Hi Everyone,
My apologies as I know this has already been discussed a lot already.
But I'm still confused by the conflicting information about whether to use snubbers or not. Hoping someone can give me a definite answer on this.
I am building a F5 amp am using Nelsons power supply design (attached).
Nelson's schematic doesn't show caps across the rectifiers or a snubbing network etc. I assume this is because he uses slow type diodes that don't need snubbing?
I have already purchased FEP30DP ultrafast diodes and am using the universal power supply board.
The advice from Randy Slone and others is to put 0.01-0.1uf caps across the diodes to reduce RF emissions.
AndrewT advises that this can make things worse and that you should use a properly calculated RC network across the diodes instead.
Unfortunately I don't have the maths ability to make sense of the Hagerman article. (http://hagtech.com/pdf/snubber.pdf)
Out of the following options what should I do?
1. Use no diode snubber
2. Use 0.01uf caps across the diodes
3. Use a RC+C network.
If I go with option 3 can someone suggest values to use for the RC+C network?
Also the F5 schmatic also doesn't show any output snubber. Would adding the 1R0.1uF output snubber be advised?
Thank you
My apologies as I know this has already been discussed a lot already.
But I'm still confused by the conflicting information about whether to use snubbers or not. Hoping someone can give me a definite answer on this.
I am building a F5 amp am using Nelsons power supply design (attached).
Nelson's schematic doesn't show caps across the rectifiers or a snubbing network etc. I assume this is because he uses slow type diodes that don't need snubbing?
I have already purchased FEP30DP ultrafast diodes and am using the universal power supply board.
The advice from Randy Slone and others is to put 0.01-0.1uf caps across the diodes to reduce RF emissions.
AndrewT advises that this can make things worse and that you should use a properly calculated RC network across the diodes instead.
Unfortunately I don't have the maths ability to make sense of the Hagerman article. (http://hagtech.com/pdf/snubber.pdf)
Out of the following options what should I do?
1. Use no diode snubber
2. Use 0.01uf caps across the diodes
3. Use a RC+C network.
If I go with option 3 can someone suggest values to use for the RC+C network?
Also the F5 schmatic also doesn't show any output snubber. Would adding the 1R0.1uF output snubber be advised?
Thank you
Attachments
Build the F5 using no diode snubber...don't sweat it for now. When you have the F5 completed and your're listening to music check out the QUASIMODO THREAD.
Do you have an Oscilloscope?
From the world of Switch Mode Power Supplies... Or maybe I looked at what someone else said in respect of it and it melted my brain so I worked something out like this.
Pre-Mumble.
Your diode gets reverse biased but suffers from recovery time and pretends to be a short circuit. During that time it sucks current with the rise limited by circuit, transformer, inductance. Then it switches off and you end up with the diodes reverse biased capacitance ringing with the circuit, transformer, inductance.
You can measure the transformer, secondary leakage inductance and the reverse biased capacitance of the diode or they might be available from a data sheet... assuming the manufacturer has been kind to you or if you have the necessary test equipment.
If not then what to do?
Let's say you just have your unit on the bench and an Oscilloscope and you will be careful. Down in the world of lumps of iron on the, 'low voltage' secondary side you might almost be safe.
The Gotcha will be whether or not your secondary side has some form of hard reference to mains ground and your Oscilloscope has the same. Let's say the secondary side does not and you can rely on the transformer in the unit for isolation.
You might be able to get away with your probe ground lead assuming you are not going to be measuring 'high' frequencies but solder a female BNC across one of the bridge diodes and plug your probe, set to x10 and make sure you are not going to ding its voltage limits, into it.
Turn things on and if they do not go bang synchronise the oscilloscope to 'line'... or one of the other options to get a stable display and mess about with the other knobs to see what might be going on.
You should see a forward biased part, about 0.7V, and the reverse biased bit the other way, half sine wave with amplitude of the transformer output. If there are ringing problems then you will see them when the diode turns off and it is possible that they will decay or not be there at all.
Let's say they are there and are quite horror show. Measure the frequency of the ringing and call it FR1. A piece of paper is useful to write it down and for doing some sums later. You will also need a pen or pencil.
Your first sum is...
FR1 = 1/2.PI.SQRT(LT.CD)
LT is the inductance thing to do with the Transformer, perhaps its secondary leakage inductance. CD is the capacitance to do with the rectifier Diode, presumably its reverse biased capacitance.
