Hi! - replacing tubes in an old amp/preamp. The EF86 and 12AX7 sockets have metal skirts that restrict the maximum size, and the new tubes are too wide.

I imagine there are adapters/risers to make this possible? Would one of these "socket savers" do the trick?:

Adapters

I think I want a Noval/B9A socket saver, yeah? So the $5 "9 Pin Miniature Socket Saver"s about halfway down the page are not correct, as "miniature" is a different socket type? I can find some Noval/B9A's on eBay, i'm sure.

And: the metal skirt around the tube prevents interference in some sense, yeah? Will my amp/preamps be more vulnerable to noise?

Thanks!
-c

The new tubes are too wide? Are you sure they are the proper tubes?
EF86/12A*7 are ordinary "9-pin miniature" tubes aka "Noval" and "B9A".
"Magnoval" is a different socket and is NOT compatible with EF86/12A*7
Without a photo we don't know what YOU mean by "metal skirts".
If your socket is like the one in the photo below, its purpose is to hold the bayonet-mount tube shield (also shown in the photo).
The metal skirt provides little or no shielding by itself. Its main purpose is to hold the full-length shield.
Without more information about your equipment it is not possible to predict vulnerability.
You could consider cutting off the skirts with a rotary hand tool ("Dremel") and an abrasive cutoff wheel.

Thanks for the reply -

Yeah, the new tubes are just slightly too large. I'm replacing an old EF86 with a new EF86, and an old 12AX7 with a new 12AY7 or 12AV7 (got one of each, both are too large). This is in an AKAI/Roberts mod amp, which is the pre section from a 1960's reel to reel -- see my page at Roberts/Akai Preamp Mod ... I've read about new tubes not fitting into old amps, and figured that might be happening here? E.g. old EF86 is ~26/32", new one is ~27/32". Some of the amps have old EF86's and 12AX7's that I got on eBay, and those all fit fine. It's these new ones that don't. I just glanced in the other amp pair, and the "skirts" look wider - there is some wiggle between them and the old tubes, so maybe these new tubes will fit in that amp.

I see that the "skirt" is as you have guessed, apparently the bottom of a heat shield that I don't have the top for (none of the four amps I've modded came with the heat shields on all 8 of those tube sockets...)

Oh, I'll be darned. So is Wikipedia wrong re: Noval here: Tube socket - Wikipedia, the free encyclopedia ? "This base type was used by many of the United States and most of the European tubes (e.g. 12AX7, EF86, EL84) produced commercially towards the end of the era..." Or maybe "9 pin miniature" is just another reference to Noval?

-c

Those are not "heat" shields. If anything, they make the heat worse. The shields are typically used for electrostatic and EMI/RFI shielding.
They are also used to hold the tubes in place in equipment that is subject to vibration and/or shock.

The Wikipedia page says that "9-pin miniature", "Noval" and "B9A" all refer to the same socket design.

12AY7 and 12AV7 have lower transconductance (gain) than 12AX7. There are not really interchangable in many circuits. Especially those that need high gain.

Ah, so it does, thanks, missed that.

So, given that I don't have the top of the RF shield, I don't suppose the bottom half is doing much for me anyway? So raising the tube up on a socket saver shouldn't be a big deal, I guess.

I just need to find socket savers that aren't themselves too big...

-c

Adding the "socket-saver" doubles the number of contacts (and potential noise).
It also makes a longer "lever arm" which makes it more susceptible to vibrating loose.
It also increases the height which may (or not?) be a problem physically to fit in the space.
I would first consider cutting off the skirts before inserting the "socket-savers".

Very true.

Standard tube shields reflect radiant heat back into the tube making the heat situation worse, diminishing the life of the tube. Good alternatives: (1) remove the shields if they are not needed, or (2) use IERC shields which increase cooling by absorbing some of the heat.

OP, if your new tube doesn't fit the socket, simply buy a NOS tube that does. Or remove the shield base so that the larger tube fits. Adding socket savers is lame because it increases the length of wiring to the tube, thereby increasing the chance for parasitic oscillations and noise. You're also adding another point of potential failure to the circuit.

Socket savers were invented for devices like tube checkers that were constantly subjected to tubes being inserted and removed, greatly increasing the chance for springing the sockets. The idea was to spring the socket saver first and not the socket in the checker.

Hi
It seems some manufacturers decided to use thicker glass or enlarge the valve for other reasons, some EL84 output valves won't fit in some guitar amps where they are fed through the chassis!
Chopping the 'skirt' down or off completely is the easiest 'bodge' and would be preferable to a socket saver jobbie.
The EF86 has it's own built in shield so not likely to need an external shield fitting.
The 12AX7 may or may not have needed one depending on the way it is being used in that specific circuit position.
Matt S

I have noticed that the cheap ceramic sockets coming out of the orient are not the same diameter as the older nos ones.


I have notice this with the tube sockets with 1" mounting centers 3/4" chassis hole.

if you have to replace and it has a 3/4" hole then use one designed for EL84 tubes as their envelopes are the "fatter" ones.

