If wanted, an RC snubber could be retrofitted on the rear for each AC input, although a spare pair of connected vias would help to anchor the midpoint. Also, a rear resist layer that left two lands for placing a smt cap in parallel with each 1,000uF electro cap at the shortest loop opportunity may be a benefit.
What is the purpose of RC snubber at AC input? What value of R and C I should use?
About SMT cap parallel with 1,000uF, check this link:
http://www.diyaudio.com/forums/power-supplies/126697-bypassing-psu-capacitors-effective.html
Not sure then so I decided to leave it out. If needed, we can solder it direct to big cap legs.
About SMT cap parallel with 1,000uF, check this link:
http://www.diyaudio.com/forums/power-supplies/126697-bypassing-psu-capacitors-effective.html
Not sure then so I decided to leave it out. If needed, we can solder it direct to big cap legs.
Looks pretty good now. However, what I meant about moving the tracks between fuses and pass transistors, I was imagining something like the following...
Attachments
Well, I did misunderstand your point. Here is an update layout in order to correct it.
Attachments
Layout updated!Attachments
A couple of things you may want to re-consider (I'm being a bit 'picky'!)
Snubbers should be mounted as close to the transformer secondaries as possible, so move them to the AC input terminals (and I assume the missing Cx is mounted underneath?)
The main charging current pulses ideally go directly from diodes to caps, don't go past 'GO', (close as possible) or a deviation thru a 'via' etc - there really isn't a good electrical reason to use the green layer at all here.
Also, that expanding 'ground plane' (green layer) is a source of radiating high current pulse noise, etc and making the audio ground point (to chassis earth) between the caps [STAR GND] is about the worst place you can put it because all your other current returns from other parts of your amp & speaker returns have to pass thru this same point of maximum noise also - this is why the 'audio ground point' as Andrew calls it, is located at the output of the filter, not at the input of the filter.
Just details, I know, and before you conclude your design ....
Snubbers should be mounted as close to the transformer secondaries as possible, so move them to the AC input terminals (and I assume the missing Cx is mounted underneath?)
The main charging current pulses ideally go directly from diodes to caps, don't go past 'GO', (close as possible) or a deviation thru a 'via' etc - there really isn't a good electrical reason to use the green layer at all here.
Also, that expanding 'ground plane' (green layer) is a source of radiating high current pulse noise, etc and making the audio ground point (to chassis earth) between the caps [STAR GND] is about the worst place you can put it because all your other current returns from other parts of your amp & speaker returns have to pass thru this same point of maximum noise also - this is why the 'audio ground point' as Andrew calls it, is located at the output of the filter, not at the input of the filter.
Just details, I know, and before you conclude your design ....
There is little spare room to put the PT snubber, and may be too difficult to place between diode and 1000uF and terminals. The wiring to the PT would be twisted, and could well be very short, depending on layout. The cap probably only needs to be a 100V poly, but if anything would be the part that could do with a larger footprint. If anything, the resistor could mostly sit on top of the related AC1/2 trace, and the C could span the AC1/2 trace in a direction back to the AC gnd terminal.
Can you elaborate on "The main charging current pulses ideally go directly from diodes to caps, don't go past 'GO', (close as possible) or a deviation thru a 'via' etc - there really isn't a good electrical reason to use the green layer at all here." as I don't appreciate the concern, and maybe others don't.
I think the thought that the STAR GND point is a source of radiating noise is a little ambiguous. That STAR GND point doesn't have to be used, and a savvy installation wouldn't use it, but increasing the land area away from the 1000uF cap pads is not putting that land area in the path of the current pulses. Any really high frequency current can be constrained to be very close to the AC terminals and diodes by using judiciously placed smt caps underneath.
Can you elaborate on "The main charging current pulses ideally go directly from diodes to caps, don't go past 'GO', (close as possible) or a deviation thru a 'via' etc - there really isn't a good electrical reason to use the green layer at all here." as I don't appreciate the concern, and maybe others don't.
I think the thought that the STAR GND point is a source of radiating noise is a little ambiguous. That STAR GND point doesn't have to be used, and a savvy installation wouldn't use it, but increasing the land area away from the 1000uF cap pads is not putting that land area in the path of the current pulses. Any really high frequency current can be constrained to be very close to the AC terminals and diodes by using judiciously placed smt caps underneath.
Hi Mr Robbins,
Sorry, the phrase "don't go past 'GO'" is from the 'Monopoly' board game - it just means 'go directly with no deviations' - in tracking terms, this really means that the pathway from diodes to capacitor terminals is as short as possible with no/small side paths if possible (same thing with the snubber & secondary windings).
The 'via' from one side of the pcb to the other is a possible source of trouble, relying on the solder wicking up thru the via - to be avoided if possible, particularly if the supply is carrying substantial current for a power amplifier, for example (I often add a separate extra copper wire from diodes pins to cap terminals - perhaps a bit extreme but a simple habit these days!)
