Fair enough but lets look closer, as for mje200 no ft data is shown but its more than likely that the figure given is the max ft as this is how these type medium power transistors are classified, so we have 65 Mhz at 100ma, at 1a ??, after Max comes the drop. It will drop like a rock too. So to not lose the speed and be able to say that they are much faster we have to run them at max IC of 100ma, then how many would we need ?? None or very little benefit here if we aiming at keeping the price down. We should also not forget the cascode devices will cause some deterioration as well as they arent perfect transistors either.
As for the hfe the same argument can be made, after 2a the hfe drops like a rock for mje200. At this current the ft will be terrible btw. So to get better hfe results than the larger part we have to keep below 1 amp to reap any rewards not forgetting to also keep below 100 ma to keep the ft rewards. It just adds up and in the end you have so many transistors in parralel that the output transistors + boardspace + cascode devices + emitter resistors + base resistors to keep the whole lot from oscillating cost 3 times the price of a single device to be able too have any benefits.
Indeed one can have benefits but you should carefully regard all the facts before indulging in such an adventure, is it really worth it ???
I use this method for drivers all the time, but here no cascodes are needed as their are many parts that only need to be doubled for SOA reasons. Despite the minor price increase I do get substancial benefits to justify it. The above example, I dont think Id take that route.
Lets see for 1 amp wed need 10 mje s (100ma) for 9 amps it would be 90, ouch. 0.13$ x90 = $11,70 + cascode devices, +........ I think Ill stop here. Cob of 90 x mje200 cob ?? Ouch. Price no object, sure go ahead. These are only proximates but carefully calculate all involved, think about it ......
Last edited:
Yes to both answers, but carefully weigh the reasons for this improvements, its not all rosy. My previous post shows why. If one can get away from using the cascodes than its benefits are rosy indeed.
In fact how is a high power output BJT made ?? It consists of many smaller BJTs on the die.
"for mje200 no ft data is shown "
ON Semiconductor shows speed peaking at 1A (5A for five paralleled), and still good at 2A (10A for five paralleled).
OTOH, the Semelab MG6332 is stone slow at low currents, it will require a high bias for some speed (300mA for 30Mhz).
ON Semiconductor shows speed peaking at 1A (5A for five paralleled), and still good at 2A (10A for five paralleled).
OTOH, the Semelab MG6332 is stone slow at low currents, it will require a high bias for some speed (300mA for 30Mhz).
OT as it is:
People buy high powered amps mostly because most quality speakers sold in the USA are in the low to mid 80dB/SPL range and you will use quite a few watts getting to "normal" listening levels in the low 90 dB/SPL range. Then you need headroom.
Also many ESL speakers need raw voltage swing, a lot of it.
Other speakers present a low Z load, so you need a high current amp to run them, and in the world of commercial amps that often means an amp with lots of output devices, and they tend to make a lot of power (Krell for example).
Of course with the advent of the "DHT craze" in the 90s, people also found that you can play pretty loud with only a handful of watts, if the clipping is soft... but that led to compression on peaks. Which in turn led to the "high efficiency speaker craze" that followed...
So bottom line is to size the amp power to the load it is driving and the room it is all in.
_-_-bear
I realized my mistake when I got home and opened the datasheet for mje253, the given ft is not the transition frequency but the gain bandwith product so transistion is in fact higher up the Ic curve.
The question is how much difference it makes performance wise and how it sounds. I dont know. Looking around hi end and other manufacturers no one seems to use it and it begs the question as to why. Price wise I dont think its cheaper when all is taken into consideration. Ive tried finding pass s comments about this but no luck yet, there is however a small discussion going on over at pass about the amps he designed for threshold, the cas1 and cas2 using this scheme.
Some members there claim it to be the cats meoww........
The question is how much difference it makes performance wise and how it sounds. I dont know. Looking around hi end and other manufacturers no one seems to use it and it begs the question as to why. Price wise I dont think its cheaper when all is taken into consideration. Ive tried finding pass s comments about this but no luck yet, there is however a small discussion going on over at pass about the amps he designed for threshold, the cas1 and cas2 using this scheme.
Some members there claim it to be the cats meoww........
Leach used the MJE243/253 in his original design as pre-drivers, and went to another higher voltage device for the later higher powered versions.
Janneman used the MJE200/210 in his cascode design.
Both look interesting.
I am currently looking at using a ThermalTrac device as the driver in a CFP pair with Baker clamp diodes for a Pass S-style output stage (no global feedback). I like the way those amplifiers sounded.
Janneman used the MJE200/210 in his cascode design.
Both look interesting.
I am currently looking at using a ThermalTrac device as the driver in a CFP pair with Baker clamp diodes for a Pass S-style output stage (no global feedback). I like the way those amplifiers sounded.
- Status
- Not open for further replies.