DHT 4P1L PSE Amplifier
The time has come. I was ignoring DHT tubes for a very long time indeed. But after measuring all of my tube-stash and seeing first hand how spectacularly linear those puppies are I was instantly charmed and the writing was on the wall. I have to build and an AMP with them! Go here and press table sorting by THD to see what I mean. Nothing can touch them and 4P1L is not the most linear out there.
I bought a bunch of 4P1L’s back in late 2000’s when I stumbled on traced curves by klausmobile. These tubes then where available at all sorts of flea-markets around eastern-central Europe and would cost cents. Not anymore it seems, but prices are still reasonable. They are now at the spot-light mainly because of all the work by Ale Moglia and his great blog. You cannot walk by it when googling 4P1L now and nor could I. So this project is heavily influenced by his work and I was not trying to reinvent the wheel here. It’s hardly possible anymore with tube circuits. Everything was already thought out by the end of 50’s and then meticulously nit-picked again by a numerous audio community’s with the explosion of the internet. And same can be said about Bartola’s adventures with 4P1L, although I’m greatly thankful for introducing me to a 6e6p-dr driver tube. What a magnificent sleeper!
But is it really worth it? All that steel and all that heat for what? Some miserable 2W of amplification power? And than again, why we should even care about tube linearity? After all, I have stated numerous times on this website that THD is meaningless number and doesn’t correlate with sound quality. And I stand by that. But from all my adventures with all types of tubes, I can state this:
“All things being equal, more linear tube will always sound more natural.
The emphasis here is on “all things equal” and this is not easily achievable in some circumstances. But even with small amounts of negative-feedback the trend stays the same. Until we arrive to the land of technocrats with deepest feedback levels possible where everything suddenly sounds the same. So yes, it’s worth it. And as all my audio-loving friends who heard this amp in their systems can testify, 2 Watts can be enough even with a “daft” speakers of 88dB sensitivity. It wont do anything spectacular and you will not be blown away by it’s sound, but it’s sure enough for a regular listening levels without making your neighbors mad.
I was still very skeptical on the success of the project. There where two road-blocks ahead. It’s AC heating and output transformers. Please read my ramblings and the solution to AC heating problem here. After hearing how good can DC regulated DHT sound I was more optimistic. On to the output transformers then. This is the sole reason I’ve never went to a full tube-based amplifier in my main system for longer than a couple of weeks. All transformed-coupled amplifiers I have ever build/borrowed sounded bandwidth-compromised! And yes, all output transformers sound different and they all get better the more zeros there are at the price tag. I liked amps with Hashimoto/Kondo OPT’s and recently I heard 300B with Monolith Magnetics SX-9, very impressive! But I’m still missing all the micro-details that I’m expecting from my speakers. Hereby my 53.05’s from reohrentechnik.de that I scored from flea-market amp wasn’t inspiring much confidence despite only positive comments from my German tube-audio nut-heads. I guess there is only one way to find out. Build the damn thing! And so the rat-nest prototype above was born.
I have tried a lot of driver tubes. 6Z38P-EV, 6Z52P, 6Z49P-DR, 6P36C, 6E5P and even D3A. They all sound different and I didn’t expected that at all. Usually with electronic-load like CCS or gyrator all good linear tubes sound similarly benign. But here, driver tube and 4P1L distortion products mix with each other and because DHT sounds really different than any IHT, the end result is always unpredictable. From all the tubes above I liked D3A best and was about to order a fresh pair, but after reading Ale’s appraisals of 6E6P-DR I change my mind and ordered this one instead. And I wasn’t disappointed. Hands down 6E6P-DR wins as a 4P1L driver and by a some margin. Once again it just shows how important is to actually listen to what you’ve build.
After playing around with BIAS points and measuring the essentials I came-up with a final schematics above. Driver tube load is a fixed-voltage mu-follower or a gyrator if you like. I don’t get hang-up on exact terminology, as it does what you think it will – provides high-impedance load for tube at a constant voltage set with 250kΩ pot. If you take output from anode, impedance will be high in to the 1 kilo-ohm range (anode resistance). If you take it from a source of the BS170 – than natural feedback loop of C1 – R1 will lower output impedance to ~40Ω. This is very handy. Now two grids of 4P1L can be driven into class A2 with standing current of 14mA very effectively on music peaks. For a real class A2 you’ll have to add a beefy buffer capable of holding the DC point though. My driver load can be easily breadboarded but for other SMD parts Ale has PCB’s to sell. Bandwidth of this stage with BS170 is 160kHz (-3dB) and is mainly limited by a volume pot. Driving 6E6P-DR with low impedance will extend that into MHz range so I would recommend a 10kΩ volume pot.
Two 4P1L’s are connected together hence the Parallel-Single-Ended (PSE) topology name. Both tubes are grid-biased and above is a circuit for both channels. Name of the game here is voltage stability. You don’t want your BIAS voltage wondering with temperature or your DHT’s might turn into simple light bulbs pretty quick. I know purist among you will not be happy with TL431 shunt-regs. But come on, this is not a line stage! Grid is driven with 40Ω and will swing some 40Vpp so if there is something that TL431 can do with it’s 100k impedance – I can’t hear it.
Filament is DC regulated and circuit is the same as final schematic here. I have tried a series filament connection of 4.2V to lower the current from 1.3A to 0.65A but wasn’t satisfied with a result. So it stays parallel connected at 2.1V. As I mentioned in a filament article, there is also a huge difference which side of the regulator to ground with positive being a preference. Nevertheless I can’t hear a huge difference between grounding different sides of a parallel-filament. Center tap pin8 or open sides pin1,7 – sound pretty similar and you need to do a quick A/B to say something definitive. Feel free to experiment with that. I’ve left pins 1 and 7 grounded in a final amp.
