I want to replace the TIP41C outputs in a Class A headphone amp with TTC004B. They will be biased at about 120mA and require heatsinking, just as the 41C's do. I have two concerns/questions.
1. The TTC004B is epoxy-encased. The data sheet warns not to rely on that for insulation, but says nothing about its impact on heat transfer. Is this something to worry about?
2. The pinout of the TTC004B is ECB, the reverse of the 41C. Given #1, will turning it around and attaching the face to the heatsink provide adequate heat transfer?
Here's the link to the data sheet.
1. The TTC004B is epoxy-encased. The data sheet warns not to rely on that for insulation, but says nothing about its impact on heat transfer. Is this something to worry about?
2. The pinout of the TTC004B is ECB, the reverse of the 41C. Given #1, will turning it around and attaching the face to the heatsink provide adequate heat transfer?
Here's the link to the data sheet.
Last edited:
Sounds like a bad idea. The package is smaller, the insulation of the epoxy limits heat transfer compared to the TIP and mounting it the wrong way around is going to make it worse (the case is thicker at the front).
It may work, but surely it is not optimal. The obvious question then is "why"?
It may work, but surely it is not optimal. The obvious question then is "why"?
The answer to 'why do it' is to get the 50% increase in Hfe compared to the TIP41C, with the attendant increase in OLG and lower distortion.
Encapsulating the entire package seems like an odd choice on Toshiba's part. This thing directly replaces BD139's as drivers and low-power outputs.
Encapsulating the entire package seems like an odd choice on Toshiba's part. This thing directly replaces BD139's as drivers and low-power outputs.
Thanks, that's what the data sheet suggests. The Fairchild-branded 41C's are 150-155 at room temp. The TTC004B are 250. Hence the attraction.
So how much dissipation is it actually subject to? You mention 120 mA, how much voltage? Those encapsulated transistors are only good for about a watt, maybe two to five if well heat sinked. You will NEVER get close to the “infinite heat sink” rating. And mounting it “backwards” makes the heat transfer that much worse.
If the outputs are followers, you will not necessarily get a massive increase in OLG by increasing the hFE. It may be fairly insensitive, if competently designed. Super high fT isn’t needed either - just by going class A you eliminate 99.9% of why you “need” high fT followers since switch-off will never happen. The reason TIP41’s were specified in the first place is likely because it’s a case where one can get away with it without serious detriment to the sound.
If the outputs are followers, you will not necessarily get a massive increase in OLG by increasing the hFE. It may be fairly insensitive, if competently designed. Super high fT isn’t needed either - just by going class A you eliminate 99.9% of why you “need” high fT followers since switch-off will never happen. The reason TIP41’s were specified in the first place is likely because it’s a case where one can get away with it without serious detriment to the sound.
Vce is ~ 11.5V so it is dissipating ~ 1.4W. That's right at the knee of the TTC004B SOA curves in the data sheet without a heatsink, and is what made me think this is doable. Definitely don't need the higher Ft.
Thats what I figured - it was down pretty low. 1.4 watts certainly is do-able.
I would be more concerned about the change in hFE across its operating region than the actual magnitude of it. It’s probably not any worse on the TIP than the TTC. TIPs get their bad rep from the high capacitance (not an issue here) and the gain fall off to about 10-15 at full rated current (again, not an issue here). High voltage transistors like the TTC’s tend to have worse quasi-saturation (reduction in gain at low vce). Not as much an issue when swinging 70 volts as a driver/predriver, but if your supply is +/-11 volts it matters more. In the application, I would be looking for a part that maintains good hFE down to 2 volts or less (at 2X the 120 mA bias) with everything else being secondary. The TIP may actually win that battle between the two, but I bet you can find others that beat both. I use the D44 for this sort of thing. The original Philips BD139 would too (careful, not all BD139’s are created equal).
I would be more concerned about the change in hFE across its operating region than the actual magnitude of it. It’s probably not any worse on the TIP than the TTC. TIPs get their bad rep from the high capacitance (not an issue here) and the gain fall off to about 10-15 at full rated current (again, not an issue here). High voltage transistors like the TTC’s tend to have worse quasi-saturation (reduction in gain at low vce). Not as much an issue when swinging 70 volts as a driver/predriver, but if your supply is +/-11 volts it matters more. In the application, I would be looking for a part that maintains good hFE down to 2 volts or less (at 2X the 120 mA bias) with everything else being secondary. The TIP may actually win that battle between the two, but I bet you can find others that beat both. I use the D44 for this sort of thing. The original Philips BD139 would too (careful, not all BD139’s are created equal).
That looks like a 30 volt version of the D44 (Right down to the poor DCSOA, which is of little concern). I think they used to make a 30 volt version at one time.
Curves in the center figure says it all - gain stays high right up till it clips. Exactly what is wanted where the power supply voltage is low.
Curves in the center figure says it all - gain stays high right up till it clips. Exactly what is wanted where the power supply voltage is low.
Thx for the input. I think I'll go ahead and try the TTC's on one channel and watch temps carefully while it warms up, before testing it.
Unless the Hfe @ 120mA comes in at the high end of spec, I don't see the SD882 offering a significant improvement over the TIP41C for this experiment. The Chinese Fairchild-branded 41C (who knows if it's 'genuine') is really not bad, outperforming the OnSemi data sheet in many respects, and amazingly consistent unit-to-unit.
Unless the Hfe @ 120mA comes in at the high end of spec, I don't see the SD882 offering a significant improvement over the TIP41C for this experiment. The Chinese Fairchild-branded 41C (who knows if it's 'genuine') is really not bad, outperforming the OnSemi data sheet in many respects, and amazingly consistent unit-to-unit.
The most obvious replacement for the BD139 / BD140 pair is the TTC015B / TTA008B pair. The TT devices are much better than the modern BD devices though. I doubt than any modern BD are like the old Philips devices.
Have you considered the 2SC6144SG / 2SA2222SG already? Seems like an odd proposal at first, but may be worth consideration. Also, they are directly pin compatible.
Although there is no line item for Thermal_Resistance_Junction_to_Case in any of the TTC004B datasheet tables, with a little ingenuity you can extract it from the plots. It's between 70 and 80 degrees C per Watt.
And of course you can use two of them in parallel to cut the per-device dissipation in half.
_
And of course you can use two of them in parallel to cut the per-device dissipation in half.
_
