It doesn't seem to work that way in practice, though! 100% negative feedback is being applied to the base of Q8. I've tested it in isolation and sure can't get it to oscillate or anything. In simulation it acts as a unity gain buffer with 2nd harmonic down -70 dB.
Essentially I wanted to experiment with a low noise, low distortion, high bandwidth input stage coupled to a low distortion unity gain output stage. I thought the cascode with an active "gyrator" load might be a good way to accomplish this, and Q8, Q9, and Q10 would act as the unity gain output buffer, along with a current source instead of R15 so it could both source and sink current.
Unfortunately, the problem is that coupling the Q1/Q4 gain stage directly to the output buffer compromises the output swing pretty severely, so an intermediate stage is needed. That's Q5 and Q6, and I am unhappy with the way that has been "shoehorned" in there. I'm trying to think of a better solution.
Obviously the output capacitor is too large. I was experimenting with different sized loads. It's not a final schematic, after all...
t e n
You seem to really really like emitter followers. My slant is EFs are almost never desirable. One of the few situations is when you're driving a really low impedance, and you can use some damping components to compensate for their negative impedance characteristics. Judging from the 100ohm pull-down at the output (no active pulldown), you're not driving a super-low output Z.
Simplify! All those stages upon stages are a source of misery!
Have fun!
I'm finding that to drive the headphones at the level I want to drive them at, I'm going to need some sort of active pulldown. Back to the drawing board.
I'm beginning to discover that, as I've found that by simply changing the type of transistor in the circuit it has become impossible to stabilize the negative feedback.
As has been kindly suggested by some forum users, the original amplifier design was overcomplicated. Taking this advice to heart, I have stripped down the amplifier to its essentials and have indeed found that performance and stability has been improved. Here's the revised circuit:
The amp appears to be stable and simulates nicely with a 64 ohm load and feedback. I still have an urge to add more, though! Any further suggestions would be appreciated.
To start with, you are using a cascode input stage, which will probably have Mhz bandwidth untamed. Then you feed that into an emitter follower with bootstrap feedback to increase the effective collector load to perhaps 100's K ?. Together with the 50 ohm feedback emitter resistor, means that the first stage will have an unfeasably high and probably non linear defined gain, not to mention phase response with frequency.
You then feed the signal into a 2 transistor stage, gain nominally 4, which drives a differential pair with output stage and feedback defining gain = 1. The max swing negative is limited by load R against the 100 r emitter resistor and by the voltage at the collector of Q6, which defines the emitter voltage of the differential pair. Positive swing is limited by load R, Q9 gain and Q7 set current.
To start, you need to define a gain budget and expected levels for each stage. Design so that no stage saturates with expected voltage swing open loop. Only then can you apply the negative feedback loop to define the overall system gain and tweak for desired response with caps etc.
What you have there is far too laboured for any kind of analysis for what happens at the edges. I would start again with a clean sheet of paper and sharp pencil :-)...
Regards,
Chris
Arrgh, nope, still having trouble with this circuit. I'll keep working on it.
It is left as an exercise for the student to roll your own Operational Amplifier section in discrete devices ;-) ...Jim Thompson
--
| James E.Thompson, CTO | mens |
| Analog Innovations, Inc. | et |
| Analog/Mixed-Signal ASIC's and Discrete Systems | manus |
| Phoenix, Arizona 85048 Skype: Contacts Only | |
| Voice:(480)460-2350 Fax: Available upon request | Brass Rat |
| E-mail Icon at http://www.analog-innovations.com | 1962 |
Remember: Once you go over the hill, you pick up speed
Very nice. That is the sort of idea I was going for. I am going to take the advice of the group to heart and pair down the complexity of the circuit to see if I can achieve something that's stable.
That output stage is a twist on that in the MC1554, designed around
1965, by Tom Frederiksen and me ;-)Take Tom's comments to heart ("Intuitive IC Op Amps," by Thomas Frederiksen) at the bottom of page 14, to make your partial OpAmp :-) ...Jim Thompson
--
| James E.Thompson, CTO | mens |
| Analog Innovations, Inc. | et |
| Analog/Mixed-Signal ASIC's and Discrete Systems | manus |
| Phoenix, Arizona 85048 Skype: Contacts Only | |
| Voice:(480)460-2350 Fax: Available upon request | Brass Rat |
| E-mail Icon at http://www.analog-innovations.com | 1962 |
Remember: Once you go over the hill, you pick up speed
Thanks for the advice! I was playing around with something like this:
as an output, it looks like maybe it should work? However, I can't get it to swing symmetrically, it keeps slamming into some limit when going positive. I'm having trouble understanding why.
Eh, I don't think that topology is ever gonna work. :[
For a clue, take a look at the emitter current of Q8.
You're flailing, trying to design with a simulator. Design should be done with pencil and paper, and thinking about what is happening. Then check with a simulator.
I was fortunate, started out designing without benefit of a simulator. Never had my hands on a simulator until I had been successfully designing chips for almost 20 years, with only pencil, and paper... for ten years there were not even calculators... just slip sticks ;-) ...Jim Thompson
--
| James E.Thompson, CTO | mens |
| Analog Innovations, Inc. | et |
| Analog/Mixed-Signal ASIC's and Discrete Systems | manus |
| Phoenix, Arizona 85048 Skype: Contacts Only | |
| Voice:(480)460-2350 Fax: Available upon request | Brass Rat |
| E-mail Icon at http://www.analog-innovations.com | 1962 |
Remember: Once you go over the hill, you pick up speed
Do the math. Say the supply is +15v, and you are driving +10v, positive half cycle into the 100r load.
10v across 100R = 100mA. Assuming a gain of 50 at 100mA for Q9, then you need 2mA at the base, which will be at ~+10.6v. So, 18 - 10.6 / 18k, which is less than 0.5mA. Answer, not enough base drive...Regards,
Chris
I should have seen that. Time to call it a day on this circuit and try again when I'm fresh.
Drinking always clears my mind ;-) ...Jim Thompson
--
| James E.Thompson, CTO | mens |
| Analog Innovations, Inc. | et |
| Analog/Mixed-Signal ASIC's and Discrete Systems | manus |
| Phoenix, Arizona 85048 Skype: Contacts Only | |
| Voice:(480)460-2350 Fax: Available upon request | Brass Rat |
| E-mail Icon at http://www.analog-innovations.com | 1962 |
Remember: Once you go over the hill, you pick up speed
Unfortunately, due to a certain medical condition I lost the ability to tolerate any amount of alcohol a few years back. I haven't had much luck finding any "involuntary sobriety" support groups!
Yup, I've concluded it's back to the drawing board. Trying to cram in too many ideas doesn't make a good circuit.
With a quick view, looks like a nice oscillator. You are trying to induce gain by using a feed back that is only going to assist into making it push further, into oscillation.
You need to work on that feed back idea a bit.. Feed back is to be used to monitor the output so that the input circuit knows when to stop advancing for output correction, your circuit is just making it advance more. Think voltage comparator.
That's my take on it.
Jamie
ElectronDepot website is not affiliated with any of the manufacturers or service providers discussed here. All logos and trade names are the property of their respective owners.