stacked supply from a bridge.
-
- Posts: 2092
- Joined: Sat Jan 22, 2005 3:54 am
- Location: NYC
- Contact:
stacked supply from a bridge.
we all know that if you feed a bridge with a CT transformer and ground the CT you get a +- supply. what happens if you ground the "-" side?
see attached pic and spice file. the numbers are arbitrary, but remove the .1uf cap and see what happens. Also play with the ground reference point and see what happens.
dave
see attached pic and spice file. the numbers are arbitrary, but remove the .1uf cap and see what happens. Also play with the ground reference point and see what happens.
dave
- Attachments
-
- bridge.asc
- (3.23 KiB) Downloaded 701 times
-
- Picture 7.png (15.64 KiB) Viewed 13149 times
-
- Posts: 2092
- Joined: Sat Jan 22, 2005 3:54 am
- Location: NYC
- Contact:
well i'm listening to said supply as we speak. It is a slight rework of an existing amp and it seems quiet...
the last time i tried this concept it used a different hookup (chokes in the CT leg) and things were not so good. (chokes in the upper supply current return path ??)
It is for a stacked supply monkey amp using a single CT trannie for a stacked supply with one being cap input and the other choke input.
seems to work, i get 170V and 230V for a total top to bottom of 400V...
the sim really seems to burp upon trying to figure things out, but it works on the bench.
i'm keeping it here since the burps in spice seem to interest me more than the amp itself
dave
the last time i tried this concept it used a different hookup (chokes in the CT leg) and things were not so good. (chokes in the upper supply current return path ??)
It is for a stacked supply monkey amp using a single CT trannie for a stacked supply with one being cap input and the other choke input.
seems to work, i get 170V and 230V for a total top to bottom of 400V...
the sim really seems to burp upon trying to figure things out, but it works on the bench.
i'm keeping it here since the burps in spice seem to interest me more than the amp itself
dave
- Attachments
-
- Picture 11.png (21.41 KiB) Viewed 13142 times
-
- Posts: 96
- Joined: Sun Jan 23, 2005 10:58 pm
I ran the sim (the first one) and couldn't tell what I was supposed to be noticing when removing the little cap, etc. The sim ran painfully slow, but I'm not sure what burps you're talking about.
Anyway, it hurts my head trying to figure out the current paths through the ps when used in a monkey. I mean, it should be simple enough, but convincing myself that one supply is choke input while the other isn't.... But, the measurements of the supply on the bench don't lie, do they?
-- Dave
Anyway, it hurts my head trying to figure out the current paths through the ps when used in a monkey. I mean, it should be simple enough, but convincing myself that one supply is choke input while the other isn't.... But, the measurements of the supply on the bench don't lie, do they?
-- Dave
-
- Posts: 2092
- Joined: Sat Jan 22, 2005 3:54 am
- Location: NYC
- Contact:
Hi Dave and Dave,
Sim speed... in your circuit, you have essentially 3 "current sources" (the 2 secondary inductors and the 5 Hy choke) joined together with no parasitic elements (read real world components), and spice has to go to VERY small step sizes to resolve the matrix. I suspect you had the cap in place schematically but when you didn't want to use it, you changed it to something like 1p. Note what happens to the sim speed when you delete the cap altogether. There a are a number of alternatives:
1. change transformer to much smaller L values (480 Henry is much higher than reality, by a factor of over 100). Doing nothing else other than going to 1.2H, 4.8H, 4.8H speeds the sim substantially.
2. Add parasitic resistance. Even 1 ohm in series with each of the secondary windings helps.
3. Instead of the capacitor, note that a 100k resistor in the same spot speeds the sim.
(4. Sometimes it's much better to include a non-unity coupling coefficient. (use .99999 instead of 1). This is not one of those cases, though.)
Topology: Fairly common overall. That's a convenient way to get 2 voltages about a factor of 2 apart.
Spice issues: For folks doing power supply simulation, please note that the signal level you are putting in is PEAK value (not RMS, and not peak to peak). Thus to accurately simulate 120v RMS, you should enter the value 169.7 (or thereabouts).
Regards,
Steve
Sim speed... in your circuit, you have essentially 3 "current sources" (the 2 secondary inductors and the 5 Hy choke) joined together with no parasitic elements (read real world components), and spice has to go to VERY small step sizes to resolve the matrix. I suspect you had the cap in place schematically but when you didn't want to use it, you changed it to something like 1p. Note what happens to the sim speed when you delete the cap altogether. There a are a number of alternatives:
1. change transformer to much smaller L values (480 Henry is much higher than reality, by a factor of over 100). Doing nothing else other than going to 1.2H, 4.8H, 4.8H speeds the sim substantially.
2. Add parasitic resistance. Even 1 ohm in series with each of the secondary windings helps.
3. Instead of the capacitor, note that a 100k resistor in the same spot speeds the sim.
(4. Sometimes it's much better to include a non-unity coupling coefficient. (use .99999 instead of 1). This is not one of those cases, though.)
Topology: Fairly common overall. That's a convenient way to get 2 voltages about a factor of 2 apart.
Spice issues: For folks doing power supply simulation, please note that the signal level you are putting in is PEAK value (not RMS, and not peak to peak). Thus to accurately simulate 120v RMS, you should enter the value 169.7 (or thereabouts).
Regards,
Steve
-
- Posts: 2092
- Joined: Sat Jan 22, 2005 3:54 am
- Location: NYC
- Contact:
hey steve,
thanks for the reply!
I also am getting what appears to be full choke input from one supply and cap input from the other. This buys me the operating points i need. (230 for the 2A3 and 170 minus the 40V bias drop for the 437) In this case the 60V difference between the supplies makes a big difference. Don't get me wrong, i would gladly use two transformers and simply stack the supplies, but the original goal was to use a 350VCT off the shelf toroid and that is what is in the box.
Someone on another forum suggested a possible DC offset.... While I do see two tubes glow more (explaining the difference in current draw between the supplies) i could not find an offset either conceptually by looking at the schematic or by looking at the currents in spice. In spice i could see the difference in current draw between the cap and choke input filters though but it the current appeared symmetrical and out of phase through each half.
the FFT of the current from each secondary half was close and the FFT of the primary showed only odd orders which further suggests perfect AC balance doesn't it?
thoughs?
dave
thanks for the reply!
understood. i put the 120 number in there since it would be easy to remember, in hindsight it was a mistake because people would confse it for voltage.sbench wrote: 1. change transformer to much smaller L values (480 Henry is much higher than reality, by a factor of over 100). Doing nothing else other than going to 1.2H, 4.8H, 4.8H speeds the sim substantially
are you referring to 1 ohm external to the coil?? I had 50R entered into the coil. I was using a value of 1n for the cap just to get it to sim... the actual circuit doesn;t have anything intentional there but we all now there is some C across the secondary.2. Add parasitic resistance. Even 1 ohm in series with each of the secondary windings helps.
yeah, that is what started this whole idea 4 odd years ago. it hummed like a banshee. It may very well have been a grounding issue since i simply need to move the ground reference around in spice and things change drastically.Topology: Fairly common overall. That's a convenient way to get 2 voltages about a factor of 2 apart.
I also am getting what appears to be full choke input from one supply and cap input from the other. This buys me the operating points i need. (230 for the 2A3 and 170 minus the 40V bias drop for the 437) In this case the 60V difference between the supplies makes a big difference. Don't get me wrong, i would gladly use two transformers and simply stack the supplies, but the original goal was to use a 350VCT off the shelf toroid and that is what is in the box.
duh... of course it is peak... that explains why my inductance numbers numbers didn't match up with my measured output. Entering 170V as AC made the required inductance numbers for the secondaries close enough for gov'ment work with the measured reality. THANKS!Spice issues: For folks doing power supply simulation, please note that the signal level you are putting in is PEAK value (not RMS, and not peak to peak). Thus to accurately simulate 120v RMS, you should enter the value 169.7 (or thereabouts).
Someone on another forum suggested a possible DC offset.... While I do see two tubes glow more (explaining the difference in current draw between the supplies) i could not find an offset either conceptually by looking at the schematic or by looking at the currents in spice. In spice i could see the difference in current draw between the cap and choke input filters though but it the current appeared symmetrical and out of phase through each half.
the FFT of the current from each secondary half was close and the FFT of the primary showed only odd orders which further suggests perfect AC balance doesn't it?
thoughs?
dave
-
- Posts: 96
- Joined: Sun Jan 23, 2005 10:58 pm
It's a common first-glance reaction to this kind of supply, thinking that one half-winding draws more current than the other. Obviously, that's not the case; both half-windings contribute equally to both supplies just as they would with a simple FWCT arraignment.Someone on another forum suggested a possible DC offset....
-- Dave
-
- Posts: 2092
- Joined: Sat Jan 22, 2005 3:54 am
- Location: NYC
- Contact: