post on amp design by a newbie

[ultraelite]

So as far as I can tell most of the modern(ish) tube amp designs for hi-fi audio systems are class AB transformer coupled push pull amplifiers with negative feedback, which brings me to my questions, which may have been asked before but I have been unable to find posts that completely answer my questions.

1. Has anyone tried quasi-complementary push pull amplification (using an pnp and npn transistors instead of a transformer for power inversion)?

2. I was told that NPN transistors simply aren't as good without any sort of qualifying statements why are they so bad?

3. I want to build an amplifier but buying a specific transformer is expensive (looked into building MacIntosh MC275 but transformer is $600), can anyone tell me what I need to know in order to properly design around a transformer (or is that just a terrible idea?), I understand how the turns ratio can be used to step up or down the voltage and that I'm probably looking for a center tap laminated steel core, however beyond that I'm not sure what else needs to be considered.

4. Any cheap high quality transformers that might suit my needs? I've done a little bit of searching on ebay but any other resources would be appreciated.

[lochness]

So as far as I can tell most of the modern(ish) tube amp designs for hi-fi audio systems are class AB transformer coupled push pull amplifiers with negative feedback, which brings me to my questions, which may have been asked before but I have been unable to find posts that completely answer my questions.

Hmmm...I thought Class A SET with no negative feedback was in vogue.

1. Has anyone tried quasi-complementary push pull amplification (using an pnp and npn transistors instead of a transformer for power inversion)?

I don't understand what you just said but there were big words so I'm assuming you are correct and that would work. Chances are someone has tried and posted their experince with it online. Try googling quasi-whodawatchamacallit.

2. I was told that NPN transistors simply aren't as good without any sort of qualifying statements why are they so bad?

I was always under the impression that PNP transistors weren't as good as NPN as most designes with transistors in them use NPN or N-channel.

3. I want to build an amplifier but buying a specific transformer is expensive (looked into building MacIntosh MC275 but transformer is $600), can anyone tell me what I need to know in order to properly design around a transformer (or is that just a terrible idea?), I understand how the turns ratio can be used to step up or down the voltage and that I'm probably looking for a center tap laminated steel core, however beyond that I'm not sure what else needs to be considered.

Build the design you want and then have EDCOR make the transformers for you (if there isn't an equivelent Hammond or non-custom EDCOR transformer available). They're prices are great (other places charge a lot more for similar quality custom transformers).
Personally I suggest designing you're own amp around cheap tubes (expensive tubes rarely sound better*cough*300B*cough* than cheaper tubes*cough*829B*cough*).

4. Any cheap high quality transformers that might suit my needs? I've done a little bit of searching on ebay but any other resources would be appreciated.

EDCOR

[HollowState]

I see a possible can of worms being opened here. This is a tube forum where transistors are relegated mainly to support positions. Many, myself included, feel they have non-linearity's that are worse then tubes and just don't sound as musical. (to use a very subjective term) Actually, a true complementary transistor output stage uses both an NPN & PNP pair. Quasi is two of the same polarity.

Please forget about building a MacIntosh amp mostly for the reason stated. Their design goal was efficiency by allowing the the output stage to run near class B. This was accomplished by a very special output transformer built for tight coupling through the use of bifilar windings with equal plate and cathode loading plus lots of feedback. Both local and global. I own several models and I'm of the opinion that they sound no better then "regular" designs. Sometimes less better.

So you want to build an amplifier. First decide on the power level you need for your speakers. Efficient speakers require less watts and so a single ended amp might be of use. For very inefficient speakers that need more power, then push-pull will provide that. Circuits abound on the web. Since the output transformer is the most important part, I usually start there. There are many things to consider for a transformer.

The impedance match presented to the output tube(s). Maximum power level. Frequency response extremes. Style as in pentode or ultra-linear taps for P-P. Same for S-E. Cost. And always remember, there is no perfect output transformer. Each one is a carefully chosen set of compromises for the desired outcome.

Some manufacturers of reasonable cost: Hammond, Edcore, Electra-Print, Sowter.

[ultraelite]

quote: 2. I was told that NPN transistors simply aren't as good without any sort of qualifying statements why are they so bad?

I was always under the impression that PNP transistors weren't as good as NPN as most designes with transistors in them use NPN or N-channel

Sorry about that, I meant to say "I was told that PNP transistors simply aren't as good without any sort of qualifying statements why are they so bad?"

To understand what quasi complementary is imagine a class B or class AB amplifier and instead of the transformer inverting the power for the second transistor use a PNP transistor to invert the signal (which is the negative half of the waveform) fed into a regular NPN. The schematic would look like a darlington pair in parallel with a darlington pair with a PNP transistor as the intial transistor instead of a regular NPN. ( http://en.wikipedia.org/wiki/Darlington_transistor ) I wish I could show you a picture of the schematic but can't seem to find one, if anyone is still confused I'll PSPICE one up or something.

