5U4's in parallel

[Cliff Schecht]

I want to build a 5E8 and have a transformer that is up to the job heater-wise. The HV winding is only 540V center-tapped and so I'll probably crank up the plate current and adjust the droppers to compensate.

But I've never used 5U4's, let alone two of them in parallel. What's the typical voltage drop and multiplication factor for one? How about the same for two in parallel?

[tubeswell]

The voltage drop is the same, but the DC peak current deliverable from 2 in parallel is 2 x what one tube is*, so it is a lot less likely to sag. You need a 6A rectifier winding for 2 x 5U4G in parallel

* IIRC a single 5U4GA delivers 250mA, and a 5U4GB delivers 275mA, but a 5U4G delivers only 225mA (whereas a Sovtek '5U4G' delivers 250mA - confused?)

[FYL]

But I've never used 5U4's, let alone two of them in parallel. What's the typical voltage drop and multiplication factor for one? How about the same for two in parallel?

There are *many* different 5U4s. A 5U4G isn't the same as a 5U4GB, etc.

The voltage drop can be calculated from the datasheets, for instance Sylvania gives 44 V @ 225 mA for the 5U4G (http://www.r-type.org/pdfs/5u4g.pdf), GE 50 V @ 275 mA or the 5U4GB (http://www.r-type.org/pdfs/5u4gb.pdf).

Using good ole Ohm's law, plate R is respectively 196R and 182R - close enough, let's say 190R.

A pair of 5881s in AB1 biased at 70% will draw app. 90 mA at idle, and maybe 150 mA cranked. Voltage drop will be 17 V at idle and 28 V max.

Two valves in // = shared current. Scale the results and you're done : 9 V and 14 V for the valves, plus iron losses etc.

[Firestorm]

I think you'll be disappointed by the actual B+ voltage available from this transformer into two 5U4s with a choke filter (as in the 5E8). The regulation of a tube rectifier is not linear, so Ohm's Law only gets you into the right territory. More significantly, the curves are quite different depending on whether you use a choke or cap filter; you "lose" quite a bit more voltage with the choke.

Just eyeballing the characteristics chart in RC-30 for 5U4GB (not even trying to apply the diode nomograph) looks like this: For an AC supply of 270V RMS per plate and a choke of at least 10H, DC output at the choke input is about 235VDC@50mA load current; 210VDC@100mA; 200VDC@150mA. As FYL says, with two tubes, you divide the current.

If you diverge from the 5E8 and use a cap input to filter: 270V RMS gets you DC output of about: 350VDC@50mA; 315VDC@100mA; 300VDC@150mA.

The charts assume a certain value for plate supply impedance (in the primary and secondary windings) but every transformer is different, so ymmv.

You could improve things by using 5AR4s, which have a higher perveance and should get you more like 230VDC@100mA (with a choke).

If you feel you must use 5U4s, let me recommend the Electro Harmonix version. It is the only 5U4 I've ever liked (and that includes NOS RCAs and the much-vaunted Mullard GZ37).

I don't know what the VA rating of your transformer is, but if it's sufficient you could abandon the dual parallel rectifier and wire them as a bridge instead to get voltages very much like the original (a little higher it looks like). The 5U4 should certainly be able to handle the current requirements even in that configuration.

[Cliff Schecht]

Yes! I was wondering about doing a bridge but couldn't find an example anywhere. How is this usually done?

[Alexo]

Yes... not sure why you'd want to use 2 5U4's, unless you were running an octet of 6L6's or something and needed huge amounts of current. 2 rectifier tubes = half the sag, so I would maybe just opt for a GZ34 if that's what you're after.

A bridge rect is not quite possible wihout at leqast 3 rectifier tubes - the problem is that the cathodes are always tied together, and you need them separated for the negative voltage reference.

Directly heated cathodes present another problem for the bridge.

The way to do it right would be a quad of indirectly heated rectifiers, like EZ81's - link the anodes on each one, so that each is acting like a great big diode. Personally, I would just build a diode bridge, then run its output through a 5U4 or whatever if you wanted 'real tube' sag.


EDIT: OK, so you are building a 5E8 and you may want it to be as accurate as possible, hence the 2x5U4's, even if this is an outmoded choice (what isn't?). But if you're not using an authentic PT to start, in terms of inductance, current and voltage ratings, your PSU authenticity is sorta out the window from the get-go.

[Cliff Schecht]

Yeah authenticity is not a huge concern, but I'd like to get the right feel of the amp. These are close to a Bandmaster from the same era with just 1 5U4 and it's common in the 5E8 to pull one of the 5U4's for more sag. I just thought it would be cool to have the option, plus I have tons of those massive military version 5U4's. I may end up using them in a Rocket build coming up for a friend instead. The PT is friggin large but it might fit the bill if I can't find anything smaller (I'd have to figure out how to vertically mount it too!).

