All-tube vs tube/mosfet hybrid ?

[Bigun]

In the past few weeks I had the chance again to compare three DIY amplifiers.

1st one was an all SS amp (my DIY TGM3 amp, Class AB bipolar throughout, global nfb)

2nd one was an all tube triode amp (my DIY CELLINI amp, Class A, no global nfb)

3rd one was a hybrid amp (6N1P triode in common cathode, cap-coupled via an EF buffer to heavily biassed Class AB bipolar output with no global nfb). All three through the same speakers (single cone full range driver with no cross-over of course), all with the same source (Magnum Dynalab FM tuner on a 'good' station i.e. not compressed pop)

The SS amp - the bass was very dynamic, the high's were not harsh but crisp and clean.
The tube amp - warm, bass was less dynamic, less crisp in the high's but they were very fluid sounding

The hybrid - good bass dynamics (not quite as good as the SS amp), but the high's were fluid like the tube amp, only they were cleaner.

 

[Audio_idiot]

YEAH!!! getting down to built all & listen is much better then just talking about "which is better"!

 

[E.T.]

I don't do the thing of putting together a super high-end system to listen to "audiophile" recordings of crap I don't like. I build gear to listen to stuff I do like, and most of it isn't the best when it comes to the engineering and skill in mastering. IME, tubes do a much better job of making such recordings listenable and enjoyable. Many SS amps make you want to stick icepicks in your ears after listening to these same recordings for an hour.

I figured that out one time when I got to listen to a $120,000 Krell/B&W system. It was incredibly "accurate", but it dawned on me that I'd rather listen to an ST-70.

1. I don't want to hear the mics phasing, or any other flaws in the recording.

2. I want to hear what the artist and producer wanted the album to sound like, not what they settled for, and tube amps do a fine job of getting there.

I built the Pete Millett Starving Student headphone amp, and I was flat-out shocked at how hi-fi it sounds, especially for $35. It's got a tube pre and MOSFET power stage, and tho it sounds fine, it's a SS sound mostly. IOt's really true that most of 'tube sound' comes from the power tubes. I think this is because they're being used as current amplifiers, which is a harder trick than voltage amplification? Que no?

 

[ChrisA]

2. I want to hear what the artist and producer wanted the album to sound like, not what they settled for, and tube amps do a fine job of getting there...

Reminds my of a story I heard about Neil Young. He was recording an album in the studio and wanted to know what the music really sounded like so he burns a copy to CD then took it out to the parking lot and listened on the stereo in his pickup truck. Actaully I think he started doing this with cassette tapes. I'd suppose today he'd download the mix to an iPod and test it that way. I doubt Neil was the first guy to figure out that most consumers listen to music on consumer electronics, and don't have the monitors and room treatments found on major recording studios. So they mix and master the music to sound good on consumer level playback equipment.

 

[andyjevans]

IOt's really true that most of 'tube sound' comes from the power tubes.>>

I don't know if that's right. Everyone goes on about DHT amps as if one 300b makes the whole amplification chain DHT - it doesn't! I've had the most significant gains in using DHTs in the first two stages - 26 followed by 10y or 46 or 6B4G or 31, 71a etc.

So I'm really thinking that putting a couple of Semisouth devices on the end of two DHT stages should do it. the frustrating thing for me is I don't build with ss, so I'd need a schematic of an output stage and with these Semisouth devices I don't think we've quite come that far yet. Please tell me a good output stage if you know of one. There are many experiments going on. The Pass Son of Zen would be one way to go for an output stage - no idea if it works with Semisouth devices. All help appreciated here! Would be thinking of a SE to PP interstage before the output stage, like a Lundahl LL1660/18mA

 

[Bigun]

Why use semisouth ? the parts are not cheap, they require large heatsinks. If you're going to have heater supplies and the rest for DHTs then just get yourself a good output transformer and go all DHT.

 

[andyjevans]

Yes - I can see that argument. But Semisouth is not more expensive than 300g tubes, and they don't run down and need to be replaced, so cheaper in the long run. I listen to music when I go to sleep and leave the amp on in the day quite a lot. A pair of 300b runs down pretty fast.

The issue is partly power - my present 300b amp won't drive my speakers (Alpair 10). I could run the 300b hotter and they probably would.

So I can see advantages in a ss output stage - don't know about the sound, but the DHTs in the input should already be good.

Andy

 

[pieter t]

Andy, I will build the Ciuffoli amplifier the coming weeks for someone who owned 300B SE amps but has minimonitor loudspeakers now with some 85dB of sensitivity, so he needs something a bit more powerful.
You may know Andrea's design: D3A input tube (OK not DHT....) - IT/phase splitter - circlotron Fet output.
I will build it such that I can also change the output stage to Susan Parker's transformer coupled one (Zeus) when Andrea's design would not meet the expectations.
Choice for the power fets is not decided yet. I have lateral fets (BUZ900), but the Alfet fets are an option, and Farnell stocks Semisouth fets.
Personally I don't believe that the choice of output fets will have much influence on final sound quality; these are all follower output stages with 100% local feedback.
I agree with your findings for tube amplifiers that gain and/or driver stages are at least as important as output tubes; too often 300B is used "for the sake of...." without paying enough attention to input stage, driver stage, power supply....., and the combination with the loudspeakers to drive.

