Starting a thread for this
6SL7 input tubes in SRPP configuration, directly coupled (no capacitors) to 2A3 driver tubes.
Somewhere in the 3W range into an optimal load.
18 Likes
What’s the most annoying thing that you came across building something like this, so far?
1 Like
@Polygonhell if you end up starting up an audio LLC and making these, I call dibs on offering my time to be your helper. This is like a sign in my search for finding a joyful calling to balance my day job…
4 Likes
Amp has about 30hrs on it at this point, plugged into the WiiM Pro as a DAC.
I will say I’ve never heard a piece of equipment change more over 2 or 3 days, I assume the combination of new tubes, new transformer, new capacitors. Though there are no coupling capacitors to burn in.
It went from ascerbic to rediculously relaxed to somewhere between the two with excellent macro dynamics over about 3 days.
I’ve used a variety of headphones on it, works well with the Utopia, HD6XX, HD800s, MySphere 3.1’s.
I guess I should start with what I expected
Aggressive dynamics as a result of the SS rectifier
Very linear response as a result of the transformer being designed for an 8 Ohm load
To put that in perspective the load line for an 8 Ohm load ought to look something like this, how linear an amplifier is is dictated by how evenly spaced the blue lines are on the red line, this is a pretty typical load line for a 2A3, I dropped the bias point a bit over the data sheet to 50mA because otherwise with a larger anode load you can get precariously close to the 15W maximum for the tube.
This is what it would look like into the Utopia
And what I got, pretty much what I expected.
Stage is large.
It has good treble and Bass extension, though it’s certainly nit bright or bloated.
Leading edge dynamics are impressive.
It has a touch of midrange warmth, though I wouldn’t describe it as wet.
Bass is perhaps a touch forwards (I suspect because of the transformer).
I need to hear it on a better DAC, but it’s a bit painful to relocate and it’s too far away from either of them currently.
And the Bad
It’s not completely silent, probably closer to A&S Pendant noise levels that Eddie Current or DNA.
Some of that is the fact I can’t adjust the trim pots finely enough, I have a couple of new pots arriving this week, that should solve that problem.
And I think there is a second noise, a very quiet hiss, that is only impacting the Right channel, that goes away if you turn the volume up with nothing playing. I suspect this is a ground path issue, but it’s hard to isolate until I resolve the hum pot issue.
There’s no soft start on the tube heaters, so on startup it likely puts excessive current through the tube heater filaments, I’ll address this at some point, but until then it’s not getting anywhere near expensive 2A3’s.
Things I want to try.
DC heaters for the pre tubes, and possibly later the 2A3’s.
The Lundahl output transformer, it has taps for 16/64 and 300 Ohms.
12 Likes
The metal work, I hate drilling holes in things, especially large holes in Aluminum.
I did it all by hand, and it’s just messy.
The rest of it wasn’t “Hard”, though I wouldn’t point to point wire if I were doing it again, my OCD makes making any decision as to where to run something painful.
6 Likes
What drove you choice in tubes? Amp design, desired sonic signature, other? Any chance of tube rectification? Who do i contact to put in a deposit?
2 Likes
I had a spare 2a3, I could test with…
I considered a few tubes, 45 among others, wanted a bit more output than I’d get out of a 45.
I intend to make a single power supply rather than the independent ones, when I do that, it will likely be tube rectified, the more SS aspect is the fact it’s regulated though.
No intention of selling them, was more a learning exercise, one that’s still ongoing as I mess with the design.
I have a design for a hybrid with an SE MOSFET output section I also want to build when I get through messing with this.
3 Likes
Is there a way, without too much extra complexity of making it dht switchable? I know some dacs and amps allow this but i dont know the mechanics behind it.
Ok but just warning @NickMimi has ways of getting people to “cooperate”
2 Likes
To a point, but it’s a compromise, the way lampizator does it is the “pre” tube is an active load/CCS so they get in effect a horizontal load line, they can then pick a really low bias point, that will work for a VERY large number of tubes, then they just have to deal with the heaters.
