RCA Ba6a Innerstage XFMR

[CJ]

This guy has been the subject of a lot of talk over the years by people who want to diy the infamous BA6a compressor, or fix a dead one.

Rare, expensive and prone to failure.

Thanks to Neilsk we have a dead one on the bench.

What's inside?
A lot of turns of very fine wire sitting on a butt stacked alloy L lamination core. I did not see a paper gap.

This means twin coils. And there are two pies on each coil for both the primary and secondary.

Wound sec-pri.

Ratio is 1:2 p-p plates to grids

Pri turns: 1800 each pie of #36= 7200 turns.
DCR is 475 ohms for each identical pie so 1.9 K total.

Sec turns: 3600 each pie of #44 (yikes!) so 14,400 turns.
DCR (est) 12 K ohms

The L lam is not in my Magnet catalog but is like a 4 L with a shorter leg. The nickel alloy has some corrosion so it is hard to determine the nickel content by sight. Will do inductance test with test coils to find Permeability and inductance.

1.115" stack by 1/2 inch tongue = 3.596 cm^2

Lam thickness is 0.015"
Probably started out as 0.014" but they did not dry out the core tube so moisture caused lam corrosion.
105 RMS volts primary will generate 5 K Gauss at 20 Hz.

There was no wax impregnate in the coils so the turns were free to rattle and break. I would not buy one of these expensive transformers as by now they have just about had it.

Interesting is that I believe the infamous Triad HS-29 (Pultec innerstage) has 1800 and 3600 turns on their single coil XFMR IIRC.

Pictures tomorrow,

 

[CJ]

this guy came in without the can so a lot of hassle saved ,

Neve Marinair like brass frame around the core,

notice core is butt stacked,

 

[CJ]

core will develop gap if not clamped down by the brass frame,

 

[CJ]

detective work required, why does only one coil covered with layer insulation?

any guesses?

 

[CJ]

secondary is wound first, so this is the primary, #36 , about 130 turns per layer,

fill factor is pretty low, fill factor is how close to 100 percent of maximum turns per layer you wind, determined by the wire size, 90 percent fill factor would be pretty tight as you can not approach 100 without crossovers, and tolerance of the wire diameter, this might be 80 percent fill factor right here.

no wax which is weird, you can move the wires around easily, so low tension also.

no wax was probably done to lower capacitance, which has to be watched on a coil like this with so many turns,

but no wax means reduced life span because the wires will vibrate with large signals,
could be that somebody in the processing dept. at UTC was asleep at the wheel which is a real possibillity since processing is the lowest job on the ladder which means everybody is on drugs or alcohol, at least thats the way it was where i worked, and yes, i spen some time in processing with the varnish tank and epoxy end capping in the sand pit,

 

[CJ]

1 mil copper shield between the secondary and primary, there was also quite a bit of insulation between pri-sec, about 10 mil on the outer side and 20 mil under the copper plus tape so over 1/32" between pri and sec,

copper is tarnished due to no wax,
moisture from the insulation has me wondering if the coil ever went through the dryout process,

 

[CJ]

insulation over the shield is layers of 1 mil glassine, this stuff was weird, almost like half glassine and half poly, it kind of melted into a glob on the other coil when i used the heat gun,

the primary coil slid right off the slick shield insulation, again you can see on the inside that there is only one coil with layer insulation, why is this?

 

[CJ]

unwinding the secondary, thin stuff, #44, guitar pickups usually use #42 so break strength on this stuff is minuscule, good thing no wax, or pine tar like the Langevin tube preamp stuff,

20 layers, 180 turns per, no broken wire during the unwind, very consistent wire, 0.002" exactly, the whole way through,

 

[CJ]

core tube had a big overlap on the side, don't see that too often,

 

[CJ]

ok, the answer to today's quiz is that during the winding process, with the dual pie design, one pie layer was put down before the other pie was started, in other words, they did not start winding both pies at the same time,

why? so the finish wires of each coil would meet in the middle, this way you do not have to put down a crossover tunnel with tape, so the last layer of one of the pies is wound on top of the layer insulation of the previous layers wound. this also means that the starts will both be on the outside of the coil as the traverse moves the same way for both pies.

as i always say, every time you take apart a transformer, you see something you have never seen before. so many ways to put wire around a core,

 

[CJ]

Lamination might be hard to source for this guy, have to check online,

exactly like a 4L only different,

rust is from core tube/former moisture ,

will do inductance test tonight,

 

[CJ]

pinout etc

 

[CJ]

Inductance tests on this core show the lams to be regular non grain oriented silicon steel,

2 coils of 1800 turns each were used,

so 3600 total turns produced 75 Henrys at 20 Hz.

