The C4000/5000 set in my Carver M200t seems to have seen better days.
Now I have got MJ15001/2 and MJ15024/5 that are direct swaps.
However I am planning on re jiggering the heat sink and get the transitors to go flat onto another panel.
So I'm thinking I can drop in a different transistor (something with more efficiency etc etc).
Here is the MJ15001/2 datasheet.
http://www.onsemi.com/pub_link/Collateral/MJ15001-D.PDF
Collector
Now I have got MJ15001/2 and MJ15024/5 that are direct swaps.
However I am planning on re jiggering the heat sink and get the transitors to go flat onto another panel.
So I'm thinking I can drop in a different transistor (something with more efficiency etc etc).
Here is the MJ15001/2 datasheet.
http://www.onsemi.com/pub_link/Collateral/MJ15001-D.PDF
MAXIMUM RATINGS
Rating Symbol Value Unit
Rating Symbol Value Unit
Collector
−Emitter Voltage VCEO 140 Vdc
Collector
−Base Voltage VCBO 140 Vdc
Emitter
−Base Voltage VEBO 5 Vdc
Collector Current
− Continuous IC 15 Adc
Base Current
− Continuous IB 5 Adc
Emitter Current
− Continuous IE 20 Adc
Total Power Dissipation @ T
C = 25°C
Derate above 25
°C
P
D 200
1.14
W
W/
W
W/
°C
Hopefully that is possible.
Thanks.
Srinath.
Operating and Storage Junction
Temperature Range
TJ, Tstg –65 to +200 °CTemperature Range
Hopefully that is possible.
Thanks.
Srinath.
Last edited:
Like this
2SA1216 pdf, 2SA1216 description, 2SA1216 datasheets, 2SA1216 view ::: ALLDATASHEET :::
Not that one exactly, but something like that I'm thinking I can swap in.
Cool.
Srinath.
2SA1216 pdf, 2SA1216 description, 2SA1216 datasheets, 2SA1216 view ::: ALLDATASHEET :::
Not that one exactly, but something like that I'm thinking I can swap in.
Cool.
Srinath.
I take it you want to substitute TO247 (aka TO3PL) plastic package for the T03 cans. I can't see how there is going to be any change of efficiency but I would think MJL21193/4 or MJL21195/6 will be just right and quite an improvement on the linearity of the ancient MJ150XX series parts.
Perfect. Thanks so much.
Now I thought the plastic body transistors make less heat (which is why they can be plastic LOL) I was sorta thinking efficiency as power taken in to sound being put out. Those TO-3's make more heat than sound I think.
Thanks.
Srinath.
Now I thought the plastic body transistors make less heat (which is why they can be plastic LOL) I was sorta thinking efficiency as power taken in to sound being put out. Those TO-3's make more heat than sound I think.
Thanks.
Srinath.
plastic package devices cannot take as much heat as metal packaged.
Something to do with ion migration at higher temperatures.
This is usually confirmed by reading Tjmax from the datasheet.
To3 are usually 200°C
Plastic BJT are usually 150°C & FETs 175°C
But there are exceptions. Read the datasheet.
The higher operating temperature allows a smaller heatsink or a fan that turns on later or a temp switch that turns off later.
Swapping in plastic package deserves a redesign, or could risk a blow up.
Something to do with ion migration at higher temperatures.
This is usually confirmed by reading Tjmax from the datasheet.
To3 are usually 200°C
Plastic BJT are usually 150°C & FETs 175°C
But there are exceptions. Read the datasheet.
The higher operating temperature allows a smaller heatsink or a fan that turns on later or a temp switch that turns off later.
Swapping in plastic package deserves a redesign, or could risk a blow up.
That sounds a little counter intuitive.
If the plastic one does run cooler - why would it need a bigger heat sink. Yes temperature difference to the 4th power is the rate of cooling. However that higher rate of cooling only exists when the big temperature difference exists. As in, the object will cool 16 times faster if it had 2 X the difference to ambient. But once it drops to the same temperature as the one that runs cooler anyway it will shed heat from there on out at the same rate.