Configurations like that can’t deliver any real power, but it doesn’t matter for a pre/DAC.
In an amp with an output transformer you also get into optimal loads, you’d ideally use a different transformer for say a 300B vs a 2A3, it’s a bit less critical given the headphone loads, but you won’t get apples to apples comparisons.
There are probably several other DHT’s that will work, but I’d have to ensure the bias point is OK. And obviously deal with different heater voltages. I’m pretty sure a 45 won’t work, because the bias point will be too high, but I’ll check it in simulation and verify.
The amp being directly coupled means a few things one of them is it’s very sensitive to component values, I can’t just swap out cathode resistors.
3 Likes
I just checked and confirmed, the Bias point is too high for a 45, would require replacing the Cathode resistors, which because it’s directly coupled are actually 3 12W resistors. Might be interesting to try at some point.
Would probably work well with a PX4, though it would require an adapter and different heater supply.
For a 300B it would be very linear, and underpowerd, but would work with an updated heater supply.
3 Likes
Small update.
Got some new 10 Turn Pots, that let me cancel out the vast majority of the heater noise.
I also resolved a ground issue, I’d deliberately grounded the inputs to the case, and it resulted in a ground loop that could be heard for the first 30% or so of the volume pot.
That was an easy fix, once someone on DIYAudio forums suggested it.
I have a couple of other ground path fixes I want to make, because I can
I still have a slight crackle audible with very sensitive headphones, which could just be a noisy tube (I have some extras on order).
As an aside, I got my Susvara back last night, so I thought I’d give it a shot on the amp, it’s surprisingly good, volume at about 50%, background obviously dead silent (thankyou low sensitivity), doesn’t sound constrained like it did on the DNA even on the Studio T. Very open sounding, possibly a bit dry, and I really now want to try it with a transformer better matched to the headphone. And a DAC that costs more than $100. But that can wait until I’ve worked through the last of the ground fixes.
8 Likes
I love the “Open” sound of some tube amplifiers, it is what attracts me to particular units more than others The Radu Tarta unit reproduced a character of this nature on HP’s and so does my Decware unit. Very nice @Polygonhell 
I am following along and enjoying living vicariously 
through your adventure on this build. Thank you for documenting it!
3 Likes
More adventures in noise removal.
I had a quiet hiss, that would start after the amp was on for an hour or two, worse in one channel that the other.
Swapped to another pair of cheap 6SL7’s and still there.
Swapped the 2 PSU’s over (I never realized what a useful tool having 2 was) and the louder side moved, so clearly PSU related.
Doing some research it’s not unusual for resistors or diodes in a PSU to “hiss”, and I figure the temperature change in the PSU was just increasing enough to be audible.
I stuck a scope on the 400V line and the Output as a baseline added a 20uF decoupling capacitor to the 400V line, measured again and it was significantly reduced.
Listened and the hiss it’s gone.
There is still a slight background hum ~1mV and barely audible, which is going to be from the AC heater on either the 2A3 or 6SL7, that’s probably in the ballpark or better than most amps with AC heaters manage.
I designed a regulated DC supply PCB that with different transformers should work for both heaters, I’m tempted just to use my bench supply to test which one it is, but both have substantial DC lifts, and I don’t know if that’s safe, nor can I come up with an experiment that will convince me it is.
7 Likes
Interesting that one psu is humming and the other is not, but thats the wonder of manufacturing differences i suppose.
1 Like
It’s not hum, it’s hiss.
Both did it to an extent, just one more than the other.
One of them runs a degree or 2 hotter than the other as well.
Hiss is apparently indicative of resistors and diodes, Hum is more about power and grounding.
I had no capacitance after the rectifier on the 400V line, and I suspect any would have resolved it.
It’s interesting because all of this is just physical implementation rather than any real change to the circuit, because as we know wires are not just wires, resistors are not just resistors diodes are not just diodes…
It’s interesting how the noise changes with load it’s dead silent with
The Susvara, that’s not very surprising it’s very inefficient.
I think I can get it there with the HD600’s which is a load way outside what it was designed to drive which results in noise if you turn the volume up very high (not that you could listen at that level).
