Some questions about an amplifier design

[Jon the Grimm]

Hi!

I thought it would be fun to build a tube amp over the summer. This design looked pretty good but I have a few questions. http://stiftsbogtrykkeriet.dk/~mcs/Williamson/ppwil002_v2.gif

My first question is, can I substitute 807's for the 1614's? From what I could find , they both seem to be glorified 6l6's, but with the 807's having somewhat higher voltage ratings.

[Eli Duttman]

If you have a nice supply of 807's, use 'em. The Williamson topology shown requires the use of PREMIUM quality O/P trafos and that's EXPENSIVE.

Are you building monoblocks?

[Jon the Grimm]

I was hoping to make one big stereo amp to save on some power supply components.

What do I need to look for in an output transformer, that say, a Hammond or an
Edcor would not have?

[Eli Duttman]

No way is a Hammond O/P trafo up to the task of Williamson style circuitry. Phase shifts will eat you alive. You might get away with Edcor. IIRC, a member did a Williamson with Edcor recently. Still, I much prefer Mullard style circuitry with budget "iron", as stability trouble is much less likely.

Edcor's model CXPP60-8-6.6K looks about right for 6L6 family tubes, like the 807s you have in mind.

If a stereoblock is to be constructed, SS rectify the B+ and avoid worrying if vacuum rectifiers can handle the B+ draw. Also, sourcing good 5AR4/GZ34s has become problematic. If your heart is set on using vacuum rectification, employ a pair of 6AU4 damper diodes.

[DrRick]

I built the amp in the article you linked to, back in 1970 when I was 13.
I used 1625's for the output tubes. The 1625 is a 12 volt heater version of the 807. It was and still is super cheap.
I also used 12AU7's in place of the 6SN7's.
I used a BUD 17'*13"*3" chassis and built a stereo version. It was the first brand new chassis I ever had. It took a month of skipping lunch in the school cafeteria to save up for it.
It worked great.
I still have it.
If you have any questions let me know.

[Gordy]

Also consider this design: http://greygum.net/files/radiotronics/PP807-A515.pdf

From the figures at the end of the text it appears to have constantly low distortion from approx. 2 to 10 W.

It is a Williamson modified by F. Langford-Smith, the gentleman behind the Radiotron Designer's Handbook.

[Jon the Grimm]

Thanks for the info.

Eli, would it be at all problematic using that 60w opt when a 30w one is specified? Would I be better off using their 45w version?
My preference for rectification would be a pair of 866a's, (because I have a lot of them), but I see that those damper diodes could run off the same filament transformer as the rest of the tubes which is a significant advantage.

DrRick, it is good to hear that you actually built it, so I assume that it must a reasonably good design.

Gordy, I saw that design too but I was hoping for just a bit more power.

-------------------------------------------------------

I need a little advice on the power supply:
To scale it up to supply power to both channels, I assume I just need to double transformer mA, halve the inductance and resistance of the chokes, and double the capacitance. I had to adjust some values to what parts are available. I am considering using the Hammond 715 which is 115v to 510-0-510v @300mA. Adjusting for 120v input, I get 510*120/115 = 532v. Using Hammond s cheat sheet I get output mA = secondary mA*1.54 for a choke input filter. This gives 462 mA which is more than enough.
So I download PSUD and simmed the "adjusted" PSU at 400mA and 280mA drain and got this:

This seems pretty good, except the voltage might be a little low depending on the current draw for which the measurements given in the schematic are given. Additionally, three pulses exceed the 2A max current capacity of the 866's on startup while charging that large capacitor, so I would probably need something to limit the current inrush, if I decide to actually use the 866's. I also tried simming it with the damper diodes but this dropped the b+ voltage by 25v, so I would probably need to buy a custom power transformer if I used them.

Because the peak to peak ripple voltage was only was only 16mV, and I read hear that pp amps are o.k. with up to about 1v of ripple, I decided to try eliminating one of the chokes. Here was the results:

Ripple voltage was now a still reasonable 500mV but $60 and a lot of space would be saved. Additionally, a resistor was added in series with the choke to drop b+ a bit which would seem to allow more tweakability. In fact, without that resistor, the damper diodes produced sufficient voltage output.

I'm new at this; my tube experience only involves repair. Am I on the right track? Is there a problem with eliminating that choke that I am overlooking?

[Eli Duttman]

In designs with GNFB, like Williamson and Mullard, magnetic headroom in the O/P trafo is essential. Otherwise, the deep bass error correction signal saturates the core. Even with the headroom, an additional precaution is in order. Add a high pass cap. at the I/P, which in combination with the 1st grid to ground resistor yields a F₃ in the 15 to 17 Hz. range.

Yes, it's possible for damper diodes and signal tubes to share a filament winding, but it's not a good idea. If you are comfortable with the hazards Hg vapor rectifiers present, you can use the 866As in your possession. Just make sure you have a well sealed container of powdered sulfur on hand. Should an envelope break, sprinkle Sulfur on the mess to sequester the liberated Hg. Don't forget to allow PLENTY of time for the rectifiers to get hot, before activating the HT winding. Frankly, I'd use damper diodes and a separate filament trafo for them. B+ rise is automatically very slow and there are no toxic waste issues.

[Jon the Grimm]

Thanks Eli, I really appreciate the help. I will get the 60w output transformers.

[richwalters]

Originally Posted by Eli Duttman

In designs with GNFB, like Williamson and Mullard, magnetic headroom in the O/P trafo is essential. Otherwise, the deep bass error correction signal saturates the core. Even with the headroom, an additional precaution is in order. Add a high pass cap. at the I/P, which in combination with the 1st grid to ground resistor yields a F₃ in the 15 to 17 Hz. range.

6dB/oct; I find that isn't enough. I came across this problem using a octave shifter keyboard, where some shifting warbling subharmonics below transformer cutoff creates subharmonic blocking. This can occur with some synthesisers.
There are various Sallen-Key types about, but I use this low noise tube version, 18dB /oct; -3dB 15Hz. It can be improved but in it's current form has made the world of difference.
Tube amps with a good LF response can be particulary prone to subharmonic signals upsetting the GNFB loop. This implies with Plitrons the situation is far worse !


richy