1-30MHz push pull small linear questions
1. I think a MPQ2222 transistor array will be suited better as it is capable of more power (2n2222 matched quad)?
2. Can the input transformer be replaced with a transistor phase splitter, such as this one ../imgqa/eboard/Antenna/rf-evpqqszasa1.jpg with 100nF DC blocking capacitors at it's output?
3. If yes, how could I then bias the bases of Q2, Q3, Q4, Q5?
You cannot place BJT transistors in parallel without emitter ballast resistors (thermal runaway).
If you replace the input transformer with a transistor have to use NPN and PNP transistors, or use only NPN but have to rid out somehow the output transformer.
I think better is to let as it is, and to increase power use up to 3 transistors in parallel, and place 1 ohms resistors series with the emitters.
1. matching the transistors will always help in that kind of amplifier but the main consideration is the MPQ2222 has all five transistors on one slab of silicon. That gives them good thermal matching. The IC will get hot and under fixed bias the output stage current will increase with temperature but using the Q1 transistor as a diode in that way ensures it also increases conduction at the same rate which in turn reduces output stage current. So it stabilizes the bias point. You can do the same with discrete components but getting a good thermal match can be much more difficult.
2. In theory yes, and in small signal circuits it will work but in a power amplifier it presents a problem - it can't produce enough bias current. Although it doesn't look like much input current is needed, when you factor in the charge on the transistor and it's junction capacitance at RF, the amount of drive signal starts to be significant. A transformer can easily convert the higher voltage from the driver stage to a higher current for the output transistors, a pair of 5.6K resistors can't do that.
3. The only way you might be able to do it is to significantly reduce the collector and emitter resistors in the phase splitter - may to 220 Ohms or lower. Then capacitively couple to the output transistors but also bias the top and bottom pairs through chokes from a bias source like the Q1. The resulting circuit looks like the one in post #1 but the two halves of the driver transformer are replaced with chokes and the signal fed in to each base from the phase splitter.
Brian.
I see...
Maybe instead of 1 ohm resistors at each emitter, I could place one 1 ohm resistor in series with the DR choke?
That is what I was thinking to do, to avoid the input transformer. Ok so I will use 220R instead of R3 and R4 in the phase splitter (maybe less). Then I couple the two phases through say 100nF caps to the bases of the push pull pairs. Then I use the 2.2k and the Q1 at it is, except that their common point is connected to the bases of the transistors through a choke (one choke for each base pair).
Have I understood that correctly?
You got it!
I won't guarantee results though but it's worth a try.
Regarding the emitter resistors, you need one in each emitter to provide degenerative feedback. If you use a single resistor it won't help to balance the currents through the transistors. Obviously don't use inductive resistors!
Brian.
Years ago designing integrated PA's for GSM in silicon process (3 stages, and about 3W output) we used 3 BJTs in parallel for the last stage. Each of these BJTs use 0.2 ohms ballast resistors in series to each emitter.
Those resistors were inside of the chip PA. Initially for the prototype die we placed in parallel few resistors on each emitter, and burn (laser) one at the time doing a kind of tuning to see where the efficiency (PAE) starts to degrade significant, but to keep thermal runaway under control.
Definitely you need to use ballast resistors (on each emitter) even if use lower power transistors all part of the same integrated circuit.
Isn't output power degrated? I am curious about the purpose these resistors serve
They help to balance the current through the transistors. When you wire transistors directly in parallel the one that conducts first will sink most current and partially 'bypass' the other transistor(s). The natural variation in manufacture means there is almost always some difference in the transistors, even when they are on the same silicon.
You could place a resistor in each base pin but the lower current would require a higher value which has other consequences, especially at high frequency. The base resistor value would have to be individually chosen according the gain of the transistor. Placing a resistor in the emitter is more effective because it also stabilizes the DC characteristics. It is the voltage between the base and emitter that creates the base current so if the current flowing through an emitter resistor rises, it raises the emitter voltage and in turn reduces the B-E voltage and hence current.
It does cause some power loss but it is minimal compared with losing operation of one or more paralleled transistor.
Brian.
It doesn't. MPQ2222 is comprised of 4 individual chips in a package, respectively the Vbe matching and thermal coupling is considerably worse than of originally utilized monolithic CA3083 array. MPQ2222 doesn't even have a matching specification.
Balancing transistors may be dispensable in the original circuit but are surely necessary for MPQ2222.
I thought any transistor array is monolithic!
Thanks for the info, thumbs up!
How about an IC for amplifying the DSB signal out of an SA612 to 1W or close to it, broadband 1-30MHz and if possible use no input and output transformers.
Something like a power video opamp operating to 30MHz or a power MMIC?
Is there such an IC out there or do I have to rely on the transistor array and external transformer?
Where to start looking for such a thing?