Tube amps - what 3 things…

I’ve got a NY Audiolab Futterman OTL3 that’s “hot-rodded” and modified to output transformerlesss triode.

Hi @ieales,

My friend does the same thing with 300B SET with Shishido schematics: ECC82 SRPP - direct coupling - 6F6 - interstage trans - 300B fix bias. The same excellent results.

 The ESR would be quite different with parallel caps...

Yes you right, but it is important haw to do it. When I parallel a bunch of big ~10000uF capacitors with one "bypass" 50-100uf capacitor it works. But when I tried to parallel 100uF BG with 50uF AN Kaisei Non Polar it causing sound smearing. 

That's a very satisfying...and comprehensible!...description Ralph; thank you.  

@alexberger 

The ESR would be quite different with parallel caps... and topology counts

The designer/builder of our custom amps told me that fixed bias provides more "snot" than cathode bias. He meant power by snot. I know you’ve stated in this thread that there is no power difference between these two biasing configurations. Can you please discuss the bias configuration elements related to power magnitude?

@xenolith 

Fixed bias does have that reputation (and FWIW, we use it in our amps as well).

The power output is the same for a given bias point, but for cathode bias you need a higher power supply voltage since a voltage is dropped across the cathode resistor and that might have to be 30 or 40 Volts in a lot of cases.

The cathode resistor has the additional issue, if its unbypassed, of creating degenerative feedback. This makes it harder to drive. The feedback means it will be lower distortion, but that might not mean much if the driver circuit makes more in order to drive it.

So there are a lot of variables and each has to be managed properly to pull it off. I would expect it to be pretty difficult to do a side by side audition if that were the only variable, on account of all the secondary variables involved.

Thanks very much for your input Ralph. The designer/builder of our custom amps told me that fixed bias provides more "snot" than cathode bias. He meant power by snot. I know you’ve stated in this thread that there is no power difference between these two biasing configurations. Can you please discuss the bias configuration elements related to power magnitude?

I’m very confused as I view both he and you as extremely well on informed on this subject so am quite puzzled at how such diametrically opposed assertions could come from you and he.

Thank you.

Hi @atmasphere ,

In my case cathode resistor was 820 Ohm, so:

with 100uF  - f = 2Hz,

10000uF - f = 0.02Hz

100000uF - f=0.002Hz

In theory, even 100uF is good enough. But in the real life the difference between 10000uF and 100000uF is very clear and significant.

Hi @ideal8592 ,

In my case 100uf/16v were Black Gate CE W (M).

10000 were the same Black Gate  + Nichicon Muze 10000uf/16v.

100000 were the same Black Gate, Nichicon Muze + Nichicon KA 9x10000uf/16v + Elna 3300uF/10v

It wasn't my "invention". At least dozen DIY people tried it many years before me. You can try yourself and it works!

I compared sound with 100uF, 10000uF and 100000uF. It was a big difference in SQ between all 3 cases.

Were they:

• the same technology
• the same size
• formed and capacitance verified

Any of the above could change the SQ... and what reference other than different?

I know that cathode bias is still cathode bias with any cathode capacitor value. But the cathode to ground impedance is realy very different. On 20 Hz with standard 100uF the impedance is Z=80 Ohm, 10000uf Z=0.8 Ohm, 100000uF Z=0.08Ohms.

@alexberger The impedance does not set the bias value- the DC resistance does. I'm sure the capacitor (and its quality) affected the sound quality! A tube amplifies due to the difference between the grid and cathode voltages- essentially as a differential amplifier. The cathode bypass allows the ground 'signal return' to be part of the gain structure of the circuit- essentially increasing its gain.

If you enjoy this sort of tinkering, here's a tip: figure out what the timing constant in the power supply is that feeds that tube. Then make sure that the timing constant of the cathode circuit is a little bit higher (by at least an octave) than the timing constant in the power supply. To this end, to prevent phase shift at 20Hz you'd want the cathode circuit to be good to 2Hz, so the power supply leg should go even lower. In this way the tube will be less able to modulate the power supply leg, which will reduce IM distortion.... Have fun!!

Hi @atmasphere ,

I know that cathode bias is still cathode bias with any cathode capacitor value. But the cathode to ground impedance is realy very different. On 20 Hz with standard 100uF the impedance is Z=80 Ohm, 10000uf Z=0.8 Ohm, 100000uF Z=0.08Ohms.
I did an expiriment by changed cathode capacitor value in the first stage of my SET amplifier. I compared sound with 100uF, 10000uF and 100000uF. It was a big difference in SQ between all 3 cases.

