6386 need any info or link.

[dale116dot7]

6BA6, G2 and G3 wired to the plate, two tubes for each spot a 6386 would have gone. Nice and simple. If you need two tubes, get ten and you can match two up pretty good. If you need eight tubes (Fairchild 660), get 20 tubes. You will need to match the tubes very carefully.

 

[rafafredd]

Did you matched for diferent bias points, or isn´t it really necessary?

 

[dale116dot7]

Yes, you have to do that otherwise it will thump and oscillate when you compress hard. It takes many hours to match the tubes so they don't do that.

 

[CJ]

Put a sawtooth on the grid.

 

[Henrik Vogel]

Sorry to start posting here by digging up an old thread.

I understand using two 6BA6's or 6BC8's for each place there should have been a 6386 requires that the tube pair is closely matched. But do all 16 tubes (if building a 670) need to be matched, or is it enough that the eight pairs are matched? Matching eight pairs sounds tedious enough, matching 16 tubes sounds...time consuming. :grin:

Best,
Henrik Vogel

 

[gyraf]

Hi Henrik,

Welcome to GroupDIY..

Yes, you're right - the matching required is for the complete group of tubes, meaning that it will be MUCH easier to obtain decent matching with individual-envelope-tubes than with dual's...

Jakob E.

 

[Henrik Vogel]

Christ almighty. OK, thanks, Jakob!

Cheers,
Henrik

 

[CJ]

time consuming and expensive.
there is a guy hanging out here sometimes who makes a commercial 670 only with a 6ba6.

pm him if you need some in depth info.

also somnecurve tracing over at PSW. in Klett's forum I believe.

cj

 

[Larrchild]

Here's some comparisons of 6386's and 6BA6's from my curve tracer.
They are similar curves. Chek it out: http://recforums.prosoundweb.com/index.php/t/6028/9858/

 

[Henrik Vogel]

Thanks for this info, people.

Larrchild,
Do you actually match 16 tubes for each unit you make?

Thanks,
Henrik

 

[Larrchild]

Yes, in pairs. It makes a stereo unit possible.

They not only have to track in push pull, but left ch. to right ch.
Either one would be plenty to worry about.

And this is a zero-feedback amp..so the push pull nulling is the only way one can get away with this. Cancellation of F2 and some free negative feedback that occurs between grid and cathode due to mysterious space charge effects.

Without matching, the THD is all over the place, even when the cathodes are balanced.

 

[Henrik Vogel]

Sorry to sound dim, I just want to make sure I get this as perhaps I'd like to try this myself (with help of my dear uncle who knows a lot more about this than I do): You are using eight pairs of matched tubes, but the pairs aren't necessarily matched to each other? So you don't do what Jakob/Gyraf thinks one should, match all 16 so that they all share the same characteristics? Wouldn't it seem advisable to at least match them in four groups with four tubes in each, allowing you to arrange them like A-A, B-B, C-C, D-D for one channel, and the same arrangement for the other?

Or would one just need too many tubes to choose from in order to make this possible?

Thanks,
Henrik

 

[Larrchild]

You don't have to go to that extreme if you can't. If they are matched as pairs side to side in push pull, even if the gains of the pairs are not the same, the groups making up each side will balance better.

Avoiding image shift in stereo limiting is where rotating matched pairs is the final tweak.. if the fact that at -18 db of gr, the left side is one db hotter, etc bugs ya.
Most can live with much less.

If you can get the first 15 db matched, thats pretty good in the real world.

 

[Rob Flinn]

Cancellation of F2 and some free negative feedback that occurs between grid and cathode due to mysterious space charge effects.

I think you also get neg feedback due to there being no cathode decoupling caps.

 

[Larrchild]

there is a big honkin 4ufd NP across the cathodes tho.
Must affect degeneration.

 

[Henrik Vogel]

When one matches tubes, is it only conductance you look for or are there other characteristics you need to have matched?

 

[gyraf]

What I ment was: both groups of eight tubes (actually four groups of four) should be selected so that they ON AVERAGE has the same characteristics. This makes the needed matching much easier, as you could swap around tubes between groups to even out individual differences..

Jakob E.

 

[Larrchild]

When one matches tubes, is it only conductance you look for or are there other characteristics you need to have matched?

The relationship of conductance to varying bias voltages as provided by a stepped ramp signal in the curver tracer. It quickly paints a characteristic set of lines that one tries to match when matching tubes.

You could get out the graph paper and do it that way. Hardly a mass-production technique, but for one unit, you could do it in a few hours.

What I ment was: both groups of eight tubes (actually four groups of four) should be selected so that they ON AVERAGE has the same characteristics. This makes the needed matching much easier, as you could swap around tubes between groups to even out individual differences..

Jakob is saying better, what I was trying to. Let's say you create pairs of matched tubes with arbitrary gain numbers 1-10. Putting a pair of 6's and a pair of 4's on one side mite equal 2 pairs of 5's on the opposite side. etc.

Also, 3 jacks beats a pair of aces.

 

[Henrik Vogel]

Thanks guys. I didn't realize Fairchild had anything to do with poker. :green:

/H

 

[PRR]

> I think you also get neg feedback due to there being no cathode decoupling caps.

No. Since it runs Class A push-pull, the signal currents cancel in the common cathode resistor. (It may have separate resistors with a capacitor tying them together.)

Also: any form of negative feedback makes gain more constant, and we DON'T want that.