7/8" hole is the standard at 1-1/8" mounting centers for the EIA 9* types. other sockets are compact variations that will or will not work with the tube depending on the there deviation to the standard.

and the compact variation is mass produced in the orient and they don't work well with nos tubes because the diameter is about 1/16 inch smaller than what it should be.


But seriously, I doubt you really need the shield to begin with.

All very helpful replies, thanks a lot. I will look in to cutting the shields off. Access is tough.

I have sets of NOS tubes that are in these amps now. They work fine, but I do have a bit more amp hiss/buzz/noise than I want. I've done a bunch of careful cable routing and the like, and I suspect what remains is either from the tubes or old capacitors. Before i replace the power filter capacitors I wanted to swap in a set of new tubes to see if it made any difference... sound reasonable?

Matt - when you say the 12AX7 may or may not want a shield, you're talking about the whole intact shield (both pieces) not just the skirt part I'm contemplating removing, right?

If anyone is curious, this is more or less the schematic:

http://lacinato.com/robertsmod/Akai%...mps_In_One.jpg

its probably the caps. tubes hiss like a badly tuned radio station, and lose gain.

Have you thought about simply replacing the old socket with a new one that is wide enough? You can buy 10 brand new sockets for a price of one "Socket Saver".

So I used a pair of tin snips and made a few vertical cuts in the collars and just fanned them out slightly -- plenty of room now.

And drtechno was right - the new tubes didn't change the noise.

Maybe I should start a new thread for the cap issue... but: I assume I should start with C25 and C26? (schematic)

Those two caps are currently in a single "2-in-1" cap body with three terminals (common ground) -- any reason I need to or should replicate that kind of configuration? (besides space)

For power filtering caps is there value in getting fancy/expensive orange caps or are those used only for in-line-with-the-signal positions?

Thanks again, all,
-c

The "cleanest" physical replacement is to replace the dual cap with another dual cap.
You can improve (reduce) the residual hum by adding some additional filtering to C-25 (the post-choke plate supply point). If you replace the dual cap with another 2X 20 uF @ 450 volts, you can then add on (if you have room) another 20uF/450 V axial-lead cap in parallel with C-25 (inside the chassis) to reduce the supply ripple. I wouldn't increase the value of C-26 at all to avoid increasing the peak current in the 6X4 rectifier.

Any good-quality aluminum electrolytic replacement will be an improvement over the original, old, possibly low on capacity, filter cap.

While you're at it, be sure to also replace the input stage bypass C-7

IMHO, the next in priority should be the cathode bypass caps: C-2, C-10 & C-15, Then followed by the coupling caps (use the best you can afford here): C-9 & C-11.

Hi
Both anodes of the 12AX7 are at quite a high impedance so may pick up unnecessary interference, either hum or mobile phones etc. If this is a bother you may need to reconsider a shield.
Matt S

Sure you can. But have you ever tried replacing a tube socket in a typical chassis? In many cases I've seen it is extraordinarily difficult and only a last-ditch option.

Yes, I've done it. It can be a bit difficult if there are more than two wires per pin connected or if the wires are hard and twisted around the pins, but I don't see it as something extraordinarily difficult.
IMO it can be even easier than removing the metal screen.

It probably depends on whether you are more handy with a soldering iron or some metal cutting equipment.

I am handy with both. But some firebottle chassis I've seen has "rats-nest" wiring so bad you can hardly even see the bottom of the socket. I would do almost anything to avoid that kind of electronic thicket. Especially with vintage, brittle, heat-damaged components.

Amen! Plus, in gear that was originally produced in any significant quantity, it's not uncommon to have the sockets riveted to the chassis. Even with a drill press and a new bit, drilling out rivets is not much fun.

Since chconnor already "opened up" the socket shield base with tin snips (a perfectly workable solution), it's academic anyway.

Much obliged, Lotus --

You mean "cleanest" in terms of noise/electrical issues or "cleanest" as in "least hacky"? :-) (I'm not worried about the aesthetics of jamming a couple caps in there in strange ways, if that's what you meant.)

What did you mean here by "inside the chassis"?

Thanks, Matt - by "shield" here I assume you mean a for-real shield -- i.e. one that actually encloses the tube, not just the base of a missing shield?

I meant "cleanest" in the aesthetic, non-electronic sense only. As long as you connect the replacement caps to the same terminations (especially where the ground is connected) there shouldn't be any change in the noise other than a likely reduction in low-level hum because of the better filtering the additional cap(s) will provide.

Multi-section filter caps are usually mounted in a can that sits in a phenolic or metal plate fastened through a large hole in the chassis surface. I'd try to install any additional parallel capacitors below the chassis top surface, not sticking up on top if there is room inside the chassis. Position is not critical, but fairly high peak currents flow through those filter caps, so the (-) lead of any add-on cap should be connected to ground at the same point that the original dual cap (-) terminal was grounded.

