[00940]

The postman brought me a nice little gift yesterday:
 

 
Btw, for the European guys in search of transformers at a reasonable price, the pair cost me 30€ shipped, from Don-Audio. It took about 5-6 weeks to get it. Looks like a good time to thank the Torpedo team for specifying a now easy to get parafeed transformer.
 
Since tinkering is a never-ending pleasure, I thought I might repurpose my current amp into a very simple parafeed one, with one tube per channel. It's currently an OTL µ-follower with ecl85/mosfet. The power supply has a regulated 280VDC@60ma B+ and a 12.6VDC/1A heaters supply (could be upped to 15V). The µ-follower mosfet section is on its own pcb and can be used as a gyrator easily, by taking the output from the anode of the tubes. Only the tubes themselves, the output caps have to be changed (the ecl85 really have too low a µ for an output transformer).
 
So what tubes ? We need something with rp under 4k, better yet under 3K, which can stand a 200Vdc Va at a decent current and has a µ above 25 or so. It's best if it doesn't break the bank too. All data following is from Tom Schlangen.
 
- As an obvious first choice, I was thinking about the ecl84/pcl84 (pentode, triode strapped). With a µ of 31, rp of 3K and Pd of 4W, it's roughly similar to two 6J6 sections in //. Bonus being I've got a quad home.
- A second possible choice from my parts box is the ef184. Triode strapped, µ of 55, rp of 3K7 and Pd of 2.5W. µ is a bit high, power dissipation a bit low... I've got a few on hand though.
- A third tube that looks interesting on paper is the 7ky6-9kx6. But it's very uncommon in Europe, let's forget about it.
- Fourth tube that looks interesting is the 6ag7 (or the equivalent 6p9 in Russia). Triode strapped, we get a µ of 25, rp of 2k2, Pd of 9W.
- And as a fifth, there's the el83/pl83. Again triode strapped, µ of 25, rp of 1k8, Pd of 9W. It could push the 7K tap easily...
 
Any other candidates I'm missing ? There's always the 6922 (in //) I guess.

 

[nikongod]

6922 can not handle 200V. Even with the sections paralleled the current & dissipation is not going to be very high. 
 
12AU7, 6SN7, 12SN7 with both sections in parallel might work. 
 
6DN7, 6EM7 are popular from the Bottlehead SEX amp. 
 
Whatever tubes you choose, you should try hooking the transformers up as faux-autoformers. Instead of grounding the primary, wire it directly to the 300ohm tap.

 

[FrankCooter]

Lots of candidates.
 
I've used and liked the E55l, C3g, and 418A in "spud" amps. The Western Electric 418A is essentially a pentode version of the now prohibitively expensive 437A. To my ears it was the clear winner of the group. They pop up on Ebay from time to time, usually in the $30.00-$50.00 price range. If you want to go cheap, try a 12GN7 in triode mode.

 

[00940]

Thanks. 
 
I'll have a go at the autoformer thing, easy enough to try.
 
At a first sight:

- 12au7 and 6SN7 are a bit too low wrt µ. But I'm gonna have a second look at the 6SN7 anyway.
- 6EM7 is interesting but the strong section has low-µ... It will need a two stage topology and it's sadly very rare in Europe.

 

[nikongod]

  - 12au7 and 6SN7 are a bit too low wrt µ. But I'm gonna have a second look at the 6SN7 anyway.
 

 
Agreed for Mu, I suggested them because they are available EVERYWHERE. 

 

[00940]

   
Agreed for Mu, I suggested them because they are available EVERYWHERE. 

 
Indeed.

 
 
Browsing through datasheets, ebay searches to see what's available and diy forums, I'm slowly narrowing my choice though.
 
 
1st option is now the 6E5P. Might be the time to snatch a lot before prices go up, it seems to slowly pick up attention. Info here: http://www.bartola.co.uk/valves/tag/6e5p/
 
2nd option is the el83/pl83/6P15P. But µ is borderline.
 
3rd option is the ecl84. But it looks clearly inferior to both tubes above.

 

[DingoSmuggler]

  ...you should try hooking the transformers up as faux-autoformers. Instead of grounding the primary, wire it directly to the 300ohm tap.

Curious as to what benefits this would offer.
A 10.3k ohm load for the tube instead of 10k?

 

[nikongod]

Curious as to what benefits this would offer.
A 10.3k ohm load for the tube instead of 10k?

You get less phase shift at high frequencies.

You also use every turn of copper all the time which I think is cool, even though I'm not totally sure how big the benefit is for this. When set to the low setting a multi-tapped transformer has a lot of wire doing nothing.

The load impedance seen by the tube does go up a bit too

 

[00940]

Allright, I've received some 6E5P, so time to go forward with this.
 
Here is what I'm considering. Operating point is 180V / 20ma (roughly). Needless to say, a lot of inspiration is coming from Dsavitsk's designs.
 

 
 
Some stuff to comment on:
 
- I'll use the transformer as transformer. The increased safety it provides is welcome in my view.
- C4-R9 are necessary because the 6E5P has a low rp (a bit less than 1k). This translates into a quite serious subsonic boost if I use C3 alone. C4-R9 flatten it almost perfectly. Sims in LTspice agree with the spreadsheet and analysis you can find here : http://www.siteswithstyle.com/voltsecond/parafeed_fun/parafeed_fun.html
- With C4-R9, one could also take the output from the source of M1 rather than from the anode, significantly reducing distortion... I'll have to compare if there is a big difference in sound. Could be a "fun" switch: tubey or not.
- The cathode bias scheme is inspired from the Torpedo, but I wanted a two terminals solution to facilitate wiring. The 2sk170 is probably the only jfet you can use with only 1.3V to operate. This might be overkill though.
 
 
Comments, ideas ?

 

[00940]

I finally received (almost) all the parts and started the work on the wiring and casework. A custom transformer is nice but the delay (almost two months) isn't. In the end, I start anew rather than re-using the old amp.
 
I simplified things, it's now extremely similar to a torpedo. The tough part, so to speak, is to fit everything into a case 4cm high. That's mainly why I needed a custom transformer, to have one flat enough. There's room enough in there for a small dac board. It might be a future evolution.
 
Schem:
 

Internals:
 

 
Quick idea of the end result:
 

 
I need to perfect my metalwork.. the holes are ok but I managed to scrap a bit the paint around the holes. So I'll probably have to stick a piece of leather or nice wood on top of the amp to hide that.

 

[dsavitsk]

Very nice project

- The cathode bias scheme is inspired from the Torpedo, but I wanted a two terminals solution to facilitate wiring. The 2sk170 is probably the only jfet you can use with only 1.3V to operate. This might be overkill though.

If you use a LM334 instead of the 2sk170, you have the bias scheme from the L-2. The LM334 is a CCS that can operate with less than 1V across it, so it works well in that spot.

 

[00940]

Ok , that layout was no good. 50hz hum was audible. Moving the power transformer out of the box fixes it. So, the transformer will get its own small external box. Connectors for higher voltages are always a bit of a pain, I think I'll go for a captive lead.
 
This said, it's a very nice performer once the transformer set aside. I'm pleasantly surprised by those little edcor (even if I shouldn't be, considering all the good feedback on the Torpedo).
 
The lm2596 switching reg is fine btw. No excessive noise (it's switching at 150khz). It seriously reduces the wasted power for the heaters and those little pcb are dirt cheap on ebay. Beware though: the pcb is very thin and bends easily which could lead to some smd parts lifting their pads.