OK I can see that the plastic black body that is a lot smaller will shed heat a lot worse than the larger metal shiny one. But 90% of the heat loss comes from the heat sink - and the transistor basically dumps heat into the heat sink. There the metal back of the plastic transistor works as well as the TO3.
I am missing something.
Cool.
Srinath.
If the plastic one does run cooler - why would it need a bigger heat sink. Yes temperature difference to the 4th power is the rate of cooling. However that higher rate of cooling only exists when the big temperature difference exists. As in, the object will cool 16 times faster if it had 2 X the difference to ambient. But once it drops to the same temperature as the one that runs cooler anyway it will shed heat from there on out at the same rate.
OK I can see that the plastic black body that is a lot smaller will shed heat a lot worse than the larger metal shiny one. But 90% of the heat loss comes from the heat sink - and the transistor basically dumps heat into the heat sink. There the metal back of the plastic transistor works as well as the TO3.
I am missing something.
Cool.
Srinath.
The plastic one does NOT run cooler. They have a lower maximum allowable temperature. The plastic molding compound actually melts somewhere below 200C which is why you can't run them that hot. Right near the transistor die they do get that hot and if the plastic melts inside it will eventually ruin the transistor. Not immediately the first time, but it is very fast comapred to other failure mechanisms.
For a given amount of watts of heat, the heat sink will rise the same number of degrees, regardless of whether a plastic or metal transistors are used. But the plastic one gets hotter inside because the heat has a harder time getting out. And at the same time it can't take as much temperature as a metal one could.
For a given amount of watts of heat, the heat sink will rise the same number of degrees, regardless of whether a plastic or metal transistors are used. But the plastic one gets hotter inside because the heat has a harder time getting out. And at the same time it can't take as much temperature as a metal one could.
The problem with the Carvers (well, one of them) is that they don't really have enough heat sink to really beat on them. Just a little more heat sinking would help a lot. And changing the commutator transistors with some high fT audiophile let's see if we can even get genuine ones today Sanken units will have no impact on the sound. The bottom transistors in the stack, maybe. But with a triple EF even that is doubtful.
wg_ski: That explaination on the heat makes sense. Thanks.
Now what transistor you suggest ?
Thanks.
Srinath.
Now what transistor you suggest ?
Thanks.
Srinath.
Yes, why not?
Anyway Cx000 is just a house number, I bet they were some commercial product of the day, relabelled for OEM use.
What does that mean?
Are they blown?
Transistors do not wear out or weaken, they just work or die.
If they are made of silicon and used in the same circuit, efficiency will not change.
It depends more on circuit operation than anything else.
Body material is irrelevant.
Power is dissipated in silicon junctions, packaging is to keep contamination away and to provide some means of mounting.
If you were making some module to be used in an Interplanetary probe, such as Voyager, and you needed to save up to the last gram possible, unencapsulated chips could be used, they will work very well in vacuum.
Well, that's why different "classes" exist.
Pick your letter 😉
If you somehow try to mean that they waste more heat than output power is produced, case is absolutely irrelevant, only operating class and load parameters matter.
Who says they run cooler under the same electrical conditions?
You mean radiation cooling.
Not the main transistor cooling mechanism by any means.
Although it applies nicely to tubes in vacuum.
Nice cut and paste but the temperatures mentioned are not what you imagine.
There you need to use absolute temperatures.
So an 80C heatsink is not twice as hot as a 40C one, what you seem to imply, but
12.8% hotter .
And anyway radiation in that case is small compared to convection.
The only way it will drop to the same temperature to its surroundings is with the amp turned off and no power is being dissipated.
If any power is being dissipated, it will be hotter than its surroundings to shed that extra heat away.
Agree that colour does not matter here compared to heatsinking.
Not really, because the metal is not the same.
99.99% TO3 are steel , although i fondly remember Motorola aluminum ones and have read about copper cased Germanium.
"Plastic" power transistors are actually copper backed, and plastic is there just to cover the actual chip.
..
Yes, they often do.
Sorry for the French but nonsense.
a) encapsulation plastic is Epoxy, a very good one by the way, which is not thermoplastic so it will decompose and crack rather than melt.
And its temperature limit is way higher than 200C anyway.
