High end diy amps

[nikongod]

Quote:

Originally Posted by MASantos /img/forum/go_quote.gif Could you guys elaborate on the parafeed concept?

To understand why parafeed you must understand a problem with transformers.

A conventional SET amp puts DC across the transformer, with current flowing through the primary. This tends to magnetize the core which leads to saturation. Saturation is bad, m'kay? Now to combat saturation the transformer is gapped. some loss is designed into the windings so that rather than hopelessly saturating the core there is room for both idle current and signal. Now the problem is LOSS: you loose coupling, and gapped transformers have lower bandwidth than what is possible with a similar un-gapped transformer.

A second problem with a conventional SET amp is that the transformer is smack in there between B+ and the tube. the transformer dosnt know if "signal" is from the plate of the tube, or ripple from B+. SET amps generally have really bad PSR: power supply rejection. Noise on B+ WILL get out. with speakers a couple chokes will usually solve the issue well enough, but with headphones things get harder fast.

Push pull amps solve these problems by forcing the idle currents to be equal in both halves of the primary, and the magnetism from idle current cancels. PP amps can have VERY wide bandwidth with a "series feed" transformer, but nobody likes how PP amps sound (OK, maybe some people do.... I do they just are not as common) so we solved 1 problem. Push pull amps also have common mode rejection which yields GREAT PSR of ripple on B+. You can make a quiet PP amp much more easily than an SET. Both problems with SET amps are gone, but lots of people really like the way SET amps sound and are all too happy to turn their noses up at PP amps.

Now that those 2 are out there, we have parafeed. aaah. It is a play on words, conventional SET is considered "series feed" because the transformer is in series with the tube, in a parafeed amp its in parallel. the key to parafeed is the "shunt element" Something must be between the plate of the tube and B+ to get a voltage output for the transformer (in both above examples the transformer is there) This can be any number of things depending on budget. It can be a big fat power resistor, a "plate choke", or a CCS (SS or tube) They all have their own things going on. The resistor is cheap and needs a LOT of voltage to work as well as a choke or CCS. The choke or CCS can have an impedance MUCH higher than any transformer you would realistically work in there. This buys a TON of PSR. It is not uncomon (from what I have read) to have PSR in the 80+dB range on a parafeed amp, its usually around 30 on a conventional SET. This great PSR winds up saving a bunch of money building the power supply. In the amp I posted above, I used a CRCRC power supply the caps (all 3 of them added up) come out to like 175uf and nobody has ever complained about noise. none. dead quiet. To do that on an SET amp would require chokes which cost several times what the little resistors do. There is no reason not to use a choke, its just overkill

lots of people like that, go for it.

The second thing is that in a parafeed amp no DC current flows through the transformer. This is GREAT for a couple reasons. The largest is that you can get the bandwidth of a PP amp with the overal sound of a SET amp. Its really cool. The second reason this is cool is that since the transformer can be isolated from HV DC an output autoformer can be used (as mentioned in one of dsavistik's builds, and the Gary Dahl Espressivo) this is tooooo cool because it allows you a bunch of taps for output load! With a headphone which could be anything from 8 to at least 600 ohms: totally awesome.

A "disadvantage" of parafeed (its all perspective) is that the amps typically require a big fat coupling cap. Its usually between 3 and 20uf: Not so large that you cant use a film cap, but large enough that truly fap-o-matic types (teflon &gold plated silver foil.... hand rolled on the smooth stomachs of red-headed virgins under the light of a full moon) are prohibitive. Bummer. There are ways around the coupling cap, but they are a TON of work to implement so they are rarely done. IMO/IME its not that big of a downside.

I hope that was helpful, look around some if your interested.

 

[sachu]

^^ Ari, I had been putting off reading about parafeed till this thread earlier today. Just got done reading dsavistik's page and then your post above was like an awesome revision of the design elements of the parafeed amp. This is such a cool topology. Thanks!!

That spud amp of yours blew me away with how good it sounded. I know you didn't agree with me on this, but i honestly felt yours blew away the DNA sonnett. Fabulous!! Look forward to building the spud amp finally..just as soon as things settle down a bit. Life is conspiring against me to find any DIY time for myself.

 

[nattonrice]

Quote:

Originally Posted by nikongod /img/forum/go_quote.gif truly fap-o-matic types

Almost dumped a cup of tea on my laptop thanks to this =)

 

[Rescue Toaster]

Quote:

Originally Posted by nikongod /img/forum/go_quote.gif A "disadvantage" of parafeed (its all perspective) is that the amps typically require a big fat coupling cap. Its usually between 3 and 20uf: Not so large that you cant use a film cap, but large enough that truly fap-o-matic types (teflon &gold plated silver foil.... hand rolled on the smooth stomachs of red-headed virgins under the light of a full moon) are prohibitive. Bummer. There are ways around the coupling cap, but they are a TON of work to implement so they are rarely done. IMO/IME its not that big of a downside.

