Good idea, a fuse to protect the transformer secondary. I got the board down to 1 x 2.5 inches if I go 1.25 x 2.5 I can maybe get a PCB fuse holder on it.
I want to make a half wave version board also, with only two diodes.
I'm setting it up for TO220 diodes, that way the board can mount on a small piece if 1.5 x 1.5 aluminum angle stock with all the TO220 diodes hanging off the long edge, heatsinked to a little piece of angle stock. this way the whole bottom of the board is shielded from shorts and the other leg of the aluminum sinks the diodes, and its small.
That sounds like a fair bit of trouble to go to, for only your personal use -- maybe you'll have extras made and sell them on here ?? 
1,5 x 1,5 inch angle stock will place a good bit of aluminum in the thermal pathway between the TO-220 device and the heat sink though (assuming I correctly understand what you have in mind). ¾ x ¾" has worked well for me in the past, other TO-220's in different service, albeit.
Cheers
edit: Also don't forget: The flatness spec for the outside surfaces is usually much tighter, than the inside surfaces; many extrusions, anyway.
1,5 x 1,5 inch angle stock will place a good bit of aluminum in the thermal pathway between the TO-220 device and the heat sink though (assuming I correctly understand what you have in mind
). ¾ x ¾" has worked well for me in the past, other TO-220's in different service, albeit.Cheers
edit: Also don't forget: The flatness spec for the outside surfaces is usually much tighter, than the inside surfaces; many extrusions, anyway.
Last edited:
I thought of 3/4 or 1 inch, you have to subtract 1/8 inch for the stock thickness. A TO220 diode pops up over an inch once you get some !/4 inch standoffs under the board. I'll be glad to share some boards though when I finally get them back. I'll post the copper layout here for criticism before I send it off to jlcpcb (separate thread).
Last edited:
Hi guys,
I'm designing a new rectifier board for my B&M Loudspeakers with center tapped transformer. I didn't found a recommendation where to place the snubber circuit.
I have two different PCB designs and want your recommendation which one to choose. Must the snubber circuit be in line (version 1) or would you recommend to choose the parallel version 2. Or doesn't it matter?!
Version 1: AC in => Snubber => Rectifiers
Version 2: Snubber <= AC in => Rectifiers
In my opinion Version 2 could be better due to short AC tracks...
I'm designing a new rectifier board for my B&M Loudspeakers with center tapped transformer. I didn't found a recommendation where to place the snubber circuit.
I have two different PCB designs and want your recommendation which one to choose. Must the snubber circuit be in line (version 1) or would you recommend to choose the parallel version 2. Or doesn't it matter?!
Version 1: AC in => Snubber => Rectifiers
Version 2: Snubber <= AC in => Rectifiers
In my opinion Version 2 could be better due to short AC tracks...
Last edited:
Version 1 is my preference, best to snub before the bridge, but what do I know?
You should take advantage of the two capacitors on each polarity and use a low value 2 or 3 watt resistor like 0.22r between the caps to get some HF filtering. Maybe a 0.1uf in series with a 1 ohm resistor in parrallel with each electrolytic
You should take advantage of the two capacitors on each polarity and use a low value 2 or 3 watt resistor like 0.22r between the caps to get some HF filtering. Maybe a 0.1uf in series with a 1 ohm resistor in parrallel with each electrolytic
Last edited:
As far as damping out oscillatory ringing ("snubbing") is concerned, the two layouts in #2165 are equally good. #1 is not preferred over #2, and #2 is not preferred over #1.
I suggest positioning the 6.35mm FastOn blade connecters wherever is most convenient for the person who assembles the final amplifier in its chassis. Spaced far enough apart from other components (and each other) to allow unconstrained, easy access by fingers or pliers. Placed such that sharp bends in the wires are not necessary. Accessible by voltmeter probes, that sort of thing. "Designed For Manufacturability" , "Designed For Testability" , etc.
I suggest positioning the 6.35mm FastOn blade connecters wherever is most convenient for the person who assembles the final amplifier in its chassis. Spaced far enough apart from other components (and each other) to allow unconstrained, easy access by fingers or pliers. Placed such that sharp bends in the wires are not necessary. Accessible by voltmeter probes, that sort of thing. "Designed For Manufacturability" , "Designed For Testability" , etc.
