Hi-Fi Valve Amplifer Kits

[SpecialK]

"Hi-Fi" and "Valve" don't go together... people who like the latter want the distortion, people who like the former want nothing more than amplification.

Okay.  I agree in principal.  However, it is something I want to experience and will decide if there is something special or if it's wishfull thinking.

I am compelled to test the theory that it's all about being a constant current source rather than a constant votage source.  Sweeping a signal around a speaker's resonant frequency and comparing to a solid state amp would be interesting.

[trobbins]

I've had the unit up and ran it up to about 30wpc into dummy loads before clipping.  Even after getting the global feedback properly phased, it only takes 0.5Vpk-pk to drive it to clipping.  I haven't set up volume pots yet, but I suspect I might need to reduce the gain or increase the NFB.

Even if you have a very simple form of global feedback you may easily get oscillation at low and/or high frequencies unless you are pedantically following a known mature design using exactly the same parts as specified, and even then the actual wiring and implementation can sneak in and cause instability.  There are certainly DIY ways to test for stability margins, and that is recommended if you have any value in speakers.  Perhaps the KISS principle initially applies, and just use an input attenuator to ensure your signal source can't force the amp in to clipping.

I do have some questions about grounding and earthing.  ATM I have the chassis earthed.  While under test I have the unit on an isolation transformer.  However, any stereo receiver I've ever worked on did not have a chassis earth connection despite having a steel chassis.

I have not connected the DC ground to chassis.  Every piece of audio gear I've worked on used the chassis as DC and signal ground.

I suppose I should be using the chassis as both PE and ground.

It sounds like you are making this up 'on the run'.  I'd recommend following a well known, mature amp as a template as to what they do.  Safety always dictates using a protective earth to chassis (unless you have some modern equipment with compliant double insulated mains, but even then hum noise can still circulate).  Standard audio circuitry grounding approach is just one link from circuitry 0V to chassis - where the link is on the 0V distribution is open to many implementations, and can depend on how other audio equipment is interconnected.  Avoiding or suppressing hum loops is a wide topic that many have discussed.

[SpecialK]

I wouldn't say I'm making this up on the run.  The PCB I am using specifies component values and I imagine the design was copied from a popular unit.

However, I have had to make some concessions to parts availabilty and modify to suit.  For instance, the schematic shows a transformer with a 335V secondary into a bridge rectifier.  All the transformers available locally were of much higher voltage.  I opted for a 600V unit 300-0-300 whic required me to use a full-wave recifier instead.   The bias power supply was shown with a completely separate 70V winding.  Of course the Hammond has only a 50V tap with respect to the HV secondary.  I ended up going half-wave knowing the grid shouldn't flow current and sag too much.

Since my B+ was going to be a bit low, I opted for bigger OPT 4300ohm primary instead of 5.5k on the schematic.  I should be able to drive it with more current as I assume plate dissipation = E*I

The whole point was to be a learning experience.  I must say I have been having good progress, and having fun.

 

[MIS42N]

I am about 90% finished building a EL34 ultralinear push-pull.  Took the "easy" route and used a PCB from eBay.  Then spent a tidy sum on some beefy Hammond iron.

WooHoo, that brings back memories. I built the same maybe 50 years ago. I created a regulated 400V supply using a 2n3055 mounted on a massive heatsink as a voltage dropper from unregulated 450V. The heatsink was a death trap, floating at the supply voltage. The output transformer was massive (and the most expensive part by far). It delivered around 50W before distortion (1%). Wished I knew where it went in the moves since then.

You are certainly diving in at the deep end. I built a few valve amps in the day, mainly kits. Started when I was a teenager. I think the nicest was a 10W channel ultralinear push pull for my mum. The output valves may have been EL84s.

Transistor amps are 'better'. They deliver more power with lower distortion. A past neighbour had a digital processor that could be used between preamp and main amp to emulate various 'sounds'. One setting was for a Fender bass amp, it required a 500W output amp to emulate a 50W valve amp to get the overloaded sound without actually overloading.

Not really helpful, but I shall follow this thread to see if you like the results. I don't know what speakers you have, they also colour the sound. Valve amps don't have the same damping factor as transistor amps so the speakers are a little less controlled - which may bring something to the party or just sound 'muddy'.

