Small Line Amp (Tube)

[NewYorkDave]

I ran some tests today...

With 300V B+, balanced AC heaters, 50K source and 600-ohm load via an Edcor 4:1 XSM10K/600 output transformer:

The THD results are less than stellar, but I'll keep tweaking (or maybe I'll find another problem lurking in my breadboard!). For all I know, it might sound really good but at the moment, I'm focusing on trying to improve the specs.

PS: The THD results with a 12AV7 output were too ghastly to post. I really don't get it. Gotta investigate that further...

 

[NewYorkDave]

Monologue continues:
Here's what i mean when I say that I don't understand why the 12AV7 does so poorly as the ouput tube. In my One-Bottle pre, one half of an AV7 can and has done a max of +18dBM (through 4:1 into 600). Two halves of an AV7, run at about the same plate voltage and current (per side), into a similar load impedance, should do better, right? Like 3dB better. But they're not. The output clips at +18dBM. And I tried a few different samples to rule out the weak tube theory. And I also tried a decade box for the cathode resistor, to tweak the grid bias and plate current. It didn't do much.

The result is so far from what makes sense--and, for what it's worth, far from simulated results with previously very-trustworthy 12AV7 models--I must be missing something that should be obvious to me but has eluded the grasp of my sleep-deprived brain today. Grrr.. Off to bed.

 

[Guest]

Just so you don't get too lonely, I'll write something. I enjoy your monologue by the way. I agree that something must be plaguing your breadboard. The schematic seems fine but the circuit behaves twitchy. Oscillation in the first stage that was solved with shielded wire, 12AV7 parallel not performing well. Maybe something in the layout has gremlins. Does 1/2 the output tube perform as you would expect from the one bottle experience?

Don't give up yet.

 

[bovox]

Yes please hang in there... I´m enjoying you talking!
If you look at another design the R..e 4..7 have 1k in front of both
grids in paralell of the E88CC. And the Pultec one has also separated the gids with 1k.But I imagine you already looked for that.
Anyway ,cheers Bo

 

[analag]

My simulation of your circuit shows much improvement if the feedback cap is placed between the output cap and the transformer. Also the 470uF cap can be omitted, its not improving anything really.

 

[NewYorkDave]

Taking the feedback after the output coupling cap means greater phase shift and frequency selectivity in the feedback circuit, which may or may not be a problem depending on the output transformer. And bypassing the cathode of V2 increases the open-loop gain by about 6dB. I knew both these things in theory--which is why I accounted for 'em in the circuit--but after seeing your message, I went to the bench and confirmed both experimentally just to be sure.

My trevails with this circuit serve as a reminder that SPICE can sometimes mislead you. Using 12AV7 models that have proven very accurate in other circuits, the simulation said that this circuit would do +22dBM before crapping out. On the bench, I discovered that it can, but not with a 12AV7 as the output tube! The old-school (non-SPICE) calculations I did intially turned out to be closer to the truth--but SPICE was less tedious than dealing with the loadlines in-depth.. and it told me something closer to what I wanted to hear, so I listened :roll:.

But all is not lost. It actually does reasonably well with a 12BH7, especially if I increase the negative feedback by 6dB by strapping another 100K across the 100K feedback resistor that's already there. Max output has gone down by 0.5dB due to the shunting effect of the (now-reduced) feedback resistor. But the near-halving of THD across the normal operating range seems worth the slight loss at the max end.

12AU7 does just a little bit worse than BH7 at +4 and +14 but craps out about 2dB sooner. And AV7 is pretty much the same as AU7 at +4 and +14 but runs out of steam at a mere +18.5dBM output. So either the AU7 or the AV7 could be used as an emergency sub, but BH7 is definitely the output tube to use in this circuit. A 12AV7 definitely works great for V1, however.

As I mentioned earlier, V2 plate and cathode resistors have been changed to 12K and 270 ohms, respectively. I'm thinking it'd be cool to add a switch to change between 100K and 50K (adding another 100K in parallel) to give "high gain" and "low gain" modes for the output stage. The cathode bypass cap could be switched in and out to do the same for the input stage, although I have yet to check if there'll be any problems there with increased hum from an unbypassed cathode. (I'm using AC heaters).

Well, I have paying work that demands my attention so I'm gonna have to put this on the back burner for quite a while.

PS: Bo, you had a good point about grid stoppers on V2. Adding 1K there definitely helped to get rid of some "squeeging" I noticed during testing, when running the output stage open-loop with signal input directly to its grid.

 

[CJ]

Cool pwr supply had the same one in hi skool no never used the c minus that is all for now.

 

[analag]

AC heater....hmmm, haven't tried that one in awhile, that throws a different light on the matter of the cathode bypass cap. Overall excellent circuit analysis from a fellow tube maniac.

What are your thoughts on the 5687WA as a buffer.
analag

 

[NewYorkDave]

I'm not familiar with the 5687WA.

I couldn't resist messing with the breadboard a little during lunchtime today. I plugged 12AV7s in both spots and fired it up; I wanted to make sure I could reproduce the results I got yesterday. But I discovered something else... Holy crap, there are definitely some gremlins living in this thing! I could make a substantial change in the level/content of the distortion product by touching various components or leads, moving wires around, etc.

Although max output level remains unchanged, I can reduce THD at lower levels by a pretty significant amount just by putzing around with the routing of the leads, the orientation of a component and so on. Don't let anybody tell you layout is "non-critical" in an audio circuit.

