use of a faster device does not increase the GAIN through that device.
The extended fT reduces the gain droop.
This allows the gain from other parts of the circuit to extend further because the droop is not as severe.
This may sound like the same thing, but it is handled quite differently inside the open loop gain curve.
One advantage of moving one of the device induced "gain droops" is to move a single pole higher up the frequency band. That leaves one less pole at the original frequency. This can make the amplifier more stable by allowing the gain slope at the unity gain frequency to fall below the critical -9dB/octave. Whereas with the slower device that slope may exceed that -9dB/octave at the unity gain crossing frequency.
Last edited:
Starting this quest many years ago on a given circuit and on given set of semis i noticed that behavior between types of capacitors if used as a miller in a driver or a vas .
In the circuits i built for me i always use silver mica 500v which happens to creates the less problems
I actually tested in a position that original vas miller cap was 100pf with a multilayer cap try to lower it to 82pf things remained fine up to point with a reasonable load ..When the load was tough there was spikes and signs of oscillation.
Lower than that at 68pf things was pretty messy
Used 82 pf silver mica and resulted no problems at the same test conditions as the multilayer ...
That is 5 years ago i need to dig to find the waveforms i will see if this is possible at the weekend
My drivers and VAS are always at 250MHZ and my outputs at 30mhz with no problems
Kind regards
Sakis
In the circuits i built for me i always use silver mica 500v which happens to creates the less problems
I actually tested in a position that original vas miller cap was 100pf with a multilayer cap try to lower it to 82pf things remained fine up to point with a reasonable load ..When the load was tough there was spikes and signs of oscillation.
Lower than that at 68pf things was pretty messy
Used 82 pf silver mica and resulted no problems at the same test conditions as the multilayer ...
That is 5 years ago i need to dig to find the waveforms i will see if this is possible at the weekend
My drivers and VAS are always at 250MHZ and my outputs at 30mhz with no problems
Kind regards
Sakis
Then buy or build more 4ohms dummy loads.
Add on some inductors to emulate L+R loading. (ESP describes a test circuit that requires the L+R loading)
Add on some capacitors to emulate C||R loading. Everyone that is serious about testing their builds, test with a few capacitor values.
Whether we need to go into loop gain and bandwidth to explain results correctly or not, my comment is about a false assumption based on the available HF device gain in a P3a but as surely as night follows day, substituting faster transistors generally, will lead to stability problems. This may often involve compensation, as when a friend fitted a BF469 (yet another obsolete TV video driver) as VAS transistor to his P3a. It "worked fine" as he claimed but the output stage became very hot and fried after about 30 minutes use - just one recent example.
There is also something quite different about Philips/NXP Ft data which you see was often specified at 100 MHz. i.e. unity gain was interpolated from results using a test frequency of 100 MHz. Many semiconductor Ft specifications are estimated with a test frequency of only 1 MHz and this makes comparisons doubtful, though I haven't yet looked at the relationship between the test frequency and the Ft for any specific devices.
There is also something quite different about Philips/NXP Ft data which you see was often specified at 100 MHz. i.e. unity gain was interpolated from results using a test frequency of 100 MHz. Many semiconductor Ft specifications are estimated with a test frequency of only 1 MHz and this makes comparisons doubtful, though I haven't yet looked at the relationship between the test frequency and the Ft for any specific devices.
When I bought multilayer ceramic 100pF NP0 caps the supplier declared them as 100V but now I see that they changed it to 50V. Could that be the cause of parasitic oscillation? The voltage rating of the caps was too low and the rating was wrong?
How good are ordinary 100pF disc ceramic caps rated for 1kV?
How good are ordinary 100pF disc ceramic caps rated for 1kV?
100V rating for cheap disc ceramic caps is rare now and the bargain Chinese sourced parts have only been available in 50V rating for some years now. I can't use them because the voltage is too low. I'm not sure of the failure mode but I don't think they would cause oscillation as long as the signal voltage hadn't caused dielectric breakdown. Use a new one at low levels only and see if oscillation persists. If it does, look elsewhere for the problem. An open cap failure here could certainly cause oscillation, as you would have no compensation cap at all. Measure it if you can, then see if there is any change by simply omitting it (only briefly).
High voltage ceramic disc caps will almost certainly not be NPO, unless very large. Don't use them, as they could be X7R, Y5 types etc and not recommended for audio for reasons often explained here and described by Self, Curl and many others. This paper describes the different applications and available ranges for SMT parts but the principles are the same: http://www.koaspeer.com/pdfs/TN-196R14NPO-Y5V.pdf .