Parafeed really does control the current loops well. With a good CCS load and returning the output transformer to cathode instead of ground, you end up with just a few parts in the signal path. Tube, parafeed cap, transformer, and the headphones themselves.

That said, I personally gave up on it because of the cap & transformer sizing problems. Better transformers have much higher inductance, better tubes have relatively low Rp, and the more you go in those directions, the larger the parafeed cap must be. Undersized caps can give a large hump in the bass response.

Personally I'm thinking that a really, really good series-feed output transformer combined with high quality power supply regulation might be the way to go (as far as SE goes). But, lack of good preamp tubes (expensive NOS 5687's, and only ECC99 in current production) and cost of good transformers has turned me off of tubes lately.

 

[Omega]

Quote:

Originally Posted by nikongod /img/forum/go_quote.gif Fortunately there isnt even a foggy consensus for the best DIY tube amp project. I think this is actually a benefit because it encourages diversity and spurs the inclusion of outside though in new amps, and for people to build old things which is great.

I wanted to quote this, because I think it reinforces a very valuable perspective, especially in DIY. The discovery of your taste and pursuit of new experiences, should be the enjoyable goal of this hobby. Rushing to what everyone else considers "the best" loses sight of the journey. Getting great "bang for the buck" is a nice side effect. Of course, we all have limited time and money, and want to spend these maximizing enjoyment...but I find time and again that the experience of trying something new, and also the sharing of the experience, easily trumps any incremental gain in audio performance.

 

[c12mech]

Thanks for the replies. I'm really enjoying everyone's input. I love the way tubes sound and want to build one of those no-holds-barred amps. I even plan to include the fap-o-matic caps. Parafeed sounds like something that I should look into a little more.

Nikongod,
Since it was pointed out that you had a lot of DT880 experience what can you recommend for a next project that will compliment them. These will be my first good cans and I'd like to get the most out of them.

Bryant

 

[nikongod]

My guesses would be (on head-wize)
The kurt strain OTL Mu Follower
And the ever popular Morgan jones clone

The Morgan Jones has proven quite versatile to me, I have not tried or hear the other one but it seems like it could be neat.

 

[dBs]

Insightful, all of it.

 

[dsavitsk]

Quote:

Originally Posted by nikongod /img/forum/go_quote.gif I hope that was helpful, look around some if your interested.

Excellent post, Ari!

A couple of things I'd add -- First, a CCS loaded parafeed can only swing as many voltage as the CCS will allow. For instance, if B+ is 200V and the plate idles at 160V and the CCS needs 10V across it to operate, the tube can only swing 60Vp-p before clipping -- and that clipping is CCS clipping which is nasty. That's 42V RMS -- and into an 8K transformer primary, that's 220mW before copper losses and such. Now, ~200mW is more than plenty for a headphone amp, but it is not the 5W you see quoted from other designs. You can increase this by increasing B+, but that is expensive and requires dissipating a lot heat. The point is that power is often limited in comparison to other designs.

With single feed, on the other hand, the transformer acts as an energy storage device and can swing higher than B+ so you can grab some extra power. So, if the plate idles at 160V, the tube and transformer can conspire to swing more or less to the tube's limit, and tube clipping is not so nasty. This is great for a power amp pushing speakers. It is really less important for headphones, and it is one reason why single feed headphone amps are kind of a silly idea.

Now, a choke loaded parafeed can do the same thing as the single feed amp with the choke allowing the high swings. But, the loading choke has exactly the same issues as the single feed transformer (namely saturation) so the advantages here are few, if any.

Second, the final cap in the power supply of a single feed amp is in the signal path every bit as much as a parafeed cap is. And, it is generally large and thus an electrolytic.

Third, not only can the CCS be thought of as providing high PSRR, but it is also a near infinite load for the tube (over 10 billion ohms is possible). What this means is that in a CCS loaded parafeed amp, the signal AC is essentially blocked from traversing the PS components so the power supply is almost inconsequential (it is still important in that it needs to keep the CCS stable, but that's another issue.) A good read on some of this is at Active loads and signal current control

 

[dBs]

Yep, CCS is exactly what I am looking to do for my parafeed =)

 

[c12mech]

Thanks for all the links so far. I try and read as much as I can about a project before I jump in. This has all been very helpful to me and I'm sure others as well.

nikongod,
I've looked at that Morgan Jones clone a couple of times and since I have Valve Amplifiers I may look into it some more.

 

[c12mech]

What books besides Valve Amplifiers are good to read on this subject? I'm planning on getting Merlin Blencowe's Designing Tube Preamps For Guitar and Bass this week. I know it's for a specific kind of amp but everything I've read on his website is really helpful. As good as the internet is, I still like having a real book in front of me to read.