Ordering guitar amp OT

[Roe]

Some Fender OT primary to secondary leakage inductance data:

- '59BM, BF SR originals and reissues: 5mH to 8mH,
- BF Twin: ~6mH,
- Vibroverb, Vibrolux Rev.: ~90mH,
- Deluxe Rev. : 15mH to 18mH,
- Super Champ, Princeton Rev. : 9mH.

Marshall OTs with 16 Ohm secondary shorted:

- '66 JTM 50 (Drake 784-128): 17mH
- JMP 50/2204 (Drake 784-139): 11mH
- '67/68 100W Plexi (Dagnall): 14mH.

The marshall OTs seem to have changed over the years.
In late 67 dagnalls take of for drakes on 100w amps (with a few rare exceptions). in the 70s Dagnall seems to have changed from paper to nylon insulation and some claim the winding pattern changed as well.

[Helmholtz]

The marshall OTs seem to have changed over the years.
In late 67 dagnalls take of for drakes on 100w amps (with a few rare exceptions). in the 70s Dagnall seems to have changed from paper to nylon insulation and some claim the winding pattern changed as well.

I only measured two pre-70s Dagnalls, so can't comment on later ones.
If they measure the same I expect them to sound the same.

[R.G.]

From https://www.aikenamps.com/index.php/out ... -explained
"The leakage inductance is proportional to the square of the number of turns, so you must decrease the number of turns to reduce the leakage inductance, but this is at odds with the need to increase the number of turns for good low frequency response. In addition, the flux density may be exceeded if you reduce the number of turns. Different winding techniques, such as interleaving, can help reduce the amount of leakage inductance, and improve the high frequency response."

This is mildly misleading. Leakage inductance is indeed proportional to the square of the number of turns, as is primary and secondary inductances. In an inductor winding, doubling the number of turns increases the inductance by a factor of four. But decreasing turns to decrease leakage is not a formula to decrease leakage inductance per se. In the Golden Age of hifi tube amps, the figure of merit was the the ratio of primary inductance (and hence bass response) to leakage inductance, not the magnitudes.

Helmholtz is correct. At a given impedance level, the primary inductance determines the bass cutoff. The treble cutoff is determined by the combination of leakage inductance and parasitic capacitances.

Leakage inductance is determined literally by the magnetic flux/field lines that leak out of a winding and does not couple through any core to another winding. It therefore acts like an inductance in series with the nominal primary winding that does not cause anything to appear on a nominal secondary. If you wind two coils and set them a meter apart on a table, they will have some flux that couples both coils, and a LOT that does not. So driving one coil will let you sense some of that drive in the second coil, but not very much. The amount you can sense in the second coil goes up and down as the square of the distance between them. The inductance you sense driving the first coil is nearly the entire isolated inductance of the first coil.

As you move the coils closer, more and more of the magnetic field from the first coil gets into the middle of the second coil, so you can sense more in the second coil. As you move the second coil literally inside the first coil, the second coil can pick up most of the first coil's drive.

But not all of it. The second (now inner) coil misses some of it because some of the first coil's flux sneaks between the two of them in a cylinder on the inside. We try to fix this by inserting iron into the middle of the two coils. Iron "conducts" magnetic field about 8 to 10 thousand times better than air or vacuum or any other non-magnetic material. Inserting iron in the middle of the two coils makes the magnetic field get "short circuited" into the iron by a factor of thousands. This rapidly leads to the iron enclosed coils we think of as transformers.

But magnetic fields are sneaky. Even with two concentric coils, some of the flux can sneak between the layers of one coil, and between the layers of the second. This rapidly leads to interleaving. By winding one layer of the first coil, one layer of the second, a second layer of the first coil, a second layer of the second coil, we can force vastly more of the M-field to be inside coils of the secondary. The ultimate is reached with bi- (and tri-, quad-, penta-...) filar winding. Two wires are wound literally side by side in a coil. This makes the smallest possible spaces and cracks for the field to sneak out of. It's used in special cases, but because it's a screaming pain to wind, it's used sparingly. It's expensive.

The received wisdom here is that the relative minimization of leakage inductance is how well and intimately you can intermix the primary and secondary windings. Multifilar winding to some degree is how you reduce the relative effect to the best possible degree. Interleaving, winding one or a few layers of primary, then secondary, then primary, then... is the way it's done in practice. I had to go through the math description of magnetic leakage as a green-eared engineerling. Compared to 1:1 interleaving (that is one primary, one secondary) leakage inductance goes down by the square of the number of interfaces between sections.

So yes, leakage goes up by the square of turns, but the turns is fixed by impedance considerations and bass response needs. Leakage inductance is minimized as much as possible by interleaveing in as many degrees as possible, economic, and practical

[Helmholtz]

One practical addition:
The primary leakage inductance is typically lowest to the largest impedance secondary tap.
So with global NFB this should be used for best stability.
In other words, the NFB signal should preferably be taken from the full secondary.
Adjust NFB voltage divider resistors for desired NFB fraction.

