For DHT single-ended (SE) topology, I have to admit that I reached to the conclusion that in my opinion either 6C4C or 4P1L are the way forward in terms of sound and cost after not being happy with the option of running the 45 in A2 mode. Both 4P1L and 6C4C sound lovely in SE despite many will say the 300B is unbeatable. Yes, won’t say a ridiculously thing such as 6C4 or 4P1L are the “best DHTs”. We all know that there are many great DHTs out there, but at a cost. Well, if cost is not a problem for you, you can chose great NOS valves from PX4, 50, 300B to 813 or 845. The latter comes with a hidden price: the power supply. I’ve been there as I’m building the 4-65a SE and most of the budget is used in the power supplies. Sound-wise, we did a side-by-side listening tests on many SE and PSE amplifiers and couldn’t find a significant different between 4P1L PSE and 300B stages. This could easily end up in loosing the tangent and falling into an endless debate about topologies, OT, driver-output stage combinations, bla, bla, bla, but in reality you can’t beat a 4P1L PSE in terms of cost and bias flexibility (i.e. you can easily get 5W from a pair of 4P1L as we will see later). I wish I could achieve the output power I like (i.e. 3W) with a 45. A 45 in push-pull is then very attractive but I haven’t listened (or build) it yet.
I have a very decent stash of both 4P1L and 6C4C, so obviously I will be inclined to get the most out of these ladies rather than continuing burning money on other NOS valves . If you are still reading this is simply because you have (or at least considering buying) 4P1Ls or 6C4Cs and you want to build a good amplifier with them.
So how much power can you get out of the 4P1L? Anatoliy did his own tests and was very pleased with the results in terms of sound. I haven’t run the 4P1L in A2 yet but here it would look like in A2:
You can get clean 5W from a pair of 4P1L running at 50mA (each) and biased at 200V. The driving requirements are only 50Vpp and we can see in the diagram above that the positive grid excursion is to just over 10-12V. Obviously the right driver needs to be used to provide the necessary grid current in A2 and also withstand the changes in grid impedance when transitioning from A1 (high impedance) into A2 (low impedance) with minimum distortion.
I don’t currently have an 2K5Ω OT gapped at 100mA, so won’t be looking at running a 4P1L PSE in A2 like this.
Instead, I have a pair of LL1623/60mA which can provide a varied set of transformation ratios: 5K6Ω, 3KΩ and 1K6Ω.
With this OT I could then easily get 2-3W out from a 6C4C or 4P1L PSE as we shall see looking at the loadlines.
6C4C in SE with LL1623
The 6C4C is preferred to be run at 60mA and 250V. I haven’t had the time to hook up the curve tracer and get the curves for the 6C4C yet, but here is my first attempt to model this triode:
The model is close enough for the purpose of my tests, so let’s look now to the performance of 6C4C with a 3KΩ load:
As we would expect, the distortion is not great when pushing the valve to deliver 2W. I will ignore as part of this analysis the potential improvement of the THD due to distortion cancellation with the driver, that is something you can play with in SPICE and make your own conclusions of the optimal topology and design. For an Ra of 850Ω, we have 3.5x load impedance with the 3KΩ load in this case, so hence the resulting THD. However, the driving requirements are only 66Vpp.
Now looking at the same quiescent point but with a load of 5K6Ω:
As expected distortion is lower but driving is higher (i.e. about 81Vpp).
How much power we could get out of this valve in A2? Let’s have a look:
We could get an extra Watt by running it in A2 and with the penalty of distortion to reach up to nearly 2%. The bias point is shifted left a bit to 220V/60mA. The driving now is increased to about 100Vpp.
Enough for now, I better go for breakfast! I will look at the 4P1L PSE curves in the next post…
Ale
Ale, very interesting information for thought! all very well for 4П1Л. Great perspective!
question: 1) what is the maximum current of the grid PSE 4П1Л (point of PSE 4П1Л Vp = 200V, Ip = 99.9ma, Vg =-13.7V) in class A2 at maximum power output? 2) which the anode current, driver’s stage, you recommend for a reliable work PSE 4П1Л in A2? thanks
Vyacheslav, I haven’t modeled this in SPICE yet as don’t have an accurate A2 grid current model. Dmitry is working on it. But I would suspect is something around 5-10mA at least.
For driving in A2, I’d go for a DC coupled mosfet gyrator as in my 4-65A design (ala Michael Koster). You need a driver that can not only provide the A2 current but also has minimum distortion when impedance changes from high (A1 operation) to low (A2 operation)….
Hi Ale,
Great perspective with 4p1l. I have a dozen and just started to use my new tube tester on them. I am excited to try. I am working on a push pull using LL1663, and I’m thinking of parallel 4p1l to get to the right load match. Did you try different filament power scenarios? Can you tell if there is difference in sound between filament bias and fixed? Is it better to combine ccs to power the filament in the bias loop? I would prefer to power filaments at 4.2v to lower the current needs.
On the other hand I am still thinking to parallel 3 or 4 if them in a SE amp, to drive a lower z transformer. Did you try simulation for more than 2 in parallel?
Sorry to ask so many questions? I am at the beginning of my journey and I trying to grasp so much.
I don’t have yet all the means to test my amps but I will post sinks impressions on the PP.
Best,
Radu
PS. I do not intend to play in A2 yet. I am happy with A1.
Hi Radu,
I always run the 4P1L with filaments in parallel: 2.1V/650mA. Better and more linear. I have tried filament bias so far but was planning to build up an SE stage with fixed bias.
I haven’t tried more than 2 valves am afraid.
Cheers!
ale
Hi Ale,
I think you are forcing too much the quiescent point/load combination. As you can see you are entering the A2 region where the curves are not very linear and quite close one to the other. In terms of current requirement from the driver I think it’s a lot. As a rule of thumb the grid current in your example is going to increase quickly. Actually you might be at the point where it start to increase exponentially, even if the positive grid swing is limited. In your case you are not using some region towards the cut-off where the valve is more linear than positive grid side!
I would try something like 220/45mA per valve at around -17.5V bias. In such case with 55V peak-to-peak you should get even more than 5W, probably 6W max Pout at 4-5% THD, stopping at +10V on one side and -45 one the other using most of the headroom. When you reach the limit it is natural that THD cannot be too low. You can hope for this only with some thoriated tungsten filament valves that are very linear even down to very low current (near cut-off).
Cheers,
45
P.S.
I am referring to the 200V/100mA with 2.5K load example.
Great feedback, thanks. A proper mu-follower for class A2 driving will be definitely required but wasn’t looking at building this setup anyway. Less demand to the 4P1L as you suggest will probably be better. I want to build a test rig but my OT is 60mA and not planning to buy another OT gapped at higher current for the time being
Cheers
Ale
I have pair of unused Lundahl LL1627 90mA in Italy. I could bring them next time I go back in June.
Cheers
That would be awesome! Thanks
Hi,
just got my hands on a quadruple of 4P1L, and a pair of SE-OPT with 5k impedance and 70ma DC allowance. I plan just to have negative grid to start with, A2 may come later.
The software to plot the loadlines, can it be had some place? familiar with paint-kit, birds of a feather i d say 😉
i could keep you posted on my progress if you are interested, i will be opening a thread at diyAudio shortly.