Snubbing is completely optional. It's your choice.
Some builders go to the trouble of listening without, and with, snubbers, then choosing the one that sounds best.
Some just optimize the snubbers with Quasimodo, solder them in, and with great peace of mind: forget about them forever after. I'm one of those.
Yet others assemble the entire audio project but leave out the snubbers. Then they connect a load, play at maximum output, and connect up a battery powered, not connected to mains earth, oscilloscope to investigate ringing, using the external highpass filter member peufeu* described in post #142 of this thread. If the unsnubbed equipment does exhibit ringing, then they turn everything off and solder in the snubbers. On the other hand if the unsnubbed equipment exhibits no ringing, then they don't install snubbers. And they bask in the glow of reduced costs, having just saved the price of the C+RC snubbing components.
*posted to diyAudio in December 2013, fyi
_
Some builders go to the trouble of listening without, and with, snubbers, then choosing the one that sounds best.
Some just optimize the snubbers with Quasimodo, solder them in, and with great peace of mind: forget about them forever after. I'm one of those.
Yet others assemble the entire audio project but leave out the snubbers. Then they connect a load, play at maximum output, and connect up a battery powered, not connected to mains earth, oscilloscope to investigate ringing, using the external highpass filter member peufeu* described in post #142 of this thread. If the unsnubbed equipment does exhibit ringing, then they turn everything off and solder in the snubbers. On the other hand if the unsnubbed equipment exhibits no ringing, then they don't install snubbers. And they bask in the glow of reduced costs, having just saved the price of the C+RC snubbing components.
*posted to diyAudio in December 2013, fyi
_
Last edited:
Yes. Any transformer feeding any rectifier can benefit.
It's the combination of transformer leakage inductance and rectifier commutation that produces the ringing. Different transformers (even different *matching* windings) and rectifiers, of course, cause more trouble than others.
If you're trying to decide whether to 'go to the trouble' on a particular piece of gear, remember every piece you have with a linear power supply may benefit. And that's now, AND in the future.
Not Mark (sorry ..),
Cheers
It's the combination of transformer leakage inductance and rectifier commutation that produces the ringing. Different transformers (even different *matching* windings) and rectifiers, of course, cause more trouble than others.
If you're trying to decide whether to 'go to the trouble' on a particular piece of gear, remember every piece you have with a linear power supply may benefit. And that's now, AND in the future.
Not Mark (sorry ..),
Cheers
Last edited:
How loud the bell rings in an application may depend on a few factors. Although the bell relates to the energy in the leakage inductance, hitting the bell relates to the abruptness of the commutation which is influenced by the peak diode current and how peaky that is. That can ne influenced by for example how much filter capacitance you use.
Schade presented a way to estimate rectifier current waveforms when operating in a simple power supply with a transformer and filter capacitor. We can use PSUD2 or LTSpice nowadays. An important aspect to appreciating how hard the bell is hit is the rate (dI/dt) at which the diode current 'hits' 0A as it commutates from on to off. The higher the diode current waveform peak, and shorter the conduction duration, the harder the 'hit'.
Schade showed that the current waveform exhibits a higher peak level, and corresponding shorter 'on' duration, as the filter capacitor value increases. The other major influences are the load current value, the mains frequency, and the ratio of transformer effective resistance to load resistance.
That form of assessment assumes that other issues are not significant, like mains voltage waveform distortion, or transformer winding leakage inductance, or non-ideal diode operation.
Filter capacitor ESR would be an influence similar to diode 'resistance', and so for many practical power supplies it could be negligible given that an ESR may be quite low compared to load resistance or effective transformer resistance. But that is one advantage of using PSUD2 or LTSpice, if you can insert valid values for parts then the simulated output can be valid for the application.
Schade showed that the current waveform exhibits a higher peak level, and corresponding shorter 'on' duration, as the filter capacitor value increases. The other major influences are the load current value, the mains frequency, and the ratio of transformer effective resistance to load resistance.
That form of assessment assumes that other issues are not significant, like mains voltage waveform distortion, or transformer winding leakage inductance, or non-ideal diode operation.
Filter capacitor ESR would be an influence similar to diode 'resistance', and so for many practical power supplies it could be negligible given that an ESR may be quite low compared to load resistance or effective transformer resistance. But that is one advantage of using PSUD2 or LTSpice, if you can insert valid values for parts then the simulated output can be valid for the application.
