API feedback circuit Q's w/ lower stepup input Xformer

[JW]

Hi folks,

I'm still learning, but I gather from reading that all the parts in regard to feedback circuit and the volume control are intricately related (in an API 312)

I have an input transformer from a PM1000 that I wired at 150ohm:2.7Kohm, and am also using it's 1:1 600 ohm output transformers.

Since this is a lower stepup, I dialed in gain to the same as a 312 I built from the usual CAPI/EA parts by adjusting the resistor on the way to the volume pot. In the 312, this is 200 ohm. I needed to adjust this to 37ohm to get the gain I wanted. A quick listen didn't reveal audible oscillating. 

But my question is, to what extent should I adjust the other parts to maintain the same lo pass and hi pass filters? Or is this possible?

I can run a sweep but I'm limited by my converters in regards to seeing above and below 20Hz-20K.

My parts currently are:
feedback: 120pF/20K ohm
series resistor: 37ohm
series electrolytic: 470uF
volume pot: 22K

Am I correct to think as you lower the series resistor, the -3dB upper knee of the lpf goes  down? So, I need to raise it back up by adjusting the feedback cap? But up or down? Yamaha uses 47pF here. I can't see where it is anyway by testing.

So I guess I need some math hand holding.

 

[abbey road d enfer]

Since this is a lower stepup, I dialed in gain to the same as a 312 I built from the usual CAPI/EA parts by adjusting the resistor on the way to the volume pot. In the 312, this is 200 ohm. I needed to adjust this to 37ohm to get the gain I wanted. 

Remember that by doing this, you change the overall gain range from ca. 40dB to about 50dB. The max gain will be the same beut the min gain will be lower. Not a concern but good to remember.

But my question is, to what extent should I adjust the other parts to maintain the same lo pass and hi pass filters? Or is this possible?

In the 312, the HF cut-off point is defined by the NFB parts, but only at max gain. At min gain the -3dB point is governed by the opamp and the xfmrs.

Am I correct to think as you lower the series resistor, the -3dB upper knee of the lpf goes  down?

Yes.

So, I need to raise it back up by adjusting the feedback cap? But up or down?

Up. In RC circuits, when one element increases, the other must decrease, for a constant product.

Yamaha uses 47pF here.

I very much doubt it. microFarad maybe, but not pico.

 

[JW]

Thanks Abbey,

I guess I need help understanding how the low pass filter is related to the series resistor.

In this thread https://groupdiy.com/index.php?topic=26948.0 Bo Hansen says:

"The cap across the feedback resistor forms a lowpass filter.
The corner frequency where the signal is dropped by -3dB is calculated by 1 / (2xPi() x Rfb x Cfb) with R in Ohm and C in Farad.
If you fill in the alternative parts couple from Bo's quote (4,7k/470pF, 10k/220pF, 22k/100pF...) your formula looks like 1/(2*Pi()* 4700 * 0,000000000470) resulting in about 72kHz.
This 72kHz is the frequency at wich your signal has dropped -3dB (voltage gain of 0,707). You most probably wont hear the 72kHz, but this filter only has a 6dB/oct. slope, so your interest might by, what is the loss at 20kHz.
Voltage gain at 20kHz for this is 1 / (squareroot (1+ (20000Hz / Lpf-frequency)^2)), giving 0,963. This loss in dB is log10(0,963)*20, giving -0,32dB. The introduced phase error at 20kHz for this Lpf is at Arctan(20000Hz / Lpf-freqency) * 180 / Pi(), giving 15,5°. (phase at 72kHz is 45°)."

So, I'm plugging my numbers in for the hi pass filter, in the same fashion according to the equation above 1 / (2xPi() x Rfb x Cfb)

So, 1/ (2 x Pi x (22K+37=22037R) x .470f) =.000015Hz? This is the high pass -3dB knee?

But how is this related to the lo pass? It looks like I'm plugging in the same parts as on the API schematic  for the feedback circuit since that what I have right now and getting around 66Khz.

 

[abbey road d enfer]

So, I'm plugging my numbers in for the hi pass filter, in the same fashion according to the equation above 1 / (2xPi() x Rfb x Cfb)

This formula is for the low-pass filter. For the LPF, it's the same formula with the 37 ohm resistor and the 470uFcapacitor.

So, 1/ (2 x Pi x (22K+37=22037R) x .470f) =.000015Hz?

You got your math wrong. 470uF is 0.000 470 F. And lower case "f" is for femto (1/1000 000 000 000 000).

 

[JW]

Okay, so for the hi pass filter, I'm calculating 9.15Hz. So this is the worst case scenario with the pot maxed out right? So 37ohms to ground.

According to Harpo in that thread, API's hi pass -3dB knee is at 3.4Hz. So if I'm worried about phase issues above that, then I should at least adjust something to get it down to API's hi pass knee.

But do I not have any issues with the low pass filter if I have the same parts in the feedback circuit? (120pF/20k)

 

[abbey road d enfer]

According to Harpo in that thread, API's hi pass -3dB knee is at 3.4Hz. So if I'm worried about phase issues above that, then I should at least adjust something to get it down to API's hi pass knee.

Phase issues with filters is a favorite of snake oil merchants.
Phase audibility at 3.4Hz? Who's joking?
Ther are two points to consider. When the LF response is governed by a dominant zero, in other words when the LF response is essentially 1st-order, the cut-off frequency determines the lowest usable frequency. A cut-off (-3dB) frequency of 3.4Hz results is -0.15dB at 20Hz. Whether it's acceptable is up to the designer; it's certainly acceptable in view of the traditional 20Hz-20kHz paradigm, but may be considered poor for purist HiFi aficionados who swear by DC-to-light response.
In the case of the API, the designer has certainlly considered it adequate, in view of the fact that capacitors loose value with age, so even if they lose 50% of their value, the LF response is still respectable.
Another aspect is distortion. Electrolytic caps are known to introduce distortion when submitted to a significant voltage across them. Since a capacitor's impedance  increases with decreasing frequency, distortion is predominant at low frequencies. In order to minimize the voltage across, the value must be as high as possible.
In conclusion, the value of the capacitor there is not governed by illusory phase issues, but by simple frequency response and distortion concerns.

But do I not have any issues with the low pass filter if I have the same parts in the feedback circuit? (120pF/20k)

That's correct, but remember that the characteristic frequency of the RC network is not the actual circuit's response, particularly at lower gain.