Remember that in a bridge rectifier + filter capacitor circuit, the ON diodes protect the OFF diodes from enormous reverse voltage. If the secondary is 75VAC (106V peak), no diode will see more than 106V of reverse bias.
And if there's a surge on the mains, the surge on the secondary is clamped to 106V by the ON diodes and by the enormous energy capacity of the filter capacitors. No diode sees more than 106V of reverse bias. Fool around with simulations and observe.
One nice thing about bridge rectifier assemblies like the GBPC3510, is that they are easier to bolt to the chassis (or to a heatsink) than discrete diodes. They take up less room. On the other hand, bridge rectifier assemblies finished dead last in a test of 48 different diodes, looking at diode-induced oscillatory ringing in the transformer secondary. (link)
On the other, other hand: the very best diode with no snubber, was not as good as the very worst diode (GBPC3510 bridge) plus a CRC snubber. For what it's worth.
And if there's a surge on the mains, the surge on the secondary is clamped to 106V by the ON diodes and by the enormous energy capacity of the filter capacitors. No diode sees more than 106V of reverse bias. Fool around with simulations and observe.
One nice thing about bridge rectifier assemblies like the GBPC3510, is that they are easier to bolt to the chassis (or to a heatsink) than discrete diodes. They take up less room. On the other hand, bridge rectifier assemblies finished dead last in a test of 48 different diodes, looking at diode-induced oscillatory ringing in the transformer secondary. (link)
On the other, other hand: the very best diode with no snubber, was not as good as the very worst diode (GBPC3510 bridge) plus a CRC snubber. For what it's worth.
Hi Mark,
You have to temper theory with real life. I wouldn't bet on surges being effectively clamped simply due to dI/dt if nothing else. Those capacitors won't do anything to clamp the surge for some small time, meaning the surge appears in all it's glory across the rectifier diodes.
-Chris
You have to temper theory with real life. I wouldn't bet on surges being effectively clamped simply due to dI/dt if nothing else. Those capacitors won't do anything to clamp the surge for some small time, meaning the surge appears in all it's glory across the rectifier diodes.
Given that a bridge rectifier is constructed with single rectifier diodes encapsulated, something is off in that study. Did they test clamped on the leads of single rectifiers directly next to the body, and used leads to the bridge rectifier? Otherwise each tested bridge should have a diode they were exactly equivalent to.
Can't dispute that one bit! I use bridge rectifier assemblies even when I only need a full wave rectifier in older tube equipment. That buys me three more tie points if needed.
-Chris
Click on the (link) to read more about the diode test article. If you decide to spend the EUR 0,99 you'll see that there was only one 35 ampere, standard speed, PN silicon, non-soft-recovery, $0.75 diode among the 48 tested diodes: It was 1/4th of a GBPC3510 bridge (which sells for $3 in qty=1). The other standard speed, PN silicon, non-soft-recovery diodes with <35A current ratings, were awful too. But the most awful of the bunch turned out to be the GBPC3510. Schottky, Silicon Carbide, Fast Recovery, HexFRED, Soft Recovery, SuperBarrier, and Stealth diodes were lots, LOTS better.
All diodes were tested in the exact same test fixture under the exact same conditions; differences in the output ringing waveforms was due exclusively to differences between the diodes. But you may find that you simply HATE the test fixture and/or the measurement methodology. Regrettably, Linear Audio magazine has ceased publishing new issues; so if you decide to re-do the experiments yourself using a different fixture or methodology, you won't be able to publish your findings in L.A.
All diodes were tested in the exact same test fixture under the exact same conditions; differences in the output ringing waveforms was due exclusively to differences between the diodes. But you may find that you simply HATE the test fixture and/or the measurement methodology. Regrettably, Linear Audio magazine has ceased publishing new issues; so if you decide to re-do the experiments yourself using a different fixture or methodology, you won't be able to publish your findings in L.A.
You should have disclosed this important piece of information from he beginning
Given the high power high voltage involved I would straight go for a large macho man rectifier bridge and call it a day.
Adding some turn on surge protection would be fine, doubly so considering your large capacitors.
Well,why not choose something .... um .... suitable for the job?
As in .... 400V rated bridges or better?
100V rated ones will only confirm your bias against monolithic bridges.
Remember to heat sink them.
