TIP141 internal resistors and diode.

[mac]

I have a power amp on the beadboard.
I'm using a darlington BC550C + TIP41C so as to avoid loading the bias network because it runs at 250ma, TIP hfe: 60.
Works and sounds fine.

I was looking for power darlingtons like TIP141 to simplify things a bit.
I noticed that this device has a couple of internal resistors across the emitters, 8k and 40, and a reverse diode from C to E.

I wonder what's the purpose of those parts, and if it's better to stick to the home-made darlington.

mac

[antonis]

Resistors serve for transistors working at sufficient current for an acceptable high beta (current gain)..
Diode, as it's wired, serves as Flyback Diode (preventing inductive spikes passing through transistor)..

Almost all power Darlingtons are internally equipped with those items..

P.S.
In case of discrete Darligton configuration, you will need those resistors..
You won't need that diode in case of not driving inductive loads but it won't make any harm..

[mac]

You won't need that diode in case of not driving inductive loads but it won't make any harm..

Hand wired 4:1 OT @ 36v

I'll add a 4007 and play with those resistors.

mac

[antonis]

IMHO, proceed with TIP141 and place 1N4007 diode across OT driven winding..

[Rob Strand]

I was looking for power darlingtons like TIP141 to simplify things a bit.
I noticed that this device has a couple of internal resistors across the emitters, 8k and 40, and a reverse diode from C to E.

I wonder what's the purpose of those parts, and if it's better to stick to the home-made darlington.

Without the low valued resistor across the BE junction of the second transistor (power transistor) the Darlington would be extremely slow and it would limit the application of the device.   Darlingtons are fairly slow already.   You pretty much need it for a power amp, even discrete implementations require BE resistors on the power device.

The resistor also means the second transistor won't turn on until a minimum amount of current flows through the first transistor, IC1 = VBE/RBE2 ~ 10mA to 20mA, which is fine for a power amp.    This is the main cause of the non-flat gain vs IC you see on Darlingtons.  The gain starts at hFE1 and rises to approx hFE1*hFE2.

The resistor across the BE junction of the first transistor is an annoyance for power amplifier design. You just have to live with it.  The value of the resistor varies across different devices.   It is convenient when the Darlington is used for a switch as it helps speed-up the turn-off time without adding your own resistor.

The resistors come about due to the structure of the monotlithic Darlington.   Here's some discussion from part of an old RCA applications note:

[mac]

With all my respects to the guys @ RCA, with those resistors it sounds horrible 
Only when the second BE resistor is about 470 ohm it starts to sound ok.
I kept the reversed diode across the darlington.

BTW, I downloaded the RCA textbook for further reading. thx Rob.

mac

[Rob Strand]

With all my respects to the guys @ RCA, with those resistors it sounds horrible 
Only when the second BE resistor is about 470 ohm it starts to sound ok.
I kept the reversed diode across the darlington.

There's plenty of scenarious but if the bias current for the output stage is too low only the driver stage is on and the main transistors are cut-off, you can get a form of crossover distortion.  The fix is to adjust the VBE multiplier on the amp and raise the bias current.

When you increase the BE resistor on the second transistor you lower the current in the driver necessary for the main transistor to turn on.  So if the VBE multiplier is set to a fixed voltage, adjusting the second BE resistor is the only way to get the output stage transistors to turn on.

Alternatively you leave the BE resistors at their low value then change the VBE multplier voltage to push more current through the second transistor.

Suppose the BE resistor on the second transistor is 40 ohm and the VBE voltage is 0.6V.  That means the first transistor needs to pass at least 0.6/40 = 15mA.   Anything less than 15mA and the bias current will come from the drivers alone.   If you target a bias current of say 10mA to 20mA through the second transistor, the Darlington *as a whole* needs to pass 15mA+10mA=25mA to 15mA+20mA=35mA. 

If your amp has diodes instead of a VBE multiplier to set the bias current then then only way to up the bias voltage and hence the bias current is to add a very small resistor in series with those diodes.  The best way to measure the bias current is to use the emitter resistors on the output stage (the 0.22 ohm to 1ohm resistors).

FYI, here's an example. They pass a total of 30mA through the Darlington.   Using a resistor alone to set the bias is a crap idea. The current source in the amp makes it less bad.

[mac]

This is what I have on the breadboard:
Class A:
36V -- Output Transformer 4:1 -- TIP41C + NPN darlington -- 3.9 ohm emitter resistor (1V to 1.2V or 250ma to 300ma) -- gnd
                              --        |< 1N400x        --
                                        |
                                       Bias

I tried a D2499 + NPN  from an old TV. It sounded bad.
The current across the 40 ohm resistor is 0.7V/40 ohm = 17.5ma, while the transistor idle base current is 300ma/hfe:60 = 5ma.
It's like the resistor becomes the front man.

