Tubelab SE / Simple45 build: Why 5842/417A? Why not 6J5?

[geomorphdog]

Note: I recently started a thread with the same name on Audiokarma.org.

I'm preparing for a Tubelab SE (or SE II) build with 45 tubes. I have a pair of NOS Tamura F-475 OPTs and couple of hunkin' PAECO PT's, though one of those should be plenty. I was getting ready to pull the trigger on a tube order from Brent Jessee, but his stock of 5842/417A tubes seems to have evaporated. I said to myself, "Well, self, we might as well get into some calculations loadlines while we ponder where to get some sweet driver tubes!" I annotated the SE II schematic and plotted up a few loadlines for the 45 tubes. Of course, the loadline for the 5842 is dead simple because the design uses a constant current source (CCS) on the plate, so the loadline is just a horizontal line. I found that the CCS is spec'ed for 12 mA, and the quiescent plate voltage is typically 175 V. Wait, what? Sure, the 5842 can take a max plate voltage of 200 V, but none of the characteristic graphs include lines for anything over 150 V. Moreover, the spec plate current is 26 mA at 150 V. That's an operating point that appears to give the tube some room to breathe, whereas 12 mA at 175 V is crammed down into the lower-right part of the plate characteristics.

Then, I look into necessary signal swings for the 45: 36 Vrms on the plate with mu=3.45 means 10.4 Vrms on the grid, or a swing of +/- 14.8 V. Even with the operating point of the 5842 crammed into the lower right, that should be fine and would require, wait for it, 0.243 Vrms on the grid of the 5842. Quarter-volt sensitivity? Jeez, seems kind of ridiculous, but OK.

Still, might another, more readily available tube do the trick? The Powerdrive stage in the design (i.e., MOSFET source follower between driver and output stages) seems to open up a lot of options. I look into the 6J5, which is the single-triode equivalent of the 6SN7. It has less than half the gain (mu=20 vs. the 5842's mu=43), but all that gain of the 5842 seems a bit over the top. Plus, there seem to be issues with the 5842: It's microphonic, and according to one post, it threatens to oscillate while sitting in the cardboard box.

So why fuss with the 5842? Why not the 6J5? It has low distortion, and mainly 2nd harmonic. With less current and substantially higher, but still reasonable, plate voltage, the operating point can be smack in the middle of the plate characteristics graph. Thoughts?

A summary of some replies to this post on Audiokarma.org:

• Some recommend ditching the Tubelab design or aspects thereof.
• I could probably use 6J5's, and it might be fine.
  • The 6J5 is pretty awesome.
  • But the lower gain of the 6J5 would make the amp less dynamic-sounding.
• Turns out the 5842 works really well with the Tubelab design.
  • Yeah, the operating point is a bit weird, but don't worry about it.
  • If you find you don't like the original operating point, the design has a lot of adjustability built in, and you can change the operating point in conjunction with testing.
  • 5842 tubes are available at reasonable prices. (Subtext: You don't have to go with the WE's or Amperex gold pins.)
• The PCB schematic is annoying to some curmudgeons, but I'll have to live with it. Or make my own. Whatever. And get off my lawn!
• I should probably put my question to a couple of other forums, including DIY-audio, where I'd likely get a response from the designer, George.

Per the last point, that's what I'm doing! Thanks in advance!

 

[Tubelab_com]

The original TSE was designed over 20 years ago. At that time 5842's could be found for under $5 each. During the initial design I did real world testing on lots of different tubes that I had or could get a decent sample size of. The design was always intended for the 45 output tube, but I also tested the 2A3 and 300B and wanted the design to be flexible enough to use any of them. After weeding out lots of tubes for the usual reasons some serious bench testing and listening test left me with the CCS loaded 5842 as the best choice. Any of the tubes with a Mu of less than 30 did not have enough gain to drive a 45 tube into clipping with a cell phone as the source. Yeah, people still do this. and even the 5842 falls short with a 300B output tube. If you have a preamp or other source with enough output and will not use a 300B tube, you might get away with a low gain input tube, but I doubt it.

I have not seen Microphonic issues with my Raytheon 5842 tubes. Yes, the tube will oscillate if it is used in a layout with long or poorly routed wiring. Any tube with a high Gm will oscillate if the conditions are right.

Several hundred TSE's and TSE-II's have been built in 20 years. The user response has been great, and the most common complaint was regarding not enough overall gain especially with 300B tubes.

The TSE became the TSE-II back in 2019 after the Sharp filament regulator chip went extinct. At that time I asked the existing user base what they would do differently for the TSE-II. If you are making your own amp, you might want to read through that thread;

https://www.diyaudio.com/community/threads/after-a-14-year-run-the-tse-must-die.331038/

During the testing done long ago the 6J4 tube was the number two choice. Most of the tubes I tested were the mil-spec version, the 8532 / 6J4WA.