With 4k transformer biased at 70mA and -21V at the grids I got really clean 2 Watts of amplification. That’s in tube-amp terms “clean” as being lower than 1%. Another nail to the coffin of sub ppm distortion hunters as this 1% sounds so much better than anything transistor based. Distortion starts rising at 2.5W and we soon run into grid current. With 3k transformer like LL1623 you can expect to push 3 Watts at 2% THD and still have some grid voltage headroom. My 53.05’s from reohrentechnik.de shown them self’s surprisingly well. With two parallel 4P1L biased at 35mA each we expect sum tube Ra to be around 900Ω and with this impedance bandwidth (-3dB) is about 55kHz. Not bad at all! Primary resonance is still out of audio band, well damped an doesn’t go above 1dB. Having wound couple of output transformers myself, I can say just by looking at the graphs above that these are really well engineered OPT’s. Of course, what’s more important is how they sound, but again – I’ve never heard anything good from OPT’s having less than 20kHz of BW.
Couple words about power supply. For 2.1V filaments, regulators will be happy with anything above 9V at their inputs. So target that if you will go with choke-filtering or different TR voltage. Also when using double-secondary transformer like above, make sure to decouple outputs with common-mode chokes. Otherwise regulators like to start oscillating when the inter-winding capacitance loop is closed by grounding both ends at common ground point. For 6E6P-DR heater I used something different than my standard LM317 soft-start voltage regulator. This time I went with a simple current-regulated CCS. So far so good. IHT’s don’t care much what’s happening with their heaters as long as it’s at right current/voltage.
For HV supply I first tried a simple Mosfet capacitance multiplier like in a schematics above. Despite driver stage having a really good PSSR and transformer inductance “shielding” the final stage, there is still a big influence of PSU decoupling. It’s not that critical like with a simple resistor loaded tube stage, but still, all caps I’ve tried sounded different and more expensive “super-caps” had a major advantage. So you’ve been warned. Throw as much money to supply decoupling and inter-stage coupling cap as you dare. This is critical for the final result. When I got tired of cap-swapping I decided to try something I never tried before – shunt power supply for a whole amp! I used the same ones I developed for my reference DAC tube stage and I will cover them in a future articles. I set them as “hot” as I could – 110mA. This is as much as my little HV transformer will go without bursting into flames. And my oh my. What I’ve heard was the most sweet 2 Watts of tube amplification ever!
I was finally convinced that this rats-nest deserved an amplifier chassis. And as soon as possible. I gathered some aluminum scrap pieces from all the corners of my shop, bought some heat-sinks and started building something that would resemble a tube amp. Now I don’t know how a tube amp “should” look nor do I have any influences to follow here. I just know that everything has to fit and function as it should. I also don’t like any bling or posh as my British friends say.
Simple Scandinavian minimalism works fine for me. Function over looks. Mind over matter. And no expensive woods that will leave our poor planet more wrecked than it already is. Front panels are made out of a cheap pine board and impregnated with natural oils. Locating second filament transformer right beneath the main one is not a standard solution to say the least. But it works. That 8mm thick steel plate is shielding volume stepped-attenuator very effectively and this amp is absolutely silent with 94dB speakers as all the other gear I build. This is very personal for me. Any hum in your system is a clear sign of poor-engineering. Here is how it all came together.
Not the cleanest of my builds, but hey, it’s a tube amp! So point-to-point component wiring is a tradition thing. It could be made much cleaner with proper PCB’s all over the place, but would it sound any different? I guess not. Possibly even worse. And talking about the sound. I will not bore you with my panegyric about what does a DHT sound like or what it does to your music. This must be experienced first-hand to really appreciate it. Only then one can try to put it into words that will mean anything to your subjective cognition. All I can say is the hype is real. Everything you ever read about them is true in one sense or another. I’ve never heard a human voice sounding on my system with such a level of realism! It’s not even funny. Blindfolded and with good vocal recordings like binaural ones I could easily be fooled into thinking someone speaks or sings actually standing right in front of me. This never happened in my room before. And I had quite a few 300B amps for evaluation costing close to a small fortune. There is just something special about this 6E6P-DR and 4P1L combo and it’s overwhelmingly addictive.
Now if your understanding of audio nirvana is a Telefunken radio from 40’s, maybe the same one that your grandma had, the one you used to sit and listen to a as a kid. And that realistically sounding “voice from a box” is now engraved into your subconsciousness to such a degree that everything else just sounds too “modern” and “alien” to you, I get it. I really do. Whatever makes you happy! This is what’s this hobby is all about. But let’s just make it clear that this retro “low-fi” sound has nothing to do with “high-fidelity” and our subject here would leave you immensely disappointed. This sweetheart just sparkles! With energy, dynamics and micro-details that I would never expect from a tube amp. It’s almost on-pair with my reference amp that I will try to cover soon in my next article.
The responsibility for that much increased resolution falls down to a super-low-impedance of a shunt regulated power supply. I always thought that output transformer is a bottle-neck of any tube-amp and it’s true to some degree. But it’s not a brick-wall type of limitation and can be further mitigated. So even an average transformer can be “transformed” by an order of magnitude given enough low-impedance capacitance or regulation. It’s not all about frequency response but also the slew-rate. You live and learn. And finally it’s a visual thing also. Looking into this honey-dripping yellowish sweetness when listening to your favorite songs is really hypnotizing! It’s like a ritual of putting your vinyl on turntable or looking at a reel-to-reel player spinning. You just have that feeling of something wonderful happening right in front of your eyes. I guess pics above speak for themselves. Drum-role please for the final gallery of daylight portraits.