Gibson BR-3 interstage transformer?

[Murrayatuptown]

I find myself back here talking to myself again.

I paid $108 minus a 10% (?) holiday coupon from Mercury + expensive shipping. I think the net was $118 delivered. That may be offensive in comparison to lesser IT's (that apparently do not work), but it's cheap for an EH-150-III compared to what Mercury's OT & PT cost. I don't think there is anything miraculous about the Gibson or other 1930's PP 6L6 OT's, and PT's from Hammond aren't terribly expensive compared to Mercury.

IT's in the hi-fi world are scarce & expensive, so I think the Mercury IT_PI is a relative bargain if the target is EH-150. If it's BR-3, maybe not.

I have a Tenma RLC meter (because I don't want to pay for a Philips or HP/Keysight to use a couple times a year). I can change the frequency to some inconveniently spaced choices...IIRC 100 Hz, 120 Hz, 1 kHz & 10 kHz. 10 kHz is useless for a guitar amp. The output is whatever the hell it wants to put out. Seems to put out higher level when driver higher-Z DUT. I put a DMM in parallel and saw 200 mV and 600 mV two different times trying to measure an EH-pickup.

Back to Hg IT. I measured almost 1900 ohms DC on primary and as little more than 3800 ohms DC for full secondary. I mention those because the inductance measurements vary with test frequency and whether the RLC meter is set to series or parallel.

I could get semi-stable measurements for inductance at 100 & 120 Hz that vary ~ 5% between Lp and Ls (60-66 H) and 192-212 H for full secondary depending whether Lp or Ls so it seems about 1:3.2 Z ratio or 1.78 n1:n2 turns ratio full secondary and 1:0.89 pri:half-sec.

At 1 kHz there was low Q and such wild variation between Lp and Ls & didn't trust anything. Unstable inductance and inconsistent compared to trends at lower frequency...leading me to conclude the Tenma doesn't like determining inductance at high values (high Z). (I hate auto-ranging impedance instruments that decide what frequency or what signal level. So I wish bought the damn Philips...that has 50 or so mV output. It's only money.

I have griped about the EH-150 Style III s/n 12828 owner-drawn schematic with 100 k plate resistor feeding the PI driver 6C5 and as-drawn 0.05 uF coupling cap connecting to IT PI primary. 0.05 uF and 63 H is a series-resonant circuit, a notch filter at 90 Hz. No. Design is licensed from an entity related to Western Electric. They would not put a notch filter at that frequency then try to boost it with the bass switch. No one else seems to ac-couple to an IT PI, so I think the author just expected to see a plate resistor there.

[Murrayatuptown]

I found a Gibson schematic for an EH-160, the AC/DC lower-powered variant of EH-150, with capacitor-coupling to the IT PI, and a 100k resistor feeding the B+ to the IT PI driver tube but versions of BR-3, BR-9 & early EH-100 & EH-150 had driver B+ fed through the IT.

They all appear in the 116-schematic Gibson PDF that's out there.

I wonder how different the build variants sounded as the circuits changed every couple years and whether people had more than one version of an amp model and noticed any difference. Or if they were as fussy as we are now. Barney Kessel was pretty hyped about the unobtainium grade of Cobalt in his 30's blade pickup magnets (from video). Listeners didn't even know what they were hearing sometimes when electric guitars were new...thinking they were hearing a horn solo but it was a guitar.

[_Steve]

I can't for the life of me understand why they would have that coupling cap there. Perhaps someone more knowledgeable can chime in?

{EDIT} looking again i guess it's to decouple the load of the IT so the driver is only loaded by the 100K load resister?

60 Henries is great for that primary!

[Murrayatuptown]

I think I will go find a post that actually refers to the T-8780 or -8790 IT PI and post what I just recorded.
The BR-3 schematic looks enough like the others to make me assume they used the same interstage transformer phase inverter the other models did.

But Mercury mentions both Gibson transformers, one being 'the larger', being replicated by their ToneClone IT PI offering.

[Murrayatuptown]

If you can't sleep, read this.

If you don't like science experiments, your choice.

I think I will go find a post that actually refers to the T-8780 or -8790 IT PI and post what I just recorded. (This looks like the best one).
The BR-3 schematic looks enough like the others to make me assume they used the same interstage transformer phase inverter the other models did.

But Mercury mentions both Gibson transformers, one being 'the larger', being replicated by their ToneClone IT PI offering.

I haven't yet tested a Gibson IT-PI, but have one. Looks like I can probably test it through the tube sockets by removing tubes on each side of the IT PI and inserting a pin that makes contact. I have only does some inductance ratio (which I am always hesitant about but use for a 2nd opinion) and voltage ratio measurements for the aftermarket one. I have more of those on order and don't want to get too nosy but it's just curiosity.

I just pushed aside the Tenma LCR meter and replaced it with a Fluke PM6304 RLC meter. I had to buy an offshore aftermarket Kelvin lead set with the not-yet-identified Lemo 8 -PIN CONNECTOR because people want $2000 for a Fluke cable, which is ridiculous, but they are not available from Fluke.