[Firestorm]

A bridge rect is not quite possible wihout at leqast 3 rectifier tubes - the problem is that the cathodes are always tied together, and you need them separated for the negative voltage reference.

Plus, three heater windings, I believe (though indirectly heated cathode tubes might simplify it as long as they don't arc over).

But the fix is to construct the bridge with the tubes in their conventional place (with B+ coming off the cathodes), but implement the other half of the bridge with solid state diodes.

That way you could still have your one tube/two tube sag variation. Although if you use the choke filter, the PS is kind of "pre-sagged." DC output curves are significantly flatter than with a cap filter.

[Alexo]

But the fix is to construct the bridge with the tubes in their conventional place (with B+ coming off the cathodes), but implement the other half of the bridge with solid state diodes.

I'm trying to visualize this... I think this would give you half the sag of a full-wave tube rectifier (ignoring the second 5U4 at the moment).

[Cliff Schecht]

I'll probably end up sticking with the 5U4 and do a hybrid tube/ss bridge network. I think with the right combo of power transformer and tube/ss bridge configuration I can nail the ~420V B+ with plenty of current on tap. I'll probably setup the amp for dual 5U4's still so that I have the option of pulling one of the rectifiers for more sag. I might rig something up this weekend to find which massive transformer and rectifier configuration will get the voltages about right.

[tubeswell]

But the fix is to construct the bridge with the tubes in their conventional place (with B+ coming off the cathodes), but implement the other half of the bridge with solid state diodes.

I'm trying to visualize this... I think this would give you half the sag of a full-wave tube rectifier (ignoring the second 5U4 at the moment).

Its on Merlin's website about 1/2 way down this page:

http://www.freewebs.com/valvewizard/bridge.html

[tubeswell]

I'll probably end up sticking with the 5U4 and do a hybrid tube/ss bridge network. I think with the right combo of power transformer and tube/ss bridge configuration I can nail the ~420V B+ with plenty of current on tap. I'll probably setup the amp for dual 5U4's still so that I have the option of pulling one of the rectifiers for more sag. I might rig something up this weekend to find which massive transformer and rectifier configuration will get the voltages about right.

Go for it Cliff. Build a classic amp using classic stuff and you can see for yourself. I like the 5E8A with 2 x 5U4G.

On the Qn of brands of 5U4, the Sovtek coke-bottle shaped 5U4G not only look uber cool, but I've never had any problems with them (touch wood) they seem to be ultra reliable and I use then in 3 of my own amps at present. They are not really a 5U4G, but more like a 5U4GA in operating parameters (which is what the 5E8A originally ran). At 12cm tall, they are bigger than a standard 5U4GB, but have very rugged construction, and don't make any strange humming noises, that you sometimes get with EH 5U4GB etc.

[Cliff Schecht]

I definitely will and I'm pretty excited about this build. Building classic amps is more fun nowadays than working up my own designs and a lot less work. I've wanted a Twin forever and although this isn't a traditional 4-6L6'er with LTP PI, I do think it will be characteristically very much like a Twin.

I do have a good amount of US 5U4's and (unfortunately) a lot of these types of 5R4's (see pic below). Any words on these guys? I know they can handle lots of voltage at about 250mA but how much do they drop? Are they usable in place of a 5Y3, 5V4 or other rectifier types? [img:290:508]http://images.buyitsellit.com/246373.jpg[/img]

[FYL]

The 5R4WGY aka "the potato masher" was designed for aerospace applications and can work in extreme conditions. Quite high voltage drop with 67 V @ 250 mA, doesn't like capacitive loads unless suitable resistors are inserted between the PT and the recto, very heavy.

Datasheet of the similar -B version: http://www.r-type.org/pdfs/5r4wgb.pdf

[Firestorm]

I do have a good amount of US 5U4's and (unfortunately) a lot of these types of 5R4's (see pic below). Any words on these guys?

They drop more voltage than 5U4s and have worse regulation, but they can withstand 15,000 takeoffs and landings! (Aviation tube joke).

Mostly good for high AC supplies (and they can deal with huge PIV -- like 3KV), but they really should be operated into a choke instead of a cap to prevent exceeding the peak plate current rating. Most data sheets show "typical operation" with a 4uF cap filter (though some have 20uF), but all stipulate an effective plate supply impedence of as much as 575 ohms per plate (in high voltage use). Not sure how little you could get away with in an MI amp, but it's unlikely your transformer windings will provide anything close so you'll be hanging resistors onto your PT leads.

Still, you could build an amp called "The Potato Masher" and actually have potato mashers in it!