Hi all,
I am powering some digital circuits with various power supplies and it's becoming too big to put in one box. I am looking to move all the transformers, rectification and reservoir/filter capacitors into box 1 and have the regulators and all the stuff I'm powering in the second box. This appears to conform to the widely recommended route of keeping the connection between the regulators and the load as short as possible.
The power supplies are pretty basic and look something like this:
Having done some reading on here I came across an interesting post on here that says "you might have 100uf (low ESR) at the PS... and then put a 0.1 - 1.0 uF film foil, or even ceramic, RIGHT across the power pins of an op-amp. The op-amp gets a supply of power free from inductance and behaves better..." (full post here: http://www.diyaudio.com/forums/solid-state/70006-super-capacitor-power-supply-filter.html#post793595)
So, assuming that no one shouts my general plan down and that the regulator and everything after it will definitely go into box 2 (near to the load) then my question is that first 0.1uF capacitor. Should it go into box one or box 2? Or, Should I leave it in box 1 and add a second low value ceramic or polypropeline film in as the first component in the second box...just before the reg?
Thanks for reading and any help you can offer,
Crom
I am powering some digital circuits with various power supplies and it's becoming too big to put in one box. I am looking to move all the transformers, rectification and reservoir/filter capacitors into box 1 and have the regulators and all the stuff I'm powering in the second box. This appears to conform to the widely recommended route of keeping the connection between the regulators and the load as short as possible.
The power supplies are pretty basic and look something like this:
An externally hosted image should be here but it was not working when we last tested it.
Having done some reading on here I came across an interesting post on here that says "you might have 100uf (low ESR) at the PS... and then put a 0.1 - 1.0 uF film foil, or even ceramic, RIGHT across the power pins of an op-amp. The op-amp gets a supply of power free from inductance and behaves better..." (full post here: http://www.diyaudio.com/forums/solid-state/70006-super-capacitor-power-supply-filter.html#post793595)
So, assuming that no one shouts my general plan down and that the regulator and everything after it will definitely go into box 2 (near to the load) then my question is that first 0.1uF capacitor. Should it go into box one or box 2? Or, Should I leave it in box 1 and add a second low value ceramic or polypropeline film in as the first component in the second box...just before the reg?
Thanks for reading and any help you can offer,
Crom
At a minimum, the 0.1uF film capacitor should go in the 2nd box very close to the regulator input. I would also consider a second capacitor added to the 2nd box in parallel with it, maybe a 680uF 'lytic.
Thanks for the feedback. In case it changes your thinking on the value of the additional capacitor, I should mention that the value of the final electrolytic in the chain has been changed in my design to 100uF (Elna silmic II) as this allows the reg to provide the best transient step response.
For all purposes, you can separate the effects of line regulation and load regulation to the pre- and post- capacitors, respectively. So no, it won't change the advice.
Generally the more capacitance pre- regulator the better the line regulation, so in that respect the more capacitance the better. There is a law of diminishing returns on that advice, and too much capacitance just taxes the rectifiers, increases current crest factor in the transformer, and causes higher harmonic ripple current (bursts of peaked current) in the capacitor. PSUD is helpful to see just how much capacitance pre- regulator is sufficient.
You could probably put in 47uF locally in the 2nd case and it would perform the same overall function, which is to provide the regulator with a low impedance source.
Oh yeah, maybe install a diode from regulator output back to input for accidental input short circuit protection.
Generally the more capacitance pre- regulator the better the line regulation, so in that respect the more capacitance the better. There is a law of diminishing returns on that advice, and too much capacitance just taxes the rectifiers, increases current crest factor in the transformer, and causes higher harmonic ripple current (bursts of peaked current) in the capacitor. PSUD is helpful to see just how much capacitance pre- regulator is sufficient.
You could probably put in 47uF locally in the 2nd case and it would perform the same overall function, which is to provide the regulator with a low impedance source.
Oh yeah, maybe install a diode from regulator output back to input for accidental input short circuit protection.
One pitfall can be the charging pulses into the smoothing capacitors.
These modulate the zero volt line.
Keep the power supply ground and digital grounds separate and join at one point at the smoothing capacitors.
These modulate the zero volt line.
Keep the power supply ground and digital grounds separate and join at one point at the smoothing capacitors.
95% of the benefit in using two boxes is the physical separation of the transformer from the rest of the circuit. Even running post-transformer AC into the audio chassis has enormous benefits.
Raw DC is better - just rectify and put a cap or two in the PSU chassis, then run unregulated DC to the audio chassis, and regulate near the audio circuits.
Examples -
http://www.diyaudio.com/forums/analogue-source/238618-technics-sl-1200-dc-power-supply.html
http://www.diyaudio.com/forums/pass-labs/204336-building-pearl-2-a.html
Raw DC is better - just rectify and put a cap or two in the PSU chassis, then run unregulated DC to the audio chassis, and regulate near the audio circuits.
