products
2N3055HV, 2N3055 MA, and 2N3773.
Also 741, 747 and 723 in round cans.
Military grade stuff.
2N3055HV, 2N3055 MA, and 2N3773.
Also 741, 747 and 723 in round cans.
Military grade stuff.
These spice models can be found on ON Semiconductor's webpage:
https://www.onsemi.com/pub/collateral/tip142.lib
https://www.onsemi.com/pub/collateral/tip147.lib
I don't understand why there should be thermal stability issues with Darlington BJT. Normally BJT behave thermally much more stable than power MOSFETs. I have often used TIP Darlingtons in output stages without any problems, even in common emitter amplifiers, even without emitter resistors, which tend to be less thermally stable. Thermal instability is mostly an issue of thermal management. People often operate them on heat sinks with unsufficient Rth and too much quiescent current.
TIP122 / 127 142T / 147T - They are slow (3 mhz). Hfe varies greatly depending on the current (hundreds of times). First, it is very low (from 10 ma), a maximum of 1-2a, then a decline.
Last edited:
@OldDIY 3MHz is fast enough for 20 Vrms @ 20 kHz. hfe vs collector current and vs temperature (more dominant) doesn't matter in a closed feedback loop if you have enough control current.
More critical is Vbe vs temperature, especially for common emitter amplifiers, but this is much more critical for MOSFETs with respect to their Vth. In eighter case you are safe if you keep the quiescent current low, work with emitter / source resistors and a sufficient heat sink. Mostly you can even avoid thermal compensation.
To keep the quiecent current under control you can also use the LT1166 from Analog Devices for emitter / source follower amplifiers. This device is designed for power MOSFETs but you can also use it for Darlington transistors if you spread the base voltage difference, seen by the control loop by use of a few silicon diodes.
More critical is Vbe vs temperature, especially for common emitter amplifiers, but this is much more critical for MOSFETs with respect to their Vth. In eighter case you are safe if you keep the quiescent current low, work with emitter / source resistors and a sufficient heat sink. Mostly you can even avoid thermal compensation.
To keep the quiecent current under control you can also use the LT1166 from Analog Devices for emitter / source follower amplifiers. This device is designed for power MOSFETs but you can also use it for Darlington transistors if you spread the base voltage difference, seen by the control loop by use of a few silicon diodes.
Last edited:
These darlingtons are no different than any other epitaxial base output transistors. They don’t have a high fT driver that you would typically use in hi-fi application, but that’s not really a problem for an instrument amp. The beta fall off at high current is the same as that if a 3055. The reason you see it so low at low current is because of the internal base-emitter resistors (they steal base current). The effective beta of discretes does the same thing when the resistors are added. The base-emitter resistors are normally connected differently, from base to output in most Hi-if gear, and this is a limitation of darlingtons. The crossover distortion will always be worse as it is physically impossible to get the optimum bias in both the output and driver simultaneously. The solution is usually to run the bias too high, and hence the “thermal stability” issues. Use a bigger heat sink and that problem magically disappears. Or turn the bias down and put up with 0.2% distortion. Instrument amps will normally size the heat sink for the full maximum possible dissipation (vcc^2/(10*RL)) per device so naturally it will be thermally stable with a proper vbe multiplier.
They do handle full power at 40 volts, same as 3055. The only epi-base devices that beat that are only available in TO-3 and cost a bundle these days. These devices are limited to 40-ish volt supplies at 8 ohms per pair. Four ohms takes two, two ohms takes three. Don’t believe it when someone says one pair will run 4 ohms (unless you drop the voltage).