Sherman & the Center Tap Mystery

[ofb]

in an older sub-thread, Sherman described a method for using a center tap transformer with two bridges. he made it with the briangt pcb kit. the described circuit is illustrated below, and it doesn't work because PG- and PG+ connect at the power star: AC1 and AC2 short. which is indeed what Steve found when he tried to follow Sherman's directions.

yet Sherman got it going somehow. it seems there must be missing information. does anyone know what that would be?

the sub-thread,
#386
http://www.diyaudio.com/forums/showt...876#post380876
#402
http://www.diyaudio.com/forums/showt...085#post382085
#414
http://www.diyaudio.com/forums/showt...245#post382245
#425
http://www.diyaudio.com/forums/showt...366#post383366
#444-#455
http://www.diyaudio.com/forums/showt...348#post388348
#464
http://www.diyaudio.com/forums/showt...148#post393148

and elsewhere
http://www.diyaudio.com/forums/showt...279#post598279
http://www.diyaudio.com/forums/showt...382#post548382



i've wondered if Sherman has flipped some diodes around on the pcb, instead of orienting them all in the same direction. perhaps it's possible to end up with what's essentially a single bridge CT circuit with some harmless run-around between the 'halves' of the bridge, but i haven't figured a combination that does that.

so, i'm dry of ideas. can someone else solve the mystery?

[lilolee]

[ofb]

thank you lilolee, but i'm afraid this is not the same thing.

the ESP circuit is running parallel bridges; one bridge per channel. the circuit i am describing does not have this separation. the briangt is series bridges that supply both channels. PG- and PG+ meet at the ground star of each channel.

Sherman's description and his drawings do not indicate he has changed this to one bridge per channel. he shows all lines from bridge PCB to channel PCBs as parallel.

[lilolee]

Sherman wrote the following

I have built a few power supplies with 2 bridges using CT transformers. It does seem to matter how you use the CT though. It would seem that (using the Brian GT boards nomenclature) you would wire one AC line to AC1H and the other to AC2H and then use the CT for AC1N and AC2N. When I tried that I blew fuses.

What I found that has worked for me is this-
AC1 from the transformer to AC1H on the board
CT from the transformer to AC1N on the board
CT from the transformer to AC2H on the board
AC2 from the transformer to AC2N on the board

diagram here

[ofb]

Originally posted by lilolee
Sherman wrote the following

What I found that has worked for me is this-
AC1 from the transformer to AC1H on the board
CT from the transformer to AC1N on the board
CT from the transformer to AC2H on the board
AC2 from the transformer to AC2N on the board

yes. this is what i have drawn.

diagram here

yes. this illustration shows parallel leads after the bridge PCB.

below is the PCBs in question. i have overlain the bridge schematic on a duplicate bridge image. there are two independant bridges with dual outputs. all four outputs go to each channel PCB as the illustration you have linked shows. then on the channel PCB you see PG+ and PG- join at the same ground. so these bridges are in series.

the ESP drawing is of parallel bridges; one per channel.



i'm afraid i do not see what you are trying to point out to me. this is a most uncomfortable feeling.

[megajocke]

That circuit should not work. The transformer is shorted for half of the cycle.

[ofb]

yes. that circuit should not work. that is indeed the point.

as i said in my opening post, this should not work. the curiosity is that this is what Sherman describes in detail, and at some length, as working for him. as i said, there must be missing information: something must be different in Sherman's actual arrangement that he did not see.

which is pretty odd, because there is not much to see.

we know he has a CT tx. we know there are eight diodes on the briangt bridge PCB. we know that four wires go from the bridge PCB to each channel PCB. we know PG- and PG+ meet on the channel PCBs.

which means this circuit pops the fuse or releases the magic smoke. yet his didn't. instead it powered his gainclone.

my best guess is he's done something odd with the arrangement of the diodes and gotten some sort of single bridge functioning out of it, but danged if i can replicate that either.

hence my question is, what did he do?

rest assured, i'm not asking how to make a series bridge arrangement work with a CT tx. you can't.

[garbage]

if i'm not mistaken, there is nothing wrong with the circuit.

like lilolee mentioned,

AC1 from the transformer to AC1H on the board
CT from the transformer to AC1N on the board
CT from the transformer to AC2H on the board
AC2 from the transformer to AC2N on the board

the above should just work.

let's say you have a 20/0/20 supply.

first 20-0 will be rectified to something like 28vdc, measured from V- and PG-.

second 20-0 will give 28vdc, measured from V+ and PG+.

PG- and PG+ would be the 0 in +28/0/-28vdc. V+ being the positive 28 and V- being the negative 28.

[richie00boy]

I've not built the circuit, but I did try simulating it by chance when investigating power supplies a while ago. I got output from it but the ripple on the lines was far worse than acceptable from other configurations.

[ofb]

garbage, PG- and PG+ meet at the channel PCB. hence AC1 meets AC2. also, AC1 meets CT, and CT meets AC2 -- all tx lines are shorted together when you join PG1 and PG2.

the circuit becomes identical to laying a diode from AC1 to CT, and a second diode from CT to AC2.