I wonder if anyone has any suggestions on this problem?
I have built a pair of these on one PCB, with seperate supplies.
Both seem to suffer from oscillation, mainly on -ve half cycle
when presented with a load, loudspeaker or resistor.
Supply leads are 4" or less, I have tried extra decoupling of supplies on the board,
base stopper resistors on the output pair. increased C6, etc etc.
The transistors are MPSA42's, MJE15031 VAS, MJE15034/5
drivers and MJL4281A/4302A outputs.
Increasing the bias current solves the problem for a given drive
level, but increasing drive brings the problem back.
I think what I'm saying is that it's fine in class "A", but not in
class "B", and I don't have a big enough room for an 80 watt
class "A" heatsink!
Help!!
Thankyou,
Clive
(Forgot to say, these amps are built on Rod's excellent PCB.
The heatsinks are earthed and the supply is +- 30v)
I have built a pair of these on one PCB, with seperate supplies.
Both seem to suffer from oscillation, mainly on -ve half cycle
when presented with a load, loudspeaker or resistor.
Supply leads are 4" or less, I have tried extra decoupling of supplies on the board,
base stopper resistors on the output pair. increased C6, etc etc.
The transistors are MPSA42's, MJE15031 VAS, MJE15034/5
drivers and MJL4281A/4302A outputs.
Increasing the bias current solves the problem for a given drive
level, but increasing drive brings the problem back.
I think what I'm saying is that it's fine in class "A", but not in
class "B", and I don't have a big enough room for an 80 watt
class "A" heatsink!
Help!!
Thankyou,
Clive
(Forgot to say, these amps are built on Rod's excellent PCB.
The heatsinks are earthed and the supply is +- 30v)
The extra .1uf goes between collector and emitter of Q9.
My board is REV-A.
I have since noticed that when approaching clipping with
square wave drive, (probably irrelevant!) there is some
rubbish appearing on the bottom of the waveform.
I seem to have fixed this by changing the VAS from MJE15031
to BD140.
I just wonder if the OnSemi devices are a bit fast for this amp??
My board is REV-A.
I have since noticed that when approaching clipping with
square wave drive, (probably irrelevant!) there is some
rubbish appearing on the bottom of the waveform.
I seem to have fixed this by changing the VAS from MJE15031
to BD140.
I just wonder if the OnSemi devices are a bit fast for this amp??
clive299087 said:
Rod's website shows an MJE15035 for the VAS stage. Specwise, the 035 has half the current rating of the 031, and is a shade faster in fact at 100mA of Ic. I'd guess that the 035 has a smaller die and has less capacitance.
This may translate to requiring a larger miller capacitance (C4) on the amp to keep it stable if an MJE15031 is used as a VAS.
Cheers!
JojoD818 said:
Increasing the feedback cap will lower the amp's f3 for that portion, which with the values that Rod uses will be at 1.59Hz - pretty much as low as needed; ditto with the bootstrap capacitor.
The f3 of 10Hz claimed for the amp is mostly from the input capacitor and the impedance seen driving the first transistor...
Cheers!
ps: I think the non-polar capacitors for the feedback (and the input coupling too) is a good idea - i.e. film-type capacitors. However replacing them with inexpensive electrolytic non-polars, such as those used for tweeters etc may not be good. Back-to-back standard electrolytics are pretty much the same thing... May be better to use back-to-back electros and biasing them.
clem_o said:
No, no, no... not those caps used in tweeters please.
I have 4 pieces of 470uf blackgates that I would like to tie togother to make a non-polar, and a 220uf Pana FC which is intended for the bootstrap. However, original values for both caps is 100uf only though.
I'm also pondering on Wima MKS4 4.7uf for the input cap and other bypassing (0.1uf) duties. For C2, C4, C6 I plan to use polystyrene caps. But since I will be ordering these parts, I am thinking if I should order 100uf FC for the feedback and bootstrap cap or just use the blackgate and pana fc that I already have.
Hi JojoD818,
I haven't had the chance of trying out the P3A for myself, but based on the circuit topology there are limits to how 'good' this amp can get. Certainly putting in good parts will "up" its performance, but there's a point of diminishing returns as you begin to reach the potential of the topology itself.
If I were you I'd go ahead and order the 100uF Pana FCs, give a listen. Then switch over to the fabled Blackgates and give it a try as well. The trick here of course is, don't cut the leads of the capacitors, just tack-solder them into place with the long leads.
Then, perhaps you'll wind up pulling out the blackgates and putting them into some other 'more advanced' project in the future...
Cheers!
I haven't had the chance of trying out the P3A for myself, but based on the circuit topology there are limits to how 'good' this amp can get. Certainly putting in good parts will "up" its performance, but there's a point of diminishing returns as you begin to reach the potential of the topology itself.
If I were you I'd go ahead and order the 100uF Pana FCs, give a listen. Then switch over to the fabled Blackgates and give it a try as well. The trick here of course is, don't cut the leads of the capacitors, just tack-solder them into place with the long leads.
Then, perhaps you'll wind up pulling out the blackgates and putting them into some other 'more advanced' project in the future...
Cheers!
Yes, increasing C4 Miller cap to 150pf does solve the problem,
with MJE15035 or 15031, BUT my square waves aren't square anymore (lots of slope on leading and trailing edges of waveform
as compared to the input)
Changing the feedback and bootstrap cap values/types makes
no difference to my amp at all.
It seems to be better with a BD140 in the VAS.
Thanks for all the ideas!
with MJE15035 or 15031, BUT my square waves aren't square anymore (lots of slope on leading and trailing edges of waveform
as compared to the input)
Changing the feedback and bootstrap cap values/types makes
no difference to my amp at all.
It seems to be better with a BD140 in the VAS.
Thanks for all the ideas!
Goes to show that there are 'optimal' and suboptimal choices for parts at a particular position in an amp!
I wonder if a small amount of capacitance in parallel with the FB resistor would help in stabilizing the amp... typical values 10pF to 47pF. But, that would still affect the slew rate of the amp.
Cheers!
I wonder if a small amount of capacitance in parallel with the FB resistor would help in stabilizing the amp... typical values 10pF to 47pF. But, that would still affect the slew rate of the amp.
Cheers!
Thanks for the suggestions clem_o.
Fitting a 10pf across the feedback resistor certainly
does stop that remaining bit of oscillation without
the square wave degradation that increasing the
Miller cap gave me.
However, using a BD140 or MJE350 (rather than the
MJE15035 specified) as VAS seems to achieve the
same effect, so I think I'll stick with that for now.
Fitting a 10pf across the feedback resistor certainly
does stop that remaining bit of oscillation without
the square wave degradation that increasing the
Miller cap gave me.
However, using a BD140 or MJE350 (rather than the
MJE15035 specified) as VAS seems to achieve the
same effect, so I think I'll stick with that for now.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.