[drussell]

"Hi-Fi" and "Valve" don't go together... people who like the latter want the distortion, people who like the former want nothing more than amplification.

"That's like, your opinion, man." 

While I do happen to prefer the sound of my Dynaco Mark VI tube amps to that of my Bryston 4B, I would certainly say that both of them count as high fidelity!!

(Especially in these days of highly psycho-acoustic compressed music and crap playback devices that people think sound "good.")   

[Gyro]

Yes, I thought that was a gross over generalisation too. There have been many excellent, low distortion valve amplifier designs / products.

Hi-Fi pre-dates transistor designs by a significant margin (and good transistor designs by an even bigger one).

[Zero999]

Yes, I thought that was a gross over generalisation too. There have been many excellent, low distortion valve amplifier designs / products.

Hi-Fi pre-dates transistor designs by a significant margin (and good transistor designs by an even bigger one).

It's true though. There's no point in going for valve if you want Hi-Fi. Before power transistors existed, there was no alternative, but thermionic glassware no longer makes any sense for Hi-Fi from a technical, or practical point for view. Fair enough if you want to have fun, go for it. There's nothing wrong with that. There are still lots of steaem enthusiasts around.

[RandallMcRee]

Yes, I thought that was a gross over generalisation too. There have been many excellent, low distortion valve amplifier designs / products.

Hi-Fi pre-dates transistor designs by a significant margin (and good transistor designs by an even bigger one).

It's true though. There's no point in going for valve if you want Hi-Fi. Before power transistors existed, there was no alternative, but thermionic glassware no longer makes any sense for Hi-Fi from a technical, or practical point for view. Fair enough if you want to have fun, go for it. There's nothing wrong with that. There are still lots of steaem enthusiasts around.

Oh my god, can't believe I am reading this BS on eevblog.
Valves can do anything that transistors can do--its all in the implementation. Go over to diyaudio and talk to some experts.

It's just that:

• Valves are much much more expensive.
  Valves are cool, because retro and retro is fashionable.
  Valves are less reliable (and thus even more expensive)!

Just remember that speakers are your weak point for distortion and artifacts, not the amplifier
Valves can easily reach 0.1% THD which is pretty much golden ear territory.

[drussell]

Just remember that speakers are your weak point for distortion and artifacts, not the amplifier

Indeed.  The choice of speakers will always make far more of a difference to the overall sound quality than the amplifier choice (unless it is spectacularly bad, I suppose... ) which is why I've always built my own.  Over the many iterations and experiments with speakers, I learned far more about music reproduction and acoustics than my various amplifier endeavors over the decades.

Plus, there's nothing wrong with building a nice ultralinear... 
I expect the OP will be pleased with the sound from his warm glowy sound reproduction apparatus.   

It's wintertime here, you're allowed to live by the glow... 

P.S. To the OP, what style of circuit did you choose for the phase splitter?

[Zero999]

Yes, I thought that was a gross over generalisation too. There have been many excellent, low distortion valve amplifier designs / products.

Hi-Fi pre-dates transistor designs by a significant margin (and good transistor designs by an even bigger one).

It's true though. There's no point in going for valve if you want Hi-Fi. Before power transistors existed, there was no alternative, but thermionic glassware no longer makes any sense for Hi-Fi from a technical, or practical point for view. Fair enough if you want to have fun, go for it. There's nothing wrong with that. There are still lots of steaem enthusiasts around.

Oh my god, can't believe I am reading this BS on eevblog.
Valves can do anything that transistors can do--its all in the implementation. Go over to diyaudio and talk to some experts.

It's just that:

• Valves are much much more expensive.
  Valves are cool, because retro and retro is fashionable.
  Valves are less reliable (and thus even more expensive)!

Just remember that speakers are your weak point for distortion and artifacts, not the amplifier
Valves can easily reach 0.1% THD which is pretty much golden ear territory.

What BS? I didn't say, that valves can't do Hi-Fi, just that there's no point, when cheap solid state devices can do the same.

Room acoustics are also a problem, unless earphones are used, which I personally dislike, especialy earbuds.

[drussell]

Room acoustics are also a problem, unless earphones are used, which I personally dislike, especialy earbuds.

Yes, room acoustics are also a huge factor in most cases, which is why I've always loved running my full big PA rig outdoors the few times I've done that. 

Earbuds are definitely all crap but decent headphones can be an absolute dream.  My favorite ones are still definitely my original late-70s style Sennheiser HD-400 phones, the real, original 600Ω ones with the yellow foamies...  not that new crappy model series they sadly sell as an HD 400.  You do need something with some serious voltage swing to drive them, though.  They're not going to play at any kind of level on today's typical ultra-low-power headphone jacks intended for 32Ω

[RandallMcRee]

Yes, I thought that was a gross over generalisation too. There have been many excellent, low distortion valve amplifier designs / products.

Hi-Fi pre-dates transistor designs by a significant margin (and good transistor designs by an even bigger one).

It's true though. There's no point in going for valve if you want Hi-Fi. Before power transistors existed, there was no alternative, but thermionic glassware no longer makes any sense for Hi-Fi from a technical, or practical point for view. Fair enough if you want to have fun, go for it. There's nothing wrong with that. There are still lots of steaem enthusiasts around.

Oh my god, can't believe I am reading this BS on eevblog.
Valves can do anything that transistors can do--its all in the implementation. Go over to diyaudio and talk to some experts.

It's just that:

• Valves are much much more expensive.
  Valves are cool, because retro and retro is fashionable.
  Valves are less reliable (and thus even more expensive)!

Just remember that speakers are your weak point for distortion and artifacts, not the amplifier
Valves can easily reach 0.1% THD which is pretty much golden ear territory.

What BS? I didn't say, that valves can't do Hi-Fi, just that there's no point, when cheap solid state devices can do the same.

Room acoustics are also a problem, unless earphones are used, which I personally dislike, especialy earbuds.

Hey did not mean you specifically, I meant the  first post #3? (which I flagged to moderator as a troll) and the general notion that valves *cannot* be Hi-Fi.

To me, to clarify, valves are like the passion for vintage automobiles or flying old jets. The technology has moved on but the old tech can still do what needs to be done and has that nostalgia value, to boot. Nothing wrong with that, right?

I really think we should presume, unless told otherwise, that the OP *does* want the old tech, because he specifically asked about valve amplifiers. Surely he would not do that unless he knew about their drawbacks and perceives them as virtues?

[SpecialK]

Just remember that speakers are your weak point for distortion and artifacts, not the amplifier

Indeed.  The choice of speakers will always make far more of a difference to the overall sound quality than the amplifier choice (unless it is spectacularly bad, I suppose... ) which is why I've always built my own.  Over the many iterations and experiments with speakers, I learned far more about music reproduction and acoustics than my various amplifier endeavors over the decades.

Plus, there's nothing wrong with building a nice ultralinear... 
I expect the OP will be pleased with the sound from his warm glowy sound reproduction apparatus.   

It's wintertime here, you're allowed to live by the glow... 

P.S. To the OP, what style of circuit did you choose for the phase splitter?

To clarify a bit, the motivtion to make this wasn't to build the greatest Hi-Fi amp money could buy.  I do realize that the bulk of your hi-fi budget should go to speakers.  I don't really care get geeked into audio to that level.

The idea here is to find out waht tubes are and what they are not.  Basically an exercise to satisfy my intellectual curiousity.

I'm not sure what kind of phase splitter I'm using.  I didn't lay out the design. I didn't have enough education on the topic to formulate an opinion. Here is the PCB:



https://www.ebay.com/c/1133273523

Mind you I bought the bare board for $16.50 USD.  It's quite well constructed.

 

[MIS42N]

I'm not sure what kind of phase splitter I'm using.  I didn't lay out the design. I didn't have enough education on the topic to formulate an opinion. Here is the PCB:

It's a common cathode phase splitter. Economical but not regarded as the best way to split phases. It is difficult to get the same amplitude in both phases.

I note also that there's no negative feedback. It looks like the circuit is minimal cost for good results rather than for all out best performance.

[trobbins]

There are L and R pcb terminals with 'FB' marking (in the linked ebay item).