Just melt a drop of solder which is much hotter and apply it to the top of any power transistor, leaving the soldering iron tip in contact with it to keep temperature very high.
b) What gets destroyed above 150/175/200C is the actual atom lattice/structure inside the chip crystal which makes it a transistor or whatever instead of a plain piece of metal.
Irreversible damage inside the chip , specially second breakdown, can be much faster than any hypotethical plastic melting.
Here you have maximum allowable dissipation, detailed by time.
Curves go from DC (as in "all the time you want") to 1 micro second 😱
An externally hosted image should be here but it was not working when we last tested it.
True
Not true.
The chip is soldered to a copper plate and plastic "on the other side" is irrelevant.
Irrelevant.
Who cares that Epoxy may crack at 350C and burn (with difficulty, it has flame retardants incorporated) at, say, 400C or higher while steel melts at, say, 1200C?
The chip inside dies above 150/175/200C , way below any casing material does.
Typical molding compounds for plastic semi's have a glass transition temp somewhere between 150 and 200C. They don't truly melt, but the goop can migrate into the semiconductor material and passivation and contaminate things. It won't outright kill the device, but it seriously reduces the mean time to failure. Same thing for going above 200C. That won't kill a device either, but it reduces the reliability to some set threshhold. Unfortunately, that threshhold varies between classes of devices and isn't one set number of hours. You can run to 300C, but maybe only for 20 hours or so. At 150-200C it will usually be between 10,000 and 10 million. Manufacturers will provide rel data if there is enough money riding on it. You *can* get higher Tg overmolds, but they don't use these for cheap commodity parts and they are still not much above 200C.
True enough, a copper base has a lower thermal impedance than a steel one in a TO-3. But manufacturing tolerances and flatness are sloppy on the TO-264 resulting in a higher case to sink thermal resistance. Junction to case is a bit lower on the TO-264 but in practical application it's still going to run at a higher Tj. And again, the TO-264 has that additional failure mecahnism of the epoxy compound which limits Tj to 150C with the same die - for the same reliability numbers.
The problem with second breakdown is that it doesn't require high junction temperatures to occur. You get very localized hot spots that are much smaller than the "peak junction temperature" calculated from the thermal model of the die. A run of the mill IR imager won't even pick it up till its too late. A good one with a very small spot size is needed. You could be running fat and happy with a 100C hot spot distributed over a reasonable area, and then push a 100 volt pulse just a bit too long. You get a hot spot a couple microns wide, Vbe drops in that region, and all the current wants to flow through it. You know how that will end and all this happens before you can even measure in increase in junction temperature. The die layout and canstruction is very important in keeping this at bay. The SOA curves for this are based on a lot of destructive testing and statistical confidence limits. Some manufacturers are more conservative with the ratings than others, so there's always some gray area. I've found Toshiba parts to be very tolerant of going above s/b ratings compared to others. The more modern On stuff is more tolerant than the old JEDEC parts that were available in the 70's. What they will do and what they will guaranteee are often two completely different answers. If there weren't any wiggle room in the ratings, power amps like the PLX3402 wouldn't even survive the first gig bridging into a 2x18 sub. Analyze it sometime. I have.
True enough, a copper base has a lower thermal impedance than a steel one in a TO-3. But manufacturing tolerances and flatness are sloppy on the TO-264 resulting in a higher case to sink thermal resistance. Junction to case is a bit lower on the TO-264 but in practical application it's still going to run at a higher Tj. And again, the TO-264 has that additional failure mecahnism of the epoxy compound which limits Tj to 150C with the same die - for the same reliability numbers.