Examples -
http://www.diyaudio.com/forums/analogue-source/238618-technics-sl-1200-dc-power-supply.html
http://www.diyaudio.com/forums/pass-labs/204336-building-pearl-2-a.html
Fantastic, thank you very much - that's really interesting reading. I don't have a bleeder resistor in my design but think that I need to add one now. I also like the point to point wiring idea I might try that for the next build.
Regarding the earthing Nigel, I'm not sure what you mean. Currently, the plan was to supply the earth to the second box via the shielding on the umbilical cord. This would earth the chassis of the second box. I also had a plan to not connect the umbilical cord shielding at the second chassis end so that it acted as a drain back to the first box. I would then earth the second box to the mains earth separately (however, they would obviously end up meeting somewhere in the mains system).
It sounds like what you are saying is that it would be better to do the second plan (ie not connect the 2 chassis' earths) but also run a separate earth connecting the second chassis back to the first. I guess it should go back to the transformer earch and chassis earth because I don't have an earth connection at the smoothing capacitors...just +/-.
Apologies for the probable numptiness of the above...I'm learning ;-)
Regarding the earthing Nigel, I'm not sure what you mean. Currently, the plan was to supply the earth to the second box via the shielding on the umbilical cord. This would earth the chassis of the second box. I also had a plan to not connect the umbilical cord shielding at the second chassis end so that it acted as a drain back to the first box. I would then earth the second box to the mains earth separately (however, they would obviously end up meeting somewhere in the mains system).
It sounds like what you are saying is that it would be better to do the second plan (ie not connect the 2 chassis' earths) but also run a separate earth connecting the second chassis back to the first. I guess it should go back to the transformer earch and chassis earth because I don't have an earth connection at the smoothing capacitors...just +/-.
Apologies for the probable numptiness of the above...I'm learning ;-)
If you have enough wires inside the shield (screen) to have V+, GND and V-, then by all means make the umbilical shield mains safety earth. It will be the best grounded for noise. Earth the PSU enclosure, and treat the shield and audio enclosure as that same potential safety earth.
Or just float the audio chassis completely...
If your supply includes RFI filtering (common-mode chokes, ferrite beads, CLC ladders, etc), I'd recommend you keep it inside the first box, the one containing the transformer. If possible you'd like to prevent RFI from entering the second box, the one with the sensitive audio circuitry.
OK, I'm making headway guys. One practical question. When I build this thing and the PSU at that point ends in a large capacitor, am I OK to turn it on for testing without a load? I'd like to play with various combinations and hook them up to a scope to see the ripple, noise etc (or lack of it hopefully) but not if I'm going to fry anything.
Yes, absolutely fine to test it with no load attached. It would be good practice to add a bleeder resistor acros the supply (output to 0v) at this point to drain the cap when the raw supply is turned off. Something like 4700ohms and 1W rating is fine for voltage levels your sketch proposes.
Picking up a couple of other points raised above - you really do want to keep the rectifier and caps close to the transformer. The charging loop is the noisiest thing in the whole supply, and extending raw AC into a second chassis for rectification is a sure-fire way to end up with more noise than a one-box solution - [because you've sent peaky charging currents around a much greater large loop area]
Secondly, the earlier suggestion of a larger cap at the input to the regulator(s) in the second chassis is a good one for a subtle reason - it turns the incidental slight resistance of your DC interconnect and socket contacts into a C-R-C filter. So don't grossly oversize the cable between the two chassis! A few tenths of an ohm and this local bypass cap (1000uf is convenient) will provide useful, beneficial HF filtering ahead of the regulator. If you wish, you might add an ohm or two at the output of your raw supply after the big reservoir cap to enhance this effect.
Picking up a couple of other points raised above - you really do want to keep the rectifier and caps close to the transformer. The charging loop is the noisiest thing in the whole supply, and extending raw AC into a second chassis for rectification is a sure-fire way to end up with more noise than a one-box solution - [because you've sent peaky charging currents around a much greater large loop area]
Secondly, the earlier suggestion of a larger cap at the input to the regulator(s) in the second chassis is a good one for a subtle reason - it turns the incidental slight resistance of your DC interconnect and socket contacts into a C-R-C filter. So don't grossly oversize the cable between the two chassis! A few tenths of an ohm and this local bypass cap (1000uf is convenient) will provide useful, beneficial HF filtering ahead of the regulator. If you wish, you might add an ohm or two at the output of your raw supply after the big reservoir cap to enhance this effect.
Last edited:
Martin, these are great points - thank you. I now understand the charging loop point that was raised about - this will all happen in box one in my design.
