RCA Ba6a Innerstage XFMR
- Thread starter CJ
- Start date
Help Support GroupDIY Audio Forum:
vinyvamos
Well-known member
vinyvamos
Well-known member
yes unloaded,
here is a B-H shot of one primary wind, i needed to drop the turns in half to get saturation, thus,one primary being used, 3600 turns
took 340 volts to get saturated so plenty of headroom on this innerstage xfmr,
curve shows medium core loss (area inside loop is core loss)
so probably 50/50 ie radiometal
here is a B-H shot of one primary wind, i needed to drop the turns in half to get saturation, thus,one primary being used, 3600 turns
took 340 volts to get saturated so plenty of headroom on this innerstage xfmr,
curve shows medium core loss (area inside loop is core loss)
so probably 50/50 ie radiometal
Attachments
Last edited:
now we can check Gauss vs saturation,
cross section of the core is 3.596
100,000,000/(4(K1 stacking) x 3.596 (area in cm^2) x 60 Hz (test freq) x 3600 turns (1/2 pri))
25,000,000/3.596 x 60 x 3600 = 32 V
we apply 340 volts to get full saturation so Bmax is 32 x 340 = 10.9 K Gauss
10.9 KG is right about where 49 permalloy saturates so we do have a nickel alloy core
80 NI Supermalloy saturates at about 5 KG, and would not be a common application of this expensive alloy.
the high turns count on this RCA xfmr was used to keep the nickel core operating in a linear region of the B-H loop. the down side is the xfmr has to be carefully designed for HF response as the many turns degrade performance, as seen by my failed first attempt at rewinding.
cross section of the core is 3.596
100,000,000/(4(K1 stacking) x 3.596 (area in cm^2) x 60 Hz (test freq) x 3600 turns (1/2 pri))
25,000,000/3.596 x 60 x 3600 = 32 V
we apply 340 volts to get full saturation so Bmax is 32 x 340 = 10.9 K Gauss
10.9 KG is right about where 49 permalloy saturates so we do have a nickel alloy core
80 NI Supermalloy saturates at about 5 KG, and would not be a common application of this expensive alloy.
the high turns count on this RCA xfmr was used to keep the nickel core operating in a linear region of the B-H loop. the down side is the xfmr has to be carefully designed for HF response as the many turns degrade performance, as seen by my failed first attempt at rewinding.
Last edited:
vinyvamos
Well-known member
Nice work! And how does the butted gap affect the BH curve? Wouldn't the curve be wider with interleaved lams? Therefore higher core loss.
So why did they choose to gap this core? Because in doing so it looks like they made their task a lot more challenging. But not impossible
So why did they choose to gap this core? Because in doing so it looks like they made their task a lot more challenging. But not impossible
I will try a BH test with the core stacked Lap 1 and see what happens, and do an inductance test.
I guess they wanted to protect the nickel core from unbalanced DC.
And it is quicker to lace up a butt stacked core, interleaving L lams is a real pain.
I guess they wanted to protect the nickel core from unbalanced DC.
And it is quicker to lace up a butt stacked core, interleaving L lams is a real pain.
vinyvamos
Well-known member
Yes true, much quicker to assemble the butted L... the Sowter is not gapped.
Here are photos of inside of my can. The inner can looks to be similar shape to another series of UTC, here I'm showing my example which is an S10 interstage.
Here are photos of inside of my can. The inner can looks to be similar shape to another series of UTC, here I'm showing my example which is an S10 interstage.
Attachments
Inductance measurements and Permeability for butt stack vs Lap 1 interleaving for the Ba6a core>
Butt Stack
20 Hz 300 Henrys. Uac 5800
40 Hz 300 H. U 5800
100 Hz 285 H U 5600
200 Hz 285 H U 5600
Lap 1
20 Hz 1500 H U 30,000
40 Hz 1000 H U 20,000
100 Hz 900 H U 17,500
200 Hz 850 H U 17,000
You can see the effect of the gapped core in stabilizing permeability at the expense of Uncle Henry.
The Lap 1 core sees a 5 fold increase in inductance at 20 Hz.
But inductance drops in half at 200 Hz and tolerance to DC imbalance will be compromised.
What will the saturation voltage be on the Lap 1 core?
BH test tomorrow...
Butt Stack
20 Hz 300 Henrys. Uac 5800
40 Hz 300 H. U 5800
100 Hz 285 H U 5600
200 Hz 285 H U 5600
Lap 1
20 Hz 1500 H U 30,000
40 Hz 1000 H U 20,000
100 Hz 900 H U 17,500
200 Hz 850 H U 17,000
You can see the effect of the gapped core in stabilizing permeability at the expense of Uncle Henry.
The Lap 1 core sees a 5 fold increase in inductance at 20 Hz.