I am looking at a driver circuit for an even more demanding amp design than the TSE. This one uses a TV sweep tube, or a paralleled pair for 15 to 35 watts of SE power. The output stage uses UNSET technology to make triode curves on a TV sweep tube. The driver stage uses a pentode in UNSET combined with a circuit I designed for squeezing maximum gain (over 1000 V/V) from a pentode in a guitar amp application to generate large drive voltages at near zero distortion. See posts #110 to 119 here:

https://www.diyaudio.com/community/...he-way-i-always-wanted-to-do-it.410333/page-6

 

[geomorphdog]

The original TSE was designed over 20 years ago. At that time 5842's could be found for under $5 each.

Sob! But perhaps the fact that the WE versions of this tube are now going for $100-200 each is some testament to the 5842's utility in your design.

After weeding out lots of tubes for the usual reasons some serious bench testing and listening test left me with the CCS loaded 5842 as the best choice.

It's hard to argue with actual bench and listening tests, and I'm happy to benefit, as so many apparently have already, from your hard work.

Any of the tubes with a Mu of less than 30 did not have enough gain to drive a 45 tube into clipping with a cell phone as the source. Yeah, people still do this.

I get it. I could be a purist and insist that a more proper sensitivity is ~2 Vrms, but all else being equal, it's nice to be able to use just about any preamp or receiver's or integrated amp's pre-outs to drive the power amp. Without the need for GNFB, then as long as the driver makes beautiful music, more gain is more better!

The TSE became the TSE-II back in 2019 after the Sharp filament regulator chip went extinct. At that time I asked the existing user base what they would do differently for the TSE-II. If you are making your own amp, you might want to read through that thread

I've read some of it and will give it a complete and careful read before building. May I ask, though, does the thread suggest any changes from the TSE-II BOM you've posted? (attached here for reference is TSE-II_V1.2_BOM_7-26-19.pdf) Last I checked, all of the solid state components listed there are still available.

During the testing done long ago the 6J4 tube was the number two choice.

Can you comment on what made the 5842 the ultimate winner? I see that the 6J5 has even higher gain, but maybe the CCS cuts a better loadline through the characteristic curves of the 5842.

I am looking at a driver circuit for an even more demanding amp design than the TSE. This one uses a TV sweep tube, or a paralleled pair for 15 to 35 watts of SE power. The output stage uses UNSET technology to make triode curves on a TV sweep tube. The driver stage uses a pentode in UNSET combined with a circuit I designed for squeezing maximum gain (over 1000 V/V) from a pentode in a guitar amp application to generate large drive voltages at near zero distortion.

That sounds wicked cool. I have a smidgen of experience with TV sweep tubes; I got a Sansui 1000 last summer. It uses 25E5's. They're pretty cheap when you can find them!

I greatly appreciate your response to my post. I will proceed with Plan A, which is to follow your design and use the 5842. I have an acquaintance who is an SET purist sort of guy, and he pointed me to your Simple45 pages a few years ago. I recently received, free of charge, a pair of NOS Tamura F-475 OPTs from a local "collective" of tube enthusiasts. These OPTs were originally purchased new in 1998 but were not used before the original owner passed away and left the OPTs to the collective. I feel a certain burden, or at least obligation, to do these OPTs justice and build the best possible amp with them. I've been building and modding tube amps just long enough to recognize and appreciate the utility of your Powerdrive design. It's an elegant solution and reminiscent of Alexander's cutting of the Gordian Knot. Make the best driver stage. Make the best output stage. Problem coupling the two? No problem, because we have MOSFETs!

If you're at all interested and do the YouTube thing, I've made a video about your design:

Thanks again,
Stephen Lancaster

 

[OldHector]

There is another tube that appears to be part of the same family as 5842/417A and that is 5847/404A. It is a pentode but is high gain with nice triode curves. They are still plentiful and cheap - is it a like a 5842 when triode wired?
The image comes from VT52.com, but I cannot navigate it on my PC.

The curves for the 5842 are not quite so steep, but if you ignore the half volt curves and match the axis, aren't so far away.

 

[gupsta]

I picked 8 NOS 5847's with the intention of using them as a cheap substute. There was quite a spread in the test results on my curve tracer. Not had the need to fit them yet, but the pin connections are somewhat different so not any easy swap being pcb mount.

 

[sunil]

What about the 6E5P in triode, with a converter?

Would it be enough to drive a 2A3 or 300b ?