If by 'overdamped snubber' you mean the resistor in the snubber is too large then this does not necessarily lead to overdamping. The reason is that the formula presented earlier in this thread assumes a simple LCR circuit with all items in series, and all capacitances in parallel. This is simple to analyse and gives some insight into circuit behaviour but it can also lead people astray. In reality some of the capacitance is in series with the resistor and some bypasses the resistor (e.g. the transformer capacitance).
Think about the simple situation where you have an inductance (representing the transformer leakage inductance) with some capacitance in parallel (representing the transformer capacitance). Now add a CR snubber in parallel. If you have R=0 then you have simply added some capacitance, so not changed damping. If you have R=infinity then you have not added anything, so not changed damping. For all other values of R you have added some damping. There will be a value for R which maximises the damping, but what this maximum damping will be depends on the ratio of the two capacitances. All other values of R will give less damping, although still more than for extreme values of R.
In reality the transformer will have some losses so there will be some damping without a snubber. This is why simply adding a capacitor can help.
Think about the simple situation where you have an inductance (representing the transformer leakage inductance) with some capacitance in parallel (representing the transformer capacitance). Now add a CR snubber in parallel. If you have R=0 then you have simply added some capacitance, so not changed damping. If you have R=infinity then you have not added anything, so not changed damping. For all other values of R you have added some damping. There will be a value for R which maximises the damping, but what this maximum damping will be depends on the ratio of the two capacitances. All other values of R will give less damping, although still more than for extreme values of R.
In reality the transformer will have some losses so there will be some damping without a snubber. This is why simply adding a capacitor can help.
zeta>1 ("overdamping") isn't dangerous in itself. However the capacitor in series with the damping resistor does need to be sufficiently large. The series capacitor's impedance at the RLC resonant frequency needs to be very low. Infinity farads would be best*. In my opinion the lower limit is aboutFor the critically damped case, zeta=1, this means Cs ≥ 15 * Cx , which coincidentally lines up with the frequently chosen component values Cx=0.01uF , Cs=0.15uF.
Unless you're using extremely high voltage secondaries and need extremely high voltage capacitors, I recommend choosing a much larger Cs, such as 1.0uF or 3.3uF. The extra cost is low, it isn't much bigger on the PCB, and it is more forgiving when an aggressive lab tech accidentally or deliberately dials up extreme amounts of overdamping, such as perhaps zeta=2.0
The lower limit above is from a derivation found in the Appendix of this article in Linear Audio magazine.
_
- Cs ≥ 15 * Cx * zeta^2
Unless you're using extremely high voltage secondaries and need extremely high voltage capacitors, I recommend choosing a much larger Cs, such as 1.0uF or 3.3uF. The extra cost is low, it isn't much bigger on the PCB, and it is more forgiving when an aggressive lab tech accidentally or deliberately dials up extreme amounts of overdamping, such as perhaps zeta=2.0
The lower limit above is from a derivation found in the Appendix of this article in Linear Audio magazine.
*which is a short circuit. The damping resistor is now connected directly across the secondary, meaning it dissipates (Vsec_rms * Vsec_rms / Rdamping) watts of power. Immediately we appreciate that the job of Cs is to reduce power dissipation in Rdamping at the mains frequency, while also becoming a near-perfect short circuit at the RLC resonant frequency, a few decades higher than f_mains.
_
Last edited:
I did a google search on this subject and found this article.
One size fits all???
Good linear PSU(s) for CAPS music player that won't break the bank - Page 12 - Music Servers - Audiophile Style
One size fits all???
Good linear PSU(s) for CAPS music player that won't break the bank - Page 12 - Music Servers - Audiophile Style
I'm not sure than "one size fits them all", because it seems to me that there are a little bit small RC-values in that article (but I'm not sure).
I have my own simple rule for RC-snabber:
1) R equal to voltage (in volts), or R - little larger than transformer winding DC-resistance (both values are near-optimal), or R between those two values.
2) C >= 0.022 uF. (0.022uF <= C <= 1uF). The higher the voltage - the lower the value.
For example:
1) We have secondary winding 18 V (1 Ohm winding DC resistance) - I'l take R = 15-20 Ohm, and C= 0.22-0.47uF.
1) We have secondary winding 180 V (30 Ohm winding DC resistance) - I'l take R = 100-200 Ohm, and C= 0.033-0.1uF.
Overdamping is a problem only if resistor becames hot (for 50-60 Hz transformer).
I have my own simple rule for RC-snabber:
1) R equal to voltage (in volts), or R - little larger than transformer winding DC-resistance (both values are near-optimal), or R between those two values.
2) C >= 0.022 uF. (0.022uF <= C <= 1uF). The higher the voltage - the lower the value.
For example:
1) We have secondary winding 18 V (1 Ohm winding DC resistance) - I'l take R = 15-20 Ohm, and C= 0.22-0.47uF.
1) We have secondary winding 180 V (30 Ohm winding DC resistance) - I'l take R = 100-200 Ohm, and C= 0.033-0.1uF.
Overdamping is a problem only if resistor becames hot (for 50-60 Hz transformer).
Last edited:
By all means, try other ideas and investigate other approaches. Just be sure you have a way to actually measure and evaluate the improvement you're getting, before versus after. One measurement technique that can work, is to use a completely floating oscilloscope -- i.e., a battery powered scope. Now you can attach your probes to any two circuit nodes whatsoever, without worrying that probe #2 has created a 20 ampere short to ground.
There are other ways to do the measurements too, some of which I consider unacceptably risky & dangerous, that you will have to invent or discover independently. If you are killed or maimed, it happened without my assistance.
In my experience, battery powered scopes with less than about 20 MHz of bandwidth, seem to have cheap and crummy triggering systems. So these scopes are not able to trigger upon, and display, the waveforms you want to see (HF oscillatory ringing riding on top of mains sine wave). Yes, even though those waveforms are less than 1 MHz damped sinewaves. I don't really know why a 10 MHz battery powered scope can't do the job, but I do know: it can't. Maybe the competitive pressure to produce the cheapest "10 MHz" scope in all of China, forces unhappy compromises between price and performance.
There are other ways to do the measurements too, some of which I consider unacceptably risky & dangerous, that you will have to invent or discover independently. If you are killed or maimed, it happened without my assistance.
In my experience, battery powered scopes with less than about 20 MHz of bandwidth, seem to have cheap and crummy triggering systems. So these scopes are not able to trigger upon, and display, the waveforms you want to see (HF oscillatory ringing riding on top of mains sine wave). Yes, even though those waveforms are less than 1 MHz damped sinewaves. I don't really know why a 10 MHz battery powered scope can't do the job, but I do know: it can't. Maybe the competitive pressure to produce the cheapest "10 MHz" scope in all of China, forces unhappy compromises between price and performance.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.