Basically, the audio distortion is significantly higher in types other than NPO or COG, because their capacitance is not constant unless the voltage across it remains fairly constant. Unfortunately, audio is anything but constant. There are other unwanted effects related to piezoelectric properties of ceramics but that's another topic.
High voltage ceramic disc caps will almost certainly not be NPO, unless very large. Don't use them, as they could be X7R, Y5 types etc and not recommended for audio for reasons often explained here and described by Self, Curl and many others. This paper describes the different applications and available ranges for SMT parts but the principles are the same: http://www.koaspeer.com/pdfs/TN-196R14NPO-Y5V.pdf .
Basically, the audio distortion is significantly higher in types other than NPO or COG, because their capacitance is not constant unless the voltage across it remains fairly constant. Unfortunately, audio is anything but constant. There are other unwanted effects related to piezoelectric properties of ceramics but that's another topic.
Last edited:
The colour of the cap is no guarantee of any particular grade but they are usually 1-3 kV rated and are probably X7R to reach it. They seem to be widely available, perhaps because they're cheap and small. It's a pity that 100V NPO disc ceramics aren't around any more but I'm surprised kit suppliers don't supply MLCCs instead, as they're not expensive and better suited.
My LJM L12-2 modules came from Along1986090 and are also fitted with blue 2kV ceramics just branded "B". I have changed one channel to MLCC type and can't hear much difference but that's hardly a test. I still hope to find time and bench space to check the results properly with a QA400 USB spectrum analyser.
My LJM L12-2 modules came from Along1986090 and are also fitted with blue 2kV ceramics just branded "B". I have changed one channel to MLCC type and can't hear much difference but that's hardly a test. I still hope to find time and bench space to check the results properly with a QA400 USB spectrum analyser.
Hi, today I start populating an original ESP P3a board that I buy from RE some years ago.
While inserting the resistors in their places on the board, I noted that R18, R19 and R17 on one channel, and R18, R19 in the other channel, are not figured in the original schematic.
I suspect these resistor places are related to the optional attachment of a "Sound Inpairment Monitor" circuit, and can safely left unpopulated, Am I right?
While inserting the resistors in their places on the board, I noted that R18, R19 and R17 on one channel, and R18, R19 in the other channel, are not figured in the original schematic.
I suspect these resistor places are related to the optional attachment of a "Sound Inpairment Monitor" circuit, and can safely left unpopulated, Am I right?
If you bought from ESP, you would also have received notes with the PCB, explaining assembly and the SIM feature. The design has since been revised, eliminating the SIM and discussion on the P3a project webpage. However, it is still shown as a separate project, P57.
The SIM adaptor was not much more than a pair of 10k resistors, one from each base of Q1,Q2 to external test connection pins. Yes, you can assemble it just as you see in the pic of the earlier version, where the resistors and pins are also omitted. As the dual amplifier board is also intended to be cut in 2 for separate boards, I can't imagine why R17 is on one board and not the other. As it is not shown on the latest version, perhaps you should check the circuit by inspecting the traces, which will soon reveal where it is connected and whether it is necessary at all.
60-80W Power Amplifier
The SIM adaptor was not much more than a pair of 10k resistors, one from each base of Q1,Q2 to external test connection pins. Yes, you can assemble it just as you see in the pic of the earlier version, where the resistors and pins are also omitted. As the dual amplifier board is also intended to be cut in 2 for separate boards, I can't imagine why R17 is on one board and not the other. As it is not shown on the latest version, perhaps you should check the circuit by inspecting the traces, which will soon reveal where it is connected and whether it is necessary at all.
60-80W Power Amplifier
Yes, thanks, R17 is only in one channel and go to R5, the other end connect to nothing, so I will ignore it all. BTW. at the time I received a key to enter the construction section of Rod site, but it was long lost/forgotten, due to many time and PCs passed under the bridge..
Attachments
Last edited:
today i built a p3a and while testing, the bias was suddenly jumping from zero to very high (full brightness on the the 60W bulb). i suspected oscillation, and started removing and testing each transistor and reflow of joints, etc.
still same thing kept on happening, i replaced the o/p transistors too and just before testing again, just thought to check the resistance of the bias pot. the pot resistance, after turning the screw, kept jumping to zero after some value.. (Rod specifically warns on keeping the resistance of the pot to highest value before applying power)
replaced the bad pot..now everything is fine, sounds too good and extremely quiet too!
just thought to share here..
reg
Prasi
still same thing kept on happening, i replaced the o/p transistors too and just before testing again, just thought to check the resistance of the bias pot. the pot resistance, after turning the screw, kept jumping to zero after some value.. (Rod specifically warns on keeping the resistance of the pot to highest value before applying power)
replaced the bad pot..now everything is fine, sounds too good and extremely quiet too!
just thought to share here..
reg
Prasi
Just for my reference, may I know what type of o/p transistors, drivers, and the vas did you use for your build p3a?