I think one of the best things about tube amps is the fact that they're fussy. I love the sound of my Pass XA-25 but it just sits there. Looking at glowing tubes, swapping 'em out for others, adjusting bias (I used to do that with my Jolida) before my current tube amps no longer required it. Tubes, although often hot, are cool. I still really enjoy my little Dennis Had amp from time to time, and at least always have a quartet of old GEs glowing in my preamp. My guitar amps have tubes but you can't see them (the tubes, not the amps).

Until they aren't. Tube amplifiers with output transformers, regardless of how good they are wound are not as transparent as their OTL or direct drive counterparts (given a level playing field). I've have all three design types and it's pretty clear what an output transformer adds to the mix. Getting rid of the capacitor at the output is also a benefit.

i agree with this expressed by @clio09 

witness ralph's otl amps, or my personal fave... the linear tube audio zotl40 ref+ and its remarkable clarity and palpability

no doubt output transformers do something to the sound heard out of an amp... whether that something is positive or negative is highly situational

@atmasphere 

+1

Cathode bias is not a good thing in input and driver stages, not just in the output stage.

I tried one solution in my DIY SET 300B amp. There is 6sn7 input tube and 6F6G in triode mode driver tube with coupling capacitors between stages. The solution is very big, around 100000uF cathode capacitance in the input and in the driver tube. It gave a huge improvement in everything: bass, soundstage, separation, high resolution.
It is a pity the same solution can't be used in the output tube cathode, because of high current start-up issues. The Maximum capacitor I can put in my 300B cathode is 5000uF. It improved sound but it can't exchange fixed bias in terms of SQ. 

@alexberger I've yet to see a voltage amplifier or driver circuit that isn't cathode biased. When a cathode bypass capacitor is used, its still cathode bias. Its not cathode bias when you have a manual bias adjustment to make, or the amplifier is employing some form of autobias (both are forms of 'fixed bias' since the bias adjustment is made via the grid).

I've read controversy around the sonic differences between cathode bias and fixed bias but IME its really about design and how well the circuit is executed. Both types work just fine.

Cathode bias is not a good thing in input and driver stages, not just in the output stage.

I tried one solution in my DIY SET 300B amp. There is 6sn7 input tube and 6F6G in triode mode driver tube with coupling capacitors between stages. The solution is very big, around 100000uF cathode capacitance in the input and in the driver tube. It gave a huge improvement in everything: bass, soundstage, separation, high resolution.
It is a pity the same solution can't be used in the output tube cathode, because of high current start-up issues. The Maximum capacitor I can put in my 300B cathode is 5000uF. It improved sound but it can't exchange fixed bias in terms of SQ. 

Hidaleeho.  Having computer troubles, so just seeing the status of this thread.  Good exchange of info...good.  Ralph, I'm sure that I'm wrong because I'm even more sure that you're right.

Got some 8 Henry chokes on the power supplies of the amps at chez xeno, and wow, they're lurkers.  Carry on...

My Firebottle is from late 2016 (it has its birthday on the bottom with Had's signature) I and  bought it in early 2017 for 1100 bucks or so. I'll never sell it.

I have a selection of amps including Naim (solid state), Vincent pre / power (Hybrid) and recent Cayin (tubes). Currently despite being the cheapest the Cayin clearly wins in my system (but with EL34 tubes not KT88's). Three lessons, 1, forget the technology listen to the amplifier. 2, some speakers suit some amplifiers, some do not, listen to speakers first then buy an amplifier that suits them. 3, don't be hung up on price, I auditioned at least 20 tube amplifiers before buying I am perfectly happy with the cheapest, by a long, long way. Features, don't pay for ones you don't need, they add cost and complication. Think cars with multiple 'drive' settings, the 1st month you try them all, 95% will never change them after that. 

I agree with the previous comment on transformers being the most important component.

Until they aren't. Tube amplifiers with output transformers, regardless of how good they are wound are not as transparent as their OTL or direct drive counterparts (given a level playing field). I've have all three design types and it's pretty clear what an output transformer adds to the mix. Getting rid of the capacitor at the output is also a benefit.

I agree with the previous comment on transformers being the most important component.

