Hi,
I badly need some advice before ordering parts......
I am about to recap my Cyrus PSX unregulated supply that powers my Cyrus 2 Solid state stereo amp.
Can I safely change from 15000uF caps to 22000uF?
I have no idea how to work it out.
The transformer is 500VA.
The rails are +/-40V DC.
The fuse on each rail is 4 Amps.
The original total capacitance was 60000uF.
I would like to make it 88000uF!!
The rectification uses pairs of parallelled diodes which I believe to be 3 Amp each but I am inexperienced.
PFR852 Datasheet
Do you think a 47% increase in capacitance is safe in this case? Is the inrush current a concern for the diodes?
Thanks in advance,
Martin.
EDIT: The regulators on the right of the picture are for a low voltage supply that I never use. Cyrus intended that this part be used with a C.D. player.
I badly need some advice before ordering parts......
I am about to recap my Cyrus PSX unregulated supply that powers my Cyrus 2 Solid state stereo amp.
Can I safely change from 15000uF caps to 22000uF?
I have no idea how to work it out.
The transformer is 500VA.
The rails are +/-40V DC.
The fuse on each rail is 4 Amps.
The original total capacitance was 60000uF.
I would like to make it 88000uF!!
The rectification uses pairs of parallelled diodes which I believe to be 3 Amp each but I am inexperienced.
PFR852 Datasheet
Do you think a 47% increase in capacitance is safe in this case? Is the inrush current a concern for the diodes?
Thanks in advance,
Martin.
EDIT: The regulators on the right of the picture are for a low voltage supply that I never use. Cyrus intended that this part be used with a C.D. player.
Attachments
Hi Sonus,
if you want to play safe then lift one end of each rectifier and series connect 0r1 back into the empty hole.
Since each bridge component is a parallel pair of diodes you gain twice from this modification.
Firstly, parallel diodes don't share current too well and the one that runs hot carries more current and then it runs hotter etc. Adding current sharing resistors helps the diodes to run at nearer equal current and thus not overload any single one.
Secondly, the added resistance 0r1//0r1 =0r05 in each leg of the bridge adds 0r1 to the total circuit resistance and this reduces the peak charging current. It also widens the charging pulses slightly, reducing the Isquared R losses in the transformer resulting in slightly (probably unmeasurable) cooler running.
Now you are ready for the smoothing capacitor value increase.
Personally, I would go for it without the extra resistors, but that is the way it should have been done by Cyrus (unless they found that it sounded nicer without). I have 15mF//4m7F fed from 1n5404 in discrete bridge and no problems (it was originally 4m7F//4m7F) but you are proposing more than double that.
You could add a soft start in the space at the mains cable end.
if you want to play safe then lift one end of each rectifier and series connect 0r1 back into the empty hole.
Since each bridge component is a parallel pair of diodes you gain twice from this modification.
Firstly, parallel diodes don't share current too well and the one that runs hot carries more current and then it runs hotter etc. Adding current sharing resistors helps the diodes to run at nearer equal current and thus not overload any single one.
Secondly, the added resistance 0r1//0r1 =0r05 in each leg of the bridge adds 0r1 to the total circuit resistance and this reduces the peak charging current. It also widens the charging pulses slightly, reducing the Isquared R losses in the transformer resulting in slightly (probably unmeasurable) cooler running.
Now you are ready for the smoothing capacitor value increase.
Personally, I would go for it without the extra resistors, but that is the way it should have been done by Cyrus (unless they found that it sounded nicer without). I have 15mF//4m7F fed from 1n5404 in discrete bridge and no problems (it was originally 4m7F//4m7F) but you are proposing more than double that.
You could add a soft start in the space at the mains cable end.
Thank you, yet again, for all your advice Andrew.
I have ordered the 22mF caps tonight!! Mundorf M-Lytic 50V from HIFI Collective.
I find it interesting that adding such a small resistance to the diodes can make a difference. I have so much to learn. For the sake of neatness and simplicity I'd rather not add resistance to the diodes at the moment.