The problem with second breakdown is that it doesn't require high junction temperatures to occur. You get very localized hot spots that are much smaller than the "peak junction temperature" calculated from the thermal model of the die. A run of the mill IR imager won't even pick it up till its too late. A good one with a very small spot size is needed. You could be running fat and happy with a 100C hot spot distributed over a reasonable area, and then push a 100 volt pulse just a bit too long. You get a hot spot a couple microns wide, Vbe drops in that region, and all the current wants to flow through it. You know how that will end and all this happens before you can even measure in increase in junction temperature. The die layout and canstruction is very important in keeping this at bay. The SOA curves for this are based on a lot of destructive testing and statistical confidence limits. Some manufacturers are more conservative with the ratings than others, so there's always some gray area. I've found Toshiba parts to be very tolerant of going above s/b ratings compared to others. The more modern On stuff is more tolerant than the old JEDEC parts that were available in the 70's. What they will do and what they will guaranteee are often two completely different answers. If there weren't any wiggle room in the ratings, power amps like the PLX3402 wouldn't even survive the first gig bridging into a 2x18 sub. Analyze it sometime. I have.
Carver used the MJ15015/16 marked C4000/5000 in the M400 (cube).
Crown also used this part in the Power Base 1.
The SOA is rated at 180W at 60V, but the gain drops like a rock above 4A and the driver transistors frequently blow up.
The MJ1502x series generally work fine with no changes required.
Crown also used this part in the Power Base 1.
The SOA is rated at 180W at 60V, but the gain drops like a rock above 4A and the driver transistors frequently blow up.
The MJ1502x series generally work fine with no changes required.
Having had a metal can TO3 endowed SMPS blow up where molten metal flew over my shoulder leaving a pair of glowing topless TO3's on the heatsink, I prefer plastic.
sBrads: LOL, thanks for that mental image - and now I'll be doing my best to avoid it if you dont mind.
djk: Mj15015/6 eh, good to know.
MJ15024/5 is a substitute, as is a mj15001/2 - which was what I was gonna use cos they have a gain of 25 vs 15 for the mj15024/5. However - since I have no clue what gain is and if more is actually louder - and if its just louder @ full volume or louder all across - no idea.
wg_ski: You're saying the modern transistors are going to fare better - and the plastic case one are more modern - I'm a bit lost with this post of yours.
JMFahey: I have to read your post over many times. I should stop making sweeping statements huh.
Thanks guys. This saga continues.
Srinath.
djk: Mj15015/6 eh, good to know.
MJ15024/5 is a substitute, as is a mj15001/2 - which was what I was gonna use cos they have a gain of 25 vs 15 for the mj15024/5. However - since I have no clue what gain is and if more is actually louder - and if its just louder @ full volume or louder all across - no idea.
wg_ski: You're saying the modern transistors are going to fare better - and the plastic case one are more modern - I'm a bit lost with this post of yours.
JMFahey: I have to read your post over many times. I should stop making sweeping statements huh.
Thanks guys. This saga continues.
Srinath.
Not louder. More feedback and thus less distortion.
The amplifier must be able to meet the PEAK transient current demand of the speaker.
This peak current is shared between the output devices of the parallel sets.
eg
200W into 8ohms amplifier can output 40Vac and 5Aac into a resistor load.
Attach a reactive speaker and the current demand when fast transients are applied and the current demand increases spectacularly.
I design for 3 times the current.
Those 40Vac and 5Aac become 56Vpk and 7Apk into a resistor.
Into an 8ohms multi-way speaker and applying a 3times demand results in 56Vp and 21Apk through the upper devices or the lower devices.
Assuming you have a 3pr output stage, the output devices will see a demand of ~7apk through each.
look up the datasheet for gain (hFE) @ 7A to calculate the base current. add on the driver current into it's load resistor to find the total driver current. Now look up the datasheet to find the hFE of the driver for that emitter current you calculated.
This can be around 700mApk to 2000mApk. That imposes a mighty SOA on the driver and pulls an enormous current from the VAS. This is why I recommend a pre-driver for low impedance loads, or for high power amplifiers.
When you have the peak transient currents and the appropriate hFE for those peak currents you can use the datasheet SOA transient curves to determine which component blows up first, the VAS, or the Driver, or the Outputs.
The amplifier must be able to meet the PEAK transient current demand of the speaker.
This peak current is shared between the output devices of the parallel sets.
eg
200W into 8ohms amplifier can output 40Vac and 5Aac into a resistor load.
Attach a reactive speaker and the current demand when fast transients are applied and the current demand increases spectacularly.
I design for 3 times the current.
Those 40Vac and 5Aac become 56Vpk and 7Apk into a resistor.