I recently posted about some coupling capacitor upgrades on some chifi integrated tube amps and it made a big different in sound quality.

But just recently I gave some Icon Audio equipment a try again (had one several years ago). In direct comparison to the chifi gear the resolution doubled. I am hearing elements of the music at are either veiled or completely missing on the chifi gear even after upgrades of tubes and capacitors. This is primarily due to the better transformers on the Icon and of course the actual circuit design.

Next most important thing is do not listen to Youtube reviews. I think many of those guys are being paid off to promote average equipment or they are half deaf and have not experienced truly good equipment.

Once you find a good piece of equipment you can tweak the tone with some nice tubes or upgrading a few coupling capacitors. But it would be unrealistic to try to upgrade transformers.

Today afternoon I was spinning "Get On Board", Taj Mahal and Ry Cooder. Can't get enough of this album, but "Packing Up Ready To Go"... that song is something else. They sure found their sweet groove. Should you not feel the presence of that bass drum in your room, change your system. 
Oh, yes: recording is great, mastering is very, very good - and appropriate. It is all captured and preserved, from sweet rawness that usually get conjured during casual jam in a forgotten roadhouse, to the reverberation of the room they recorded it in. Absolute treat for music lover and audiophile.

My 3 most important things - for my system, not only for an amp:
1. True to the recording - reproduce what is recorded, no more, no less. If there are "spatial cues", I want to "see" the room. 
2. As close as possible to the real voice(s) and instrument(s) - reference is the live event and, in my case, years of being musician
3. Goosebumps. A lump in my throat. On some recordings happens. On most - it does not. I need to be taken, transported to the event, even for just a moment. System is the vehicle. Music is the fuel. Rare are the pieces do have such a beauty that can transcend poor reproduction. Conversely, if music is crap, no system can make you suddenly hear The Masterpiece. 
In too many years I'd care to mention I had many different systems. In early eighties, I had one ghetto blaster in my apartment. I even tried D class Bel Canto with Dynaudio speakers. Now, I am on my second Audiomat Prelude Reference. I sold the first one, to "try something different"... I also have Audiomat Phono 1.6 - which would classify as SS. My MD806t uses two 6992. Speakers are Capriccio Continuo Auralea 309, very good match with the amp, and the room. 
I prefer good sound, and don't care is it tube, SS, D class... nor I care about features. Less is usually more: I prefer integrated with passive preamp - power amp with input selector and volume control. 
I heard two different Zu speakers, one with Decware EL84 and another one with Trafomatic. If I would have your dilemma, I would try to hear those manufacturers. Trafomatic was more to my taste. 
Personally, the only integrated amp, besides Audiomat, that I would buy is Ayon.

Also, tube is just a part of the design. I heard EL84 design that was well balanced, with a nice and tight LF output. Also heard 6L6 design with mid-bass hump, and quite flabby down low. Good mids, but nothing else. So, which tube should be followed by which iron, which caps etc. Avoid all of that and go out and listen. Ask Zu people. Visit dealers. We can all give you great insights, but ultimately, those are only our, ultimately subjective observations. Hearing is very subjective, just like any other sense: taste, touch, even vision - just read the studies on crime eyewitness reliability. 

Unless you are unlucky, trying tube amps will not result in you going back to SS.

Cary Audio used to have a kit division, Audio Electronic Supply, and it disappeared as soon as Dennis Had retired. Presumably it was too good a bargain. I had their phono stage and pre-amp (both in standard form and 'Dennis Had Special Edition') along with their Sixpac monoblocks and their SE-811 SET triode monoblocks. Also used a Copland CTA-301 and CTA-305 pre-amp, Quad 24p phono stage, Quad QC-24 pre-amp, and Quad II/forty monoblocks. Even used a NuVistor phono stage.

Currently the Quads are still in use, but I do miss the addictive musicality of the SET monoblocks. Sure, it was distortion-laden, but I loved it. I wish I still had them.

All of these were better than the best SS power amp I ever had, a YBA-1, from the era when they were built in France rather than China. I still have it, but it just doesn't begin to compare.

There are many good Tube Amps on the market for sure, I have had many being a Tube Amp only guy since I was a teenager, (60) now. I like and currently use the McIntosh MC275's run in parallel through McIntosh C2600 Tube preamp. These units are built like a tank, take plenty of abuse and keep playing excellent music. The other 5 Tube Amps I currently have, sound great but seems there's always something little going on with them that annoys me. These others are all the name brands we hear about, and yes they are good but not quite as bullet proof as the 275's in my opinion. Which must be about the general consensus on McIntosh 275's because they are very hard to find new or used for the past 5 years or so.

Hybrids are wonderful. A few nice AH integrated out on USAM now. 

Anything Jim White at Aesthetix make is finely engineered by a guy with ears. The Mimas is fantastic value…

IF a person believes point to point is inherently better than a well executed multi layer board in the right places…. might want to reconsider , ….

Best in music

Jim

I wanted to mention Ive been using a Aestheix Mimas for a couple of years now. It has a tube preamp with a solid state power amp. It’s an awesome integrated amp. Lots of features great metal remote, tons of power but it can sound really delicate if the music calls for it. Great reviews out there: 

I’ve owned several ARC power amps Classic 120 monos were awesome except when they blew a tube.

My recommendation is the Ayon Spirit V I had the Spirit lll the Ayon amps are exceptionally well made. Auto bias, beautiful looking, top notch parts. Excellent protection circuitry. Not sure how much power you need. If you need more the Ayon Triton is a 120wpc in pentode and 70wpc in triode, it’s a beast!

In general, power supplies are important on all pieces of equipment. I think point to point wiring helps determine quality of build and sound. I believe separates are critical for improved sound, so mono blocks would be part of my choice. Past these few generalizations you then need to figure in the room size and acoustics along with the efficiency of your speakers, then you could begin to narrow the choice down.

Well designed autobias is precise. The tubes idle @ the setpoint very shortly after turn on and after several hours of robust use. It's probably an order of magnitude more precise that a pot and resistor in fixed bias designs.

The biggest advantage with a quality CT OPT is both halves idle identically, thus reducing distortion and noise.

ralph, curious to know what is your take on auto biasing... now increasingly employed in high level tube amps, audio research’s newer amps for instance

the feature certainly offers a major convenience factor for users, but what do you see as the tradeoffs or downsides sonically or operationally, if any? (of course we note the obvious add’l cost and complexity of the feature)...

also, does auto biasing get the power tube biasing as precisely correct as manually biasing via multi meter (as arc used to do it for years?)

I see automatic bias as a good thing. The less the user has to fiddle with the amp the better- makes it easier to live with.

If designed right there's no downside except the additional circuitry itself. How precise it is depends on the design; in some circuits the precision is more important than others. So its a case by case basis.

The downside of manual bias is the tube can drift while you're not looking, assuming you got it right in the first place. An auto bias system takes care of that. FWIW, cathode bias is a form of automatic bias. This can makes things confusing, because many autobias systems typically use some sort of monitor on the tube and adjust the bias via the grid rather than the cathode of the tube. For this reason, most modern autobias systems are considered 'fixed bias'.

@atmasphere

ralph, curious to know what is your take on auto biasing... now increasingly employed in high level tube amps, audio research’s newer amps for instance

the feature certainly offers a major convenience factor for users, but what do you see as the tradeoffs or downsides sonically or operationally, if any? (of course we note the obvious add’l cost and complexity of the feature)...

also, does auto biasing get the power tube biasing as precisely correct as manually biasing via multi meter (as arc used to do it for years?)

My tube amps don't care if they're running 4 ohm or 8 ohm loads, have had both, and the 4 ohm loads do not cause any issues with greater loading, heating, etc, with the tube amps

I think you'll find that using a thermal imaging camera that on a 4 Ohm load, the output transformers are indeed running warmer.

I would be careful of auto biasing.  I think it tends to suck the headroom out of tube amps.

If you find this happening I would expect there's a problem in the circuit. Autobias shouldn't affect the power the amp makes.

Three things I would look for: 1. A sub out option. Some tube amps are sort of bare bones items but the sub out on mine really makes things convenient 2. Autobias feature. I was afraid of tubes due to the complexity compared to SS. If tubes can be plug and play it helps alot 3. Be darn sure your speaker efficiency matches the noise floor of your tube amp. A hyper efficient speaker (as in Klipsch Lascalas) can rake any hum or buzz out of any circuit. Depending on the volume and style of music played this could be an issue if you're OCD. Tubes by nature are noisy compared to their SS cousins. That said, tubes can create a sound beyond most SS i've ever heard (or could afford more like). I bought a Raven Audio Nighthawk since I'm a poor painter living above his means. I've really enjoyed getting over my "tube fear" and have been constantly amazed at the stunning realism it produces. My wife says the investment was well worth it. Never expected that!

