Leach opamp

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
Hi,
I agree with all suggestions on degeneration.
Add a hi ohm resistor to input side of c2 to ground to maintain zero DC on input terminal.
Consider a preset // to new emitter degen as bias adjust.
The opamp could drive full output into 500r, consider reducing the the NFB values, 2k2 might be better but retain input bias balance.
Why have you gone for a gain of only *2 (+6db), it is little more than a buffer.
The SE output dissipates 360mW quiescent and 720mW peak, consider a 1 to 1.5Watt output transistor pair with better gain and frequency. A clip on heatsink can be used to reduce case temp. Or even a sziklai pair of 600mW biased for equal quiescent current.
The R4&5 feed could come from ground and could be a ccsource.
Remember decoupling and adopt power amp topology to separate dirty & clean grounds until star earth.
Lots of room for experimentation, good luck.
regards Andrew T.
 
Mr Evil said:

Looks good. The degeneration on Q8/Q6 and Q5 is maybe a bit heavy, and some degeneration on Q1/Q2 might be a good idea. I also don't see any frequency compensation. Is it stable without it? Even if it seems to be, bear it in mind that things are often less stable in real life than in simulations.

Hi,

The degeneration on Q6/Q8 has been fixed, also I added some
for Q1,Q2....can I compensate for the DC on output using
the pot??

AndrewT said:
Add a hi ohm resistor to input side of c2 to ground to maintain zero DC on input terminal.
Consider a preset // to new emitter degen as bias adjust.
The opamp could drive full output into 500r, consider reducing the the NFB values, 2k2 might be better but retain input bias balance.
Why have you gone for a gain of only *2 (+6db), it is little more than a buffer.
The SE output dissipates 360mW quiescent and 720mW peak, consider a 1 to 1.5Watt output transistor pair with better gain and frequency. A clip on heatsink can be used to reduce case temp. Or even a sziklai pair of 600mW biased for equal quiescent current.
The R4&5 feed could come from ground and could be a ccsource.
Remember decoupling and adopt power amp topology to separate dirty & clean grounds until star earth.
Lots of room for experimentation, good luck.
regards Andrew T.

Hi,

About adding a resistor from input of C2 to ground..I don't
understand exactly what you mean...

The output stage bias has been reduced a bit for 20ma...

about the 6 db gain...well I never said that it's a buffer...;)
The NFB resistors are just for simulations, the design
should be a general purpose opamp..

I agree about the power dissapation in the output transistors,
BD139 can handle about 10 watts, with heatsinks, of course,
Is it ok?

sziklai pair... I'm sorry, I don't know what this is..

R4&5 could be a ccs, but this would complicate the design..

About a preset for bias adj. can i replace R11 with a pot??
(in the new schematic)

About the current mirror...what should the dc on the
collector of Q8 be?? 'cause as the degeneration on Q5
is higher the dc on Q8's collector is lower, does it matters??

Thanks all...

Here is the new design...is it better?
 

Attachments

  • 11.jpg
    11.jpg
    96 KB · Views: 1,200
Hi,
do not replace r11 with a pot, instead add a pot in parallel (a la Pass) or leave space for two pcb pins and add resistor in parallel to r11.
The Iq through Q5 depends on LED voltage, once this is fixed then the Q8 collector voltage is Iq5*R8 + VbeQ5. Yes D3 controls Q8 collector.
IMO do not rely on the pot alone for the LTP degeneration, try using a pair of fixed resistors in // to the pot. Connect resistors from the wiper to each emitter of Q1&2.
IMO stay with TO92 or Eline transistors for the whole opamp circuit. With careful choice of currents this should still be able to drive 600r from SE output with a small reserve. To126 or bigger is power overkill and costs you bandwidth and gain and possibly reduced phase margin.
Compound follower pair (CFP) is another name for the Sziklai.
Finally that hi ohm resistor; connect from the left end of C2 to the bottom of R2, about 1 or 2 meg would do but anything higher than 100k is ok. It stops static disharge on input plugs if your opamp becomes an input stage and a similar res on the output cap of the preamp is recommended.
regards Andrew T.
 
Thanks for your comments and tips Andrew,
I really appreciate this.


BTW: I'v noticed (in simulations) that if the impedance on the
opamp inputs are different by a factor of more than ~10, I get no output or gain...
I know this should effect dc balance, but gain?...:confused:


Udi.
 
Thanks, Now I realize it is PUSH-PULL....

About discrete opamps, I'v read Nelson pass' article and
the one in Doug Self website...but none of these mentions the
transistors' bias considerations....

How can I calculate the dc bias of the LTP as a function of
the source impedance, or the bias at the VAS stage, etc..?

I mean, I know that the heavier the bias on the output stage
more it can drive lower impedances, and lower bias
on the input means higher source impedance, but what
is the relationship between that stages??

BTW: should I check dc offset of an opamp with both inputs
shorted to ground or just floating?



Thanks, help is appreciated..

Udi.
 
Hi,

I've build the discrete opamp below to see how does it
sound, so I made a very simple split PS using only half wave rectifier (a diode followed by a capacitor) for each polarity..

I connected the opamp in a non-inverting configuration with a gain of 40 to a mic and...NO NOISE!! NO HUM!! NOTHING...
quiet as a fish....so I'm a bit confused!! :scratch:
The supply is only half rectified so it must have a lot
of noise..and discrete opamps should have lower PSRR that
monotholics...:scratch:

So how come it's so quiet???!!!
Has anyone have an explanation for this?


Udi.


BTW: It also sounds very good!!
 

Attachments

  • 1.jpg
    1.jpg
    98.3 KB · Views: 838
Originally posted by AndrewT
Hi,
why the different values of second LTP load 1k8 & 1k6?
What is the LED voltage? 470r might be a little low for 2mA.
regards Andrew T.

Originally posted by Udip
Hi Andrew,
...when I ran simulation, I found that if both values are equal
than the transistors are not balanced....

the current is different in the two O/P transitor.
the lower one also "carries" the bias current (through the 1k6 res)
of the upper one. to keep the second LTP balanced we "cheat"
a little by providing the 1k8 res.
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.