dunno ,in XA25 constellation (read - sans source resistors) pucks are just too beasty , to leave them without constant surveillance** , regarding output offset
**without NFB loop
**without NFB loop
I just did a couple of tests with tube front end. I'm using a PCC88 tube with a CCS loading the plate, operating at 85V and 8mA in anode follower configuration. Found out that I need a low impedance, low distorsion FE to drive the output stage, otherwise I'm getting HF attenuation.
I need to add a cathode follower.
I need to add a cathode follower.
For me sound-wise that would be the last resort option. Rather use a lower Rp valve or an srpp.
Modified the input stage, added quickly a cathode follower, now the frequency response increased to 100Khz(measurement taken at 21Vrms). Distortion wise I've got 0.5% at 21Vrms. Feedback resistor is 4.7Kohm.
Did some listening tests and I find out that I need to carefully choose the GNF resistor, with 4.7K resistor the sound in bass and low midrange was shy, with 10K the sound in lower region had more weight. Did anyone else who built a version of the beast noticed that?
I had the same impression some time ago.
Changing the bias also influences the sound quality, stps of 0.1A are to be heard.
Changing the bias also influences the sound quality, stps of 0.1A are to be heard.
So are you saying that you increased the gain of the amp and by virtue reduced the negative feedback by that amount?
I run 4.7K to the 330 for the feedback. If i changed that to 10K i would have a gain of 29dB and reduce the loop feedback by 6dB. This would definitely reduce the damping factor and increase the base sound but would also increase the DC output drift.
My two cents.
I remember you mentioned before that you have J2 at home, what is listening impression comparing with the beast?
On my amp I perceive a very "clean" and dynamic sound but not harsh, comparing with other amps I have.
I see, what you DC output with 4.7K? Mine is around 80mV with any value of feedback resistor. The FE is capacitor coupled with OS.
I typically only have +-10 mv offset. I'm using a different bias system than you with the thermistors on each puck. So the offset starts about 100 mv cold and then settles in after about 20 minutes.
Good to hear those results. I built some PCBs using thermistors on the pucks, but was not happy with the stability of the bias current. Offset can be handled by negate DC feedback.
Not sure. There could be some "dualing servo" issues with respect to the interactions of the bias circuits. I do not understand the control theory involved to make a decisive conclusion.
Is it posible to do it if the front end is AC coupled?
How much of DC offset is "acceptable"?
Thanks.
I noticed that many of the the lhquam simulations used +-40V rails for the OS. Might someone share the rationale for using these voltages?
Last edited: