marshall 2203 800v on B+

[jcm-jmp]

hi guys
 
im building a 2203
 
pt bogen from a chb100
368-0-368
 
put the recto in with the 50/50uf in series with the other 50/50uf.
my voltage shot up to 880vdc
the first cap which connects to the output of the bridge recto got extreemly hot. it didnt let out smoke, but the blue f&t plastic around the cap can did melt and split. cap got way too hot.
 
i disconnected the pt from the recto and caps. the voltage from ct to each leg was good
368-0-368
 
1. why would have this cap overheated and failed.
2. whats the reason for the extreemly high B+ voltage (880vdc)
 
thanks robert

[eleventeen]

It is not quite clear to me whether you have built the Marshall on a fresh "new" chassis or exactly what you have going on.


However, strictly from the (very high) voltage you report it sounds like you are feeding a voltage doubler with your 368---I think the factor is roughly 2.5, no load. 368 * 2.4 = 883 VDC. Whether you have planned it that way or not.


If your power transformer is indeed 368-0-368 then you do not want a bridge rectifier with four diodes; you want a "two diode" full wave rectifier much like a Fender Twin Reverb. That should give you 1.4 * 368 = 515 which is much closer to your ballpark. It's still a tad high, maybe 50 volts, but not completely out of control like 880 volts!  You could reduce that pretty easily with the zener diode trick.

http://el34world.com/Forum/index.php?topic=17595.0

Also, if you are going to put e-caps in series like the Marshall schematic, it's highly recommended you use "balancing" resistors in parallel with each cap. The traditional value is 220K/1 watt, though in a new build I would probably want 2 watt or even 3 watt for no good reason other than overengineering, but you are more likely to get flameproof resistors selecting new power resistors. In truth, only about 1 mil flows through those resistors: They split a nominal 460 volts (in a Twin) or 230 volts each. Throw a 220K resistor across them and that R pulls ~~1 mil.  230/220K = .001 amp = 1 mil.

Alternatively, you could use lower value resistors as your balancers (as the Marshall does) and they would drop a few more volts. But they will start to dissipate power = heat. Do the math, as they say.

With that tranny, I would say to go look at a Fender AB763 Twin reverb power supply and duplicate it, exactly. I take it your ex-Bogen power transformer does not have a bias tap. You can just steal a side of the HV winding, run it through a "backwards" diode, voltage divider, and obtain your bias in that fashion.


Hope you didn't smoke your e-cap but I suspect you did.

[DummyLoad]

caps are now junk.


368-0-368 is NOT a bogen chb100 power transformer.


if that is a 368-0-368 (meaning it's 736V winding WITH a center tap), then use a full wave capacitor input and NOT full wave bridge.


http://www.hammondmfg.com/pdf/5c007.pdf


--pete

[jcm-jmp]

it is a bogen chb100 pt no doubt about it.
 
found the problem
 
if you look at the attached schem you will see the neg side of recto being tied to ground reference.
dont do this lol not good for caps or your iron.
 
thanks for the input guys
 
lifted the ground reference from recto neg output and now she is setting at 128v steady.

[jcm-jmp]

528vdc typo sorry

[jcm-jmp]

Ok guys
 
i have built a jcm800.
 
1. Bogen CHB100 PT 368v-0v-368v
    (no seperate bias winding on PT)
2. Dagnall 100w OT out of a JCM2000 TSL100
3. 4h choke
4. All F&T 50uf/50uf 500v cans (6 total)
 
 The amp is built to the 2203 layout with exception of, I only installed 2-8 pin octal sockets and I am running 2-KT88'
 
                                 (EXAMPLE #1 FROM ATTACHED DRAWING)
When i first fired up the unit I had disconnected the B+ rail from the plate and choke input node so that the only voltage I would read was from the (strange) voltage doubler (not really a doubler)
rectofier design Marshall used on the 2203 models. i got a stable 500+ volts for about 2 minutes then the voltage began to climb slowly and eventually to 880v where at this point the top 50uf/50uf 500v cap began to break down. the PT became extreemly hot as well. I shut the amp off.
 
                                  (EXAMPLE #2 FROM ATTACHED DRAWING)
I removed the the ground reference from the negative output of the rectofier and left the CT of PT connected at the negative side of the top 50/50uf 500v can at the node that joins the top 50/50uf can negative to the bottom 50/50uf can posative. The bottom 50/50uf can's negative goes to ground.
(referance attached drawing) When connected this way the amplifier works and sounds pretty good.
The issue is that the B+ at the rectofier output is 500+vdc when no tubes are installed, but when i put the tubes in (KT88's) the rectofier output drops to right at 300vdc.This seems like way to much voltage drop? I replaced the failed capacitor mentioned in example #1
 before firing up this example (#2). The bottom cap of the 50/50uf series connected caps is getting pretty hot as is the PT in this connection configuration.
 
voltage readings in this configuration:
 
rectofier output: 520vdc (before standby is turned on with tubes installed)
                        336vdc (after standby is turned on with tubes installed)
Plates: 330vdc
screens: 321vdc (with 5W 1K grid R)
PI plates: 180vdc & 187vdc
v2a plate: 242vdc
v2a grid: 0vdc
v2b plate: 234vdc
v2b grid: 141vdc
v1a plate:163vdc
v1a grid: 0vdc
v1b plate: 132vdc
v1b grid: 0vdc
 
bias: 330vdc*.0835a= 27.5w
        35w*.7=24.5w or 74ma
so im about 10ma hot or biased at about 80% max disapation,
but this is where it sounds best.
         
 
                                  (example #3 & #4)
These are connections that I speculate could possibly be better or correct? I am unsure at this point of which connection drawing would be the prpoper one to use.
 
I would like to have around 488vdc on the plates of the KT88's (with them re-biased for this voltage of course).
I think this would help with the over heating issue I'm having with the bottom 50/50uf cap and PT.
 
 

[sluckey]

None of those rectifier options will work with a 368-0-368 PT. Look at the power supply in this schematic and copy it. You can still put your cap cans in series if they won't handle 520vdc.

http://home.comcast.net/~seluckey/amps/misc/Marshall_Dual_50.pdf

[jcm-jmp]

thanks for reply slucky
 
thats is how i had the recto the first build (not mentioned in this post had 50w OT iron in at that time) i changed OT and wanted high plate voltages. it souned good like that but just didnt put out like a 1981 2204 i service for a friend. i put kt88's in his. did the bias mods and that thing is a beast!
 
i built mine while refering to his for guidence,but it just never had the balls his did. his plate voltage is 488vdc loaded with kt88's this was the only difference at the time. so... i started studying the differences between the 2203 and 2204 models and found the the 2204 achieved higher plate voltages because the secondary on the PT was at a higher voltage. the 2203 achieved the higher plate voltages with lower secoday PT voltages 380-0-380 through that wierd recto in parralle with those caps.
 
so..... hmmmm my PT is 368-0-368. Wait for it! YEP! i did. UHG! reworked the entire B+ rail to mimic the 2203 design.
 
well the rest is what i posted above.
 
my question is how do i get those plate voltages up to 480v?
a 100v drop in plate voltage due to kt88's in my build vs a 40v drop in plate voltage due to kt88's in a true marshall 2204.
 
something is dragging that B+ rail down.
 
ps. that #1 example is drawn exactly as the 2203 power section drawing that i have ( see attached below) when i installed that way caused major melt down to the cap.
 
Thanks
 
 ps. the 2203 or 100w power section is on third page.

[sluckey]

my question is how do i get those plate voltages up to 480v?

Simply follow my advice in reply #6.

That 368-0-368 will give you 520vdc unloaded. Probably be just about right with tubes in. But you gotta use the right circuit! Don't forget the bias circuit.

[jcm-jmp]

ok thanks slucky
 
went back to the original recto circuit like in your attachment. but left the 2 - 50/50uf in series at the recto. It did get voltage up to 520vdc unloaded, but these kt's are some current sucking SOB's. The plate voltage still dropped to 423vdc when they were put in, but the amp does have the punch now.
 i did an a/b comparison with the real 2204 mentioned above and there are only very slight differences, which im sure are to do with me removing the treb clip circuit and a couple bypass caps on the first 2 gain stage's cathodes.
 
I'm happy the amp is moderatly happy lol. As happy as it can be with KT88's at any rate lol.
 
thanks guys for your input, always a good results.
 
Thank you