The Official Dual Rail Thread

[Joey Voltage]

Kinda like KO'C and his self-derived constants?

well not really, that aspect is more just a product of over simplification. But in terms of calling a grid leaked biased stage a "Mu Amp" then yes! absolutely.

[ThermionicScott]

Watching this thread with interest. Someday, I want to build a deceptively high-powered amp using fairly high plate and low screen voltages and smaller output tubes like EL84s or 6V6s. There are some precedents in the guitar and hi-fi world, so I know it's not a crazy idea.

With a bridge rectifier and a center-tapped secondary, you can use the CT to get a B+ voltage that is exactly half that of the bridge. Something to consider.

- Scott

[Wilder Amplification]

Watching this thread with interest. Someday, I want to build a deceptively high-powered amp using fairly high plate and low screen voltages and smaller output tubes like EL84s or 6V6s. There are some precedents in the guitar and hi-fi world, so I know it's not a crazy idea.

With a bridge rectifier and a center-tapped secondary, you can use the CT to get a B+ voltage that is exactly half that of the bridge. Something to consider.

- Scott

Yes...the FWB doubler circuit that isn't really a doubler.

However, I didn't want the screen voltage to be at the mercy of the plate B+ voltage, which is why I elected to go with two PTs on mine.

[black_labb]

For the record, the term "dual rail" can be used to refer to any sort of dual plate supply config regardless of polarity (excluding the bias supply since it's not part of the normal "current carrying" supply).

"Sliding screen" is the term I use to refer to the standard single rail supply that references the screens a few volts behind the plate since the screen voltage slides up and down with signal current.



This is essentially what I'll have for my dual rail build...450VAC and a 225VAC for a 630/315 setup. I just got the plate tranny wound by Heyboer...nice thing about them is that a custom wound PT from them doesn't cost much different than an off the shelf one, and they're LOADS cheaper than off the shelf stuff at MM.

The OT will have a 5K Zp-p. With the math I did on this load for peak power (5000/4 = Zp-ct = 1250...630 / 1250 = 505mA Ipk) I had them rate the plate tranny at 450VAC @ 500mA so I'm thinkin' there'll be very minimum to no sag at all depending on the % regulation of the PT.

Now according to the plate characteristics chart on the datasheet, for an idle bias at 40mA it tells me I should have roughly around -35 volts, which I'm sure is +/- a few volts. Does this seem about right?

-35 sounds about right.
i dont even know what the bias is on mine as i designed the amp to have the bias set from the exterior so i dont measure it, just the current through the cathodes of each pair of 6550's

[Steve Conner]

I think it's best to think of a so-called "dual rail" amp as just an ordinary amp with the following modifications:

Add on an extra, higher voltage supply just for the plates, and increase the OT load impedance.

You could modify any amp in this way, there are no special requirements.

If you doubled the plate supply voltage, and doubled the load impedance, you'd get twice the output power from the same tubes. But with the following caveats:

1) Too much voltage and things (OT, tubes, tube sockets) might arc and break down.

2) You have to bias it colder, because doubling the plate voltage also doubles the idle power dissipation for a given idle current. This will affect the tone.


My experience with dual rails was with my old Ninja Toaster design. I started with a PT that gave 550V DC, and thought this was too high. So I installed a MOSFET-based regulator with two outputs: 475V for the plates and 360 for the screens. I also had a 6.6k OT, and this worked great with EL34s giving exactly 50 watts with a 1k shared screen resistor.

But with 6L6s it would only give 30W. I installed a switch to bump the screen voltage up to 475, and this gave 50 watts with 6L6s. But flipping this switch with EL34s installed, they tried to make nearly 70W with an impressive light show of glowing screens.

It works with every tube I've tried, from 6V6s to KT88s, on one setting or the other. Well apart from those CV4060s I bought on Ebay, which were a complete disaster. Those would be great for dual-rail amps, because they need a REALLY low screen voltage: even 360 was too high. People sell them as 6550 replacements, but that's misleading, they're more like a 6146B without a top cap.

http://www.tubecollector.org/cv4060.htm

I'm currently using the Sovtek "Tung-Sol 6550s" on the 360V setting, and get about 60W. Oddly I get the same power whether I use an 8 ohm or a 16 ohm load, but that suits me fine because the OT only has one tap.

Lately I turned the regulated voltage down to about 400V, because since I started using the larger tubes, the regulator dropped out and sagged that low under heavy load anyway. That would decrease the screen voltage to about 310, 320.