Did the pcb come with a schematic, or a parts list with valve options etc?

PS. is this linked schematic your pcb?
https://i1171.photobucket.com/albums/r559/ling8301/EL34-PPSchematic_zpsdf8698b6.jpg

[SpecialK]

Here is the schematic. The one posted above does not show the ultralinear/triode relay. Yeah those are indeed the feedback terminals.

[trobbins]

FWIW I could recommend the following:
- solder bridge the bias header selection pcb pads and the pentode/triode switch pcb pads (as the last thing you want is a poor connection or to change selection whilst the amp is in operation or to take high signal level wiring off the pcb and near to other low level signal wiring).
- add a bleed resistor across say B2+ (nothing worse than getting belted by charged caps).
- use UF4007 if you have them (instead of 1N4007).
- the heater supply circuit is not clearly shown - I'd recommend making the B2+ bleed in to a voltage divider such that the bottom half of the divider is about 40-60Vdc, and then bypass that 40Vdc with a cap and connect it to the wiper of a humdinger pot (the arms connect to the 12AX7 heater, and the heater is not directly connected to 0V).  The humdinger pot may provide some hum nulling due to circuit layout, and the 40Vdc elevation may provide some hum reduction due to 12AX7 and 12AU7 heater-cathode resistive leakage.
- initially test for ok voltage levels with a low AC mains feed voltage (eg. variac) and no valves, and set all the bias voltages to max.  Although you could initially use a light bulb limiter circuit to avoid collateral damage if something is NQR, it is a hit or miss form of powering up (but still better than just turning the amp on straight away).
- initially test with resistor load instead of speaker, and confirm that connecting the feedback wire through say a temporary 220k series resistor reduces the output signal level (rather than increases it).
- set the bias up initially for about 70-80% of 25+8W dissipation for EL34 (ie. the cathode current is set to 0.7 x 33W / B+ volts), and check a few times that each EL34 has the same cathode current so that each output transformer primary has balanced dc currents.

There is some neat performance and distortion testing that can be done nowadays with a PC and soundcard, some free software and some custom connections to the amp.  If you go down that path then at least you won't have to rely on your golden ears to know that the amp is operating correctly.

[SpecialK]

- I did solder all connections as I did not like the idea of a 400V wire coming loose and shorting
- I added a 1MΩ resister accoss all caps.  I noticed without tubes thaey would stay charged with HV for a long time
- I did not know about UF4007.  I did install 1N4007 but also have 1000V Hexfreds to subsitute if necessary.
- I'm just running floating 6.3 to all filaments. Thanks for the humdinger suggestion, I will try to incorporate.
- I did start by powering w/o tubes on a variac.  Adjusted bias for -40V at 120V on the variac. 
- I did power into dummy loads.  I haven't connected to a speaker.  Dialed the bias pots until I got 40mA 0.4V accross the 10ohm cathode resistors.  - Then after applying a signal ran the feedack from the OPT secondary to the front end.  The wave form was distorted and larger so I swapped the -OPT primary wiring.  I wanted the black wire on the secondary yo be the common and ground.
- I think my bias is low at 40mA.  Both channels driven I get ~25WPC into clipping.  B+ sags to around 345V.  I will use your formula, but I was thinking I need around 50mA bias current.

Thanks for the software suggestion.

[wizard69]

Rectification method has no 'sonic' implications.  When tubes are working at their best, they almost equal a good solid state circuit.  Single ended isn't as good as push pull due to the inherent absence of even harmonic distortion in the latter.

I have been active in hifi and stereo for many years, including both tube and semiconductor circuitry.  The biggest lie is that tubes are better.  You may like them better but they reproduce sound material less accurately than good solid state gear.  No exceptions.

You can fall for the various myths such as gold plating, special cables, various brands and colors of tubes, and a whole lot more.  Or you can honestly get a decent amplifier and enjoy it.  Whether tubes or transistor.  Good gear sounds good and you fool yourself if you think that one method of achieving the result is better than some other.

I've always believed that SS is better.  I'm not gullible to the audiofool nonsense.  But I am curious to see if I might like a tube setup. 

My hearing is so bad it doesn't matter.