 

[tmblack]

I was thinking of using a conventional triode like 12ax7 but with a DN2540 depletion MOSFET as a current source in the Anode end.
I haven't tried it, so I don't know if its still a tube sound or what?

 

[andyjevans]

Hey Pieter,

Very interesting!! Susan is a friend, and lives very close. What are you going to use for an output transformer if you use the Zeus? I'm interested in this!!

The Nemesis is another interesting design, and Pass Arch Nemesis with OPT and Semisouth parts, and again the SIC Puppy is similar. All need OPTs

Andy

 

[pieter t]

Andy,
Different options for output transformers (core material), but one thing for sure: c-cores.
Checking these fet output stages there are differences:
- Ciuffoli's design: circlotron output stage, no output transformer, not a distortion "champion" when I look at the FFT's. Can work in class A and AB;
- Susan Parker's Zeus: very good PP design, step-down output transformer (therefore better load for the output stage confirmed by less distortion when comparing distortion plots with Ciuffoli). Can work in class A and AB;
- Nemesis: Single ended (therefore class A by nature). Not a source follower output stage but with step-down output transformer in the drain circuit. Will be much more sensitive to power supply quality. Actually no option when looking for more power than 300B SE;
-Various Pass designs which might be "tube front end" adapted, but I have no experience with these.

 

[Bigun]

I was thinking of using a conventional triode like 12ax7 but with a DN2540 depletion MOSFET as a current source in the Anode end.
I haven't tried it, so I don't know if its still a tube sound or what?

I've never used this tube, but personally it doesn't interest me much. With a mu of 100 it's a bit extreme and with a CCS anode load you're going to have an awful lot of gain on hand - unless it's part of a nfb loop. And it will have a high output impedance in common cathode configuration which requires careful consideration of what it will be driving. By the time you add feedback to tame the gain you're not going to have much tube sound left over - at least that's my guess.

I would suggest that you might be happy with a lower gain, a medium mu tube and try it without nfb. There are a lot of all-tube pre-amp designs around that could go on the front of a mosfet source follower - food for thought ????

-Various Pass designs which might be "tube front end" adapted, but I have no experience with these.

The First Watt F4 was specifically touted for use after a tube pre-amp if I remember.

 

[andyjevans]

I looked at the F4 but as it stands it looks too complex and has electrolytic caps in the signal path. Something simpler?

Andy

 

[E.T.]

IOt's really true that most of 'tube sound' comes from the power tubes.>>

I don't know if that's right. Everyone goes on about DHT amps as if one 300b makes the whole amplification chain DHT - it doesn't! I've had the most significant gains in using DHTs in the first two stages - 26 followed by 10y or 46 or 6B4G or 31, 71a etc.

I was thinking of the typical IDHT preamp tubes. I'm sure I'd agree with you about DHT preamp tubes except that I've never heard 'em.

But everything I've read about DHT's sez they have a 'liquid' sound that's pure catnip for us tube crazies.

If I ever get the $ together I'm going to build van Waarde's 6as7 headphone amp. The specified pre/driver is a 6922 (low Rp), and I'm thinking of replacing it with a pair of 3a5. My problem is that I'm one of those many people who can't design a circuit but can build one competently.

The reason I like the 6as7 idea is that there is a whole generation of nifty little tube headphone amps that have been made obsolete by the new low-impedance cans. A preamp tube can drive 600 ohms, but apparently not 75-ohm cans. 6as7 has some real horsepower, and the Rp is only 280 ohms.

Any comments or advice will be greatly appreciated.

 

[pieter t]

A preamp tube can drive 600 ohms, but apparently not 75-ohm cans. 6as7 has some real horsepower, and the Rp is only 280 ohms.

Not all preamp tubes can drive 600 ohms; that very much depends on the circuit. I'd say that with common preamp tubes one needs a cathode follower to drive these loads, or huge amounts of feedback
The 6AS7 could drive 600 ohms with a "normal" circuit, but not much less without some tricks.
I once heard my 600 ohms cans (AKG K240) driven with 6AS7's; sounded excellent.

 

[john29302]

google moskido amp and viola the ultimate stereo amp. the moskido is fast, accurate and detailed. use the push pull preamp version foind later on thread. but what will i do with my dynaco amps?

 

[andyjevans]

I don't think the Moskido is the ultimate, though I'm sure it's up there. My latest thinking is a couple of DHT stages - probably 26 into 46 or 4P1L - into a step-down SE to PP transformer. I have LL1689 which can do 4.5:1+1.

With this setup I'm now looking at Hypex UcD180 output stage which can be wired directly to the outputs bypassing the opamp first stage. There's a thread here called something like "hybrid hypex" which does exactly this, and I'm thinking that this may be at least as good as a Moskido. I'm waiting to try a build. The Hypex modules are being updated with nCore technology so it's also a solution that holds the potential for improvement.

andy

 

[Bigun]

Andy, how did you get along with your tube + Class D explorations ?

 

[andyjevans]

I didn't get into hybrid territory - at least not yet! Bought some stuff to build and it's still in boxes.

I became very pleased indeed with my PSE 4P1L amplifier. 4P1L > LL1660/25mA > PSE 4P1L > LL1620/80mA. Everything in filament bias, just 2 stages.

Haven't built anything for 2 years as a result. But who knows....