A very simple single-ended parafeed design

[Sodacose]

Here's the schematic for a low parts count parafeed amp I built recently:
 

 
Hammond 119DA would also be suitable for output transformer duty.  The 6C45pi (6S45pe) is in current production but there's probably a good amount of NOS out there in the market too.
 
You can find a more detailed write-up and construction pics here:
 
https://wtfamps.wordpress.com/papa-rusa-headphone-amplifier/
 
edit: Note this is a free hosted WP site and I make nothing on ads, traffic, etc.  Not selling anything, just passing on the DIY love.

 

[Sodacose]

Yes I'm aware that this isn't the first parafeed and I'm sure it won't be the last  The topology has been around as long as tubes and transformers have.  Glad to see that both Dsavitsk and I landed on the same tube, too.  I hadn't seen the L2 before now.  Bottlehead Mainline also uses the 6C45pi, I believe.  Very cool tube that is fairly unique.

 

[Sodacose]

Couple of pictures of the finished amplifier:
 

 

[tomb]

Gorgeous work!

 

[Budgie]

Beautiful! And an excellent write up on the build!

 

[Sodacose]

Thanks, tomb and budgie!

 

[iperv]

Very beautiful building and project!
 
I'm on a plain to design/build a similar amp, with sowter ot and 6n6p tube.
 
Thank you for sharing.

 

[iperv]

Very beautiful building and project!

I'm planning to design/build a similar amp, with sowter ot and 6n6p tube.

Thank you for sharing.

Edit/grammar

 

[tubey1]

Wow, Total cost for BOM including tubes comes to $150. Very zen-like! If building an OTL like Crack, only the output caps alone will be worth $100 and it will still sound worse than this amp. Thanks for sharing your design.
 
This amp can be a giant killer if a proper CCS and a more refined HV power supply are installed. As is, already being a parafeed, I'm sure it sounds way better than Bottlehead Crack or Darkvoice.

 

[Sodacose]

Proper CCS?!

 

[tubey1]

Proper CCS?!

I meant no disrespect. 10M45 is nothing to laugh at. I made an amp kit that uses it and I like it a lot. But, I was reffering to C4S. I suppose using that will increase parts count and defeat the purpose of simplicity and low cost.

 

[dsavitsk]

Nice design I am a little concerned about the unbypassed cathode resistor - you are adding ~2500 Ohms to the plate resistance with it, which sort of defeats the point of the low Z tube.

I have a write up of such designs which might prove useful: http://diy.ecpaudio.com/p/parafeed-tutorial.html

But, I was reffering to C4S. I suppose using that will increase parts count and defeat the purpose of simplicity and low cost.

The C4S is not a particularly good CCS. Actual impedance is β*β*R where R is the current setting resistor. Multiply it out and you'll find the Z not that high. A single10M45 is generally better. A Cascode that has a FET in it is way better. On the other hand, the parts for the C4S cost about $1, so it is actually the cheaper option.

 

[Sodacose]

Thanks, Dsavitsk!  I've read your site and designs many times over the years and appreciate your comments.  
 
The cathode resistor could be bypassed if desired (something like 1kuF, at least).  I tested with and without the cap and preferred without (subjective decision, and I could have been fooling myself).  Like you said, the cap would lower output impedance by approximately (Mu+1)*Rk divided by the impedance ratio of the output transformer.  Works out to something like 28 ohms output impedance without cap or 8 ohms with cap.  I wouldn't stop anyone building this from trying both to see what they prefer.  The caps are cheap, I just went for the fewest parts.
 
If anyone wants to read up on CCSs, look through Walt Jung's Audio Express articles.  PDF link warning:
 
http://waltjung.org/PDFs/Sources_101_P1.pdf
 
http://waltjung.org/PDFs/Sources_101_P2.pdf
 
The Part 2 covers the 10M45s.  Although it can be improved upon with some complication (notably figure 13C), the raw performance as a CCS is pretty dang impressive.  Big improvement over a resistor load and probably most chokes (and choke-loaded would end up a very different design and much more $$$).

 

[proid]

Hi Sodacose,
That's really nice design and your blog is very helpful for me as well as others who are starting to learn about tube. I have read throughly your Papa Russia headphone amplifier design and i'm confused at this:

Power output on 32 ohms headphones with 1V input is approximately 150mW.  With 300 ohm headphones, power output is approximately 50mW.  The amp is perfectly capable of forever ruining your hearing.  Careful with the volume knob.

Assume that you use the 75 ohm secondary tap for 32ohm headphone, that mean we have to devide the voltage gain for sqrt(10000/75) = 11.54, then 1V in result in 52/11.54 = 4.5V out, current avaiable at low ohm tap is 15ma*11.54 =  173ma. 4.5V at 32ohm = 4.5V/32 = 140ma < 173 ma so the output power is 4.5*4.5/32 = 632mw, that's very far from 150mw. Did i calculate it wrong ?

 

[Sodacose]

Proid,
 
Your math is correct in theory, but for these calculations you can't use Mu (52) as the stage gain.  The finite load and un-bypassed cathode resistor reduce gain to:
 
(Mu * Rload) / (Rp + Rload + (Mu +1) * Rk)
 
or approximately 27 with the values in the schematic.  Consider the CCS as infinite impedance but in parallel with the load reflected by the output transformer, so the Rload above is just that reflected by the OPT (about 4200 ohms with 32 ohm HPs).
 
If you plug 27 into your turns ratio equation (27 gain * 1Vrms input / 11.5 turns ratio), you get 2.3V output or about 165mW into 32 ohms:
 
2.3V² / 32ohms = 165mW
 
For the write up you quoted, I calculated based on the loadlines and was conservative with my rounding.  A 2V ptp signal (0.7Vrms) swings about 75V ptp on the 4200 ohm loadline with a quiescent point of 150V and 30mA on the anode and -1.5V on the grid:
 
75V² / (8 * 4200) = 167mW
 
In contrast, a 3V ptp signal (about 1Vrms) is the largest the amp will want before the grid goes positive.  Again using the plate curves and loadline, this looks like about 110V ptp, or 360mW.  The amp would not be able to do that without adding some (possibly significant) distortion.
 
In reality, the output with 32 ohm headphones on the low impedance tap and a 1Vrms input is somewhere in between the above calculations.  Headphones are not a constant impedance (planars notwithstanding) and music is not a sine wave, so I ballparked the number to 150mW and felt like it was probably a fair estimate for real headphones, real music, low distortion power.