But now it's time to put away the toys and get back to work...

 

[Michael A]

Do you think 100 ohm or 1k plate stopper resistors hard on the plate pins would do anything for parasitics etc. I seem to recall reading high gm tubes in parallel can use these.

Michael

 

[analag]

The 5687WA is an excellent tube both as a voltage amp and especially as a CF. ( a super 6DJ8) It's got the biggest cathodes in any 9 pin tube I've seen so far. These tubes are very popular in the audiophile community and very cheap but that's how "sleeper" tubes are.

The layout induced THD is very interesting, noise and oscillation problems I have always considered, but THD is new....hmmm. Another great obsvervation, Dave. Tweaking up performance through optimized layout technique.

analag

 

[NewYorkDave]

A friend recently hooked me up with real-time FFT and I'll tell ya, it's amazing. I used to measure the harmonics using a wave analyzer (very old school!) but being able to see, instantly, the harmonic levels and distribution change as you make adjustments is opening up a whole new world. I know this stuff is old news for some of you, but it's still a new toy to me and I haven't gotten over the novelty of it.

Hmmm... I was thinking tonight that it might be cool to try 6SN7 (or its Mini-Me, the 6CG7) as the output tube. I wanted to stay away from tubes audiophiles fiend after ('cause of the $$ factor, natch); but the SN7 has so many good qualities that it's hard to resist seeing how it'll do in this circuit. So many experiments to do, so little free time...

 

[Larrchild]

I'm suprised you are not seeing rf bursts on the scope cause it sure sounds like it's on the edge of the runway.

1.3pf input and .6pf output capacitance put this 12AY7 in the realm of an RF tube. Since you see THD moving around, when shifting components, I think that is just a symptom of something of higher altitude present.

If plate current moves around with this component shifting also, thats a sure bet. Snubbers and ferrite worry-beads on leads as they go to the tube is a plan.

You have to tell it it's not a radio circuit, Dave..It's just doing it's job! =)

 

[NewYorkDave]

Actually, it's 12AV7: 1.9pF grid-plate, 3.2pF grid-h/k, 0.5pF output (plate-h/k). Description: "The 12AV7 is a miniature, medium-mu twin triode for use as a radio-frequency amplifier or as a combined oscillator and mixer in VHF television receivers. It may also be used as an audio-frequency amplifier."
:wink:

I did see some RF bursts before I added a grid stopper to the output tube. I don't have any ferrite beads here, but I'll try stoppers on the preamp grids as well.

 

[Larrchild]

Actually, it's 12AV7:

Doh!
This is where beer and retention diverge, oops! Sorry.

But yeah, plenty of gain at non-audio freq's eh?
Do the tighten-up. :grin: Clip it's wings.

That WE 407 circuit I built was most uncooperative without 220 ohm grid stoppers in the 1st stage grids. It would take off at about 20-30 mhz. I should have connected it to my dipole in the yard and called "CQ".
Use carbon comp resistors for the grid-stopper.

 

[NewYorkDave]

An old mentor of mine called that "dumbing down"... That is, deliberately restricting the bandwidth of a circuit or component.

 

[alk509]

[quote author="Larrchild"]Use carbon comp resistors for the grid-stopper.[/quote]

Ooo! May I ask why?

Peace,
Al.

 

[Larrchild]

Cause the carbon innards are not actually acting as a resistor as much as a broadband RF choke. Other resistor materials will either be inductive or less effective as a choke.

 

[NewYorkDave]

I know that changing two variables at once is a no-no in a proper experiment. But I couldn't resist starting lunch a little early today wink and tacking on a pair of octal sockets to try a 6SL7/6SN7 version of the preamp. The result: sh*t, it specs out great! THD+N: 0.1% at +4DBM, 0.3% at +14dBM, 1% at +20dBM. IMD+N: 0.3% at +14dBM. Noise: -98dB. It clips at around+21dBM but distortion is low up to that point, with "textbook" distribution of harmonics (i.e., progressively smaller amounts from 2F to 8F, with nothing above the noise floor from 8F upward).

The only bummer is the higher input C of the 6SL7. With a 50K source impedance, response is 2dB down at 20kHz. I confirmed that the input C of V1 was to blame by removing the build-out resistor on the signal source, at which point the response went to +1,-0dB from 20Hz to 20kHz (with the 1dB bump at 20Hz being due to the output transformer).

Before things start getting too busy this afternoon, I'd like to un-tack one of the octal sockets and try 12AV7 as V1 and 6SN7 as V2. That'll give me an idea of how 6CG7/6FQ7 would work as V2 without me having to rewire one of the nine-pin sockets for the slightly different pinout.

(These experiments are a lot of fun. I wish I could get paid to do this).

 

[ciminosound]

That opens up a whole other can of worms. Could you then sub the 5755/WE420 for the 6SL7 as some folks do? Good SN7s are getting expensive. A 5755/6CG(FQ)7 version? Keeps the tubes in the miniature range instead of octals. 5755 has

I just happen to have a bunch of 5755's from a tube lot I bought. Seems some people like these computer tubes for audio if used properly. They never seem to be direct subs based on the 12AV7/5965 comparisions.

http://tdsl.duncanamps.com/pdf/vm342.pdf

(Of course with all tube subs your milage may vary.)

(These experiments are a lot of fun. I wish I could get paid to do this).

I know our appreciation doesn't pay the rent but... :sam: