A year is a long time... it is time for a project.

[shoshin]

A year on and I'm really happy with my last build. I'm missing building though and would like a new project. I'd like to try an all-DHT SE amp using current manufacture tubes. I don't need more than a quality 2W output; 4W would be nice. I'd also like to set up a PC-based test rig while at it. So, some fun learning ahead.

Having tossed a few thoughts around today, I'm thinking either:

1. EML 30A transformer-coupled (1:1) to EML Type 50; or
2. EML 20A/B transformer-coupled (2:3) to EML Type 50

My preference is for 1. but understand that it might take some work to get the 30A tube to "behave" (think Cu shields, low noise filament supplies, and compliant mounting). For boh options, my main concern is the transformer coupling: can I obtain decent bandwidth and dynamics with a driver having circa 7k Rp (30A; Tx gapped for 15mA) or 4k (20A/B, but 9k source impedance after the 2:3 Tx; Tx gapped for circa 30mA) into a Type 50? I expect to go custom for this, but even the best winders can only do so much. Thankfully, the Type 50 has a Miller not much more than a 45, and significantly less than a 2A3 (tough!) or 300B. The driver needs to swing some voltage, but should not need ultra-low source impedance. Experience? Thoughts?

Another option is to directly couple (monkey) a 20A/B to a 2A3, 300B or even EML's AD1 - the voltages are getting into my discomfort zone (circa 600V), but manageable. I've done direct-couple though and would like to do something different... but if my preferred different is not likely to work I will look at other options. I may also consider a 3-stage design, but would probably want a bit more power as compensation for the extra complexity, and cost!

I guess the sky will not fall in if I don't get this right first time; I will just need to factor in my options at the design stage (multiple PS transformer taps, optional top plate/chassis designs etc.) - I wish I had the space and $$$ to breadboard... and my own CNC machine (yes, I have considered it).

Thanks all for reading my out loud thinking. I'm throwing this out there for suggestions - any thoughts on designs that might meet my criteria? What type of builds would you like to see here?

Cheers.

 

[paul_b]

For the EML20 and EML30, you can use a plate choke and cap couple, though some people seem to be morally opposed to this.

While the 50 doesn't have miller capacitance to worry about like the 2A3 and 300B, the EML20 and EML30 aren't going to struggle driving any of these DHT finals. The 50 does require a considerable amount of drive voltage though. For the voltages involved with the 50, a little extra HV current and a 300B will get you a hefty bump in power and the iron requirements aren't so tough.

I wouldn't do a monkey-feed amp with that much B+ either.

If you do end up with Lundahl interstage iron, do not ignore the recommended compensation components in the datasheets!

 

[shoshin]

Thanks Paul.

Good to know those EML tubes will be fine driving moderate power, low-mu DHTs. Your comments about the 50 versus 300B are similar to my own: the 50 is less efficient and limited in power for the B+ involved.

A quality 300B at 350V and 50mA, 5K load would sound pretty decent with 6 low distortion Watts. By comparison, a Type 50 at 450V and 55mA, 4k4 load might give 4.5W, but requires greater voltage drive and is at max dissipation. The big "but" is that the 50 is supposed to sound rather nice.

I was looking to do something different to a cap couple, but I will not write it off just yet. Is it possible/desirable to use a grid choke when also using a plate choke on the preceding stage? And I guess there is the parafeed transformer couple option if I need step-up. Could do an all parafeed amp - that would be different for me!

For iron, I have had good experiences with J&K Audio Design in the past and would use them again - no problem with custom winds and although their better stuff is not cheap, I've found it to be excellent for what costs. Still, gapped interstages are a challenge.

 

[paul_b]

Yes, gapped interstage transformers are a bit of a pain, especially if you try to go for a high loading impedance. A plate choke by comparison is a little easier, but the high mu EML driver is going to want a 200H/20mA choke. I know Sowter could make something like that, but it seems rather inconvenient! I do not see any reason that you couldn't reduce the operating current and lower the plate voltage considerably, then use a constant current source. I have a feeling that this is about as inline with your goals as a coupling cap would be, but it would give you every drop of gain and keep the loading impedance very high!

5K output transformers are also if you want good bass at higher power! This is part of why we stay around 3K if possible. Heck, if 4W is enough, many new production 2A3s are perfectly happy with 20W of continuous plate dissipation, and finding a good 2.5K output transformer is far less challenging.

If you have old stock #50 tubes, the vacuum is purportedly rather weak, and the maximum grid leak resistance is pretty low on account of that. I really doubt that the new versions of the #50 copy this defect, but if you're running the old ones, a grid choke would be very helpful to get around the maximum grid leak value and not load down your driver tube. Other than that, they help a bit at maximum operating level when the finals are trying to draw grid current.

I think it would be fun to do a 300B amp that could be switched back and forth from cathode bias to fixed bias with the same B+ to change the available output power and the character of the amp.

 

[shoshin]

Thanks again Paul.

I'm not ruling out the CCS load on the driver - good idea. The EML 30B grid -V drops pretty quickly as plate V drops, so it gets a bit tricky at lower voltages. I will take another look. Are you thinking CCS load the driver, cap couple, and use a grid choke on the output?

No NOS #50s here - my future builds will all use current production tubes. The EMLs are reportedly as nice sounding as NOS globes, costabout the same (or less), and are warranted for 5yrs (original owner)... I will take the EML option.

I tend to like my EML 2A3s at higher Rp operating points, which unfortunately means higher loads. My outputs with 4k2 primary get me 46Hy (at 40mA). At 45mA, 280V B+ outputting 1W in circuit (a bit lower Rp than my amp), the transformers manage 44Hy, -3dB points at 5Hz and 80KHz with smooth roll off (I've seen the FFT screenshot). They are "20W" transformers and I assume can manage similar results with the 4W+ I might smash them with. A 2K5 transformer would spec better, but I think the 4k2 is okay enough. I hear and appreciate what you are saying.

I've contacted the transformer winder to get an ideal of what they can do for interstage - once this is known I will have a clearer way forward. Quietly, I suspect interstage transformers are best for low Mu, low Rp power tubes driving big output tubes into A2 and might be overkill in my situation.

Cheers.

 

[paul_b]

I'm not ruling out the CCS load on the driver - good idea. The EML 30B grid -V drops pretty quickly as plate V drops, so it gets a bit tricky at lower voltages. I will take another look. Are you thinking CCS load the driver, cap couple, and use a grid choke on the output?

That's probably what I would do if this came across my bench as a custom one-off build. 10mA/350V/4.5Vbias gives you plenty of swing but not so much CCS dissipation.

I tend to like my EML 2A3s at higher Rp operating points, which unfortunately means higher loads. My outputs with 4k2 primary get me 46Hy (at 40mA). At 45mA, 280V B+ outputting 1W in circuit (a bit lower Rp than my amp), the transformers manage 44Hy, -3dB points at 5Hz and 80KHz with smooth roll off (I've seen the FFT screenshot). They are "20W" transformers and I assume can manage similar results with the 4W+ I might smash them with. A 2K5 transformer would spec better, but I think the 4k2 is okay enough. I hear and appreciate what you are saying.

Those are good specs, I would use those too.

I've contacted the transformer winder to get an ideal of what they can do for interstage - once this is known I will have a clearer way forward. Quietly, I suspect interstage transformers are best for low Mu, low Rp power tubes driving big output tubes into A2 and might be overkill in my situation.

I've had an amp or two setup like that before, and although it does work to do that and you can setup the grid current and driver plate current to offset in the IT, I greatly prefer direct coupling for this type of amp.

 

[shoshin]

Back at work and less time to post here... or build, or...

I like the idea of a CCS load and grid choke - thanks. Saved for a future build.

For this build I would really like to try transformer coupling. I have not heard a transformer coupled amp and am curious. The 20AM/B might not be the best candidate for transformer coupling (the Rp is a bit high), though my winder reckons 20Hz - 20KHz flat, with -3 points at 10Hz and 60KHz, roughly. Not a significantly limiting frequency response and might be a good compromise. Wider bandwidth could be achieved more cheaply though.

 

[paul_b]

There is going to be a whole lot of copper if you want that kind of low frequency bandwidth, and all that copper will make the HF bandwidth difficult to achieve.

If you abandon the requirement that you must have a DHT driver that's transformer coupled to the output, then things get far easier.

 

[shoshin]

Thanks again Paul.

Yeah, it is a challenge for sure... would love to do all-DHT and transformer coupling, but trying to do it 2-stage could be too big a compromise. What the world (well, I) needs is a modern manufacture 8+ mu DHT with Rp of around 2KOhms... Maybe Jac will do an EML version of the AV8? I had thought about a PX-4 drives PX-4, with a 2:3 transformer couple. A bit weird. A bit expensive. A bit risky.

As an aside, I was looking for a rectifier tube last weekend and was reminded that I have about 20 matched pairs of 6C6 in the cupboard... in a purely financial sense, it is pretty hard for me to justify using anything other than a 6C6 as driver. Could always do another DC amp... using CCSinks (to stabilise current draws), gyrators, and SiC diode rectification for a relatively light-weight, modern take on using old tubes. Save the boat anchor for another day... when I don't have drag it up and down stairs to work on.

Too many options. Too little free time. Too little knowledge.

 

[paul_b]

What the world (well, I) needs is a modern manufacture 8+ mu DHT with Rp of around 2KOhms...

What you're looking for.

As an aside, I was looking for a rectifier tube last weekend and was reminded that I have about 20 matched pairs of 6C6 in the cupboard... in a purely financial sense, it is pretty hard for me to justify using anything other than a 6C6 as driver.

That would not be my first choice for an interstage transformer coupled amp. If you wanted to go back to something resembling the 91A, it will work.

 

[shoshin]

Ha! Are you suggesting I could run 1605 trans coupled to a 1605? The gain and Rp are right... costly, but tubes could be run conservatively for decently long life. Maybe increase voltage relative to current and get a bit more gain (would increase Rp tho... not really a problem at circa 1KOhm though). Am I considering this? Am I crazy?

I wasn't thinking of using the 6C6 with transformer coupling - directly couple it to a DHT (again) and use current sinks to set the current and stabilise the amp, use SiC diode rectification, maybe try a separate supply for screen. Same but different to last build.

 

[paul_b]

Sure, a 1605 driving a 1605 would please the late Sakuma greatly! You may not need all that much gain, you can go 1:2 on the interstage transformer and use a 1:2 or a 1:5 at the input.

6C6 directly coupled to a 300B could be interesting, especially if you control the screen grid voltage on the 6C6 to stabilize the operating current of the 300B. This would certainly not be a "purist's" 300B amp, but it would still be a cool project!

 

[shoshin]

Now I have two, very different, projects on the list. I'm still thinking transformer coupling for the next... but will purchase a trolley to move the amps from my workspace, down stairs, to the listening space. Will design chassis to be easy to move/ transport/ maneuver.

 

[paul_b]

The transformer coupled amps don't have to be that big. Something like a Hammond 126C is only 1.5kG. Making monoblocks should lighten the load adequately.

 

[shoshin]

Thanks Paul. I will for sure be building monos!

 

[Wntrmute2]

I'll be watching as you guys build some cool-ass shit. Too advanced for me but how else does one learn?

 

[shoshin]

Thanks for the kind words! With systems like yours I might be tempted to simply relax and listen. Paul is the advanced one - I just have enough knowledge to be dangerous. 😇

Yeah, wanting to do something different with this design and build. I'm learning some CAD basics when I have the time so I can more efficiently design the build parts I need. Let's see how this goes.

 

[shoshin]

Stream of consciousness warning.

I've been unwell the past week or so... and in my feverish thoughts I came back to a place I have been many times before: the 300B could just about be the most ideal all-rounder tube for folks with decently efficient speakers. What follows is based on my preferences and assumptions; yours may differ.

Sonically, the modern 300B comes in a variety of flavours, which can be voiced nicely via the operating points. Run them low and hot, loaded down, for that trippy 300B sound; run high and cool for a more 45-like sound. Practically, there are plenty manufacturers at a range of cost points. They are not too tough to implement and can produce decent power with low distortion (or high distortion, if that is your preference) at not nearly max dissipation (though I suspect for 40W is overstatement for some versions). Technically, not terribly difficult to drive, not too costly to implement, and comprehensive data is readily available.

I am still considering others... AD1 would likely be really nice, but manufactured only by EML (a critical supplier risk) and it needs about 300V on the plate to produce decent power, putting it at max dissipation; I like the gain of a PX-4 but power is limited at low distortion operating points even at max dissipation and suppliers are limited; PX-25 could be ideal, but who makes a real one (not TJ/Full Music); 2A3 is nice at right operating points (not the classic points) but unless running them at max dissipation (which apparently should still okay for long life) power is marginal; #45 is also nice, but power is even more marginal, even at max dissipation; high dissipation 2A3 is a contender... but there are only two manufacturers (from the same factory, EML and JJ; correction: Paul reminded me of Soviet and EH); #50 is supposed to sound glorious, but they take some swing to make 'em swing, power is limited until approaching max dissipation, and there are only two quality manufacturers ($$$)... and so it goes. All that said, operating points can be found for some of these tubes from a given B+ (say, 360V) and load, so you could start with one and change over if desired, with perhaps just a change to the filament supply. And I’m not building my final statement amp!

So I’m back to the 300b: can the modern 300B reach the sound quality heights? I think yes, but to reach them you may have to pay for the "better" types and find the operating points that suit your preferences.

 

[Wntrmute2]

Do it!

 

[paul_b]

Lots of the new production 2A3s are just fine with 25W of plate dissipation. The EH and Sovtek offerings do quite nicely running this hard. There's certainly the suggestion on the WE 300B operating points list that high voltage is more of a problem than high dissipation.

For me, the performance of a 300B amp is far more dependent on iron quality and circuit design compared to the actual 300Bs you're using (provided they aren't the 90s ones from China).