AKG’s are about the same noise level as the HD600’s better load for the amp, but very sensitive.
D8KP about the same.
Utopia Slightly less, but barely audible, and something else that I think is probably RF.
Despite being less sensitive with the MySphere’s plugged in it picks up a lot more noise including a radio station when the volume is very low (something acting as an antenna).
5 Likes
A little bit like back pressure on an engine. And what do you know, some exhausts hiss too. Now you need to find some iems to try.
Cant wait for your next update.
3 Likes
Designed these for the DC filament supplies on Friday.
Basically it’s an LPS board, which will cover pretty much anything to 20V with the right transformer, and a resistor change.
Arrived from China today total of 4 days, I’d considered DIYing the boards, but it would have cost more and taken longer.
I ordered the parts to put on the board from Digikey, the same day, and those won’t arrive until next week.
Certainly something I’ll be doing for my next project.
7 Likes
In the what I learned this week bucket.
I wanted to post this because it’s a good example how partial understanding (and we all partially understand everything, even the stuff we’re considered experts in) can lead to poor decisions.
My amp currently has a 100K Volume Pot in the input.
I was aware that 100K Pots are commonly used, but some designers opt for 50K, 20K or even 10K pots. I was aware that Pots were a large factor in determining the input impedance of an amp, for tube amps with the Pot on the input, the Pot basically is the input impedance.
And higher input impedance is always better, so why would you ever use anything smaller than a 100K Pot or for that matter why not use a 1Meg Pot.
The answer apparently is the “Miller Effect”, all electrical components are more than they appear to be resistors have capacitance, and capacitors have resistance. Valves and Transistors have parasitic capacitance which is not trivial.
The capacitance of the input tube on my amp will be somewhere between 200 and 600pF, probably nearer the 300 than 600. And that 100K pot means that I have a circuit in effect like this on the input
With R1 and R2 constituting the the Pot
The circuit containing R1 and C1 is a low pass filter, C1 is the afore mentioned 300pf, and R1 will vary from 100K Ohms at 0 volume to 0 Ohms at 100% volume.
That will move the cutoff frequency of the filter.
Using the 300pF number, at low volume say 1KOhm’s the 3dB cutoff frequency is 530KHz, which is well outside the range of human hearing, but at full volume 100KOhms it’s 1/100th of that so 5.3Hz which is very audible.
In practice you don’t run amps at full volume, and say 50% on the volume dial the cutoff is probably still in the 50KHz+ range
With a 20K Pot at full volume that cutoff is still 26KHz.
But there is something else here, I mentioned in my previous post that I was picking up radio channels with the MySphere connected at low volume, and there’s a good chance that because without some resistance in the circuit leading into the tube, that 530KHz cut off in well into the AM radio range.
It means I’ll be swapping out the Pot and adding another resistor to the input.
This whole thing is a learning exercise for me, and I thought this one was particularly interesting, because it demonstrates something I tell people all the time, things are always more complex than you think they are, and those complexities can make what you believe to be obvious, incorrect.
9 Likes
Some fiddling with the amp this weekend, I added some 4K7 grid stopper resistors to the input stage.
Cleaned up the rest of the Ground wiring, directly grounding everything to the central bus bar.
The combination of those seems to have removed pretty much all the noise.
I’ll have the parts for the DC filament supplies this week, and that’s the last major thing I want to cleanup before building a single power supply for it.
The gain is too high with the 6SL7’s, it’ll clip at 2V Peak to Peak in.
The first thing I’ll try is removing the cathode bypass on the input stage, this should drop the gain a bit, and decrease the overall distortion somewhat.
I was also looking at other possible input tubes, 6SN7 is the obvious alternative and my assumption was I’d have to change a bunch of parts to get any sort of optimal load line. But after doing some simulations, the current configuration was as good or better with a 6SN7 than my attempt at an optimal configuration. I think the push pull nature of the SRPP cancels out all the predominantly 2nd harmonic “noise” you get because the current is too low.
4 Likes
Is this from DIY or more like your own design?
2 Likes