Hi John,
Thanks, that is probably in the right ballpark.
Now, of course, we need to understand the resistance of the circuit surrounding the 0.001uF. Consider the following (hopefully extreme) case. Suppose the generator source impedance is 600 ohms and that the analyzer input impedance is 600 ohms. That is 1200 ohms against 0.001uF. What is the corner frequency of that combination compared to the fundamental frequency of 5kHz? That comes out to about 13kHz as the pole frequency of the HPF, well above 5kHz. With the fundamental lying on a +6dB/octave slope, I would think the THD results would be skewed.
Anyway, hopefully what I described is an extreme case that is well different from the experimental arrangement.
I think that the results would be more reliable if the HPF corner frequency was well below 5kHz.
Alternatively, one could run the experiment with a complementary LPF before or after the pcb HPF, with the complementary LPF implemented with high quality capacitors. Then the net frequency response through the experiment would be flat, reducing frequency response skew effects.
Even with that, it is still desirable to know how much attenuation is being caused by the pcb HPF capacitance at 5kHz, since a greater role played by the capacitance impedance will allow nonlinearities in the capacitance to have a greater effect.
Cheers,
Bob
The generator output of the Panasonic is 600 ohms. The analyzer input looks to be 100k ohms.
From an online calculator FR4 has a dielectric constant of 4.2.
a 12"x 8" 1/16 thick double sided copper clad FR4 is about 1.45nF.
High pass 1091kHz Fc.
I think the rule of thumb is 10 times the fc so 10910kHz. 5kHz would be on the right side of the slope. Worst case attenuation is then 3dB.
From an online calculator FR4 has a dielectric constant of 4.2.
a 12"x 8" 1/16 thick double sided copper clad FR4 is about 1.45nF.
High pass 1091kHz Fc.
I think the rule of thumb is 10 times the fc so 10910kHz. 5kHz would be on the right side of the slope. Worst case attenuation is then 3dB.
Last edited:
I dont do this stuff in an office/lab. I am at my home lab. I DO live in the middle of no fu*king-where. Not in a city nor even near one. Nor a neighbor close by. And, I do take care to not have anything on nearby in my lab. Lighting is via LEDs. Just the instrument being used is On. BTW - I can measure to -150dBv here.... if I had a source that good. There were some ground issues but that has been taken care of.
THx-RNMarsh
Last edited:
See #4715 ??
These pcb are just double-sided boards which i had stuffed away. I do not have a wide range of pcb to try and compare at this time. I was just curious so I grabbed a couple pcb and ran some tests.
Based on these preliminary tests.... I would not use either for SOTA designs at ULD levels.
-RM
Last edited:
You are drawing a conclusion based on dubious (if not straight flawed) results. I am not surprised, I've seen this MO here before.
And it has NOTHING directly to do with distortion. The same old stupid, wrong old ground. I've built -130db circiuits on Radio Shack bakelite breadboards.
Last edited:
Guys I just stuck that up there because it was the only reference to PCB hook I could find.
I haven't read the paper and wasn't pointing to anything.
I looked for the original Tek paper but couldn't find it.
I haven't read the paper and wasn't pointing to anything.
I looked for the original Tek paper but couldn't find it.
I build an oscillator on FR4. On a good day it goes to -138dB.
Richard got it down to to -140dB.
I think my wife gets that!
I'll tell her I learned more about it on DIYAudio. I'm sure she will be impressed.😡
Right. Sure.
Do your own pcb test and measurements. Have at it.
Might be an issue for ULDist circuits.... might not. You certainly cant say for sure.
I have never found better results were obtained with worse DF.
And, the thd is higher by some 'dubious' amount.
Test it and then tell us your results. If I can find my DA tester, I'll do that test also. But again, in my experience, higher DA also leads to poorer results.
THx-RNMarsh
Last edited:
It might be worth more realistically to test thin trace width with ground plane on opposite side. In a more controlled test. Isolated in an enclosure with shielded leads. Would coax capacitance be a problem?
I can do it but I don't necessarily have the needed amplitude in a low distortion source.
is 2.5Vrms enough?
Neither can you. Certainly not from anything presented here. You'll never let go of the myth that DA means distortion so be it. Same as DBT listening, too much to lose.
Last edited:
I'm not really up on the arguments here. But I'm just rereading Bob Pease's write up on DA or soakage as he like to call it. He begins by saying the DA is pretty linear thing. So if it's a linear thing then how does it cause distortion? It's a problem when resetting integrators but there is no reset in an audio test sine waveform signal.
It's a problem for Tektronix because they require really fast rise time with no overshoot at all.
The Shibasoku distortion analyzer is capacitor coupled between the input amplifier and first notch filter. This capacitor is an BP electrolytic that's as old as 30 years depending on the vintage of 725. The 725 resolves harmonics well below -140dBV.
So I don't get it.
And you are right DA and PCB "hook" have little or nothing to do with the distortion of your oscillators.
Audio Precision oscillators.
Samuel's website has measurements that show the successive reduction of oscillator distortion from the Tek 505 to the Audio Precision System 1, the SYS 2722 and finally the APx555.
The System 1 source is quite similar to the Tek 505 except for the computer control requirements.
They are both 5534 based State Variable.
Anyone worked out how the distortion was reduced?
Is it mainly improvements in the leveler subsystem?
The Sys 1 has a unity gain inverter compared to the Tek 505 where the gain is closer to (-)2, not sure the reason for that.
I can't find the schematics of the 2722 or APx555.
I recall claims they are AD797 based rather than 5534, which makes sense but I haven't confirmed this.
It would be helpful not to reinvent the wheel, so anyone have any information on the later products?
David
Samuel's website has measurements that show the successive reduction of oscillator distortion from the Tek 505 to the Audio Precision System 1, the SYS 2722 and finally the APx555.
The System 1 source is quite similar to the Tek 505 except for the computer control requirements.
They are both 5534 based State Variable.
Anyone worked out how the distortion was reduced?
Is it mainly improvements in the leveler subsystem?
The Sys 1 has a unity gain inverter compared to the Tek 505 where the gain is closer to (-)2, not sure the reason for that.
I can't find the schematics of the 2722 or APx555.
I recall claims they are AD797 based rather than 5534, which makes sense but I haven't confirmed this.
It would be helpful not to reinvent the wheel, so anyone have any information on the later products?
David
My curiosity was first raised by D.Self's comment that DA is theoretically linear and so shouldn't increase distortion.
So what property of a bad dielectric does cause distortion?
Best wishes
David