remote cut-off tube test

Following the great tips from PRR and others I decided to check out some remote cut-off tubes and share the results.
test conditions: Used a copy of the Fairchild  240V regulated supply with 10k on the plates and 100 ohms on the cathodes and all the pentodes wired as triodes.  Sig Gen 600 ohm source with 500mV on Vg1

The results show how linear the 6BA6 is, after the kink, compared to the other tubes that have been mentioned in the threads.

This is my first try at posting a link....hope it works  http://i39.tinypic.com/15zpg5w.jpg

I'd be interested to get some feedback even if its negative!
best
Dave P

Interesting.

It would have been interesting to have a 6386 on the graph for comparison, since this sems to be the benchmark valve, at least for audio compression use.

I'll never get one unfortunately Robb, I enquired about the new JJ 6386 but was quoted £90 each plus tax...too rich for me to especially needing 4 plus extras for matching.  That's why I started testing the others, I have heard the 6BA6 is practically identical to the 6386 as I guess you already know.
best
Dave P

Interesting indeed.

How did you do these graphs?

Jakob E

> with 10k on the plates

10K AC or AC and DC? Transformer (like Fairchild) or resistor?

Tube current changes a LOT. Plate voltage rises at low current, high GR. In the transformer-loaded Fairchild, plate DC voltage may rise from 230V to 240V. With a simple 10K resistor, these tubes may sit at 100V idle and 230V at high GR. Either way can work, but they are very different.

> source with 500mV on Vg1

Yeah, well, in a limiter, the essential detail is that the output level is (nearly) constant, NOT the input level. IMHO, you laboriously measured the wrong thing.

I suspect the sharp drop on your 6BZ6 is clipping, not clean gain reduction.

Distortion matters too.

We almost-always have to work a "good" limiter in Push-Pull; the undistorted signal level just gets too weak in high GR. Mega-breadboard.

A quick/cheap hack is to work single-ended and watch the output "sine". Since we will ultimately go push-pull, the wave can get quite ugly, as long as "half the wave is good", actually so the top and bottom together come out sine-like.

Start by picking the maximum GR you could ever need. The Fairchild expects well-controlled sources and actually does not try for more than 12dB GR very-clean. A pop-vocal slammer might "need" 30dB without gross farting, though 5% THD above 20dB GR may be acceptable.

Set the CV to give that GR, and determine the maximum clean output. (This will also tell you how much gain you need after the GR stage to reach your Studio Level.)

Now reduce the input, and reduce the CV to keep output nearly steady. (Most real limiters have some "rise", but it may be valid and more useful to plot "infinite" limiting.)

When GR gets to zero dB, check your input level. Compare it to your noise level (probably around a microVolt at the grid). That's your maximum S/N ratio when not-quite limiting. In non-insane uses, that's the S/N you hear most of the time. (If you limit insanely, limiter S/N is probably a non-issue, but source noise will be obtrusive any time the racket stops.)

Hi PRR & Jakob,
To answer your questions,
It was a 10k resistor on the plate, I'm still working up the circuit at this stage and haven't got the TX yet.
You could be right about the distortion, I'll check the output on a scope but I would be surprised as the Vg1 voltage is way below 500mV, I used a meter to measure the gain at 1KHz then divided by input volts to record the gain, I did a spreadsheet in excel and when I made the gain a log scale the 6BA6 curve came straight.

I repeated the test with 10 6BA6 tubes to check the spread, and I'll post the chart shortly.  Its a bit of a pain to do because Tinypics can't accept PDF files and excel can't save as jpegs so I have to print the chart first then scan it in and save as jpeg, then the quality takes a dive.

It will take me longer to digest all of your thoughts PRR, but I'll come back later.
best
Dave P

The differences on the chart are somewhat relevant as a curve tracer would derive them the same way, through a plate resistor, but as mentioned, a more constant-voltage environment would be further relevant to the sound of the 670. Very different sound when only one parameter is changing.

dave P said:

I have to print the chart first then scan it in and save as jpeg

Click to expand...

That sounds like a bit of a pain!
Have you tried Print Screen instead?
With your spreadsheet open you press PrtScr button on your keyboard, then you open your favourite graphics editor (MS Paint®) and press ctrl+V...

Hi PRR,
I've read your post now and can see I need to be clearer about what I did.
I will only be using these tubes in pp with a transformer but its easier to simply test the tube characteristics with a plate resistor to make basic comparisons between single tubes.

The reason I tested them like that was to find a tube which could still work down at -70V, this would be required for a 670 clone. This is the chart of ten  6BA6 tubes  measuring plate current against -Vg1 with constant 240v B+, 10k plate resistor, 100 ohms on cathode.  http://i40.tinypic.com/2cqbnfs.jpg
There was no ac input voltage.
Tubes 1-8 were Mullard and 9 & 10 were Brimar, the thick black line is the mean of all 10 tubes and is remarkably straight, which shows why its good to parallel them to get good balance.

I take your point PRR, about the gain, because as you say, the output stays constant in GR while the input rises, but surely the fact that the gain falls as the CV goes negative, implies that the input can rise to maintain the constant output you are describing, thats how I understand it anyway, am I right in thinking that?

I'll bear in mind your other tips when I get further down the line with my testing.

Jackies, thanks for the tip about screen printing, I'll try it next time.
best
Dave p

jackies said:

dave P said:

I have to print the chart first then scan it in and save as jpeg

Click to expand...

That sounds like a bit of a pain!
Have you tried Print Screen instead?
With your spreadsheet open you press PrtScr button on your keyboard, then you open your favourite graphics editor (MS Paint®) and press ctrl+V...

Click to expand...

Yep.  PrtScrn alone copies your whole desktop to the Windows clipboard.  Also, holding down the Alt key and press PrtScrn will copy just the image of your current window!  Saves a bit of time cropping.  Can you tell I've spent way too much of my life writing IT documentation? 

Yep.  PrtScrn alone copies your whole desktop to the Windows clipboard.  Also, holding down the Alt key and press PrtScrn will copy just the image of your current window!  Saves a bit of time cropping.  Can you tell I've spent way too much of my life writing IT documentation? 

Well done Stickjam, all that I.T. work has paid off! it worked a treat and has saved me a load of hassle.
thanks to all on this tip.
best
Dave p

stickjam said:

jackies said:

dave P said:

I have to print the chart first then scan it in and save as jpeg

Click to expand...

That sounds like a bit of a pain!
Have you tried Print Screen instead?
With your spreadsheet open you press PrtScr button on your keyboard, then you open your favourite graphics editor (MS Paint®) and press ctrl+V...

Click to expand...

Yep.  PrtScrn alone copies your whole desktop to the Windows clipboard.  Also, holding down the Alt key and press PrtScrn will copy just the image of your current window!   Saves a bit of time cropping.  Can you tell I've spent way too much of my life writing IT documentation? 

Click to expand...

Simpler is Gadwin Printscreen... Just select the part of your screen and save it as a jpg or anything else !!!

Read my previous thread with curvetracing:
http://www.groupdiy.com/index.php?topic=26330.0

Hope this helps.

Respect,
Val

Thanks Val,
we seem to be on the same wavelength,
best
Dave P