In order to "help" the TIP41C to "recover" when it goes off I tried a resistor @ 5ma/10 = 0.5ma, that is 1k2 to 1k5 ohm.
I can't say I hear any difference with/out "helping" resistor but I will leave it there just in case.

Finally I decided to add a switch to reconfigure the power amp from [darlington + BE resistor] to [NPN buffer + TIP41C - no BE resistor].
The buffer config is more creamy at the expense of vol and gain, very close to my Epi Valve Jr v.3.
Running my Red Fuzz (TS808 alternative) in front makes this thing scream.

While I was writing I remembered a couple of things I could try at TIP41C BE:
- Viktor Kempf diode trick from gate to splitted source resistors http://www.sugardas.lt/~igoramps/article68/article.htm
- Vox VBM1 BE diode+resistor
I should run some sims to check if these two tricks overcome the slow turn on.

mac

[antonis]

- Viktor Kempf diode trick from gate to splitted source resistors http://www.sugardas.lt/~igoramps/article68/article.htm

Полностью информативный и легко читается..!! 

[Rob Strand]

In order to "help" the TIP41C to "recover" when it goes off I tried a resistor @ 5ma/10 = 0.5ma, that is 1k2 to 1k5 ohm.
I can't say I hear any difference with/out "helping" resistor but I will leave it there just in case.

Finally I decided to add a switch to reconfigure the power amp from [darlington + BE resistor] to [NPN buffer + TIP41C - no BE resistor].
The buffer config is more creamy at the expense of vol and gain, very close to my Epi Valve Jr v.3.
Running my Red Fuzz (TS808 alternative) in front makes this thing scream.

You will struggle to speed-up turn off as it is limited by the resistor across the BE junction of the *second* Darlington transistor.   That's why that resistor is already low on the Monolithic Darlingtons.    A BE resistor on the first transistor can't fix that problem.

What you can do is stop the second transistor saturating in the first place. 

The first method is anti-saturation diodes.  Traditionally Schottky diodes were used to minimize the saturation voltage.  The idea goes back to the early days of electronics.  The idea was used on one of the old Marshall Valve-state amps:



And then there was this trick from 1975 (so it's an old problem),





FYI, the "Schottky" in the 74xx00 series logic gates has it's roots in the non-saturating transistor idea,
https://www.nutsvolts.com/magazine/article/understanding_digital_logic_ics_part_2

For an amp you need to prevent the driver stage trying to feed all the output current through any saturation diodes.  Something must limit the driver current.

There's one more trick for amps: Operate the driver stage at a lower voltage than the output stage then the output stage can never saturate.  You can use dropping diodes or resistors.

I should run some sims to check if these two tricks overcome the slow turn on

FYI a lot of spice transistor models don't model saturation correctly.  Spice can do it, the problem is the spice parameters in the models are very often wrong and that makes the simulations unrealistic.

[mac]

After digesting Rob info, and with data from a recent post that led me to this page
https://sound-au.com/amp_design.htm#s3
I tweaked some things and got it working and sounding fine.
Reverse EC 1N400x, 4k7 at BE Q1, and I could go as low as 220 at BE Q2.
Typical 40 ohm is too low and make it sounds bad at full drive.

mac

[Rob Strand]

After digesting Rob info, and with data from a recent post that led me to this page
https://sound-au.com/amp_design.htm#s3
I tweaked some things and got it working and sounding fine.
Reverse EC 1N400x, 4k7 at BE Q1, and I could go as low as 220 at BE Q2.
Typical 40 ohm is too low and make it sounds bad at full drive.

Cool.

Not sure why going below 220R is screwing things up.  The extra base current might be
affecting the circuit driving the output stage.

The text under figure 7 explaining all the finer points is spot on.

[mac]

Not sure why going below 220R is screwing things up.  The extra base current might be
affecting the circuit driving the output stage.

99.99%

Using this https://gabevee.tripod.com/sstubepre.html

mac

[Rob Strand]

Not sure why going below 220R is screwing things up.  The extra base current might be
affecting the circuit driving the output stage.

99.99%

Using this https://gabevee.tripod.com/sstubepre.html

mac

Got it.   It's a tweak by ear job for sure.
(A low RBE isn't optimizing or helping anything if it stuffs up something else.   The off-the-shelf/Monolithic Darlingtons don't need to consider their impact on other circuits.  They just make their own specs as good as they can.)