Another amp

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
Hi all.

I first began to play with the complementary LTP (or Rush pair) after rejecting he usual LTP in a PSU circuit. I found that the Rush pair was less prone to oscillation.

At first I was turned off because I hadn't seen the Rush pair used anywhere other than the 4QDtec site. But I eventually began to experiment again.

This circuit is the result. It simulates amazingly well, with .000475% THD at 20KHz, with 1Vpk-pk into 100 ohms. Whether this figure is reasonable, I do not know, that is why I want you to tell me. I don't have the parts to build this, unfortunately.

NOTE: for LTSpice users, the .four statement will be inaccurate at 20KHz unless you set the maximum time step to 100n.

At any rate, even if the circuit is a flaw, hopefully it is enough to demonstrate what I have learned in about three years time.

Special note should be taken of the transistors used: the 2n5769/5771 are fast switching transistors with low Cjo. and low beta.

Everything here was my idea as much as is possible, I did not post it anywhere until I had a decently working version.

thank you for your replies,
- keantoken
 

Attachments

  • comltpamp+power5.png
    comltpamp+power5.png
    39 KB · Views: 499
Q22 and Q21 are a CCS, and in this configuration it works opposite to a current mirror.

R14 decreases open-loop gain. Without it the circuit oscillates.

A gain configuration would be added at Q4's base, and the circuit works with a gain of 10 as I am experimenting with now. It has worked at gains of 20 and 40.

Because of Q4's inevitable base current, a resistor here causes DC offset problems. If you insert a gain configuration, use R2 to correct for DC offset. There should also be a 22p cap bypassing the larger gain resistor so it doesn't oscillate.

It usually starts oscillating when the output cannot cause enough voltage change to accommodate the input pair. When the voltage equalizes, it will stop oscillating. If more driving power is needed, output device numbers can be increased, and R3 and R4 can be lowered to increase bias current. The circuit doesn't work very well in class B. :\

In the version below, I have deleted C6, which turned out to only make the amp oscillate. I have done some other things, too. It is now capable of putting 1V pk-pk into 8 ohms. Image iz zipped since I can't upload images larger than 1000x1000.

- keantoken
 

Attachments

  • comltpamp+power6.zip
    41.3 KB · Views: 57
Disabled Account
Joined 2006
Hi Keantoken

Interesting circuit, I wouldnt comment on the circuit as a whole as Ive not looked at the rush pair but I can comment on a small part of the circuit that I regurlarly use and is vertually unknown here and is IMHO very good. Its those current source you have at the input. Using a baxandall pair in the current source gives a lot of improvement compaired to normal sources, the impedance of the circuit is much higher and it has a tiny output capacitance, great for PSRR. R11 and R18 I find is off the the mark though, it should be approximately 10x the emitter resistor for best performance, slightly higher to compensate for the internal emitter resistance, in your case maybe 10 ohms. Im not sure what C1 is doing in this circuit and I feel its not needed. Take a look at the last page of Bora is a nice guy thread, I have a schematic of the circuit I use as current sources. I sometimes use them in current mirrors as well. ;)
 
Thanks for schemo Homemodder.

I thinks it's funny that many good circuits are often overlooked, sometimes even just ignored for what I can see to be no good reason. Just weird.

I left C1 in because I found that the circuit would oscillate without it. Maybe I should try taking it out again.

At any rate I think I tried the 10xRe thing because I read it in a paper but I got funny results... I will have to try it again. :)

Thanks,
- keantoken
 
Kean,

Nice circuit, like the way you use the two anti-phase outputs from the Rush cascode (Christopher Rush, 1964, Englishman) to drive the complementary voltage amp. Could be used very nicely in a quasi-comp circuit too.

A lot of good circuits are overlooked because they are not understood, and because they fall outside the usual textbooks. There is fad and fashion in electronics, like everything.

The center degeneration resistor of a Rush is incredibly important to stability.

Thank you for sharing this circuit,

Hugh
 
Thank you so much AKSA, It's a nice feeling knowing that maybe I know something about electronics after all. EDIT: this sentence could have been taken horribly wrong if I had not corrected it...

I never read any textbooks fully. If I did It was mainly at the schematics and ideas in them. I was just too lazy to read any.

And all I do is goof around on the sim all day. :(
Not enough money to put something together.

I do think some of my circuits have been overlooked because they were weird. I don't stick to the conventional way. I attack new ideas with vigor (in the sim, anyway), because I want to see how they perform. It takes some brainwork to wrap my head around the concepts, but I follow through even if my brain starts to dry up.

I do like the idea of using the 5771/5769 as the output because, according to the models, these transistors are well matched.

Thank you,
- keantoken
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.