But inductance drops in half at 200 Hz and tolerance to DC imbalance will be compromised.
What will the saturation voltage be on the Lap 1 core?
BH test tomorrow...
I am amazed ! What a detective / team work… keep it going 
cheers
cheersvinyvamos
Well-known member
Great work CJ! So are you doing these L/perm measurements at a high input voltage? Just wondering as doesn't the perm change WRT current/magnetization? In my calculations I was setting the ungapped perm of the 49 permalloy to 10000 and then reducing it from there based on the gap while calculating for a 50% overhead in flux headroom (for the audio) after a few mA of unbalanced DC is applied.
Here are some frequency tests comparing two of my own interstages with the original UTC, inside of an original RCA BA6B. Moderate signal level and GR switched off. I use a 4-pole relay to switch between interstages. Here you can see that even going slightly above the 7200 turns count I am still not quite reaching the 140H target even without any paper in the butt joint of the core. With enough pressure I can, and I will be fabricating a similar clamp to the one that UTC used. Note that all my inductance measurements are done with a protek LCR meter (not with CJs method) so they are at 100Hz/1V parallel and IMO are only truly relevant within my own measurements. Comparable though, which is all I need.
Note how the LF drops off when the inductance is lower, with paper in the gap. I am not sure yet why my #1 interstage has a slight (0.5dB) HF lift, hmm... nothern lights radiation maybe??? Overall though this is looking very promising. I now need to do some sweeps with the BA6 forced into a GR state. And A-B listen a lot... blind A-B, immersive meditative A-B, Quantum innoculum A-B.....
THD measurements for all the interstages are looking very good. Somewhat heavy 3rd harmonic with pushed levels, which may be mostly coming from the steel output trafo.
I currently have 68V as the bias on the 6H6 sidechain rectifier and am struggling to get more than 9dB of GR out of this beast. Push it any further and the output amp clips. Is that normal? I can force the bias lower and then I can get up to 20dB GR, but I know that is asking a bit much of this ancient tractor...
EDIT: attached image states 40 mil paper in gap. totally incorrect, it is more like 5 micron / 2 mil (how do you measure this shit!?)
Here are some frequency tests comparing two of my own interstages with the original UTC, inside of an original RCA BA6B. Moderate signal level and GR switched off. I use a 4-pole relay to switch between interstages. Here you can see that even going slightly above the 7200 turns count I am still not quite reaching the 140H target even without any paper in the butt joint of the core. With enough pressure I can, and I will be fabricating a similar clamp to the one that UTC used. Note that all my inductance measurements are done with a protek LCR meter (not with CJs method) so they are at 100Hz/1V parallel and IMO are only truly relevant within my own measurements. Comparable though, which is all I need.
Note how the LF drops off when the inductance is lower, with paper in the gap. I am not sure yet why my #1 interstage has a slight (0.5dB) HF lift, hmm... nothern lights radiation maybe??? Overall though this is looking very promising. I now need to do some sweeps with the BA6 forced into a GR state. And A-B listen a lot... blind A-B, immersive meditative A-B, Quantum innoculum A-B.....
THD measurements for all the interstages are looking very good. Somewhat heavy 3rd harmonic with pushed levels, which may be mostly coming from the steel output trafo.
I currently have 68V as the bias on the 6H6 sidechain rectifier and am struggling to get more than 9dB of GR out of this beast. Push it any further and the output amp clips. Is that normal? I can force the bias lower and then I can get up to 20dB GR, but I know that is asking a bit much of this ancient tractor...
EDIT: attached image states 40 mil paper in gap. totally incorrect, it is more like 5 micron / 2 mil (how do you measure this shit!?)
Attachments
Last edited:
What kind of AC voltages do you get on the inner stage primary? It would be nice to know what level these guys are operating at, nice work on you too!
I was using 20 volts RMS for L tests, HP 200C.
Good to see that even the original craps out at 20 kc now I don't have to shoot myself ,
Is that a 40 mil gap or 4 mils?
I think a 5 to 10 mil gap would put you on the knee depending how much, if any DC present.
All that DCR starts to enter onto the equation as reactance drops at 20 Hz, 90 degrees out of phase so you use arctan to correct inductance readings,
I was using 20 volts RMS for L tests, HP 200C.
Good to see that even the original craps out at 20 kc now I don't have to shoot myself ,
Is that a 40 mil gap or 4 mils?
I think a 5 to 10 mil gap would put you on the knee depending how much, if any DC present.