I had sometimes daydreamed about adding a small resistance to 2 of the caps to set the charging cycles slightly apart from the others. Just a thought but I don't really know enough yet to theorise like this. This is probably a stupid idea.
Am I right in thinking that the diodes are the only things being stretched by this modification?
I'm always nervous when messing with the big stuff like power supplies.
Could I just fit better diodes (5A or 8A) and relax or are there other considerations?
I could add a soft start if necessary but would rather avoid it if at all possible.
Many thanks,
Martin.
I have ordered the 22mF caps tonight!! Mundorf M-Lytic 50V from HIFI Collective.
I find it interesting that adding such a small resistance to the diodes can make a difference. I have so much to learn. For the sake of neatness and simplicity I'd rather not add resistance to the diodes at the moment.
I had sometimes daydreamed about adding a small resistance to 2 of the caps to set the charging cycles slightly apart from the others. Just a thought but I don't really know enough yet to theorise like this. This is probably a stupid idea.
Am I right in thinking that the diodes are the only things being stretched by this modification?
I'm always nervous when messing with the big stuff like power supplies.
Could I just fit better diodes (5A or 8A) and relax or are there other considerations?
I could add a soft start if necessary but would rather avoid it if at all possible.
Many thanks,
Martin.
Hi Sonus,
the charging spike is quite short (probably of the order of 5% to 10% of the half wave cycle). If the capacitance is lowered, the cap discharges more between each recharge and the Mains then takes longer to recharge it (lengthens the duration of the charge spike).
Conversely, more smoothing capacitance shortens the charging spike. This may have repercussions for sound quality downstream. All our attempts at decoupling and bypassing are there to alleviate just this sort of thing.
Some experiments with To220 packaged, fast or slow or soft recovery or Schottky may be worthwhile, but are relatively expensive, particularly if they don't improve the situation.
RC snubbers across the diodes could be worth investigating.
the charging spike is quite short (probably of the order of 5% to 10% of the half wave cycle). If the capacitance is lowered, the cap discharges more between each recharge and the Mains then takes longer to recharge it (lengthens the duration of the charge spike).
Conversely, more smoothing capacitance shortens the charging spike. This may have repercussions for sound quality downstream. All our attempts at decoupling and bypassing are there to alleviate just this sort of thing.
Some experiments with To220 packaged, fast or slow or soft recovery or Schottky may be worthwhile, but are relatively expensive, particularly if they don't improve the situation.
RC snubbers across the diodes could be worth investigating.
Hi Andrew,
The original diodes are fast recovery so Cyrus must have seen some merit in this approach.
Fancy diodes are more expensive than I'd anticipated but I have found some BYW80200 200V, 8A, 35ns TO220 devices on Ebay.
I can get 8 of these delivered for 9.40GBP from Ebay.
I'll be using parallel pairs of these but will have to improvise the fitting. The originals are axial and the gaps in the copper trace are too wide for these to fit so I'll get some thick wire and solder the new devices to that. Parallel pairs of these may well be overkill but a high percentage of the cost is actually postage so it seems silly not too. Belts and braces!
BYW80200
R.C. Snubbers? That's a whole new can of worms.
I will have to snubberise everything!!!😉
I find the whole snubber thing very interesting though and will try it first on the diodes and then on the large caps.
Cheers,
Martin.
The original diodes are fast recovery so Cyrus must have seen some merit in this approach.
Fancy diodes are more expensive than I'd anticipated but I have found some BYW80200 200V, 8A, 35ns TO220 devices on Ebay.
I can get 8 of these delivered for 9.40GBP from Ebay.
I'll be using parallel pairs of these but will have to improvise the fitting. The originals are axial and the gaps in the copper trace are too wide for these to fit so I'll get some thick wire and solder the new devices to that. Parallel pairs of these may well be overkill but a high percentage of the cost is actually postage so it seems silly not too. Belts and braces!
BYW80200
R.C. Snubbers? That's a whole new can of worms.
I will have to snubberise everything!!!😉 I find the whole snubber thing very interesting though and will try it first on the diodes and then on the large caps.
Cheers,
Martin.
- Status
- Not open for further replies.