The phase shift will start a decade in frequency below the -3dB point 0.707 of the input signal at 15.9 KHz.
Phase shift will also arise inside the circuit and a phase margin of 45 degrees will be needed as well in the interests of stability.
The value of C2 suggested is for low impedance input sources. If a volume or balance control set up is included at the input that changes the situation
Apart from the zobel network there is no coil to stop out of phase signals entering the rear barn door. The output leads could be twisted to add some inductance however that can be a hit and miss.
The output terminal has an input function where signals are received from the load and from mother earth as an input and additionally an output for back emf from earth. If a 0.22 Ohm resistor is put in series with the output where this is in common series with the driver emitters.
This will to increase emitter feedback to vary the base voltages of the drivers. These will receive the same level of signal behind the buffer resistor.
There is a small level of non-linear resistance in emitter junctions. In series with a linear resistor the voltage drop across combined resistances is more linear overall. In effect the non linear element is swamped out.
In simulations this makes for stable square wave results with capacitors in parallel with speaker loads. It also allows reductions in capacitors C4 and C6 with the 30 MHz output transistors.These dont need the same values than might be required for 4 MHz outputs.
Common emitter amplifiers with collector to base capacitors multiply the set value of these capacitors and these increase the load and time required for Q1 to drive Q4 which slows that process down reducing slew rate performance.
Last edited:
Is it necessary to insulate the legs of the transistors from the ferrite bead?
No need at all. They are not conductive. Proposed ferrite bead will touch emitter and collector pins at BD transistors with no unwanted effects.
Hi, When you purchased the P3A board from Rod you will have been given details on how to obtain the P3A Assembly Instructions. It will be useful to carefully read and follow the section on Testing. This might help you get the amplifier working without all the previous mishaps. There is also some good advice on the selection of components.
Good question, why do others who have built this amplifier, which also works in Spice tests have better success without fitting ferrite beads around transistor emitters where this is effectively at the output terminal where larger currents prevail than in series in the input or output of a line stage where permeable issues in ferrite materials are miniscule. If these are overloaded by the level of power amplifier driver output currents THD will be seriously degraded.
If an output coil is required the universal preference is for an air core which has no permeability.
If an output coil is required the universal preference is for an air core which has no permeability.
Last edited:
I followed, the build guide and testing procedure to the letter.
We'll see how I get on in a couple of weeks with the amplifier in a more optimized implementation.
I suspect that this will clear up the noise on the sine waves but not the overshoot on the square wave.
Well if you say so. I really think your CRO and probes are very suspect. Maybe you should try and win one of these Rigol Oscilloscopes. That would make life much easier.
Quote from post 25 "The amp played music normally when connected to my test speakers. It even made my test speakers sound surprisingly good.
It was only when I started testing on the scope that the issues started." For 'scope tests into a dummy load you do this up close on a bench. In life however speakers are connected to the amplifier via cables which are part of the circuit.
The early P3 designs included an output coil this one does not. I think Rod Elliott has a mindset of keeping thing simple and saw that the inductance in speaker cables could give the same result and so this was not included for P3A. One can look up specifications for speaker cable to work out values.
It was only when I started testing on the scope that the issues started." For 'scope tests into a dummy load you do this up close on a bench. In life however speakers are connected to the amplifier via cables which are part of the circuit.
The early P3 designs included an output coil this one does not. I think Rod Elliott has a mindset of keeping thing simple and saw that the inductance in speaker cables could give the same result and so this was not included for P3A. One can look up specifications for speaker cable to work out values.
I think I made a post about stuff blowing up and acting up when my scope probe touched something.
Was advised to add a component to the end of the proble and to not use it in 1x mode.
In the end just gave up on testing with the scope. And am now working out how to use REW or ARTA to get noise and THD values scopes esp cheap ones like my Hantek are pointless in such situations even the FFT was a total flop.
Mine is clone of the P3A made by a local company with slightly different values. I have a post discussing the changes in values. In the end I built one channel as per Rods values and one as per the VASP values.
Still need to test both of them in REW or arta.
Was advised to add a component to the end of the proble and to not use it in 1x mode.
In the end just gave up on testing with the scope. And am now working out how to use REW or ARTA to get noise and THD values scopes esp cheap ones like my Hantek are pointless in such situations even the FFT was a total flop.
Mine is clone of the P3A made by a local company with slightly different values. I have a post discussing the changes in values. In the end I built one channel as per Rods values and one as per the VASP values.
Still need to test both of them in REW or arta.
If Rod Elliott intended that builders would run 'scope tests as part of the set up procedure the complexity and costs for a prospective builders would be put off by such a requirement. The costs of purchasing the test equipment would outweigh the cost of building the project.
If the dc offset and standing current can be set without any problems that should be good enough. Supra cable I use is a low inductance type probably better in that respect than a good many other brands.
The inductance of a length of 2.5metres will be 0.375uH and sufficient for stable results as I have simulated. If one feels a must need to run bench tests on square waves then twisting the internal lengths of speaker cable may give a similar level of inductance to the exterior cable length.
If the dc offset and standing current can be set without any problems that should be good enough. Supra cable I use is a low inductance type probably better in that respect than a good many other brands.