All that DCR starts to enter onto the equation as reactance drops at 20 Hz, 90 degrees out of phase so you use arctan to correct inductance readings,
BH curves for gapped vs interleaved RCA core,
elongated curve on the butt stack as saturation is not as abrupt due to the air gap,
saturation is reached at about the same voltage for both cores,
core loss is the area inside the curve and is probably the same for both stacking methods, although the shape is different,
the slope of the vertical part of the curves (Flux - B field) represents the Permeabillity, so the steeper curve with the lapped core means more perm as expected.
Magnetizing Force H , due to the flux, B field, is the horizontal part of the curve,
elongated curve on the butt stack as saturation is not as abrupt due to the air gap,
saturation is reached at about the same voltage for both cores,
core loss is the area inside the curve and is probably the same for both stacking methods, although the shape is different,
the slope of the vertical part of the curves (Flux - B field) represents the Permeabillity, so the steeper curve with the lapped core means more perm as expected.
Magnetizing Force H , due to the flux, B field, is the horizontal part of the curve,
Attachments
Last edited:
vinyvamos
Well-known member
Bah sorry the paper is 50 micron, so thats about 2 mil.
OK 20 volts is quite a bit more magneto-excitement than my 1 volt from my Protek... So i should be careful how I compare my L measurements to yours.
Good to see those BH curves, I thought there would be more difference between them but no, they are quite close.
There would probably be more difference in curves if the lams were EI shaped, there is more of a natural gap with the L lams when lapped.
vinyvamos
Well-known member
Some more gory data for yiz! I did some sweeps with the 6SK7 stage balanced (balancing is done with the A & B pots onboard) and then sweeps with the A & B pots set to extremes. This seems to give about 600uA of imbalance between the tubes on my original unit. Audio still passes ok and it still compresses fine, but...
Original interstage handles the imbalanced DC just fine, as does my interstage, due to the gapped core. On the other hand, the little secret candidate for BA6 interstage, the 900871 (found in old RCA distortion meters) with it's 500H and 1380Rdc per side seems to suffer LF loss with the imbalanced DC. The DCR and L figures hint to us that this is an ungapped core, but we don't know what material it is made of. On a sidenote there is one 900871 on ebay now for just $75.
The FTR I tested just for fun as it was a rejected candidate for another BA6 DIY build from many years ago. Surprisingly it handles the DC imbalance just fine, but generally it doesn't have enough inductance to begin with.
A rather recent shipment of the Sowter 9830 was kindly measured for me by Jordan Mandela and it has 500H (1V/100Hz) with only about 700Rdc on each side of it. It has an ungapped core and their website states it is made of 30% mumetal and 70% M6 steel. The website also states a DCR of 1680R... Something doesn't stack up right here (trafo pun??). There is no way you can get 500H on a 70% steel core with just 700 ohms of wire. The core is more likely to be all mumetal or 49-permalloy. So what I am getting at here is that this Sowter will probably work fine when the 6SK7 are perfectly in balance (as will the 900871), but any imbalance due to ageing tubes and/or difference in plate curves or just poor alignment will affect the LF response.
We did extensive listening tests last night in the studio, with all of the above trafos (minus the Sowter, which I don't have yet). My own interstage and the 900871 performed in an immaculately similar way to the original BA6 interstage. Blind A-B testing, with relay for switching between trafos.
Original interstage handles the imbalanced DC just fine, as does my interstage, due to the gapped core. On the other hand, the little secret candidate for BA6 interstage, the 900871 (found in old RCA distortion meters) with it's 500H and 1380Rdc per side seems to suffer LF loss with the imbalanced DC. The DCR and L figures hint to us that this is an ungapped core, but we don't know what material it is made of. On a sidenote there is one 900871 on ebay now for just $75.
The FTR I tested just for fun as it was a rejected candidate for another BA6 DIY build from many years ago. Surprisingly it handles the DC imbalance just fine, but generally it doesn't have enough inductance to begin with.
A rather recent shipment of the Sowter 9830 was kindly measured for me by Jordan Mandela and it has 500H (1V/100Hz) with only about 700Rdc on each side of it. It has an ungapped core and their website states it is made of 30% mumetal and 70% M6 steel. The website also states a DCR of 1680R... Something doesn't stack up right here (trafo pun??). There is no way you can get 500H on a 70% steel core with just 700 ohms of wire. The core is more likely to be all mumetal or 49-permalloy. So what I am getting at here is that this Sowter will probably work fine when the 6SK7 are perfectly in balance (as will the 900871), but any imbalance due to ageing tubes and/or difference in plate curves or just poor alignment will affect the LF response.
We did extensive listening tests last night in the studio, with all of the above trafos (minus the Sowter, which I don't have yet). My own interstage and the 900871 performed in an immaculately similar way to the original BA6 interstage. Blind A-B testing, with relay for switching between trafos.
Attachments
Hey hey, that's me! Really excited to have been able to contribute something to this. Vinny, glad you're jumping in on the discussion!
