Are you using this on stage or for hifi in your home? If the latter, then why the gobs of watt? Anyway... Your speakers are 10 dB more efficient than mine, so you'll probably want less than 100 uV of "stuff" on the amp output.
Thanks for asking. I was thinking of the residual hum on the output of the amp. I can hear 1 mV, 120 Hz from my listening position (2 m from the speakers). In my amp, I have about 300 uV of noise on the amp output and I do hear a faint hiss with my ear within, say, 50 cm of the speaker.
As your speakers are 10 dB (~3x) more efficient than mine, divide my numbers by three to get total system performance similar to mine.
B+ ripple (and other injection mechanisms of hum) create IMD components at +/-60, +/-120, and +/-180 Hz from the signal frequency. Plus many other IMD components. Note that in this context, any harmonics of the incoming signal should be treated as signal. It'll look like the plot attached to Post #6 from the "Parts Box 300B" thread. It sounds fuzzy or muddy to me. Never mind the really annoying grrrrrrrrrrrrrr sound of the ripple during quiet passages in the music. Ripple is not a nice sine wave. It makes rather rude noises when it feeds through to the speaker. Once the supply ripple has been cleaned up, the spectrum looks like the one in Post #12 from the Parts Box 300B thread. Note that there's still some ripple components left (60, 120, 180 Hz) as well as IMD around 1 kHz. I bet those originate from the signal source. Could be a layout or grounding issue as well.
Just say no to ripple. 🙂
~Tom
Home stereo
I plan to have the 100TL dissipate about 80 watt. I am not sure what the power output will be. The amp will be SE class A
The ripple voltages I mentioned are B+
Do you guys measure the ripple voltage at the amp output (OPT secondary) with a scope ?
If so, do you use AC or DC ?
Last edited:
5% to 20% applies ONLY to push-pull amplifiers. It is NOT acceptable for single-ended / triode (especially triode).
300 mV -> stepped down 35x is about 8.5 mV. If your speakers "1 watt sensitivity" is +90 dB (loud, but not yet painful) ... and they're 4 ohms, then using P = E²/R ... then that 8.5 mV is (0.0085)²/4 = .000018 watts. converting to decibals (10log(.000018)) = -47 dB
Now, adding the response of the speakers (+90 dB/W at 1meter) you get 42 dB hum.
42 dB hum worst case.
by comparison, the CLCLC filter (if it truly delivers 1 mV ripple) ... same calcs:
1 mv / 35 = 28 microvolts. 10log(.000028²/4) = -97 dB ... +90 dB/watt = -7 dB
Clearly INAUDIBLE even under perfect listening conditions. Whereas 42 dB might be annoying in a reasonably quiet living room between songs. During actual music output, it wouldn't be annoying at all... But if you were to play back, even moderately loud (like 70 dB) a continuous tone of 50 Hz or 70 Hz (etc.) then you'd hear beats between power supply hum and the tones. Not normally heard in music though.
MY RECOMMENDATION - use the CLCLC setup if you have the $$$ to shell out. Or, use CLC(reg)C with a regulator, to achieve same effect, if not better (will correct for power-line sagging0 instead. You definitely will want that final C though for the "instant juice when needed" point of view.
And that's about all I can add. Must work.
GoatGuy
300 mV -> stepped down 35x is about 8.5 mV. If your speakers "1 watt sensitivity" is +90 dB (loud, but not yet painful) ... and they're 4 ohms, then using P = E²/R ... then that 8.5 mV is (0.0085)²/4 = .000018 watts. converting to decibals (10log(.000018)) = -47 dB
Now, adding the response of the speakers (+90 dB/W at 1meter) you get 42 dB hum.
42 dB hum worst case.
by comparison, the CLCLC filter (if it truly delivers 1 mV ripple) ... same calcs:
1 mv / 35 = 28 microvolts. 10log(.000028²/4) = -97 dB ... +90 dB/watt = -7 dB
Clearly INAUDIBLE even under perfect listening conditions. Whereas 42 dB might be annoying in a reasonably quiet living room between songs. During actual music output, it wouldn't be annoying at all... But if you were to play back, even moderately loud (like 70 dB) a continuous tone of 50 Hz or 70 Hz (etc.) then you'd hear beats between power supply hum and the tones. Not normally heard in music though.
MY RECOMMENDATION - use the CLCLC setup if you have the $$$ to shell out. Or, use CLC(reg)C with a regulator, to achieve same effect, if not better (will correct for power-line sagging0 instead. You definitely will want that final C though for the "instant juice when needed" point of view.
And that's about all I can add. Must work.
GoatGuy
The only things AC coupling the input on an oscilloscope does is remove the DC component (and very low frequencies, probably <1~10 Hz depending on the scope). On the secondary of the OPT there is no DC component (by definition). So no need for AC coupling.
However, you won't be able to resolve 100 uV on a scope. Most scopes bottom out at 10~20 mV/div, some go as low as 2 mV/div. I use a multimeter for that purpose. Even a Fluke 73 handheld meter can resolve 1 mV. For more accurate measurements, I resort to a benchtop meter (HP 34401A) or my distortion analyzer (HP 8903A). If you're using a multimeter for this measurement, you do want to use AC mode as you're trying to measure the AC ripple and not the non-existing DC component. An external computer sound card could work as well.
~Tom
I have been using my Fluke 87V to measure AC noise output at OPT secondary.
I didn't know that I have been measuring ripple voltage.
I didn't know that I have been measuring ripple voltage.
Depending on your needs, you may find that an HP 3478A provides enough resolution. I picked mine up on eBay for $150 calibrated and everything. It's a 5.5 digit multimeter. Its main drawback is that it only measures up to 300 V. Another interesting quirk is that the calibration of the ohmmeter is either 2-wire or 4-wire. Not both. The 34401A (or 34410A) is tough to beat, though. It's also more expensive. I think new cost is $1200. Buy a calibrated one. Don't mess around.
Anyway... With an AC voltmeter on the OPT secondary, you measure ripple+noise. Your Fluke 87 probably has 400~500 Hz bandwidth (my guess), so you'll measure the RMS value of the ripple+noise that falls within that band. The 34401A has a bandwidth of nearly 1 MHz, so it'll measure any AC signal within that band and report the RMS value. Depending on your amp, you're either measuring ripple or noise, though, most likely some combination of the two. If you want to know exactly which frequencies contribute to the RMS ripple+noise level, you need a spectrum analyzer. I use an HP 3562A. You can also use an external computer sound card with some spectrum analyzer software. The 3562A has a much lower noise floor than a typical sound card, though. It doesn't have the same high dynamic range as a current-century ADC, though.
~Tom
Anyway... With an AC voltmeter on the OPT secondary, you measure ripple+noise. Your Fluke 87 probably has 400~500 Hz bandwidth (my guess), so you'll measure the RMS value of the ripple+noise that falls within that band. The 34401A has a bandwidth of nearly 1 MHz, so it'll measure any AC signal within that band and report the RMS value. Depending on your amp, you're either measuring ripple or noise, though, most likely some combination of the two. If you want to know exactly which frequencies contribute to the RMS ripple+noise level, you need a spectrum analyzer. I use an HP 3562A. You can also use an external computer sound card with some spectrum analyzer software. The 3562A has a much lower noise floor than a typical sound card, though. It doesn't have the same high dynamic range as a current-century ADC, though.
~Tom
Last edited:
- Status
- Not open for further replies.