The inductance of a length of 2.5metres will be 0.375uH and sufficient for stable results as I have simulated. If one feels a must need to run bench tests on square waves then twisting the internal lengths of speaker cable may give a similar level of inductance to the exterior cable length.
Well you are correct, assuming that some care is taken during construction and testing of the amplifier. That includes sourcing components from a reliable vendor. A sniffer might be useful for those concerned about oscillation. hifisonix developed a simple sniffer circuit. It can be used to detect large amplitude high frequency oscillation on the output. It is also quite good at detecting oscillations in the ether.
For me the main issue with building this amplifier is the sourcing of suitable output devices at a reasonable cost. The offerings from DigiKey and Mouser are becoming few.
For me the main issue with building this amplifier is the sourcing of suitable output devices at a reasonable cost. The offerings from DigiKey and Mouser are becoming few.
One can test transistors to check these are working properly by measuring the collector to base and emitter to base diode junctions. The positive voltage comes out of the black lead of your multi-meter. The other way to sense instability is by temperature. When things are hot or cold these produce air flows which can be detected by a moistened finger. This also is a good way to check for airtight seals or leaks in loudspeaker boxes.
Anyway back in post 382 I suggested including a 0.22 Ohm series resistor at the amplifier output. I took some trouble to give reasons for this but nobody took this seriously. Note changes in capacitor C4 and C6 in the second attachment and the lack of notches in the top of the square wave at node 14.
Since square waves are not representative of music these notches might be inconsequential in practice however these bear some resemblance to the grass Sadface scope tests, that he is hoping to nullify.
Anyway back in post 382 I suggested including a 0.22 Ohm series resistor at the amplifier output. I took some trouble to give reasons for this but nobody took this seriously. Note changes in capacitor C4 and C6 in the second attachment and the lack of notches in the top of the square wave at node 14.
Since square waves are not representative of music these notches might be inconsequential in practice however these bear some resemblance to the grass Sadface scope tests, that he is hoping to nullify.
The file is attached as the link to Cordell models is shown anyone can download Spice and run the simulation. Rod Elliott has given a list of suitable output transistors with varying unity gain specs and a one size fits all set of stability capacitors.
Sadface opted for the high spec devices which allows other capacitor values to be used. I looked at reducing the value of C4 to take advantage of these specs. When C4 is reduced the slew rate of the Vas transistor can be increased since load on the base of Q4 will be reduced. Q6 base has a similar load which causes a lag response for the bootstrap collector load which is a helps the main Vas however the improved unity gain of 30MHz devices allows a single dominant pole at the Vas to suffice. A suitable low pass filter is needed at the amplifier input to ensure the signal is within the slew rate capacity of the Vas - otherwise the result will be slew rate limiting and spikes on square wave tests. The attachments for slew rate are without the load having parallel capacitance.
You can change the BC550/BC560 transistors in the simulation. I have a good supply of these but they are better used in lower power amplifiers so BC546/BC556B would be a better choice here.
Sadface opted for the high spec devices which allows other capacitor values to be used. I looked at reducing the value of C4 to take advantage of these specs. When C4 is reduced the slew rate of the Vas transistor can be increased since load on the base of Q4 will be reduced. Q6 base has a similar load which causes a lag response for the bootstrap collector load which is a helps the main Vas however the improved unity gain of 30MHz devices allows a single dominant pole at the Vas to suffice. A suitable low pass filter is needed at the amplifier input to ensure the signal is within the slew rate capacity of the Vas - otherwise the result will be slew rate limiting and spikes on square wave tests. The attachments for slew rate are without the load having parallel capacitance.
You can change the BC550/BC560 transistors in the simulation. I have a good supply of these but they are better used in lower power amplifiers so BC546/BC556B would be a better choice here.
Attachments
Last edited:
The output issue Aditya is talking about is that Im not seeing the 70-80 watts output under load.
As per the design it says you should get 70 watts output with 35 Volt DC rails.
Im not seeing 70 watts.
My Phone outputs a 1v pp signal but when hooked up to the amp this drops to 0.4 v pp.
What me and aditya discussed is this issue.
My take is I just need to up my input signal via a pre amp if required.
I need to re-conduct these tests using my K5 Pro DAC instead of my phone. As the input into the p3a.
I have a wood stock amp. Where my output rail voltage very close to the transformer rail voltage i.e. if I use a 18v rail. I see 16 Volts RMS out on the scope.
On the P3A I only see 9 volts RMS output even with a 35v rail.
The good news is output remains the same weather I use Rods values or VASPs values. For the components.
As per the design it says you should get 70 watts output with 35 Volt DC rails.
Im not seeing 70 watts.
My Phone outputs a 1v pp signal but when hooked up to the amp this drops to 0.4 v pp.
What me and aditya discussed is this issue.
My take is I just need to up my input signal via a pre amp if required.
I need to re-conduct these tests using my K5 Pro DAC instead of my phone. As the input into the p3a.
I have a wood stock amp. Where my output rail voltage very close to the transformer rail voltage i.e. if I use a 18v rail. I see 16 Volts RMS out on the scope.
On the P3A I only see 9 volts RMS output even with a 35v rail.
The good news is output remains the same weather I use Rods values or VASPs values. For the components.