Also, remember that those caps carry high voltage, so it's not a bad idea to slide some insulated tubing over any exposed positive lead of the add-on cap.

Hi
I was meaning shield as in the full length tube that encloses the valve. They would fit 'bayonet' mounting on the short item you have just chopped!
Matt S

Thanks again, Lotus! And thanks, Matt.

Do you think C16 would also be a useful one to swap?

Is the 450V rating recommended for C25 and C26 or can I go with 350V, do you think? The old service manual for the amp (before it was significantly modded, but i don't think the power characteristics have changed) lists them as 350. It lists C9 and C11 as 400V, C7 as 300, and the others you listed as less 25V. I've measured like 250-300V in the amp in various places...

Nominal peak B+in that amp can hit over 350 volts if it's plugged into a high (127volt AC) line. The average voltage you read on a meter on the first filter cap will probably stay below 350 volts most of the time, but if you measure the real voltage (with a scope) (carefully!) you might be surprised. Also, if you unplug or "pop" the 6BQ5, the voltage will go up.

Depends on where you live, but I've seen 127-128 volts on my mains often.

You can probably "get away" with 350 volt caps for C25 and C26, but using caps rated at 450 will be safer. When re-capping old gear I tend to err on the conservative side.

Yes, definitely replace the screen bypass C16. A 350 volt cap should be fine there.

Indeed, just recently I've been measuring little spells of ~128... I was clued in by my PC power protector beeping at me. :-) Good tip, thanks.

Should I bump up the rating on any of the smaller caps, or, no big deal, there? E.g. the 25's or the 300? edit: or the 400's?

edit: geeze, right now the mains are at 131.1V...

Ok, did some pricing and I'm getting 50V caps for those listed at 25V, and I'll try to get away with a 350V for C7 (original amp manual lists it as "300V"). The filter caps will be 450V. I will report back with results. :-)

Anyone have recommendations for the coupling caps C9 and C11? The "best" I find are a couple bucks... sound about right?

For C9 I want a .02 uF, but I found a .022... seems a .022 with 5% tolerance is probably a safe/good replacement for a 50+ year-old .02, eh? And my understanding of coupling caps is that if you err a little in the upward direction it's less problematic?

The voltage ratings you're using now are fine.

C9 at 0.022uF instead of 0.02uF will change the HP -3 dB point of the inter-stage coupling from 16 Hz to 14.5 Hz. In other words, it won't make any difference at all, and you're right: With coupling caps, adding some capacity is generally better than reducing it.

Chances are the output transformer doesn't work below 30 Hz anyway.

Brilliant - thanks!

Last question, I swear: from the schematic you mentioned C2, C10, and C15. Should I include C30 in that list?

Ahh, that sure looks like an afterthought "add-on". Looks like someone wanted a pre-output-stage, output tap/insert point. It's a coupling cap that's only used if you use J5 to feed something. Otherwise it won't affect the amp's sound.

Certainly change it as long as you're in the amp. Since it's a 25uF electrolytic, it really should have a DC charging path instead of just being connected to the output jack (with no DC return). However you don't want to load that point too much because its part of the feedback loop around the output stage and output transformer.

I'd add a small resistor (20 k ohms) from the low side ((-) lead) of C30 to ground (in parallel to J5). I'd also avoid loading that output (if you ever have a use for it) with anything less than a 10k or so input load to avoid reducing the output stage negative feedback. Too low an impedance there will eliminate the negative feedback and might make to output stage unstable.

You might want to add a 1 or 2k ohm resistor in series with J5, just to prevent the possibility.

Adding an output there is definitely a "kludge" Also, there is no AC coupling to the "insert" (ring) of J5. It goes directly to the grid of the 6BQ5, so anything connected there should be AC-coupled only. Connecting something with a low impedance, DC-coupled output will kill, or at least change, the grid bias on the output stage.

Very interesting. Yeah, if nothing is in J5 then C30 is open circuit, and if you plug something in it ties C30 to the tip and also ties the grid of the 6BQ5 to ground, which I was told was done to act as a "mute" function when J5 is used. When I was working on the amps, I found that I could not use J5 as an output unless I defeated the 6BQ5 mute -- otherwise it would have overwhelming buzz/hum on the J5 output. Maybe this relates to what you're saying? (Can't remember if that was true for all 4 I converted or just 2 of them...)

That suited me fine, though: I like to take a DI from J5 while driving an amp via the 6BQ5. It has worked fine so far, but if i'm harming the amp or risking my equipment (or myself), or if I'm degrading the quality of the sound, it'd be great to know. Would adding the resistors you described improve the situation, considering that it hasn't caused me any problems that I'm aware of?

I'm out of my depth, but I take it this is the "mute" function I described... ? J5 is present in the original schematic, so it wasn't added by the guy that designed this mod, but maybe it makes more sense in the reel-to-reel preamp context that these amps were designed for? Or maybe they're just kludgey. :-)