My first CFP power amp

Jump to Latest
#21
The problem is that your current sources don't have a constant voltage reference, so the current varies as you ramp up the voltage from zero on the variac. If the LTP has insufficient current to operate normally, the output can latch up on the negative rail. You should be able to reproduce this in your simulation by running it at a lower rail voltage.

With 8V rails, the current sources only deliver 345uA and 670uA, and the output is latched up to the negative rail:

P3A-8V.png


With 30V rails, the current sources deliver 1.4mA and 5.5mA, and the amp works as expected:

P3A-30V.png
 
  • Like
Reactions: maxlorenz, indianajo and jacques antoine
#23
hboetius said:
Hi all!
It has ben many years since my last post here, I guess some 20 years, I am now retired an have some more time to my Hobby.

I am working on this CFP, which causes me som troubles.

I would be wery thankful if any of you could give me a hint to solve my little problem.

I have attached files to give you an idea of my project.

Micro-Cap schematic:View attachment 1115526

Kicad schematic:View attachment 1115527

KiCad pcbView attachment 1115528

KiCad 3DView attachment 1115529

Pcb under testView attachment 1115530

Test setup:View attachment 1115533

Here is the problem:
1. When I power up the amp without speaker connected all is fine and dc off set at the output terminals is 70mV. After powering up I can connect the speaker and hear some music.
2. Powering up with the speaker connetcted causes the amp to draw a lot of current, and I have 6V on the terminals.
3. Powering up with a cap in series with the speaker, all is fine and I have music in the speaker.
4. Powering up with an 120 ohm resistor in series with the speaker also does the trick, with les music volume of cause.

I really would appreciate som help, please!

Henrik
Click to expand...
Hi Henrik,

Is there a connection between signal and power ground? From the schematic it looks like signal ground is in the middle between the positive and negative supplies. Looks odd to me.

Best regards,
Tobias
 
#24
Rory Christ has it right. It takes about 800ua through R8 just to turn on Q4. That 800ua through R5 makes a voltage of 16V across R5. So, it takes 16V on the lower rail before Q4 will turn on and overcome the current sink of Q10. When the lower rail is under 16v then Q10 is driving the output to the low rail with an 8 ohm load. If your lower rail can supply about 16v/8ohms or 2amps then the amplifier will snap into operation because Q4 can then turn on and start controlling the output voltage. I don't see an easy fix without adding components. If R8 was removed and a PNP current mirror was added using the current of Q2 then the amplifier may start gracefully since the base current of Q4 would be available at low rail voltage. Also, the offset would be minimum since Q1 and Q2 would have equal currents.
James
 
#25
Dear All
It is quite overwhelming, thank you so much for your thoughts and constructive suggestions.
I have read your responses with great interest and now have an action plan ready for the project and as my work progresses I will post them here.
All the best from Henrik
 
#27
Bigun said:
That’s a nice looking project, good amp. Does R9/10 get warm or hot? If so, that’s a sign of high freq. oscillation.
There appears to be enough compensation caps But a scope will help. Lower frequency (100kHz+) oscillations can be simply from the bench top wiring, feedback from the speaker wiring back to input, even through grounding wires. I had such an issue once whic went away when I moved the wires about.
Click to expand...
Thanks, I will check R9/10
 
#28
BSST said:
Sometimes regulated supplies don't play well with power amps--- a turn-on transient pulls the regulator into current limiting, supply voltage drops, and the systems stalls in that state. I'm not saying this is the case here, but I recall from somewhere in memory an amp that worked OK when the regulator was abandoned.
Click to expand...
I will be aware of that, it sounds reasonable though I have never experienced it.
 
#29
EdGr said:
The high supply current is almost certainly due to oscillation.

The design has a low-impedance path from rail-to-rail at high frequencies through C3, Q4, and C9. I would remove C9 and possibly C4 as well. C3 should be the only compensation capacitor absolutely required.
Ed
Click to expand...
I will try to correct the CCS's, and if the problem is still there I will go for the oscillation with help from my scope and try out your suggestions.
 
#30
Bigun said:
C4 (lead comp) can be problematic, but the C9 helps provide confidence that the CFP is not as likely to be osillating (a similar cap can be installed on the other CFP driver as a try it and see). Shoot-through in the CFP at high frequency is another parasitic behaviour that pulls more supply current but I too put my money on oscillations.
Click to expand...
That is my pov except that I think it is the CCS's, but if not we agre
 
#31
Rory Christ said:
The problem is that your current sources don't have a constant voltage reference, so the current varies as you ramp up the voltage from zero on the variac. If the LTP has insufficient current to operate normally, the output can latch up on the negative rail. You should be able to reproduce this in your simulation by running it at a lower rail voltage.

With 8V rails, the current sources only deliver 345uA and 670uA, and the output is latched up to the negative rail:

View attachment 1116205

With 30V rails, the current sources deliver 1.4mA and 5.5mA, and the amp works as expected:

View attachment 1116206
Click to expand...
Hi Roy
Thank you so much for your effort in simulating. It is so clear to me that this is a problem, just like OldDiy pointed out, but the consequences becomes quite obvious in your simulation.
Why didn't I do this my self from the beginning?
As I mentioned earlier, I have decided to solder an LM7805 TO92 (I have a lot of them) on the back of the pcb, and take it from there.
Once again, thanks.
 
Last edited:
#34
James361 said:
Rory Christ has it right. It takes about 800ua through R8 just to turn on Q4. That 800ua through R5 makes a voltage of 16V across R5. So, it takes 16V on the lower rail before Q4 will turn on and overcome the current sink of Q10. When the lower rail is under 16v then Q10 is driving the output to the low rail with an 8 ohm load. If your lower rail can supply about 16v/8ohms or 2amps then the amplifier will snap into operation because Q4 can then turn on and start controlling the output voltage. I don't see an easy fix without adding components. If R8 was removed and a PNP current mirror was added using the current of Q2 then the amplifier may start gracefully since the base current of Q4 would be available at low rail voltage. Also, the offset would be minimum since Q1 and Q2 would have equal currents.
James
Click to expand...
Thank you for your analysis, I will certainly try to simulate your hint, but I will try to fix the amp with this PCB.
 
#38
The current through Q8 and R16 needs to be twice the current through R1. That means that you need 2 ma through Q8 and R16 and not 1 ma as noted. Q8 current needs to divide evenly into Q5 and Q6.
So R16 needs to be around 360 ohms rather than 700 ohms.
James
 
You must log in or register to reply here.
Top Bottom