Hello, I am new here, I read some guides here while I was also trying to design my amplifier and thought of making an account and have my design be reviewed here.
So, my audio power amplifier is not really mine, I made it while following along Bob Cordell's book on Audio Power Amp. design. Here it is,
It is a 15W amplifier with 8-ohm load. I am having output problems as I cannot seem to meet that 15W requirement. My closed loop gain is 28 and with 570mV input, I should be able to get 15.96Vpk output, enough to reach and go over a bit to the 15W spec requirement but I am getting a lower output voltage. Also, is my +17V -17V supply good or I should go higher? I am somewhat limited to that as I am planning to re-use a 12-0-12V 3A transformer from a previous project. I am also thinking of changing my output transistors (Q12 and Q14) to TIP2955 and TIP3055 as they seem much cheaper, but I do not know how to choose the transistor to be their drivers.
------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
The design is not far from the circuit from Cordell's, I have my CFP output stage to have a quiescent current of about 100mA, I achieved that by experimenting values on the Vbe Multiplier resistors. The 22pF Miller Cap value, I followed how is it calculated from the book. As for the LTP quiescent current it is at 2mA and the VAS stage is 7mA. I do not have much knowledge on how to do the output network (Zobel) I just followed the typical values mentioned in his book.
------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
After that, any suggestions? recommendations? or something I should do?
Attached is also the LTSpice Simulation file already included the models I used in which I would also like to thank and credit Cordell.
So, my audio power amplifier is not really mine, I made it while following along Bob Cordell's book on Audio Power Amp. design. Here it is,
It is a 15W amplifier with 8-ohm load. I am having output problems as I cannot seem to meet that 15W requirement. My closed loop gain is 28 and with 570mV input, I should be able to get 15.96Vpk output, enough to reach and go over a bit to the 15W spec requirement but I am getting a lower output voltage. Also, is my +17V -17V supply good or I should go higher? I am somewhat limited to that as I am planning to re-use a 12-0-12V 3A transformer from a previous project. I am also thinking of changing my output transistors (Q12 and Q14) to TIP2955 and TIP3055 as they seem much cheaper, but I do not know how to choose the transistor to be their drivers.
------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
The design is not far from the circuit from Cordell's, I have my CFP output stage to have a quiescent current of about 100mA, I achieved that by experimenting values on the Vbe Multiplier resistors. The 22pF Miller Cap value, I followed how is it calculated from the book. As for the LTP quiescent current it is at 2mA and the VAS stage is 7mA. I do not have much knowledge on how to do the output network (Zobel) I just followed the typical values mentioned in his book.
------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
After that, any suggestions? recommendations? or something I should do?
Attached is also the LTSpice Simulation file already included the models I used in which I would also like to thank and credit Cordell.
Thanks, I was trying to save money by using a lower rated one instead and use one I already have. But still thanks, I was aiming for 16-0-16 but now that you mentioned 15-0-15 I'll go for that one.
Oh wait, I changed my supply rails to +21 and -21. But the output is still trying to be lower than 15V. I figured there might be something in the circuit that may be attenuating my signal at 20kHz but even when I changed by input signal frequency to 5kHz, the output is still trying to be below 15V.
I then removed my input network and just directly connected V3 to Q1's base, now, the output is not being lowered.
I then removed my input network and just directly connected V3 to Q1's base, now, the output is not being lowered.
Thanks! I was not able to see that, now probing at the base of Q1, there is indeed a significant drop. As for the Zobel's C7, I changed it to 50p, my mistake the one written on the book is 0.050uF, I carelessly used 50p. I'll change it to 47pF.
Maybe a dumb question:
Most amplifiers I've seen so far have the resistors at the input connected as voltage divider like shown in the schematic here. Rarely I've seen R19 omitted and R21 defining the input impedance. Is there any specific reason or advantage to wire the resistors as attenuator?
R20 and R19 are not intended as attenuator. R20 forms an input filter with C4 and R19 is the bias resistor for Q1, for bipolar inputs, a value equal to R17 is preferred to minimize output offset. It only shows that the attenuator is formed as a result.
