Hi all, I am wonder what the purpose of using multiple Low Power transistor in parallel.
Just an analogy. For 200W, I use 4 x 50W rated power transistors. Is it any better compared to using 2 x 100W ones? I am asking because I saw 6 pairs of Sanken A1860/4886 on the Krell 400xi. Each pair is rated at 80W. If they use higher power variants, it will cut down the number (eg. 4 pairs of 130W rated ones).
So, is there any reason to use more transistors??
Just an analogy. For 200W, I use 4 x 50W rated power transistors. Is it any better compared to using 2 x 100W ones? I am asking because I saw 6 pairs of Sanken A1860/4886 on the Krell 400xi. Each pair is rated at 80W. If they use higher power variants, it will cut down the number (eg. 4 pairs of 130W rated ones).
So, is there any reason to use more transistors??
You might be able to find a single transistor powerful enough, but it means all the heat has to go through a single insulating washer. If you share the heat amongst several washers you have greater surface area and so lower thermal resistance. That could be one reason.
Only the designer can tell exactly why. I could imagine few possible reasons:
- Better spread of the heat
- Better overall performance (flat hfe, higher ft, lower capacitance) of the low power devices
- Very good negotiated price of a given transistor type
- etc
- Better spread of the heat
- Better overall performance (flat hfe, higher ft, lower capacitance) of the low power devices
- Very good negotiated price of a given transistor type
- etc
Paralleling output devices is very basically an easy way to achieve a higher output wattage for a given voltage .
It also provides a more stable circuit at lower temperatures of the active devices.
Of course this is very simply put and in BJT output devices small value emitter resistors are required to "balance them up " .
Several audio designers I know use that method and designs have appeared in EW over the years.
How deep do you want to go into this?
It also provides a more stable circuit at lower temperatures of the active devices.
Of course this is very simply put and in BJT output devices small value emitter resistors are required to "balance them up " .
Several audio designers I know use that method and designs have appeared in EW over the years.
How deep do you want to go into this?
Sorry but it does not work that way.
You are comparing amplifier power output with "datasheet spec" power dissipation, two only faintly related parameters.
Re-read datasheets: rated power dissipation assume completely unrealistic 25 degrees Centigrade case temperature and in fact is just a way to print "looks good numbers", in fact it´s an indirect way to show junction-to-case thermal resistance, which is only one of the thermal resistances involved in cooling a junction..
Given minimum case to heatsink thermal resistance, you would need a freezing temperature 😱 heatsink to achieve that: clearly unrealistic.
In practice, a 100 to 150W labelled transistor can dissipate only, say, 30 to 40W, go figure.
On the other side, a 100W amp will dissipate 30 to 40W as heat if Class AB and 100W or slightly more if Class A so only by sheer coincidence a couple 100/150W transistors *can* be used in a 100W amplifier.
Now you see why both many transistors and good heatsinking is required above certain power level.
I was reading up the spec sheets of several brands and models. I think perhaps current limit is also the key. I notice that even though the transistor is rated 130W, the max current is just 15A (2SA1860 80W one is 14A).
Then power dissipation also varies with temperature (what JMFahey mentions). At 50C, 1860 drops to 65W. At 75C, it drops to 50W.
Perhaps also limitation of each transistor. I am thinking the hfe part. Although hfe is 180 max, I maybe its not enough. Eg, for a very small 1mA current, it will only be boosted to 180mA. But if there are 6 transistors, it becomes 1080mA. Of course, other factors like balast resistors will affect this but I think overall gain will still be higher than just 1 transistor.
there are also downsides with using multiple transistors, but I think overall benefits is still higher.
Smaller transistors often have better second breakdown performance than bigger ones do. The problem is caused by current crowding/hogging on parts of the die, and gets worse the bigger you make it. Two 100 watt transistors may have better power handling than a single 200 watt, at the same case temperature, at higher voltages. And as others have already noted, it’s far easier to keep the case temperature down with two packages spreading the heat out.
Umm .. sorry escksu, but your arithmetic in post #7 has an ooopsie --
The 1 mA Base drive will be distributed across the 6 Bases, so close to the same net current gain. However, it is true that transistors have decreasing hFE with increasing Collector current, so that would be another advantage.
If you're mostly interested in power ratings, though, start with the Safe Operating Area graphs toward the end of typical data sheets. Study it carefully and plan to do a little arithmetic -- you'll be amazed at how fragile some very hefty power transistors really are.
Cheers
The 1 mA Base drive will be distributed across the 6 Bases, so close to the same net current gain. However, it is true that transistors have decreasing hFE with increasing Collector current, so that would be another advantage.
If you're mostly interested in power ratings, though, start with the Safe Operating Area graphs toward the end of typical data sheets. Study it carefully and plan to do a little arithmetic -- you'll be amazed at how fragile some very hefty power transistors really are.
Cheers
Krell even parallels drivers, predrivers etc. They do design their products to operate at higher current levels than needed ( conservatism) thus use multiple devices that are optimal in other characteristics, but need paralleling to provide the right amount of current. If you went just by data sheets, it would not make sense. Dan had a method to the madness and gave Krell that signature sound.
The guys have given you great reasons above. A couple more... Krell used TO-3 output devices in the late 90’s when most had switched to TO-3P plastic flatpaks... these devices were ring emitter derivatives of the famed 2sa1302 pair ( On/ Motorola got the rights from Toshiba) and Dan negotiated a great deal for a Million of those ( if I recall) in the early/ mid 90’s and used them extensively in the KAV/ FPB series. This is one example of parts availability playing a part in design philosophy.
Another reason is, with rising current, hfe drops on single dies, so multiple chips help relieve loading on the drivers.
The guys have given you great reasons above. A couple more... Krell used TO-3 output devices in the late 90’s when most had switched to TO-3P plastic flatpaks... these devices were ring emitter derivatives of the famed 2sa1302 pair ( On/ Motorola got the rights from Toshiba) and Dan negotiated a great deal for a Million of those ( if I recall) in the early/ mid 90’s and used them extensively in the KAV/ FPB series. This is one example of parts availability playing a part in design philosophy.
Another reason is, with rising current, hfe drops on single dies, so multiple chips help relieve loading on the drivers.
Last edited:
