autoformer
I'll toss a wrench in there. What about an autoformer, it's VA rating is twice that of the same used as a transformer. Take a 1:1 transformer of say 120 vac pri. and 120 vac sec. Now take the primary and secondary and series them as aiding each other, now your autoformer is 2*VA of the transformer. Go figure.
I'll toss a wrench in there. What about an autoformer, it's VA rating is twice that of the same used as a transformer. Take a 1:1 transformer of say 120 vac pri. and 120 vac sec. Now take the primary and secondary and series them as aiding each other, now your autoformer is 2*VA of the transformer. Go figure.
I'm still confused (I'm an electrical engineer, I'm not supposed to be confused, but I am).
If I take a 17V dual winding output transformer, and want to make a dual rail supply, there are two options open to me:
1) Connect the windings in series to make a centre tap, then use a single bridge (4 diodes) to make + and - rails in the usual way, with the CT being 0V.
2) Connect each winding to its own bridge (8 diodes total), one feeding the + rail, and the other the -.
I've read on here in the past that 2 is better than 1 for transformer utilisation.
I had worked out that this only happens if one rail is loaded more than the other (eg low frequency bass kick, or whatever usually generates high currents in an amp) because both half cycles of the winding are then used. (With option 1, each winding only supplies a half cycle, increasing I2R losses.)
But I just realised this is wrong. If only one rail is loaded in option 1, then only one winding will deliver any current at all = same I2R losses. If there was any long term chance of imbalance (not in an audio amplifier) then option 2 is actually worse because one winding will get hotter.
And of course there are extra diode losses in 2 (although they can be half the voltage rating). There is the possibility of more switching events in 2 as well, but as the diodes are in series this is only down to second order things like different reverse recovery times of the diodes etc. Blah de blah.
Therefore I'm (also) left thinking they're electrically identical apart from the diode drops, including transformer and everything. Can anyone shed any more light on this or provide real-world experience, including sonic differences? Of course if your transformer is hardwired with a centre tap you're stuck with option 1 anyway.
Thanks - Antony.
If I take a 17V dual winding output transformer, and want to make a dual rail supply, there are two options open to me:
1) Connect the windings in series to make a centre tap, then use a single bridge (4 diodes) to make + and - rails in the usual way, with the CT being 0V.
2) Connect each winding to its own bridge (8 diodes total), one feeding the + rail, and the other the -.
I've read on here in the past that 2 is better than 1 for transformer utilisation.
I had worked out that this only happens if one rail is loaded more than the other (eg low frequency bass kick, or whatever usually generates high currents in an amp) because both half cycles of the winding are then used. (With option 1, each winding only supplies a half cycle, increasing I2R losses.)
But I just realised this is wrong. If only one rail is loaded in option 1, then only one winding will deliver any current at all = same I2R losses. If there was any long term chance of imbalance (not in an audio amplifier) then option 2 is actually worse because one winding will get hotter.
And of course there are extra diode losses in 2 (although they can be half the voltage rating). There is the possibility of more switching events in 2 as well, but as the diodes are in series this is only down to second order things like different reverse recovery times of the diodes etc. Blah de blah.
Therefore I'm (also) left thinking they're electrically identical apart from the diode drops, including transformer and everything. Can anyone shed any more light on this or provide real-world experience, including sonic differences? Of course if your transformer is hardwired with a centre tap you're stuck with option 1 anyway.
Thanks - Antony.
Long ago, I tried to simulate including several parasitics and component mismatchesa a power supply with 1 or 2 bridges. Here is what I observed
2 bridges provides:
+ slightly better power factor
+ less stress on secondaries and bridges
+ in case of constant/symetrical current sink on both rails, nearly no current flowing in grounds => less risk to saturate a toroidal transformer?
+ less "accidents" and lower pikes in return currents
+ secondary voltages ride nearer to ground
= 100Hz ripple looks the same
= current ripple in capacitors looks the same
= with non-constant and asymetrical current sink, no great diff in ground currents
= no great diff versus 1 bridge for small variations of 1 transfo winding vs the other
- voltage lost in rectification is doubled
- power dissipated by bridges is doubled
My 2 cents...
2 bridges provides:
+ slightly better power factor
+ less stress on secondaries and bridges
+ in case of constant/symetrical current sink on both rails, nearly no current flowing in grounds => less risk to saturate a toroidal transformer?
+ less "accidents" and lower pikes in return currents
+ secondary voltages ride nearer to ground
= 100Hz ripple looks the same
= current ripple in capacitors looks the same
= with non-constant and asymetrical current sink, no great diff in ground currents
= no great diff versus 1 bridge for small variations of 1 transfo winding vs the other
- voltage lost in rectification is doubled
- power dissipated by bridges is doubled
My 2 cents...
Referring to the diagram,
http://www.diyaudio.com/forums/attachment.php?s=&postid=143874&stamp=1047429336
The true electrical equivalent between the two schematics would be if Rw1 and Rw2 (in the bottom) were connected (as they are in the above diagram). Then you get the exact same concerns about ground loop and whatnot. In both cases, the solution is simple -- ignore Vgnd3 (or its doubled equivalent on the dual circuit) altogether, using only Vgnd4, 5, 6, ...
Which you are supposed to do anyway. Basic grounding technique. Anyone worrying about this needs a lot help on layout.
Tim
http://www.diyaudio.com/forums/attachment.php?s=&postid=143874&stamp=1047429336
The true electrical equivalent between the two schematics would be if Rw1 and Rw2 (in the bottom) were connected (as they are in the above diagram). Then you get the exact same concerns about ground loop and whatnot. In both cases, the solution is simple -- ignore Vgnd3 (or its doubled equivalent on the dual circuit) altogether, using only Vgnd4, 5, 6, ...
Which you are supposed to do anyway. Basic grounding technique. Anyone worrying about this needs a lot help on layout.
Tim
Thanks to all. It makes more sense in the greater scheme of things now - there is no fundamental difference in transformer utilisation.
The ground current thing is why I was most intersted in the 2 bridge option. The biggest advantage that I could see is that it allows pulsing supply feed current to be led directly to the terminals of each cap, which reduces supply ripple a little:
There will always be a finite resistance connecting the caps (RW1 and RW2 in Fred's single-bridge diagram). Given a certain amount of copper in the cap ground (RW1-4), a single bridge will generate a set amount of extra ripple across the supply: If you take your ground from Vgnd3 (ground star) all the ripple appears on the loaded rail. If you use Vgnd4 (a ground "diamond" of the type I've been playing with recently) then half the ripple appears on each rail (the total across both rails is the same).
Note this isn't ground ripple - it only appears on the supply rails, because you choose your ground. I can't think of any reason to use more than one of the grounds shown (in a mono amp...).
But given the ease of connecting supply caps together with a resistance significantly smaller than their internal resistance, the ripple advantage of the 2 bridge option is now looking insignificant to me.
There are potentially a lot of 'second order' differences and practicalities in a real amp of course, but I'm now happy that the 1 and 2 bridge options are essentially electrically identical.
The ground current thing is why I was most intersted in the 2 bridge option. The biggest advantage that I could see is that it allows pulsing supply feed current to be led directly to the terminals of each cap, which reduces supply ripple a little:
There will always be a finite resistance connecting the caps (RW1 and RW2 in Fred's single-bridge diagram). Given a certain amount of copper in the cap ground (RW1-4), a single bridge will generate a set amount of extra ripple across the supply: If you take your ground from Vgnd3 (ground star) all the ripple appears on the loaded rail. If you use Vgnd4 (a ground "diamond" of the type I've been playing with recently) then half the ripple appears on each rail (the total across both rails is the same).
Note this isn't ground ripple - it only appears on the supply rails, because you choose your ground. I can't think of any reason to use more than one of the grounds shown (in a mono amp...).
But given the ease of connecting supply caps together with a resistance significantly smaller than their internal resistance, the ripple advantage of the 2 bridge option is now looking insignificant to me.
There are potentially a lot of 'second order' differences and practicalities in a real amp of course, but I'm now happy that the 1 and 2 bridge options are essentially electrically identical.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.