Thank you for the link, that is indeed a good resource and pretty much what I was looking for.

[ultraelite]

The impedance match presented to the output tube(s). Maximum power level. Frequency response extremes. Style as in pentode or ultra-linear taps for P-P. Same for S-E. Cost. And always remember, there is no perfect output transformer. Each one is a carefully chosen set of compromises for the desired outcome. Some manufacturers of reasonable cost: Hammond, Edcore, Electra-Print, Sowter.

I'm looking at the Edcore website and not sure if there is a frequency response or maximum power level, I haven't really google it very hard but is there a transformer ratings site or maybe a site explaining the differences in pentode and ultra-linear taps. I'm assuming that this amp I want to make will be P-P because of the inherent inefficiencies in S-E.

I (incorrectly) use the term transistor and vacuum tube interchangably since they are very similar in function, I understand tubes are better and that's what I really mean when I say transistor, I will try to be a little clearer henceforth.

[lochness]

I'm looking at the Edcore website and not sure if there is a frequency response or maximum power level, I haven't really google it very hard but is there a transformer ratings site or maybe a site explaining the differences in pentode and ultra-linear taps. I'm assuming that this amp I want to make will be P-P because of the inherent inefficiencies in S-E.

Unless you are using a pentode or tetrode you don't need ultra-linear taps...though if they are there it doesn't matter don't use them. Edcore does list power level (given in watts and/or max amps) and the frequency response as xHz-xKhz.

Also while PP is more efficient than SE, SE is more linear (less distortion in general).

What is your speaker set up like? It will help us help you a bit.

[ray_moth]

1. Has anyone tried quasi-complementary push pull amplification (using an pnp and npn transistors instead of a transformer for power inversion)?

I'm struggling to understand what you mean by this. Quasi-complementary PP is normally taken to mean using an NPN driver transistor and NPN power transistor on one side of the PP 'bridge', with a PNP driver transistor and NPN power transistor on the other side. It gives rise to crossover distortion, because of unequal transfer characteristics in the two sides. It has nothing whatever to do with 'power inversion' or with transformers, unless you mean phase-splitting transformers.

[kevinkr]

There seems to be quite a bit of confusion over complementary and quasi-complementary output stage definitions. On this side of the pond and on this coast it has always had a clear meaning that has nothing to to with the topology of the preceding stages - it refers solely to the output stage topology. (I spent almost 20yrs designing SS amplifiers for Hifi and MI so hopefully I am not all wet.)

A quasi-complementary output stage uses two devices of the same gender whether it be a pair (or more) of NPN, PNP, Nchannel or Pchannel mosfets in a "totem pole" or "half bridge" configuration. The phase inversion occurs in the preceding stage driving the lower device in totem pole. Note that you can also use a transformer with two independent secondaries to drive a pair output transistors in a quasi-complementary configuration. (Early hybrid Harman Kardon receivers and an expensive Fisher console did just this, with germanium power transistors yet.)

A complementary output stage uses two devices of opposite gender in a totem pole configuration. Phase inversion is usually, but not always accomplished in the actual output stage.

Transistors and tubes aren't the same and with the exception of JFETs do not operate similarly either. Please don't use the two terms interchangeably. (To original poster)

Hybrid amplifiers with tube VAS (voltage amplifier section) and solid state whether mosfet or bipolars is old hat, and IMO in most cases does not sound as good to me as a comparably priced well designed all tube design. (Counterpoint amplifiers being one example of a hybrid with tube voltage amplification and complementary mosfets.)

PNP power transistors are generally regarded as inferior due to much higher internal capacitances (Cmiller + bulk substrate) due to the more complication fabrication geometry required to make them. Note however that Japanese companies like Sanyo and Toshiba make very good high power complementary pairs with FT>10MHz in some cases.

[kevinkr]

Ultraelite, it strikes me that something like Sy's redlight district amplifier design would be a good project to try. Parts readily available and lots of help is available here from both the original designer and the tube forum crowd in general.

Also check out Eli's El Cheapo design. Lots of people have built this one too.

Both designs are well regarded.

Also if you are not that familiar with tube design and construction practices the books "Building Valve Amplifiers," and "Valve Amplifiers" 3rd Edition by Morgan Jones are indispensable.

Also read the newbies thread, and particularly the safety thread here and understand it...

[kenpeter]

I believe he may be asking if a PNP current mirror could be
used as the top device in a push pull??? Yes, but it still
wouldn't do anything to match the high plate resistance
of the tubes to the loud-speaker. A transformer is needed.

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

Now, if you go search Berning: I'll save you the trouble..
http://www.davidberning.com/

And read his patents, yes you can abuse transistors to
switch a cyclotron load at high frequency and greatly
reduce the size of the transformer needed, but it is still
needed.

You may wanna research "Cyclotron" before you go read
Berning. Will make understanding his output stage a lot
easier..