Really, I would like to run some experiments.  Do tubes sound good?  If so why?  Is it the front end?  Is it the low damping factor and transformer output?  Is it the even harmonics?  The non-flat response?

I'm pretty sure I could run some practical experiments and A-B comparisons.  Might need a couple pair of identical speakers and a pre-amp with multiple outputs.

From my perspective the best reason to build a tube based amp is that they look cool!!!    Seriously that glow and a row of tubes standing up right for all to see is not something you can duplicate well with a transistor based amp.   There is also a bit of nostalgia involved in tube based hardware.   

As for item #6 I'm not sure why you believe it would be more expensive.   It really comes down to your ability to source parts, fabricate chassis and other components.

Note that we are talking HiFi equipment here.   The story is a bit different in guitar amps where the musician leverages the behavior of the amp to create a specific soundscape.

[trobbins]

Great start.

Do you know the B+ voltage at idle?

Was the clipping symmetrical ?  One check is if the output waveform clipping is due to just the output stage load and B+, or due to the PI/driver stage voltage swing.  There are a few ways that could be determined, but it probably depends on what measuring equipment you have such as a scope and probes.

If clipping is asymmetric then the PI/driver stage has 33k and 39k anode loads that may benefit from some value tweaking.  You could also do some 12AU7 tube-rolling, and/or swapping the EL34 in case there are valve gain differences.

Are you operating triode or UL mode?  The schematic has an ambiguous screen stopper value of either 10 or 100R.  In UL mode a 100R may perhaps curtail the max output power a titch.

The B+ 47uF value also seems a bit light on for 2x 25W output.

Ciao, Tim

[SpecialK]

My notes say B+ was 368V at idle.  Could stand to double check as this was at a conservative 40mA bias.

Clipping is quite symmetrical.  I have a Siglent scope and could probably upload some pics.

Those are 100Ω resistors IIRC. I am operating in ultralinear, I haven't built a 5V supply for the coil voltage nor fitted a switch. The Hammond has a 5VAC winding for a tube rectifier filament, so it wouldn't be difficult. I doubt you should switch while the unit is powered on and I'm not sure triode would be desirable. I was actually hoping the relays were for a B+ delay before seeing the schematic.  Bummer.

I do have 100uF capactors I can use for B1+ and B2+, but I used them for the B+ not being able to find larger 450V rated electrolytics at Digikey.

[trobbins]

That sounds good about the symmetry, and that you appear to have uprated the B+ cap value by default.  Perhaps just worth leaving it as is, as any power output 'gains' are effectively illusory as you wouldn't hear the difference if you tweaked to say 30W.

368V x 40mA(cathode) = 15W per tube is very conservative, and yes B+ will sag as you increase bias current, so it can be wise to let cathode current settle for minutes between tweaks, and then recheck after a run of music.  If they are vintage EL34, then your B+ for UL could be increased to circa 400V.  Is the heater voltage sitting on 6.3Vrms, and are there primary taps on the power transformer?

[Alex Eisenhut]

If you can read French and can find them, I recommend these great reference works

https://www.renaud-bray.com/Livres_Produit.aspx?id=894280&def=Tubes+audio+anciens+et+r%C3%A9cents,IBRE,+FRANCIS,9782866611552

https://www.amazon.ca/TRAIT%C3%89-MODERNE-AMPLIFICATEURS-HAUTE-FID%C3%89LIT%C3%89/dp/2866611608

[SpecialK]

I finally decided it was time to hook up to speakers last night.  There was a hum, and routing the input and feedback cables didn't seem to affect it.  Wasn't sure if it was 60Hz heater or 120Hz from insufficient power supply filtering.  I disconnected the AC heater supply and tried DC 6.3V from a bench supply.  Noise was gone.

Today I reconnected the heater winding but grounded one leg.  The hum is gone.  Measured the ground current and it was negligible.\

Sounds great BTW.  I'm not set up for critical listening ATM, using small two-way bookshelf speakers with 5.25" woofers that reside in workbench.  The waterheater and furnace not being far away no to mention my benchtop multimeter has a stupid fan.  When all that is quiet it might have a tiny bit of hum or maybe that's my flourescent light or isolation transformer humming.  So far, so good!