The document isn't written in a very comprehensible way (at least for me). I could probably explain it much easier. Things like this trick you into thinking that electronics is much more complicated than it really is. 
How interested are you in amplifier design, what makes it tick and tick well? Just curious.
- keantoken
How interested are you in amplifier design, what makes it tick and tick well? Just curious.
- keantoken
MJL21193 said:Sometimes the problem is a good thing. I have learned something here - you are correct, it is a current mirror.
Discussion makes us dig deeper and in that we learn.
PS: But that means Dan was right - there is a CM and AndrewT didn't see it??
Hi MJL21193,
I agree. It forces one to get out the text books and reread in a different light. My references always assume "matched" transitors with identical parameters to mathmatically prove the ciruit behaves as a current mirror.
I realise one of the transistors in a current mirror is connected as a diode, but in the circuit in discussion will the parameters of transitor X and diode Y be close enough, can you consider them matched?
regards
MJL21193 said:Sometimes the problem is a good thing. I have learned something here - you are correct, it is a current mirror.
Discussion makes us dig deeper and in that we learn.
PS: But that means Dan was right - there is a CM and AndrewT didn't see it??
This is rare. Perhaps it is a success from his many efforts on increasing my understanding? You both helped me so much. I thank you both.
On this thread, I have recently conveyed information without understanding it. That's inappropriate, and I sincerely apologize.
What I saw:
Simply put, the roles of both items are the same. One of them is documented as a current mirror. Both are visually similar, in the same role, and in the same spot in the circuit. I know for sure that one of them works and I believe that the other works, and they have the same role.
This is why I think that both are a current mirror. IE: If it walks like a duck then. . . it isn't a schoolbus.
Greg Erskine said:
Hi Greg,
I think the reasoning behind this mirror is to increase gain in the LTP without emphasis on sharing. Or at least that's what I've gotten from the discussion and the paper Dan posted.
I would like someone to jump in and explain if there is a particular advantage to this as opposed to the typical 2 T version.
keantoken said:The document isn't written in a very comprehensible way (at least for me). I could probably explain it much easier. Things like this trick you into thinking that electronics is much more complicated than it really is.
How interested are you in amplifier design, what makes it tick and tick well? Just curious.
- keantoken
Interested in amplifier design? Oh absolutely not. However, I am very interested in what it can do.
MJL21193 said:
Intuition tells me. . . less ringing.
This is because, in the actual amplifier product from the 70's the passive input circuit (before the amp) that leads into it, is absolutely awful and would cause most power amps to misbehave (probably ringing midbass).
Something was responsible for the amplifier's success, and whatever it was overcame the hindrances.
In other words, this thing was fed on the audio version of poison, its nfb was compromised with a baxandall, and the amp shown still pulled off a quality performance. Hmm. . . Somewhere, there is a reason for that, and I think that you may have spotted it. I hope so. That would be cool.
Well , the final design you choose and how simple you want it
is your choice as you will be purchasing the parts.
I was in the same stage a few months ago and opted to
use a CM on the advice of many (including D. self).
for $1.00 the whole input stage of FA1 is a deal, and not
too complex.
Yes , the mirror gives 5-6db more gain allowing for a non-ef'ed
1381 (works for a mje 350 with less Re degeneration)
I've actually tried all 3 "flavors" of CM's (attachment at end)
and the board I'll make for FA1 will have pads for all 3.
OS
Attachments
homemodder said:
I mean the mismatched LTP currents. If an LTP is driven too hard while it's balanced, it will spit out 3rd order harmonics. But if unbalanced, 2nd order harmonics predominate. According to what many DIY'ers say, 2nd order harmonic content is more pleasant than 3rd order.
still interesting though.
How will you decide which one to stick with?
- keantoken
keantoken said:
Hi kean,
My simulation shows a big drop in THD when using this mirror at high output, mainly in the third and higher harmonics. Interesting indeed.
It behaves differently at low output, with 3rd staying the same and the second reduced over the typical Widlar. Still, lower THD.
MJL21193 said:
Cool. I was simming too, and I came up with this.
.0001%THD at 1KHz, 1V pk-pk.
I had a funny thought: note the different LTP devices. The 2n5769 has 1/3 the beta of the other transistor, so even though the currents are different, the base currents are the same. I wonder what would be the advantages of using different transistors in the LTP? Different transistors for NFB and input. Hmmm.
The circuit has a problem with oscillating though.
- keantoken
Attachments
Duhhh, I have been using this diode Cm for ages..
I just didn't realize it because it was being used at the
input instead of where I use it...
It does have the ability to "absorb" high order transients
and when I use it in my FA2 VAS , it kills the saturation
of the stage as well as TOTALLY eliminating H3/5.
OS
I just didn't realize it because it was being used at the
input instead of where I use it...
It does have the ability to "absorb" high order transients
and when I use it in my FA2 VAS , it kills the saturation
of the stage as well as TOTALLY eliminating H3/5.
OS
Attachments
Hi keantoken,
Matched differential pairs will always yield lower distortion as long as the circuit is designed well. Also, matching the transistors in any current mirrors will help to minimize distortion. With complimentary diff pairs, matching the beta of all four will give you lower DC offset and lowed THD.
What simulators need is a random number generator to shift the transistor parameters around with each device selected. Then you will get more realistic results. Your "matched pairs" would still have different parameters, just not as widely spread as the other transistors of the same part number. Also shift the emitter resistor values a bit. This will represent another factor that real life imposes.
-Chris
The differential pair subtracts the difference in signal between the input and output. A mismatched pair will increase distortion and DC offset voltage because the signals are not handled equally. There will be less distortion removed from the signal through the amp.
Matched differential pairs will always yield lower distortion as long as the circuit is designed well. Also, matching the transistors in any current mirrors will help to minimize distortion. With complimentary diff pairs, matching the beta of all four will give you lower DC offset and lowed THD.
What simulators need is a random number generator to shift the transistor parameters around with each device selected. Then you will get more realistic results. Your "matched pairs" would still have different parameters, just not as widely spread as the other transistors of the same part number. Also shift the emitter resistor values a bit. This will represent another factor that real life imposes.
-Chris
Thank you for being helpful Daniel.
However I'm sure those amps Ostripper posted still simulate quite well without any additions. As far as lower distortion at high power levels, I might say that it is worth using this new current mirror.
It all depends on whether you think this discussion will help you reach your goal. We could move the discussion to a new thread, if you want.
I just don't want to be wasting your time.
- keantoken
However I'm sure those amps Ostripper posted still simulate quite well without any additions. As far as lower distortion at high power levels, I might say that it is worth using this new current mirror.
It all depends on whether you think this discussion will help you reach your goal. We could move the discussion to a new thread, if you want.
I just don't want to be wasting your time.
- keantoken
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.