I like the idea of the 'bonus' CRC filter. I'd actually measured the resistance in the wire I was proposing to use and thought it a negative and was thinking of keeping it as short as possible but I guess as long as it's shielded from stray external noise then I don't need to worry so much about the length. I will also experiment with some resistors at the raw supply output too.
I like the idea of the 'bonus' CRC filter. I'd actually measured the resistance in the wire I was proposing to use and thought it a negative and was thinking of keeping it as short as possible but I guess as long as it's shielded from stray external noise then I don't need to worry so much about the length. I will also experiment with some resistors at the raw supply output too.
Things are progressing here. I have both ends of the power supplies built into the 2 cases and am starting to test and connect things up. So that I can safely assume that the capacitors have discharged I think adding a bleeder resistor is the next thing to do. Is there any reason why I shouldn't calculate an appropriate resistor value for an LED and add the LED in series with the resistor? This will then add a bleeder resistor into the circuit that also shows me what rails are working.
" if you have enough wires inside the shield (screen) to have V+, GND and V-, then by all means make the umbilical shield mains safety earth. It will be the best grounded for noise. Earth the PSU enclosure, and treat the shield and audio enclosure as that same potential safety earth"
Unless you want someone to die you absolutely MUST NOT use the umbilical shield as a mains earth for the second chassis. Not only is it totally unfit and unreliable for this purpose it also breaks just about every mains wiring regulation you may care to think about. The only permissible mains earth chassis connection is using a bolt, nut and shake proof washers connecting the crimped or soldered mains earth wire lug DIRECTLY to a clean metal part of the chassis.
Poor Andrew T will have a two headed baby if he reads this.
Unless you want someone to die you absolutely MUST NOT use the umbilical shield as a mains earth for the second chassis. Not only is it totally unfit and unreliable for this purpose it also breaks just about every mains wiring regulation you may care to think about. The only permissible mains earth chassis connection is using a bolt, nut and shake proof washers connecting the crimped or soldered mains earth wire lug DIRECTLY to a clean metal part of the chassis.
Poor Andrew T will have a two headed baby if he reads this.
Last edited:
The mains earth should be connected to the chassis close to the mains input to the chassis.
The umbilical shield should be connected to the first chassis to help shield the DC lines. Otherwise, the shield will not be as effective. The shield at the entrance to the second chassis should be connected to its chassis with a small cap of around 10nf or so, preferably a physically small ceramic with short leads. This will rf ground the second chassis and prevent a ground loop.
Since the 2nd chassis is transformer isolated from the AC mains, I doubt the fear of killing someone in the previous post is legitimate.
It's already been said, but splitting the 4700uf cap between the two boxes uses the resistance AND inductance of the umbilical line to help filter. The additional 1ohm series resistor between these split caps is also a good idea.
Thanks for the learning guys but is it OK to use the bleeder resistor in series with the LED as per question in #14 above?
Regarding the ubilical shielding, I was going to connect the shield to the first box and run the shield to the end of the umbilical but isolate it from the second box itself. For clarification, pooge, are you suggesting that I separately mains earth (ie an earth lead from the chassis to the earth pin of a mains socket)? Or, were you thinking that I connect the earth from box 1 to 2 with the 10nF ceramic?
I am quite keen to have a good earth from the second box as there are a few DC-DC convertors on the digital PCBs that I'd like to faraday cage or shield in another way and then earth for best shielding.
Finally (and thanks for your patience guys) just to check the wiring of the 1R resistors, I'm aiming for this right?
Please ignore values but in my psu 500uF would be in box 1 and 1000uF in box 2 with the 1R resistor we are talking about being represented as 50R in the above. I understand that we are looking to create an RC filter but I wanted to check whether I should mimick the 1R for the ground/neg line or just use it on the positive line as shown above...and if so, why?
Many thanks for ongoing help.
Regarding the ubilical shielding, I was going to connect the shield to the first box and run the shield to the end of the umbilical but isolate it from the second box itself. For clarification, pooge, are you suggesting that I separately mains earth (ie an earth lead from the chassis to the earth pin of a mains socket)? Or, were you thinking that I connect the earth from box 1 to 2 with the 10nF ceramic?
I am quite keen to have a good earth from the second box as there are a few DC-DC convertors on the digital PCBs that I'd like to faraday cage or shield in another way and then earth for best shielding.
Finally (and thanks for your patience guys) just to check the wiring of the 1R resistors, I'm aiming for this right?
Please ignore values but in my psu 500uF would be in box 1 and 1000uF in box 2 with the 1R resistor we are talking about being represented as 50R in the above. I understand that we are looking to create an RC filter but I wanted to check whether I should mimick the 1R for the ground/neg line or just use it on the positive line as shown above...and if so, why?
Many thanks for ongoing help.
- Status
- Not open for further replies.