Sorry, I corrected my erroneous PIV value earlier (you must not have seen that post). It's 1000 PIV, not 100...
I have designed a transformerless power supply driven transformer inrush limiter that I plan to use in this project. This will just be a 15R, 50W resistor in series with the main that will be shorted out after 250msec or so. Since my cap bank will use four 21kuF caps (per channel) I am trying to come up with a way to slow the inrush to the cap bank as well. In addition, with the cap arrangement I can throw in a power resistor around 1R or less to create CRC filtering on the output.
I will just use the GBPC5010 (50A 1000PIV) bridge for rectification. After reading Mark's article in Linear Audio I would like to add a snubber across the secondary to reduce/eliminate ringing and radiated EMF. Do I use one per leg of the secondary (the transformer is center tapped) or just one snubber across the whole thing? How do I measure the L, C values for the transformer? Is this as simple as using an LCR meter? No oscilloscope here, unfortunately... I vaguely recall a thread on this topic, but don't have the link to it.
I have designed a transformerless power supply driven transformer inrush limiter that I plan to use in this project. This will just be a 15R, 50W resistor in series with the main that will be shorted out after 250msec or so. Since my cap bank will use four 21kuF caps (per channel) I am trying to come up with a way to slow the inrush to the cap bank as well. In addition, with the cap arrangement I can throw in a power resistor around 1R or less to create CRC filtering on the output.
I will just use the GBPC5010 (50A 1000PIV) bridge for rectification. After reading Mark's article in Linear Audio I would like to add a snubber across the secondary to reduce/eliminate ringing and radiated EMF. Do I use one per leg of the secondary (the transformer is center tapped) or just one snubber across the whole thing? How do I measure the L, C values for the transformer? Is this as simple as using an LCR meter? No oscilloscope here, unfortunately... I vaguely recall a thread on this topic, but don't have the link to it.
Bring it over to my house in Area Code 408 and we'll Quasimodo it together. It will take 5 minutes to set up the experiment, 5 minutes to run the experiment, 5 minutes to write down the results, and 45 minutes to knock back the agave bevs + eat the guacamole & gazpacho. Did this same thing a year ago with a DIYA member from Denmark who happened to be in the area, and it was a lot of fun. I learned to pronounce "Copenhagen" the Danish way: Koeehh - bin - HOBB - ann. Don't shoot me, Karsten!
Last edited:
I found this article on snubbers in linear power supplies that is easy to read (or so it seems to me). They give some ballpark values and explain the rationale behind them. This would not be as good as measuring, but might be better than no snubber at all:
snubbering | DIY-Audio-Heaven
snubbering | DIY-Audio-Heaven
If you actually use the Wing And A Prayer procedure from DIY-Audio-Heaven, I suggest you make certain that your DIY-Audio-Heaven-snubber's capacitor is big. I mean, really REALLY big. At least 680 nF {the value I used on the PSU linked in my sig} and perhaps even bigger, since you're not bothering to actually measure your transformer.
Then size the resistor so that the 60 Hz power dissipated in it equals 1/3 watt (and buy a 1 watt resistor!!). You'll need to do a little vector summation to find the current (hence the power), because the secondary's AC voltage is across the SERIES combination of Rsnub and Csnub. Or just run a quick LTSPICE simulation at 60 Hz and plot the resistor heat -- the icon with the thermometer. Jack around with the simulated R value until the heat is 0.33 watts.
The result from this bastardized, schlocko lazy procedure will be seriously and horrifyingly non-optimum. And you'll use math, unlike the "No Math" procedure mentioned above. But the odds are favorable that the resulting circuit which pops out, will do more good than harm. Probably.
Then size the resistor so that the 60 Hz power dissipated in it equals 1/3 watt (and buy a 1 watt resistor!!). You'll need to do a little vector summation to find the current (hence the power), because the secondary's AC voltage is across the SERIES combination of Rsnub and Csnub. Or just run a quick LTSPICE simulation at 60 Hz and plot the resistor heat -- the icon with the thermometer. Jack around with the simulated R value until the heat is 0.33 watts.
The result from this bastardized, schlocko lazy procedure will be seriously and horrifyingly non-optimum. And you'll use math, unlike the "No Math" procedure mentioned above. But the odds are favorable that the resulting circuit which pops out, will do more good than harm. Probably.