I am not a EE...

Please don't post myths. I can assure that autobiased amps don't suck the headroom out of amps. As long as the bias time constant is long enough, the speakers will be toast long before the autobias affects the gain... assuming you're not trying to drive 80dB speakers with a 20w amp.

I am not a EE and no where in the "Ralph" league but I would be careful of auto biasing.  I think it tends to suck the headroom out of tube amps.  Listen to Ralph and to ensure you don't end up with tube amp trying to sound like a solid state amp.

If you don’t have huge power demands I’d recommend Dennis Had Inspire Fire Bottle KT88 SEP amp, and an Inspire LP 3.1 preamp. These products won’t break the bank, especially used, yet their performance is very difficult to top, especially at that price point. For $3k you can get outstanding sound for cheap compared to other products. I’ve had Inspire products since 2016, and happy to have discovered them. Audiogon is well aware of Inspire products!

Only if you keep digging...

I have to say that I am enjoying this thread alot, and I am learning quite a bit, being a newcomer to tube amps (I have owned my first for a couple of months now), I get the feeling this rabbit hole runs deep!

Checking bias monthly is OK, but knob dicking the adjustment pots will shorten their life. Tweaking for the last fractional mA is pointless. Great amps were biased with moving coil voltmeters for decades before digimania took over.

The bias is going to shift slightly with line voltage, time on and ambient temperature variations. Home users would do better to check bias at the end of the listening session with the amp and components are fully warmed.

For me autobias isn’t critical as long as it’s easily adjusted. I’ve owned Unison, Allnic, Quicksilver and Audio Research. My favorite sounding was my AR which is a pain to bias but the others provided an accessible dial. Biasing once tubes are settled should be monthly. Stick with a company that has been on the market for a while and has a solid support network. Buy an extra set of tubes or two and you should be set for a decade provided the amp isn’t left on. Good tube amp hygiene, no turning on and off in quick succession and allow a 15 minute warm up or more before hammering the system.

A lot of good advice from the group here, although it doesn't necessarily square 100% with my experience, especially those who commented on impedance issues, tube life, and noise.  I have two primary systems designed around the use of high and low-medium efficiency speakers, Altec 604 and Magnepan 3.6.  I use tube amps from 1.5 wpc to 35 wpc with the Altecs, and tube and SS amps with the Maggies, tube preamp in both cases, and power output from the mid 50s to 120 wpc.  The two SS amps I have range from 150 - 200 wpc.  They all sound great because they're the keepers of the dozens of amps I've vetted. 

The key is choose your speakers first.  Then let the speakers drive the amp options.  My tube amps don't care if they're running 4 ohm or 8 ohm loads, have had both, and the 4 ohm loads do not cause any issues with greater loading, heating, etc, with the tube amps, but yes, the SS amps run hotter with the 4 ohm load.  As far as tube life, I run primarily old stock and NOS tubes and since I only listen seriously a couple hours a day, they seem to last forever.  I can't recall the last time one of my tube amps was noisy for any reason.  With the Maggies' lower sensitivity,  tube amps are dead quiet.  With the Altecs, there's a tiny bit of hum with the SET amp, and the rest are nearly silent.

Also great advice from those who said to avoid amps with proprietary or single source, or otherwise rare tubes.  If an audition with your speaker choice is possible, then you'll be sure you're getting the best matchup for your audio investment.  Have fun shopping!

From what I hear DHT is right, but budget and such matter. Rule 1 though is GIGO, garbage in, garbage out. My system with an old PS Audio 200C or Acoustat 3l200 watt amp would sound good, but my Mark Levinson smokes them both. OTOH, if I had a $100 Schiit DAC in my system it would sound far worse than a TNT in it with the Lampizator DAC. The thing that I like though may not be what you like, so if I give you the correct answer it's dumb luck that we have similar taste in what we seek from music.

I think there are great answers here.  OP don't be intimidated.  I have two tube amps now.  My main system is a PL Evo 400.  Nothing but good things to say about this AMP.  My second is a 'vintage' Grant Fidelity Rita - in my office and more of a hobby piece for me - would be just as happy SS there.

