Maybe the way schematic is drawn is fooling me but R20 and R21 are incorrectly connected. The connection should be from the first transistors emmiter to the second transistors emiter, before the 10 ohm resistor. It could be that you could get away with it as it is but I have my doubts.
Some compensation is indeed needed like janneman suggests.
The idea of using a darlington transistor as vas on this design is ill concieved, it will worsen performance instead of improving it, a single part like ksa1381 and its complement will improve performance drastically, what works here is cascode, preferably hawksford cascode. In this case youll end up having a poweramp being a copy of the commercial preamp hawksford designed.
Some compensation is indeed needed like janneman suggests.
The idea of using a darlington transistor as vas on this design is ill concieved, it will worsen performance instead of improving it, a single part like ksa1381 and its complement will improve performance drastically, what works here is cascode, preferably hawksford cascode. In this case youll end up having a poweramp being a copy of the commercial preamp hawksford designed.
hm... probably I should give up darlington VAS ? it is already instable in Spice, and could be more unstable in reality involving more factors.
Could someone tell me how to improve VAS ? I'm hoping for less compensation instead of heavy compensation.
homemodder, could you tell me more about hawksford cascode ? didn't heard before. Interested in it.
Could someone tell me how to improve VAS ? I'm hoping for less compensation instead of heavy compensation.
homemodder, could you tell me more about hawksford cascode ? didn't heard before. Interested in it.
Hi guys, It seems that the problem is within Cdom (compensation) incorporation. I tried another simpler circuit, then replace the VAS to BC550/560 will make it oscillate. By incorporating compensation, it stable and improved THD.
Seems that BCs are too fast, making those higher frequency easier to pass X.x
Seems that BCs are too fast, making those higher frequency easier to pass X.x
Never use Darlingtons for the VAS. Instead, use so called 'beta enhancement' trannies, that is, connect the collectors of them (Q7 & Q9) to ground.
The main problem with your schematic is that there probably isnt enough current running through the first transistor of the darlington, otherwise it should work although the benefit youd get would be next to nothing and it would only be if you were using bc550/bc560 as both transistors. Those MJE s would make things disasterous. The vas is push pull configuration, with this it is its very important that you use high early voltage transistors with low Cob, go for mpsa42/92 which should be easy to find.
You can linearize the circuit by 150 % if you make use of a beta enhancer before the vas, the improvement is drastic but youll have to do some Hfe matching between NPN/PNP.
You also need a input filter to curb high frequency bandwith as asking big output transistors to conduct at MHZ frequencies will only bring about oscillation. Maybe You can get a sim to show some sort of stability at 1 Khz but now try at 20 Khz, probably mass oscillation. The cascode on the input transistors only make things worse, both linearity and stability, this is a audio amplifier not a RF amplifier.
For the hawksford paper you can go to this website and download paper J10 which explains the concept.
Malcolm Hawksford - Publications
You can linearize the circuit by 150 % if you make use of a beta enhancer before the vas, the improvement is drastic but youll have to do some Hfe matching between NPN/PNP.
You also need a input filter to curb high frequency bandwith as asking big output transistors to conduct at MHZ frequencies will only bring about oscillation. Maybe You can get a sim to show some sort of stability at 1 Khz but now try at 20 Khz, probably mass oscillation. The cascode on the input transistors only make things worse, both linearity and stability, this is a audio amplifier not a RF amplifier.
For the hawksford paper you can go to this website and download paper J10 which explains the concept.
Malcolm Hawksford - Publications
LOL because less words X.x Thanks for your explanation, it feeds me in alot of information. Anyway, why would the cascoded input would worsen this ? I though able to do it fast is good ?
I'm thinking why not just put that BCs to VAS (standalone), which haves high gain and nice figures, only drawback is dissipation but i think have methods to counter that.
With some compensation (local HF VAS feedback), it should work nicely.
The explanation would be a bit long, but you could read up on miller effect, early voltage and look at transistor datasheets at Cob - Vcb curves. Its all about compromises. If you go to the TGM5 thread a member called Elvee posted some sims about this, maybe a full explanation behind it, I cant recall now.
What kind of figures are you getting ??
20 years ago when I played with various versions of this circuit I could get THD at 20Khz figures of .006% at 80 percent of full output using beta enhanced vas. Using highish vas current hamonics beyound 2nd and 3rds are very low.
What kind of figures are you getting ??
20 years ago when I played with various versions of this circuit I could get THD at 20Khz figures of .006% at 80 percent of full output using beta enhanced vas. Using highish vas current hamonics beyound 2nd and 3rds are very low.
hm..... what about both transistor in beta-enhancement VAS using BCs ? is it wise ? FOr clarification, only the first transistor need to have it collector at ground ?
That would be much better than your current version, Drastic drop in THD. You dont need then worry about using the BC as first transistor, this you should run at around 1 to 2 ma but it cannot be done with the main vas transistor. The BC simply cant handle the SOA, how about bd139/bd140.
haha, i have those, but seems they came from different manufacturer and seems characteristic quite different. and if i'm not mistaken, now BD139 and BD140 doesn't have any ft response value in datasheet anymore.
They would be better though as they wont blow up. Those BDs are BC639 and bc640 but put in TO220 package so you can consult those datasheets. However the spec will be a little worse but youre best option here if you cannot get any other transistors.
Ding! Winner!
Playing with this in simulation, it is immediately apparent that it is underbiased and the first zero crossing is what kicks off the oscillation. This is obvious on inspection of the schematic (well, for me it was obvious only in hindsight after simulation.) It needs 4 Vbes of bias minimum, the two diodes plus R15 only provide about 1.4V. Increasing R15 to 470 ohms gets it reasonably biased and the oscillation goes away. Or you could crank up the VAS current. You should still look at the loopgain/phase margin.
- Status
- Not open for further replies.