Inductance measurements on iron (steel) core devices are dependent on too many things...including frequency, amplitude, excitation current and the meter configuration for series or parallel equivalent impedances. There's a mathematical conversion because impedance is literally 'complex', and measured components are not purely R, L, or C with text book phase angles. It's all 'in the literature', any decent test instrument manual, and the frustrated minds of people who run into this. The component values therefore change between series and parallel. If you have a fancy enough meter, you can decide or allow the meter to decide automatically, which measurement is most accurate. It's based on Q.

A transformer maker told me to do measurements at the lowest test amplitude an instrument can do, and to do amplitude measurements with all windings interconnected in series. I don't remember why, but I think it keeps voltages from floating & being influenced by stray flux. Or I'm wrong. Not confident enough to try that yet.

I started trying to take measurements on the aftermarket transformer at one amplitude and TBD multiple frequencies. The inductance varied with frequency, and drifted uncomfortably over time, even with measurement averaging, with the Fluke, and I saw the same trend with the Tenma, which lacks as many user choices. So I chose a frequency and amplitude that gave me stable inductance measurements. That was at the lowest frequency (50 Hz) and highest amplitude (2 V) I could choose. In my experience that makes sense...the transformer is operated at much higher AC voltages and flux levels than modern test equipment can provide. An RLC or LCR (seems to be either chronological or geographic preference) meter is basically an automatic impedance bridge, so there is too much information available (perfect for me) to study. A measurement 'depends'.

At a high enough frequency the phase angle (I think it was about +79 degrees at 50 Hz) gets closer to 0 degrees and the meter sees resistance as the dominant component. I could not get data I liked over 1 kHz in parallel Z. I think that was approaching the self resonant frequency where the inductive and capacitive reactance cancel. In parallel Z the primary it looked like 1.7 Megohm /_15 degrees . The current from the RLC meter test signal was in the small single digit microamp range and that might be why measurement were driftier.

I'll give a single frequency/amplitude inductance ratio (proportional to Z ratio, theoretically turns ratio squared, but I find is usually not perfect), then voltage (turns ratio)

SE Primary brown-black 1881 ohms DC. 2V drive selected at 50 Hz Added DMM and measured 2.000V and it did not alter the inductance measurement. (I had to note whether it did or not).

Parallel Z inductance component 98.25 H Q = 5.8 FYI, this dropped to about a drifty 55 H at 300 Hz, so I stayed with 50 Hz
Series Z inductance component 95.4 H Q = 5.8

Secondary yellow-yellow = 3813 ohms DC. Y1-B = 1804 ohms and B-Y2 = 2009 ohms.

The higher number indicates the second half of the secondary is wound on top of the first, making each turn progressively longer and higher resistance. Just like you see on many other center-tapped transformers.

Y-Y
Parallel Z inductance = 363 H, Q=7.16
Series Z inductance = 356.5 H Q= 7.16
The inductance ratio secondary/primary for parallel Z = 3.695
The inductance ratio secondary/primary for series Z = 3.737

Old DMM secondary voltage was 4.01 V, newer scopemeter read 3.992 V 1:2 turns ratio
Y1-B 2.013 V DMM, 1.997 scope meter pri: half-sec "1:1"
B-Y2 2.012 V DMM, 1.995 scope meter pri: half-sec "1:1"

So the primary to secondary turns ratio is 1.996. "1:2"
The two half secondaries are extremely similar in turns count (do the math with the voltages). As you've read, the resistance ratio of the two half secondaries due to different lengths of wire, do not affect the turns ratio. (Might affect something else we don't care about today). Pri-half-secondaries are both 1:1.

The Fluke has a 2V built-in DC source for DC bias, but the manual says it's for (voltage dependent) capacitances. Doesn't mention DC-biased inductance. I had hoped 2V/1812 ohms was pretty low current but the meter complained 'OVER' or similar. There is an external DC input, but failing at 2V indicates it would be stupid to continue trying to get DC biased inductance. There are ways to measure it but they are messy than a pair of clip leads, and can be dangerous to the 'operator' and test equipment due to transient voltages created by interrupting the DC current.

Leakage inductance at 2V 50 Hz:
Primary (brn-blk) was displaying 94 H inductance (series impedance measurement)
shorted Y-Y: 333 mH (series)
shorted Y1-B: 518 mH
shorted B-Y2: 800 mH
shorted Y1-B-Y2: 330 mH (essentially same as 333).

[Murrayatuptown]

I see people on antique radio forums who do some really weird stuff to avoid spending money and get something sort of working. Or a antique part isn't available.

One was to alter an IT PI with an open primary by disconnecting it, feeding DC through a resistor, and coupling the driver output through a capacitor and leaving the split secondary as an auto transformer to feed PP power tube grids. That just doesn't seem capable of producing the same audio quality. I guess if you're listening to AM broadcast band and shortwave with static, there's not much of a degradation with your vintage headphones.

The other was to remove the IT altogether and make the driver RC coupled. I guess you lose the IT PI step-up but can change the driver components...not the same amp anymore.