Catalog says

L= (.1224 x 10^-8)K1N^2 uac

75 H = 0.015069888 uac

So Permeability is. 5000 at 20 volts and 20 Hz.

Which translates to non grain steel.

Primary Inductance should turn out to be about 300 Henrys

And secondary inductance 1200 Henries.

Bmax is now going to be half of saturation level for non grain steel so about 10 k Gauss.

So figure about 210 V RMS at 20 Hz.

I doubt the 6SK7 tubes are going to get anywhere near that .
so low distortion the bass frequencies would seem to be expected.
But what is our XL at 20 Hz?

300 H x 6.28 x 20 =
37,680 ohms not to shabby.

 

[Rob Flinn]

Hello Chris. The question is, are you gonna wind a bunch ?

 

[CJ]

I'm afraid I do not have the machinery to wind that fine of a coil, and then there's the lam problem.

That type of coil has to be wound on a multiple winding machine, like 20 at a time , that is the only way to keep the 0.5 mil glassine paper to lay flat, it comes of a wide roll maybe two feet wide. If you try to lay down 1 1/2 inch strips then the coil gets all messed up after 20 layers of #44. UTC was amazing in their ability to do this coil and the A 10 etc.

I could transfer the design to an EI core like Sowter, but if the inductance of the Sowter is up to snuff then I see no reason to do that.

Since every inductance meter is different , I have no way of knowing what theirs really does.

But if somebody has a Sowter then they could use the same test I did which is apply 20 v RMS to the primary and read the current,

There are other UTC models that might make a cool BA6a, HS or LS innerstage etc.

 

[CJ]

inductance test 1800 t each coil

this is a scrapless lam, short leg dimension picked for no waste during stamping,
lams are 0.015" after scotchbrite cleanup , cold rolled barn roof scrapless = cheap
nickel lams do not rust like that,

 

[emrr]

Awesome and interesting......digging for the data I have.....I don't have the transformer to measure.

Must say I've almost never seen an RCA or UTC (or really, anything) over about 7K5 DCR, 12K DCR sounds crazy. Everything I've seen has suggested it being a 1:1 also, but we've somehow never gotten DCR measurements of a good working example.

The preceding 86 limiter interstage is 2:1 and 5320:3790 DCR, 1860:930H with a handheld meter at 120Hz.

Brian Sowter told CJ this way way back:

https://groupdiy.com/threads/api-sowter-transformers.1454/#post-18470
here's Lassoharp's measurement of the Sowter 1:1 replacement, followed by a bunch of tests we ran:

https://groupdiy.com/threads/ba6a-interstage.55693/page-2#post-714226

 

[CJ]

I did not verify the turns ratio as the leads could not be traced so it very well could be a 1:1 if the secondary coils were hooked up in parallel.

All would it would take is a DCR reading on the secondary.

The turns count is similar to the HS 29 which has 1800 + 1800 pri and 3600 + 3600 sec for a 1:2 ratio , 20K to 80K.

The HS 29 does use a nickel core so it won't take the same level or DCR imbalance.

 

[directdriver]

I have a broken one with one open half primary. I opened the terminal plate to see for myself. It is indeed 1:1 ratio. The two secondary windings are wired in parallel, hence no center tap. I saw terminals #10 and #12 each has two wires soldered together. The two 22K resistors connected to ground form a virtual center tap.

The secondary DCR I measured is 2.6K Ω.
One side of primary to CT is 1.17K Ω.

I have a hand held LCR meter that outputs 100Hz square wave and here are the readings:
Half of primary to CT is 33H.
Secondary reading is 134H.

The above appears to match lassoharp's finding in another thread just like Doug said.

I hope the above info helps and please advice on a alternative replacement or method. I tried cap coupling a la Sta-Level style but I got motorboating. I used the two 22K resistor as loading and the motorboating disappear but the coupling cap would have to be enormous to have decent bass. I suspect the two 6J7 grids like low DCR and or very picky on balanced PP signals. I think a center tapped grid choke like in Altec 322 or WE 126 might work... Maybe UTC A-30?

Will a UTC HA-107 work or CG-133 or CG-233?

 

[emrr]

This should all work, ignoring the gain difference.