I have a question involving current gain bandwidth product and performance of the LTP. Current gain bandwidth product in a BJT is affected by how much current is biased. Let's say you operate BC846c at 10uA, current gain bandwidth product is somewhere around 10MHz, where as at 1mA it is at 100MHz. Will having a larger current gain bandwidth product produce less distortion at full swing? If so, this would suggest that the larger bias is more desirable.
What is a good percentage of current needed to drive VAS to be taken from leg of LTP? My VAS Ib is about 500nA. I was experimenting with LTP current just for the sake of experimenting to see how little LTP current I could get away with. Why, I don't really know. But it works well with 12uA per leg.
Would 1mA per leg produce better results at full swing and not introduce extra noise? If not, does it make any difference? Anyone done any experiments refering to this?
What is a good percentage of current needed to drive VAS to be taken from leg of LTP? My VAS Ib is about 500nA. I was experimenting with LTP current just for the sake of experimenting to see how little LTP current I could get away with. Why, I don't really know. But it works well with 12uA per leg.
Would 1mA per leg produce better results at full swing and not introduce extra noise? If not, does it make any difference? Anyone done any experiments refering to this?
Most people have their main pole around the VAS, the LTP has to push this cap. I usually run about 3mA in the LTP for 5mA~10mA in the VAS. This generally gives adequate speed.
I run a full comp diff input so the input bias isn't too bad, only the difference between the Ns and Ps vs all of it having to come up the base resistor in a single LTP input.
My next project is going to be a minimalist single LTP pair, VAS, both with resistors to a high voltage supply for passive CCS action. I may have to bootstrap the input resistor like the Pass A40 to keep the input impedance high.
I run a full comp diff input so the input bias isn't too bad, only the difference between the Ns and Ps vs all of it having to come up the base resistor in a single LTP input.
My next project is going to be a minimalist single LTP pair, VAS, both with resistors to a high voltage supply for passive CCS action. I may have to bootstrap the input resistor like the Pass A40 to keep the input impedance high.
Hi djk,
I am using BC846C and BC856C for diff and current mirror in the LTP. The VAS stage is FJV992E, and it's current source, FJV1845E. (all SOT-23) This VAS drives an intermediate EF stage using KSC2784E and KSA1174E. Then a set of BD139/140's drives the output devices. Bias current for the VAS is around 120uA right now, and seems to work very well. I had to use a little feedback capacitance to kill the occilations when driving a RCL load, or 4 way speaker cabinet, and because of high impeadences a 3pf capacitor in parallel with the 1MOhm negative feedback resistor put a stop to that. I also noticed that drastic temperature change on the VAS transistor drifted the DC output(as expected) by +/- 50-100mV. That change represents the extreams, hot and cold, as far as operating temp. Still, I just mounted a small resistor to it as a "heater" and DC drift is no more than +/- 10mV even when playing. That just made my day.
:
Initially I was reluctant to use another EF stage in the feedback loop because of phase and occillation problems, but this did not seem to be a problem.
Using your figures in my situation, LTP bias would be like 30uA or 15uA per leg?? or am I missing something here...(very likely). Does the lessor phase gain bandwidth product(due to lower current operations) produce more distortion at full swing? or compensate less?
If the current in the LTP leg is 1mA, or ~2000 times the required current to drive the VAS, would this cause any ill effects? or better?
So far my "made from scratch" circuit performs quite well as is, but this is part of the never ending quest to learn and improve.
Thanx, Chris
I am using BC846C and BC856C for diff and current mirror in the LTP. The VAS stage is FJV992E, and it's current source, FJV1845E. (all SOT-23) This VAS drives an intermediate EF stage using KSC2784E and KSA1174E. Then a set of BD139/140's drives the output devices. Bias current for the VAS is around 120uA right now, and seems to work very well. I had to use a little feedback capacitance to kill the occilations when driving a RCL load, or 4 way speaker cabinet, and because of high impeadences a 3pf capacitor in parallel with the 1MOhm negative feedback resistor put a stop to that. I also noticed that drastic temperature change on the VAS transistor drifted the DC output(as expected) by +/- 50-100mV. That change represents the extreams, hot and cold, as far as operating temp. Still, I just mounted a small resistor to it as a "heater" and DC drift is no more than +/- 10mV even when playing. That just made my day.
:Initially I was reluctant to use another EF stage in the feedback loop because of phase and occillation problems, but this did not seem to be a problem.
Using your figures in my situation, LTP bias would be like 30uA or 15uA per leg?? or am I missing something here...(very likely).
Does the lessor phase gain bandwidth product(due to lower current operations) produce more distortion at full swing? or compensate less? If the current in the LTP leg is 1mA, or ~2000 times the required current to drive the VAS, would this cause any ill effects? or better?
So far my "made from scratch" circuit performs quite well as is, but this is part of the never ending quest to learn and improve.
Thanx, Chris
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.