If you do have a data sheet for your bridge or the diodes that tells you something about their reverse biased capacitance then cool. There may be a graph. Pick a value at something like 20% of your secondary transformer voltage and then dig about in your box of bits for a capacitor about three times that value.
If you do not have that information just pick anything within reason whatever reason might be and see what happens. Guess 1nF is not going to kill things.
Solder your capacitor across your diode. I assume you did switch things off. Switch things on again and see if there is ringing. It will/should be at a lower frequency. Measure that one and write it down as FR2...
FR2 = 1/2.PI.SQRT(LT.(CD + CA))
Where CA is the value of the capacitor you Added.
Now you have to do some maffematics. In fact I have to do some a well in order to work out what I've forgotten I did some time ago.
Move the 2.PI
2.PI.FR1 = 1/SQRT(LT.CD)
2.PI.FR2 = 1/SQRT(LT.(CD + CA))
Take reciprocals...
(2.PI.FR1)' = SQRT(LT.CD)
(2.PI.FR2)' = SQRT(LT.(CD + CA))
Square them...
(2.PI.FR1)'^2 = LT.CD
(2.PI.FR2)'^2 = LT.(CD + CA)
Call them X and Y
X = (2.PI.FR1)'^2 = LT.CD
Y = (2.PI.FR2)'^2 = LT.(CD + CA)
Divide one by the other..
Y/X = LT.(CD + CA)/LT.CD
LT disappears..
Y/X = (CD + CA)/CD
Multiply both sides by CD
CD.Y/X = CD + CA
Subtract CD from both sides..
CD.Y/X - CD = CA
Factor out CD
CD(Y/X - 1) = CA
Rearrange..
CD = CA/(Y/X - 1)
We have found out what the reverse bias capacitance of the diode is. Don't forget to do the substitutions for X and Y.
Go back to...
FR1 = 1/2.PI.SQRT(LT.CD)
Almost as before..
2.PI.FR1 = 1/SQRT(LT.CD)
(2.PI.FR1)' = SQRT(LT.CD)
(2.PI.FR1)'^2 = LT.CD
but you already know that
(2.PI.FR1)'^2 = X = LT.CD
So..
LT = X/CD
Now you need to know the resonant impedance/resistance and I'll cheat and just tell you it is,
Rres = SQRT(LT/CD)
Rumour from elsewhere has it that 'critical damping', or close, is achieved if you select a capacitance of three times the value of the diode capacitance, 3CD, and place it across the diode in series with Rres.
You do have to worry about power dissipation in Rres but, for the moment, I have run out of steam.
...
From the world of Switch Mode Power Supplies... Or maybe I looked at what someone else said in respect of it and it melted my brain so I worked something out like this.
Pre-Mumble.
Your diode gets reverse biased but suffers from recovery time and pretends to be a short circuit. During that time it sucks current with the rise limited by circuit, transformer, inductance. Then it switches off and you end up with the diodes reverse biased capacitance ringing with the circuit, transformer, inductance.
You can measure the transformer, secondary leakage inductance and the reverse biased capacitance of the diode or they might be available from a data sheet... assuming the manufacturer has been kind to you or if you have the necessary test equipment.
If not then what to do?
Let's say you just have your unit on the bench and an Oscilloscope and you will be careful. Down in the world of lumps of iron on the, 'low voltage' secondary side you might almost be safe.
The Gotcha will be whether or not your secondary side has some form of hard reference to mains ground and your Oscilloscope has the same. Let's say the secondary side does not and you can rely on the transformer in the unit for isolation.
You might be able to get away with your probe ground lead assuming you are not going to be measuring 'high' frequencies but solder a female BNC across one of the bridge diodes and plug your probe, set to x10 and make sure you are not going to ding its voltage limits, into it.
Turn things on and if they do not go bang synchronise the oscilloscope to 'line'... or one of the other options to get a stable display and mess about with the other knobs to see what might be going on.
You should see a forward biased part, about 0.7V, and the reverse biased bit the other way, half sine wave with amplitude of the transformer output. If there are ringing problems then you will see them when the diode turns off and it is possible that they will decay or not be there at all.
Let's say they are there and are quite horror show. Measure the frequency of the ringing and call it FR1. A piece of paper is useful to write it down and for doing some sums later. You will also need a pen or pencil.
