I'm developping a new amp and I followed members advices given in a previons post. In my first amp it work perfectly to get rid of the dc offset, but on my second amp (I've change only few things) I still get a DC offset of -0,36 V.
I've written all the voltages on the file attached. If somebody could give me advices to get ride of this dc offset and explain me where it comes from.
(I've tried to match Q5 Q6 gain (188 and 186) , both 560 ohm resistor, change a lot of components but nothing change .... spend a lot of time ...)
Thanks for any advices.
David
I've written all the voltages on the file attached. If somebody could give me advices to get ride of this dc offset and explain me where it comes from.
(I've tried to match Q5 Q6 gain (188 and 186) , both 560 ohm resistor, change a lot of components but nothing change .... spend a lot of time ...)
Thanks for any advices.
David
Attachments
Thanks for your advice,
I've tried without the MOSs (I've taken the NFB from R9) and the offset is still there.
Concerning Q17, I've written a mistake, its emitter voltage is -0,736V and not -0,0736V, so the Vbe junction is about a diode threshold. Furthermore i"ve tried whithout Q17 and R36) and the offset is still there too....
I've tried without the MOSs (I've taken the NFB from R9) and the offset is still there.
Concerning Q17, I've written a mistake, its emitter voltage is -0,736V and not -0,0736V, so the Vbe junction is about a diode threshold. Furthermore i"ve tried whithout Q17 and R36) and the offset is still there too....
Perhaps it is due to input bias current. If it is, then you can minimize it by sizing R18 to be equal to R17||R19 (I can't make out what R17 is supposed to be from the pic you posted). The magnitude of the offset is consistent with what I would expect from the bias currents in that amp, so it seems a likely solution.
Can you please try something? Keep the 8 ohms in (the loudspeaker substitute ;-) and only take out the two MOSFETs. Then please measure all the same voltages again.
Perhaps even try with some Darlington power transistors instead of the MOSFETs - just for testing - and then measure everything again.
Edit: I cannot see the name of the other MOSFET?
Wrong current source config
Your measurements of -0.665 V at Q17 collector and -0.077 V at base means the C-B junction is almost forward biased. Q17 is barely working and it is no minor wonder the circuit stabilizes at all!!
This is because base bias for Q17 is taken at 0V through R38. A better arrangement for the differential pair (Q5-6) emitter current source should be to bias Q17 base from a voltage divider (add a resistor between Q17 base and the negative rail). A decouplig cap betweed 0V and base should be a good idea to minimize influence of negative rail variations on Q17 collector current.
I suggest you change the 12K emitter resistor for Q17 to 6K8 and place a 2K preset (tap and one extreme tied) between Q17 base and the negative rail. Adjust to midpoint before applying power and to zero output offset when powered up.
There is no need to change R19, it should be equal to R18 for there is no DC current through R17 (blocked by C10) so for equal base currents in Q5 and Q6, the voltage drop should be the same with the output at 0V.
Rodolfo
Your measurements of -0.665 V at Q17 collector and -0.077 V at base means the C-B junction is almost forward biased. Q17 is barely working and it is no minor wonder the circuit stabilizes at all!!
This is because base bias for Q17 is taken at 0V through R38. A better arrangement for the differential pair (Q5-6) emitter current source should be to bias Q17 base from a voltage divider (add a resistor between Q17 base and the negative rail). A decouplig cap betweed 0V and base should be a good idea to minimize influence of negative rail variations on Q17 collector current.
I suggest you change the 12K emitter resistor for Q17 to 6K8 and place a 2K preset (tap and one extreme tied) between Q17 base and the negative rail. Adjust to midpoint before applying power and to zero output offset when powered up.
There is no need to change R19, it should be equal to R18 for there is no DC current through R17 (blocked by C10) so for equal base currents in Q5 and Q6, the voltage drop should be the same with the output at 0V.
Rodolfo
Hi !
I am not sure what q17 was intended for. In this configuration, it
acts like some cascode, but can't work as emitters of diffamp
have and idle voltage of -0.6v.
I suggest to skip q17 completely, now it puts trouble to the diffamp.
Also your outputstage is underbiased, means r34 needs to be lowered.
You should use a 1k-pot here. Adjust it until the voltage across
r24 gets above ~20mv. (also check r25)
You could replace the 12k with a 10k and put a 5kpot in series,
then you can adjust dc with this pot.
Also r17/18/19 might be higher, i suggest to increase them factor 4.
Now the c10 is a bit too small, you might have bad bass-response.
And c11 might be smaller, 100nF for example. This avoids the use
of electrolytics.
Mike
I am not sure what q17 was intended for. In this configuration, it
acts like some cascode, but can't work as emitters of diffamp
have and idle voltage of -0.6v.
I suggest to skip q17 completely, now it puts trouble to the diffamp.
Also your outputstage is underbiased, means r34 needs to be lowered.
You should use a 1k-pot here. Adjust it until the voltage across
r24 gets above ~20mv. (also check r25)
You could replace the 12k with a 10k and put a 5kpot in series,
then you can adjust dc with this pot.
Also r17/18/19 might be higher, i suggest to increase them factor 4.
Now the c10 is a bit too small, you might have bad bass-response.
And c11 might be smaller, 100nF for example. This avoids the use
of electrolytics.
Mike
MikeB said:
Mike:
Q17 acts as a current source. A classical differential amplifier has a constant current source feeding the emitters so a change in current by the driven device reflects 100% on the other. A resistor may be substituted (frequently done) but then impedance goes down from infinite (for an ideal current source) to the actual resistor value, lowering to less than 100% the current transfer gain. My recomendations were in line to turn Q17 into the active region and work as intended, a good current source.
Rodolfo
Yes, i'm familiar with currentsources to diffamps, but this way q17
simply passes the current from the 12k to the diffamp, at least tries
to. So without q17 the same current is fed into the diffamp.
For feeding a diffamp a simple resistor is not too bad, but introduces
ripple from psu. The impedance compared to a ccs is no big problem
here, as the voltage across the 12k does not vary much.
The AKSA for example uses a simple resistor, but has a "stabilized"
voltagesupply to the frontend.
Mike
simply passes the current from the 12k to the diffamp, at least tries
to. So without q17 the same current is fed into the diffamp.
For feeding a diffamp a simple resistor is not too bad, but introduces
ripple from psu. The impedance compared to a ccs is no big problem
here, as the voltage across the 12k does not vary much.
The AKSA for example uses a simple resistor, but has a "stabilized"
voltagesupply to the frontend.
Mike
Definitely Bocka's idea is what I'd do.
With the base of Q17 going to 0V through a resistor and the collector loaded by the differential amp, the CB junction is nearly forward-biased as mentioned before.
Certainly changing in a 270ohm emitter resistor and adding the diodes forward-biased to V- and a 12k resistor will make a proper current source.
With the base of Q17 going to 0V through a resistor and the collector loaded by the differential amp, the CB junction is nearly forward-biased as mentioned before.
Certainly changing in a 270ohm emitter resistor and adding the diodes forward-biased to V- and a 12k resistor will make a proper current source.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.