Wonderful, send me a PM when you're ready to visit. Need a few days advance notice so that The Boss Of The House is appeased. 3.6 miles from red arrow (intersection of CA-85 and CA-17) in attached image.
BTW on workdays, commute flow in the morning is: (North towards Oakland on CA17) and (North towards Mountain View on CA85).
Commute flow in the afternoon of workdays is: (South towards Los Gatos on CA17) and (South towards Morgan Hill on CA85)
The density of vehicles is (forward commute)/(reverse commute) = 3.0X.
BTW on workdays, commute flow in the morning is: (North towards Oakland on CA17) and (North towards Mountain View on CA85).
Commute flow in the afternoon of workdays is: (South towards Los Gatos on CA17) and (South towards Morgan Hill on CA85)
The density of vehicles is (forward commute)/(reverse commute) = 3.0X.
Attachments
Last edited:
Many here do not know how to formulate a question.You should have disclosed this important piece of information from he beginning
They think we are magicians.
But one needs to know enough to enable the question to become clear.
That's where beginners (in any subject/topic) need guidance.
Last edited:
OK, I have a question for you Mark...
I took a look at your Ring Not power supply board schematic and read over your Linear Audio article. In the Ring Not schematic I see that you place a capacitor of 3.3 nF across the secondary, and then have the snubber (series RC) in parallel with that. These components are in between the transformer and the bridge.
Following your paper and other sources, the model of the transformer used to illustrate ringing is a parallel inductor (Lt) and capacitor (Ca). From what I can tell, the capcitance in this model is likely to be in the tens of pico Farad range. If you put another capacitor of in parallel with the secondary (labeled Cx in your paper) like you do in the Ring Not board, and make the value of Cx>>Ca, the equivalent parallel capacitance value can be approximated by Ca||Cx=Cx. Given this value, it seems one would you only need measure the leakage inductance to be able to determine the optimum value for the RC snubber from equations.
Not that I have ever done one, but it seems possible to do a leakage inductance measurement by shorting the primary and using relatively simple measuring equipment (e.g. LCR meter) instead of a scope and the Quasimodo board. I'm just thinking about the relative access of a more novice DIYer here, who might not have a scope or want to spend the money to buy one, etc.
Is this a procedure that you think could work or are there pitfalls that I don't know about that would prevent it from being successful?
I took a look at your Ring Not power supply board schematic and read over your Linear Audio article. In the Ring Not schematic I see that you place a capacitor of 3.3 nF across the secondary, and then have the snubber (series RC) in parallel with that. These components are in between the transformer and the bridge.
Following your paper and other sources, the model of the transformer used to illustrate ringing is a parallel inductor (Lt) and capacitor (Ca). From what I can tell, the capcitance in this model is likely to be in the tens of pico Farad range. If you put another capacitor of in parallel with the secondary (labeled Cx in your paper) like you do in the Ring Not board, and make the value of Cx>>Ca, the equivalent parallel capacitance value can be approximated by Ca||Cx=Cx. Given this value, it seems one would you only need measure the leakage inductance to be able to determine the optimum value for the RC snubber from equations.
Not that I have ever done one, but it seems possible to do a leakage inductance measurement by shorting the primary and using relatively simple measuring equipment (e.g. LCR meter) instead of a scope and the Quasimodo board. I'm just thinking about the relative access of a more novice DIYer here, who might not have a scope or want to spend the money to buy one, etc.
Is this a procedure that you think could work or are there pitfalls that I don't know about that would prevent it from being successful?
I thought so too. And then I tried it. And then I read about Morgan Jones's miserable results when he tried it (link).
Necessity was the mother of invention of Quasimodo. The "simple" measurement was not simple and it gave ambiguous results / nonsense results / no results at all. Thus was born the Bellringer concept, which is the fundamental underpinning of Quasimodo et al.
However I am happy to admit that I might have fouled up my dozens of attempts to perform the "simple" measurement. I am less likely to accuse Morgan Jones of fouling up. Perhaps somebody, somewhere, is able to directly measure the leakage inductance of the secondary of a super-excellent, modern, toroidal transformer like the Antek AS-2222 (link 2). But I doubt it. And I would point out that a Quasimodo + USB scope with 2 MHz bandwidth, costs 5X less than a 2 MHz capable LCR meter / impedance bridge, even a used HP instrument on eBay.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.