Right, exactly as explained in Cordell's too. I've also read that C3 and R19 serves as a high pass filter, there he mentioned it is usually configured to below 5Hz cut-off. As for it attenuating the input signal I guess there is no way around it other than raising the supply voltage?
No, that's way to small 🙂 Use 100nF (0.1uF) like most designs.
ah dang, I just realized everyone else was talking in nano range while I was in the pico range 😳. Anyway, thanks!
I believe I really asked a stupid question and figured out the answer meanwhile:
The benefit of wiring the resistors is to maximize the input impedance that the signal source is seeing, thus lowering the load the signal source needs to drive and also minimize the bias resistor value the amplifier is seeing, thus lowering DC offset a bit. The only downside is a bit attenuation.
I illustrated this using a slightly different component arrangement:
As discussed, the attenuation just needs to be compensated by the gain of the amplifier.
The benefit of wiring the resistors is to maximize the input impedance that the signal source is seeing, thus lowering the load the signal source needs to drive and also minimize the bias resistor value the amplifier is seeing, thus lowering DC offset a bit. The only downside is a bit attenuation.
I illustrated this using a slightly different component arrangement:
As discussed, the attenuation just needs to be compensated by the gain of the amplifier.
The difference between the two is miniscule in subjective terms. In practical terms the second version unbalances the DC conditions a little if you are wanting to keep the feedback resistor and input resistance equal to minimise DC offset.
Also any solid state preamp or source will easily drive loads down to 1k or less (much less in practice because of the small voltage swings).
Also any solid state preamp or source will easily drive loads down to 1k or less (much less in practice because of the small voltage swings).
I have a question, how are you measuring the input & output voltage swing in order to calculate your 15 Watt? Start there.
15W should be across the load resistor (R24). I have set the gain of the amplifier, through ratio of R17 and R18, to 28. As such, an input of 0.55Vpeak should be able to give me 15.5Vpeak at the output which is across R24. With a 15.5Vpeak.
15.5Vpeak because using Ohm's law of V = sqrt(P * R) = sqrt(15W * 8ohms) = 10.95Vrms = 15.5Vpeak
In the second configuration shown in post 12 you can balance the input resistances by using a 24k for the input bias and 3k for the series filter. Just means using E96 series values instead of E24.
Assuming all the usual losses.
18 volt rails
you would get in or around 10v rms on output
So could put roughly 10 to 12 watts
into 8 ohms
With 4 ohm load and usual slight drops.
be around 22 to 24 watts.
theoretically should be higher. but in real world
or good model. results are different
Assuming the power supply has enough current.
Usual rule of thumb is transformer VA to be roughly twice
the wanted output wattage for AB amplifier
far as I can tell your transformer 60 to 70 VA / 60 to 70 watt
18 volt rails
you would get in or around 10v rms on output
So could put roughly 10 to 12 watts
into 8 ohms
With 4 ohm load and usual slight drops.
be around 22 to 24 watts.
theoretically should be higher. but in real world
or good model. results are different
Assuming the power supply has enough current.
Usual rule of thumb is transformer VA to be roughly twice
the wanted output wattage for AB amplifier
far as I can tell your transformer 60 to 70 VA / 60 to 70 watt
The reason for asking how you measure the output instead of assuming what it should be is simple. Using a scope you would normally see the peak voltage, using a true RMS voltmeter you should see the RMS voltage and many cheap DMMs would measure the average voltage. It is also true that many DMMs cannot accurately measure at frequencies much higher than 200 Hz. Although your formula may be correct, you will end up with three different conclusions. My rule of thumb when predicting amplifier output I say the peak output voltage swing would normally be 2V less than the DC rail voltage measured and calculate the RMS as 0.7x the peak then apply this to your formula.
You can work the process in reverse to estimate the DC rails required for the amplifier and the transformer output voltage (remember the diode drops in the rectifiers) and then calculate the VA rating of the transformer taking the transformer regulation into account. The rail capacitance also needs to be considered to make sure that the rails don't drop significantly at low frequencies.