Piling on - I could live without the UL/Triode switching.  I stay on UL 99% of the time.  If i only had triode that would be fine too.

Auto-biasing a must for me.  I have many things to fill my time. 

I turn it on. Listen and love it.  Turn it off.  There is not stress, and limited tinkering if you don't want that.

About a year in I started getting a cracking noise out of one channel.  Pulled the tube and reseated it - problem solved (with AG community help).  A 20 minute exploration and deepened my understanding and learning...fun.

You can turn this into a full time rabbit hole hobby - but tubes do not need to rue your life.  

But I have a system that gives me chills, makes me laugh out loud, and is generally the most special "me time" thing in my life next to my cars.

Have fun with it.  with some modest research on amps and speakers there really are very few bad choices.

OP's  question is very good question, and it should go with any speakers. Don't limit the range of speakers when we reply the questions.

Here is my answer if you want to have a good setup.

1, tube power amplifier

2, tube crossover in the speakers.

3,tube preamp 

Choice 3 is optional, you might get SS preamp instead.

Here is the point with the speakers: some speakers has  separated crossover and could replace easily and if the crossover is inside the speakers, it is very different to replace.

The crossover has very strong point to the sound which can compare with preamplifier ,but in long time, we ignore it.look back the vintage UK speakers, they always have outside  crossover which let us to modify easily later.

The following might seem to be have a air of Negative as reply.

Note I have had a Tube Pre Amp > Power Amp in use since the 90's.

Tube Amp's command a slight alteration to the disciplines of use that can be used with a SS Amp'.

There is a concern for a occurrence where a fire could be the result.

A Drastic Statement it is, but anybody with a Long-Term experience around individuals with a Tube History, will pick up on the need to heed the above and express a 'Vigilance' when using a circuitry that has Tubes.

In a nutshell, when using Tubes it is best to know any early warnings something is amiss with a Tubes Function, it is most reassuring to be able to see the Tube as a operating component with the Power On.

'Vigilance' means being present and observant, aware of developments that suggests a Speedy Power Off Interception is for the best.

I have disciplined myself to be in full time attendance in the Listening Room , with only very short periods of absences during periods that the Tubes are Powered On.

If an individual like to have extended periods without a observation of the equipment in use, the use of Tubes might not be ideal for such practices.

All who I know with Tube Devices are very similar disciplined to myself, which has resulted in not personally knowing anybody who has had a fire in their set up system. I even know one individual who seen an Element start to glow more red than orange and did not suffer any serious outcome as a result of the immediate Power Off that was actioned. 

The same is not said for those I know that extend their use of Tubes to the place where they work with Tube Devices on the Workbench, I know of smoking circuits occurring. 

I do not ever see myself without a Tube Device in use, but recent years have certainly introduced myself to SS Amplification that in my assessment could be used in place of my Tube Equipment. I have already pretty much decided a retirement system to be created, may have a Second Power Amp and a few devices that are SS, as I don't have to be so attentive to these when in use.      

Ralph, stated another way, if the grid voltage remains constant as the transconductance changes, the current in a bipolar [CT] transformer will be unequal.

Seems like a greater evil... unless one is OTL 😉

For example: does any manufacturer use an inductor filtered power supply instead of the much cheaper (and lighter) capacitance filtered power supply? None that I know of.

@xenolith As far as I’m aware, chokes are always used in tandem with capacitors. If you know of a topology that uses only inductors, I’d love to learn about it (Google failed to find any such thing). We’ve been using chokes (inductors) in our power supplies for decades and we are by no means the only one- the use of chokes seems to be pretty common IME.

FWIW, if the amplifier is class A and also ultralinear, you won’t need any adjustment since the correct bias can be automatically obtained using a cathode resistor. Regardless of the transconductance of the tube, if the cathode resistor is correct, the tube will always operate in the A region.

Of course, if you want to manually bias the circuit and run a bias supply (and no cathode resistor) you can, but the tube will be set at the right/same current anyway. One thing that will happen though is that as the tube ages and the transconductance drops off, if you always set to the same current the tube will be over-biased in time.

Push/Pull (I don't like really hot amps)

Don't spend too much on a tube amp. Some chinese companies (not all) are making excellent tube amps with warranty/ repair centers in USA

Include the cost of the better NOS tubes you will want to roll in

Better electronic components beats fancy features