Your first sum is...
FR1 = 1/2.PI.SQRT(LT.CD)
LT is the inductance thing to do with the Transformer, perhaps its secondary leakage inductance. CD is the capacitance to do with the rectifier Diode, presumably its reverse biased capacitance.
If you do have a data sheet for your bridge or the diodes that tells you something about their reverse biased capacitance then cool. There may be a graph. Pick a value at something like 20% of your secondary transformer voltage and then dig about in your box of bits for a capacitor about three times that value.
If you do not have that information just pick anything within reason whatever reason might be and see what happens. Guess 1nF is not going to kill things.
Solder your capacitor across your diode. I assume you did switch things off. Switch things on again and see if there is ringing. It will/should be at a lower frequency. Measure that one and write it down as FR2...
FR2 = 1/2.PI.SQRT(LT.(CD + CA))
Where CA is the value of the capacitor you Added.
Now you have to do some maffematics. In fact I have to do some a well in order to work out what I've forgotten I did some time ago.
Move the 2.PI
2.PI.FR1 = 1/SQRT(LT.CD)
2.PI.FR2 = 1/SQRT(LT.(CD + CA))
Take reciprocals...
(2.PI.FR1)' = SQRT(LT.CD)
(2.PI.FR2)' = SQRT(LT.(CD + CA))
Square them...
(2.PI.FR1)'^2 = LT.CD
(2.PI.FR2)'^2 = LT.(CD + CA)
Call them X and Y
X = (2.PI.FR1)'^2 = LT.CD
Y = (2.PI.FR2)'^2 = LT.(CD + CA)
Divide one by the other..
Y/X = LT.(CD + CA)/LT.CD
LT disappears..
Y/X = (CD + CA)/CD
Multiply both sides by CD
CD.Y/X = CD + CA
Subtract CD from both sides..
CD.Y/X - CD = CA
Factor out CD
CD(Y/X - 1) = CA
Rearrange..
CD = CA/(Y/X - 1)
We have found out what the reverse bias capacitance of the diode is. Don't forget to do the substitutions for X and Y.
Go back to...
FR1 = 1/2.PI.SQRT(LT.CD)
Almost as before..
2.PI.FR1 = 1/SQRT(LT.CD)
(2.PI.FR1)' = SQRT(LT.CD)
(2.PI.FR1)'^2 = LT.CD
but you already know that
(2.PI.FR1)'^2 = X = LT.CD
So..
LT = X/CD
Now you need to know the resonant impedance/resistance and I'll cheat and just tell you it is,
Rres = SQRT(LT/CD)
Rumour from elsewhere has it that 'critical damping', or close, is achieved if you select a capacitance of three times the value of the diode capacitance, 3CD, and place it across the diode in series with Rres.
You do have to worry about power dissipation in Rres but, for the moment, I have run out of steam.
...
Attachments
You may find that the supply does not ring in the presence of fast starting/stopping transient currents. In this case you don't need to add any snubber.
Use the quasimodo technique to find out how well behaved your transformer inductance+diode capacitance is.
Then decide whether to add on transformer inductance snubbering.
Use the quasimodo technique to find out how well behaved your transformer inductance+diode capacitance is.
Then decide whether to add on transformer inductance snubbering.
I'm putting together a very similar power supply board for another of the First Watt Class A amplifiers. I have selected an AnTek AS4218 transformer (400VA toroid, 2x18VAC secondaries @ 11 amps each). I've decided to lay out my PCB including two (C,R+C) snubbers, one for each of the secondaries. In addition to the fuse in the primary circuit, I'm also laying out the PSU PCB to include fuses for each of the secondary circuits.
Since I have an oscilloscope and a Quasimodo test jig (assembled on a Protoboard solderless breadboard), discovering the correct values of the (C,R+C) snubber components will be a simple 15 minute procedure.
Since I have an oscilloscope and a Quasimodo test jig (assembled on a Protoboard solderless breadboard), discovering the correct values of the (C,R+C) snubber components will be a simple 15 minute procedure.
Last edited:
Thanks for your fast and detailed feedback.
I think I will go without the diode snubber for now and then look into the Quasimodo solution at a later date.
I think I will go without the diode snubber for now and then look into the Quasimodo solution at a later date.
- Status
- Not open for further replies.