I've had it for 10 years and I've never seen it blow a power tube yet. I guess one day it will happen and, as per Murphy's law, take the MOSFETs with it.

People say that an amp with a regulated power supply must sound bad, but I've always enjoyed playing it.

[Wilder Amplification]

Been meaning to ask you on that Ninja amp...are you using an LTP PI circuit in that one?

[mooreamps]

The 36 watt SE amps I've built run dual rails for plate volts ; switchable between 300 and 600 vdc. Voltage for the preamp stages remain at 300 vdc.

-g

[Joey Voltage]

Been meaning to ask you on that Ninja amp...are you using an LTP PI circuit in that one?

The whole point of the design process is for you to make that decision, and put in the bench time ironing it out... figuring out what works best for the circuit in it's entirety, Especially since this is an amp you have quite openly stated you want to Commercialize and sell to people.

Plus you have already recieved affirmation here from people that have utilized this topolgy before with the results you hoped for, what else are you looking for?

you can make the decision based on the purpose, and then change it. For example If you are going for that "Textbook Perfect" jive (in so much as there is such a thing), and are not comfortable, or have no desire to use active circuitry to help things along, go for a cathodyne.... it's not set in stone, and you can always change it.

[ThermionicScott]

High plate B+/low screen B+.

Bruce...the SVT was the amp that actually got me curious about dual rail technology. That's one amp I never hear about people having problems with in regards to eating tubes, blowing OTs and whatnot. One badass amp.

I know they had trouble with the screens when using 6146s, but the 6550 ones seemed to be bulletproof.

[Merlinb]

I'm very interested to know things like what phase inverter would work best with them (paraphase, cathodyne/buffered cathodyne, LTP) as well as valve life expectancy in a dual rail amonst other things.

To me, those questions don't make a lot of sense. All you're doing is running the power tubes' screen voltage "a bit different from 90% of the other guitar amps out there". What's the PI or lifespan got to do with it?

[Steve Conner]

Yeah, +1, there's no reason why you'd have to use some special type of PI.

My Ninja amp started out with a floating paraphase PI, but I changed it to a LTPI. The supply voltage to it is a bit low, because it runs off the screen supply, but I still get as much output power as before.

[Wilder Amplification]

Yeah, +1, there's no reason why you'd have to use some special type of PI.

My Ninja amp started out with a floating paraphase PI, but I changed it to a LTPI. The supply voltage to it is a bit low, because it runs off the screen supply, but I still get as much output power as before.

I guess in terms of the PI I was noticing that very few of the dual rail amps I see use an LTPI. The SVT uses a buffered cathodyne, which I figure the buffer stages are to minimize the loading of driving 3 tubes per side. Plus I know an LTPI can kick out signal with a higher amplitude than the negative voltage at the grid so was worried I'd overdrive the grids too hard, but then again you could always drop the plate resistors to lower the gain if need be.

On my design I'm using a separate screen transformer. The only three ways I can think of to HT fuse this setup and kill both supplies at the same time -

1) Place the HT Fuse between ground and the (-) side of both supplies (or place it between ground and the supply circuit itself) or..

2) Fuse both the plate and screen supplies separately, but rate the screen fuse low enough so that if the plate supply fuse were to blow, screen current would increase enough to take the screen fuse out as well

3) Fuse the output tube cathodes on the same fuse

I know method #1 is not standard practice, but it would work as it would kill the (-) feed to both supplies. A cathode fuse would only kill the tube circuit...it wouldn't protect in the event of a shorted filter cap.

[Bruce / Mission Amps]

Yeah, +1, there's no reason why you'd have to use some special type of PI.

My Ninja amp started out with a floating paraphase PI, but I changed it to a LTPI. The supply voltage to it is a bit low, because it runs off the screen supply, but I still get as much output power as before.

Steve, what does floating paraphase PI mean?
I'm working on a new harp amp that uses a vintage paraphase type driver but also built with fixed bias power tubes... so the power tube grid load resistors are -v biased at -40vdc to -60vdc.
The voltage divider driven grid on the out of phase preamp triode is DC block and has it's own gird load resistor... similar or something totally different?

[imaradiostar]

The Valve Wizard

Sorry to jump in there Steve.

jamie

[ThermionicScott]

Floating paraphase is the "D" series Fenders. The lower resistor of the divider is at the bottom of both grid-leak resistors on the power tubes.

- Scott