look up the data posted by Eva from nearly a decade ago.
Unfortunately many of his pics were not attached and links have died.
We still have not learned.
Ceramic cpacitors are best avoided in signal path (except NP0 type which is not available in compact size) as they distort orders of magnitude more than electrolytic caps. You can use fine polyprop instead. You can also increase the value of the inner 24k9 resistors (DC connected to input nodes) to e.g. 100k which will allow for even lower cap values. Yet don't go too low (e.g.couple of uF) to avoid increased noise.
Last edited:
The break frequency of 500ohm and 220uF is 1.45Hz. It is over a decade lower than 20Hz. Distortion is more prominent around the break frequency.
Reactance of 220uF at 20Hz is j36ohm. So the two in series is 500+j36 and the magnitude of the series combination is sqrt{5002+362}=501.3ohm. The contribution of the j36 is very small. It is not as if you have 500+36=536!!!
You still have to look at the % deviation of the capacitance.
D.Self has published data showing the distortion rising above the noise floor way above the F-3dB of the RC.
That does not exclude the increase in distortion that is below the noise floor.
We can hear artifacts well below the noise floor. Distortion may be some of the audible artifacts that our brain/ear can pull out from the signal.
Then we have a problem using 220uF bipolar electrolytics then!!! Then we are back to square one!!! So what are we talking about in this 5 pages if that's not good enough? Back to metal polypro?
Putting an input high pass does not exactly solve the problem, you still need another cap that have the same issue.
I don't follow what you try to say. How do you put the input filter? Do you mean just a RC high pass filter at the input to attenuate the low frequency?
It lowers LF eposure by <6dB (90ms vs 140ms), not that it completely eliminates it.
Yes, RC passive single pole filtering at the input to set the passband of the amplifier.
Then you have less signal across the NFB capacitor.
Low signal already ensures low distortion.
Less signal ensures even lower distortion.
Thanks. I want to verify with you that's what you mean. Attached is the opamp with input RC high pass that has break frequency of 6.3Hz using 1uF and 25K resistor. The closed loop path has the 1.3K resistor and 220uF cap that gives the break frequency of 0.56Hz. This is one decade lower than the input HP filter.
It is easy to get a small metalized 1uF polypropylene cap for the input HP filter.
Thanks
Attachments
Add RF attenuator at the PCB input.
Add a grounding resistor before the DC blocking cap on the input.
remove the 2k from the two CCS in the LTP tails.
Use another 4148, or use green led (green = ON/OK)
but red is possibly less noisy and lower dynamic impedance.
Very Leach Lo Tim !
You could adopt his CCS since you already have the Zeners and save about ten components.
Add a grounding resistor before the DC blocking cap on the input.
remove the 2k from the two CCS in the LTP tails.
Use another 4148, or use green led (green = ON/OK)
but red is possibly less noisy and lower dynamic impedance.
Very Leach Lo Tim !
You could adopt his CCS since you already have the Zeners and save about ten components.
Why a LED instead of 2K resistor to set the current? Is it for temperature compensation?
I looked a Leach Lo Tim. It's not quite the same. I use current mirror on both side of the LTP. Also, I use darlington VAS to increase input impedance. In fact, IPS and VAS of Leach is very similar to my Acurus. The major difference is Leach uses 3EF output.
I wonder why amps sound so different even if the circuit is so similar. Leach is well respected, my Acurus is only a so so amp!!! Is the 3EF that important?
I updated the schematic and finished the layout. Here is the updated version with reference designators.
I am reading Self's distortion of power amp paper. Looks like I did a few things right. I did not realize imbalance of the LTP IPS cause distortion. I put in the 300 ohm emitter resistor to linearize the transistor, that actually takes care of the imbalance to a great extend. Also the EF buffering of the VAS supposed to help.
Thanks
Attachments
Last edited:
look at the components Q24, R25, D69 & R28.
The current through D69 and R28 set up a voltage.
The current through Q24 base to emitter and R25 set up a voltage.
These two voltages MUST be EQUAL.
i.e. Vfd69 + Vr28 = Vbeq24 + Vr25
For constant current through the LTP, Q21 & Q22 the current through R25 must be held constant. This requires that Vr25 is held constant.
Vbeq24 is nearly constant, in that it hardly varies over a very wide range of Ie.
For CCS operation Vbeq24 + Vr25 should be held as near constant as possible.
Your Vr28 defeats this constant voltage requirement for CCS operation.
You must remove R28 and replace it with a near constant voltage device, i.e. a device that holds near constant voltage as current through D69 varies.
Replace R28 with a diode
or
replace R28 & D69 with a red, or green, LED.
OR
copy the Leach arrangement for CCS action using the Zeners you already have in the schematic.
The current through D69 and R28 set up a voltage.
The current through Q24 base to emitter and R25 set up a voltage.
These two voltages MUST be EQUAL.
i.e. Vfd69 + Vr28 = Vbeq24 + Vr25
For constant current through the LTP, Q21 & Q22 the current through R25 must be held constant. This requires that Vr25 is held constant.
Vbeq24 is nearly constant, in that it hardly varies over a very wide range of Ie.
For CCS operation Vbeq24 + Vr25 should be held as near constant as possible.
Your Vr28 defeats this constant voltage requirement for CCS operation.
You must remove R28 and replace it with a near constant voltage device, i.e. a device that holds near constant voltage as current through D69 varies.
Replace R28 with a diode
or
replace R28 & D69 with a red, or green, LED.
OR
copy the Leach arrangement for CCS action using the Zeners you already have in the schematic.
Last edited:
- Status
- Not open for further replies.