Happy holidays!!!
Vinski80
Hi, I'll use components as per original BOM in ESP site, to the letter, except the OPT's, since I found original Toshiba complementary pairs
Happy Holidays to You too
hi guys
i just built P3A its sound is amazing and clear i just want to ask for two things:
1 - the voltage drop across each 0.33R resisotr is not equal one is higher than the other with about 20mV is this due to matching of drivers (BD139/BD140) where they are not matched?
2 - consider it a newbie question .. the power output i think is lower that 90w according to the specs at dr.elliot's site my supply si 42v and speaker is 8R what could be that?
Regards All
i just built P3A its sound is amazing and clear i just want to ask for two things:
1 - the voltage drop across each 0.33R resisotr is not equal one is higher than the other with about 20mV is this due to matching of drivers (BD139/BD140) where they are not matched?
2 - consider it a newbie question .. the power output i think is lower that 90w according to the specs at dr.elliot's site my supply si 42v and speaker is 8R what could be that?
Regards All
Hi Vin Diesel. First, more questions to be clear.
1 - Do you mean that one resistor measures 20mV and the other is different by some small amount or do you mean that the voltages are different by 20mV? The resistors will likely only be +/- 5% tolerance and this could make a difference.
2 - How are you measuring signal voltage input and power output? Note that a speaker is not a fixed load and cannot be used as a reference for measurements. Neither can the sound level produced. You need a large, non-inductive fixed 8 ohm resistor of at least 100W rating in order to measure power output. BTW, you should be aware that 43V power supplies are the absolute maximum allowable with suitably rated genuine semis and should not be used with less than 8 ohm loads. Many budget soeakers marked with 8Ω are now actually 6 ohms or less - beware the nominal impedance ratings used to cheat on sensitivity specs.
Another factor can be the power supply. Most transformers are capable of their rated voltage at their maximum power rating but at light loading, the secondary AC voltage will rise according to the regulation of the transformer used. In other words, the power supply may measure +/-43V when the amplifier is idle, but at full power output with a sinewave signal just before clipping, this could dip very briefly to maybe +/- 39V and that produces significantly less power.
Unless you are trying to prove something, such as parts quality and construction, there is not much point in power measurement. It is already defined by the design, input voltage, output voltage, load impedance and power supply voltages.
1 - Do you mean that one resistor measures 20mV and the other is different by some small amount or do you mean that the voltages are different by 20mV? The resistors will likely only be +/- 5% tolerance and this could make a difference.
2 - How are you measuring signal voltage input and power output? Note that a speaker is not a fixed load and cannot be used as a reference for measurements. Neither can the sound level produced. You need a large, non-inductive fixed 8 ohm resistor of at least 100W rating in order to measure power output. BTW, you should be aware that 43V power supplies are the absolute maximum allowable with suitably rated genuine semis and should not be used with less than 8 ohm loads. Many budget soeakers marked with 8Ω are now actually 6 ohms or less - beware the nominal impedance ratings used to cheat on sensitivity specs.
Another factor can be the power supply. Most transformers are capable of their rated voltage at their maximum power rating but at light loading, the secondary AC voltage will rise according to the regulation of the transformer used. In other words, the power supply may measure +/-43V when the amplifier is idle, but at full power output with a sinewave signal just before clipping, this could dip very briefly to maybe +/- 39V and that produces significantly less power.
Unless you are trying to prove something, such as parts quality and construction, there is not much point in power measurement. It is already defined by the design, input voltage, output voltage, load impedance and power supply voltages.
in a single pair output stage with the speaker disconnected the current through the upper emitter resistor is equal to the current in the lower emitter resistor plus the current in the NFB resistor.
If you ignore the NFB current, then the upper current is equal to the lower current.
If you measure 20mV difference, then that tells you you did not match your emitter resistors before assembly.
Take out all your emitter resistors and solder them in series. Pass a fixed and constant current through them and measure the voltage drop across each resistor. Select resistors (aim for better than 0.5%) that have close matches for Vdrop. Use them as pairs/quads in each amplifier.