Into an 8ohms multi-way speaker and applying a 3times demand results in 56Vp and 21Apk through the upper devices or the lower devices.
Assuming you have a 3pr output stage, the output devices will see a demand of ~7apk through each.
look up the datasheet for gain (hFE) @ 7A to calculate the base current. add on the driver current into it's load resistor to find the total driver current. Now look up the datasheet to find the hFE of the driver for that emitter current you calculated.
This can be around 700mApk to 2000mApk. That imposes a mighty SOA on the driver and pulls an enormous current from the VAS. This is why I recommend a pre-driver for low impedance loads, or for high power amplifiers.
When you have the peak transient currents and the appropriate hFE for those peak currents you can use the datasheet SOA transient curves to determine which component blows up first, the VAS, or the Driver, or the Outputs.
So more gain = less distortion ?
I am wondering what is wrong with this amp.
I have one channel essentially 6db down from the other. That was after I put higher hfe (of my c4000/5000 set) in the lower power channel. It was 3db down, and swapping transistors made it 6db down.
I am trying to find why its doing that, which was what made me conclude the transistors were wearing out. Maybe not ...
Cool.
Srinath.
I am wondering what is wrong with this amp.
I have one channel essentially 6db down from the other. That was after I put higher hfe (of my c4000/5000 set) in the lower power channel. It was 3db down, and swapping transistors made it 6db down.
I am trying to find why its doing that, which was what made me conclude the transistors were wearing out. Maybe not ...
Cool.
Srinath.
You got some fake transistors. Real TO-3 cans won't fly off no matter how badly you blow them. You just end up with a shorted device and a burnt PCB trace. The tops pop off real easy on the fakes - you can almost do it without tools. And a 'normal' load will blow a fake that isn't up to spec.
TO-264/247 can throw flying bits of plastic at you from a relatively benign overload. If it were encased in a TO-3 (a real one) it would be contained. Most SMPS devices are in some sort of plastic case these days. Device cost is lower and so is the cost of the next level assembly.
The advantage of metal is that you can get to a higher temperature, but in practical consumer applications it's pretty much moot. You're never going to get the required safety certs if the heat sink temperature gets above 90C (fire hazard) so you may not be able to take advantage of a higher junction (and therefore case) temperature. And EVERYTHING is driven by manufacturing cost these days. You don't get that down, you can't compete - period.
When transistors wear out they just plain die. There are seveal wear out mechanisms, and ultimately they end in a blown out device that doesn't work at all. You can get a degraded hFE from excessive reverse bias on the B-E junction but that just isn't going to happen in a working class AB output stage. Degraded hFE here would only just blow your drivers, not result in a drop in volume! The volume (gain) is dominated by the feedback and not dependent on which transistor is used.
If you get a drop in volume when you change a transistor look for something OSCILLATING. This can fool your front end and change the effective feedback factor. This would also be accompanied by a significant increase in distortion and usually overheating.
You must have something else wrong, definitely not directly related to output transistor substitution.
Typical SS amps have huge open loop gain, which gets tamed down by almost as huge NFB , which evens out gain so tightly that it practically depends only on the negative feedback network values.
As in: suppose the amp has 60dB open loop gain, you use 10K/1K feedback resistors which means 20dB effective gain, you still have 40dB margin.
Now suppose some internal stage in your "bad" amp loses 6dB gain for whateverv reason.
Total gain will still be 20dB (10K/1K) , just with "only" 34dB gain margin ... which still is a lot.
What you'll notice is that distortion, which is usually swamped by that huge NFB, is now a little higher, measurably but doubt bit can be heard.
But system (amplifier) gain will not vary by 6dB , no way.
So feed some signal into it and start tracing it to find where you lose signal.
Which I thing isn't in the amp but in them connectiors/pot/swit ching/etc.
Or somehow when replacing something else you messed with the NFB.
You don't get fakes in 20yr old military gear. The tracks, wires all went south as well. It was powered by transformerless 115v 3 phase, huge smoothing caps with no fuses or any input protection. Not your usual home stuff where I agree TO3 cans would be extremely unlikely to ever go into meltdown.
- Status
- Not open for further replies.