Last edited:
I have a question for Mark and others on secondary snubbing using Quasimodo. I have been using this design successfully in my amplifiers for some time now, thank you for it, it works very well!
I am working on a NOS R2R DAC design, first foray into digital for me (not an EE). What I am seeing in many power supply schematics for these low-voltage supplies is a lack of transformer snubbing. Rather than a snubber, often a 1uF or 2.2uF capacitor is placed across the secondary to provide some filtering of HF noise coupled from primary to secondary.
I would imagine placing a 1-2.2uF cap on the secondary would complicate the use of Quasimodo.
So my question is this - is a snubber beneficial on these low-voltage supplies? And if so, what is the best approach to calculate the snubber using Quasimodo in parallel with a 1-2.2uF filter cap?
Thanks for the help.
I am working on a NOS R2R DAC design, first foray into digital for me (not an EE). What I am seeing in many power supply schematics for these low-voltage supplies is a lack of transformer snubbing. Rather than a snubber, often a 1uF or 2.2uF capacitor is placed across the secondary to provide some filtering of HF noise coupled from primary to secondary.
I would imagine placing a 1-2.2uF cap on the secondary would complicate the use of Quasimodo.
So my question is this - is a snubber beneficial on these low-voltage supplies? And if so, what is the best approach to calculate the snubber using Quasimodo in parallel with a 1-2.2uF filter cap?
Thanks for the help.
Last edited:
This is the no-math transformer snubber thread -- if you prefer to calculate snubbers, I recommend using the equations in Jim Hagerman's technical paper. There's a link to it in the bibliography / references section of the Quasimodo pdf.
If you want to use no-math Quasimodo: (i) disconnect all wires from the transformer and, ideally, remove it from the chassis. (ii) optimize a C+RC snubber for each secondary winding, using the Quasimodo procedure. I recommend you choose your Cx no larger than 0.01 microfarads (10 nanofarads); otherwise your optimum snubbing resistor could be very small, and your optimum series capacitor Cs will definitely be very large.
For inspiration and reassurance, you might have a quick look at the VRDN voltage regulator, which was designed to provide low noise, stable DC power to DACs and other line level equipment. VRDN contains two C+RC snubbers, one for each of two secondary windings. Spoiler: Cx is implemented as a Metal Oxide Varistor (!) instead of an Epcos film capacitor.
VRDN: bipolar regulator PCB for line level ckts: ±11V to ±20V @ 1.5A with "De-Noiser"
_
Sorry, that was a misnomer, I do not wish to calculate the snubber, my question is related to using the Quasimodo technique in combination with / in lieu of a secondary capacitor to provide filtering of mains coupled noise, as I have seen used in low-noise DAC supplies. However, based on your suggestion to use a Cx no greater than 10nF, I don't think it will be practical given typical filter cap values of 1uF. Thanks for the input.
I would like to thank Mark Johnson for his work and awesome tool he gave to the community. Thanks Mark.
I've just built mine and it worked like a charm. I always wanted to be able to measure the efficiency of snubber network. Now I can.
I just have one question.
I know you have to short the primary when you measure the ringing to a secondary winding.
But is it ok to leave the other secondary windings open, so should I short all other secondary windings too?
Thank you.
Brice.
I've just built mine and it worked like a charm. I always wanted to be able to measure the efficiency of snubber network. Now I can.
I just have one question.
I know you have to short the primary when you measure the ringing to a secondary winding.
But is it ok to leave the other secondary windings open, so should I short all other secondary windings too?
Thank you.
Brice.
Hi Brice,
I'm not the expert here (plenty of physics, even more soldering, but no EE), but I'd say "short 'em all".
Anyway, Mr Johnson has been supporting this brilliant bit of kit with grace and good humor for 7+some-odd years. Once in a while I like to stick my big snout in and spare him the need to answer (when I think I know ..).
Cheers
I'm not the expert here (plenty of physics, even more soldering, but no EE), but I'd say "short 'em all".
Anyway, Mr Johnson has been supporting this brilliant bit of kit with grace and good humor for 7+some-odd years. Once in a while I like to stick my big snout in and spare him the need to answer (when I think I know ..).
Cheers