In reality, there are quite a number of different ways of doing a supply board - the pcb method above is a common way of fitting up power supplies of all shapes and sizes with little trouble but, as I mentioned above, I'm just 'nit-picking' here and I like to encourage people to design the best they can once they're aware of some of these things, and not knocking the design as a whole - it's quite okay.
The STAR GND point on the pcb - this is is often a confusing subject with quite a number of different opinions/interpretations - AndrewT has written about this quite a lot and his clear explanations make a lot of sense
I like the way you said using judiciously placed smt caps - some people just add these caps everywhere and it used to be a very common practice to add them directly across the diodes.
Sorry, the phrase "don't go past 'GO'" is from the 'Monopoly' board game - it just means 'go directly with no deviations' - in tracking terms, this really means that the pathway from diodes to capacitor terminals is as short as possible with no/small side paths if possible (same thing with the snubber & secondary windings).
The 'via' from one side of the pcb to the other is a possible source of trouble, relying on the solder wicking up thru the via - to be avoided if possible, particularly if the supply is carrying substantial current for a power amplifier, for example (I often add a separate extra copper wire from diodes pins to cap terminals - perhaps a bit extreme but a simple habit these days!)
In reality, there are quite a number of different ways of doing a supply board - the pcb method above is a common way of fitting up power supplies of all shapes and sizes with little trouble but, as I mentioned above, I'm just 'nit-picking' here and I like to encourage people to design the best they can once they're aware of some of these things, and not knocking the design as a whole - it's quite okay.
The STAR GND point on the pcb - this is is often a confusing subject with quite a number of different opinions/interpretations - AndrewT has written about this quite a lot and his clear explanations make a lot of sense
I like the way you said using judiciously placed smt caps - some people just add these caps everywhere and it used to be a very common practice to add them directly across the diodes.
I've had supply cap 'via' connection 'go bad' only the once on a double sided supply pcb, but it was a real PIA to find and I've been a bit shy about it since.
I wonder a bit about all this extended ground plane enthusiasm in power supply pcbs - it doesn't automatically get you better performance - I try to keep the 'rectifier end' well separated from the regulator end (including this CMx) and sometimes the ground plane just does the opposite - just my way, I guess.
I saw AndrewT writing about the way the transformer wires are best twisted & connected in the "Simple, no-math transformer snubber using ..." thread - quite timely.
I wonder a bit about all this extended ground plane enthusiasm in power supply pcbs - it doesn't automatically get you better performance - I try to keep the 'rectifier end' well separated from the regulator end (including this CMx) and sometimes the ground plane just does the opposite - just my way, I guess.
I saw AndrewT writing about the way the transformer wires are best twisted & connected in the "Simple, no-math transformer snubber using ..." thread - quite timely.
I guess a leaded part 'via' could go bad if the solder wet only the 'underside' pad, and didn't penetrate the hole and also wet the component side pad, and then the plated through hole plating went bad all around the circumference - that would indicate both a bad pcb and a poor soldering process.
On balance, I would see the extended land area more as a mechanism for heat dispersal than for 'emi' management, or voltage drop minimisation. To that end, the blue and red lands could be increased to flood each side of the board around that general diode/electro area - the only concern then being a to have resist layers, and to be careful about the diode heatsinks as they may not be insulated, and the diodes may come in different polarity packages, and a heatsink may inadvertently touch the component side tracks.
On balance, I would see the extended land area more as a mechanism for heat dispersal than for 'emi' management, or voltage drop minimisation. To that end, the blue and red lands could be increased to flood each side of the board around that general diode/electro area - the only concern then being a to have resist layers, and to be careful about the diode heatsinks as they may not be insulated, and the diodes may come in different polarity packages, and a heatsink may inadvertently touch the component side tracks.
I still think not to have a place for film cap in layout. I read trobbins && jameshillj posts and decide to look in the old forum posts for a while. Here is the changed I made.
- I removed extra copper panel in middle, am not sure about this?
- I moved status led + resistor to above and make its ground join with star ground.
- Change ground schema to T-Ground <- any one can check if it is ok now.
Any one could point me the schematic of snubbed PSU with Cx, Cs, Rs ?
- I removed extra copper panel in middle, am not sure about this?
- I moved status led + resistor to above and make its ground join with star ground.
- Change ground schema to T-Ground <- any one can check if it is ok now.
Any one could point me the schematic of snubbed PSU with Cx, Cs, Rs ?
Attachments
Just trim back that blue area and tuck them between the AC input pins (AC1/AC GND & AC GND/AS2) and the power caps - can be at an angle - you could stand the Rs vertically - Rx sits directly across the series combination of Cs + Rs - can easily add it to the AC input terminal pins/legs on the bottom - just an extra 'gilding of the lily' as your design is pretty good as it is, IMHO.
If using the Quasimodo values, the Cs is 150nF and the Cx is 10nF (IIRC!) and spacing for 63 volt caps are fine
If using the Quasimodo values, the Cs is 150nF and the Cx is 10nF (IIRC!) and spacing for 63 volt caps are fine
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.