With multi-section 4-8-16 Ohm secondaries I've seen highest leakage inductance with the 8 Ohm winding shorted, but shorting the full secondary always results in lowest leakage inductance.

[nuke]

R. G., From a practical standpoint how does one survey an unknown output transformer for these characteristics?

I have a reasonably decent digital LCR meter, which has a range of frequencies available, 100hz, 120hz, 1khz, 10khz and 100khz and a pretty range of measurements. I've used it to make basic measurements of pickups and what not, and the Bode plots I've made of the pickups using an exciter coil and integrator match well with the measured values.

I've heard some conflicting opinions on how this applies to measuring chunkier lumps, like output transformers and chokes.

[Helmholtz]

You can use your LCR meter to measure leakage inductance.
It's not well suited to measure primary inductance because the measuring voltage is low and doesn't represent realistic operating conditions.
Means the LCR reading might be lower by a factor of 10 or more than the effective inductance at some medium output with a primary voltage of maybe 200Vrms.

[R.G.]

Yep. leakage inductances are small enough to measure with LCRs and so on. You short the secondary, so it reflects nearly a zero impedance to the primary, but the leakage inductance is the part that cannot be affected by the secondary conditions. So the inductance read on the primary with the secondary shorted is the leakage inductance.

Then you measure the primary inductance with the secondary open. This can be done with a pulse inductance test ( V = L * di/dt, so L = Vdt/di) or by resonating the primary with a capacitor and computing the inductance from the resulting inductance frequency. Primary inductance is a little funny, as the inductance varies with the size of the exciting signal. It's best measured with a medium sized signal as the iron BH curve gives a lower inductance at both low and high field strengths.

Oh, yeah. You can use a known capacitor/resistor and measure phase shifts caused by the inductance.

[Helmholtz]

Primary inductance varies with V/f.
I determine the inductance from the impedance (Z = Vrms/Irms) at mains frequency. Secondaries unloaded, of course.
Ignoring the comparatively small DCR we can write Z= 2*pi*f*L, so L= Z/(2*pi*f).
My voltage source is an isolating (!) mains variac.
Fortunately our mains voltage is 230V/50Hz and the variac gives me up to around 270V. With 120V mains you could use a step-up transformer after the variac.
Varying the voltage across the primary while measuring the current allows me to see how the inductance varies over the "corresponding output power".
With the voltage applied to the full primary, the corresponding output power is found from P = V²/Raa.
Example: V= 240V, Raa = 4k gives P = 14.4W. Under these conditions a decent guitar amp OT should have an inductance >50H.
I've measured values up to 500H.
The Marshall Drake 784-139 measures 95H.
At voltages lower than 240V/50Hz the inductance typically drops.
This means that the bass response gets worse at low output.

[pdf64]

... I think Pete Farrington / @pdf64 might have some knowledge in this area.

Thanks, I suppose I can come across as convincing, but this discussion shows me up as being something of a dilettante

Helmholtz's secondary leakage inductance measurements are great. When combined with a global feedback loop, the high reading from the Vibrolux OT may show why stability in those models with that 022848 OT can be somewhat marginal.
I wonder why it's so much higher than the Deluxe Reverb OT, which looks to have a similar simple, non interleaved design?

[Helmholtz]

I wonder why it's so much higher than the Deluxe Reverb OT, which looks to have a similar simple, non interleaved design?

Without dissecting, all I can say is that you can't guess leakage from size.

[nuke]

I went rumaging around my junk box and retrieved two output transformers, a "NSC022855" (Electroharmonix Fender 50-watt, 2-ohm output transformer) that I took out of a 66 Super Reverb and a Vox AC30 output transformer with a barcode sticker on it, which I took out of a vintage AC30, I think it was a Korg era OPT, but I don't know for sure.

Both were replaced about 20+ years ago, for the same reason, they both sounded like poo. They went into my junkbox, just in case... Really need to purge more often.

I did a quick check for leakage inductance and both were quite high. Little wonder they sounded so bad.

The AC30 was a big resto, and the transformer made a huge difference. It went from blah to wow and the owner was floored by it. As I recall it was a Mercury, but I'm not certain after so many years. I remember how great it sounded when I got it fired up the first time.

[Colossal]

.
The AC30 was a big resto, and the transformer made a huge difference. It went from blah to wow and the owner was floored by it. As I recall it was a Mercury, but I'm not certain after so many years. I remember how great it sounded when I got it fired up the first time.

Do you happen to recall which era of Vox AC30 OT you used? Haddon, Albion, Woden? Just curious...

[martin manning]

Hammond's data sheets include lots of information that others do not, including leakage inductance, so I compiled a table of several of their Guitar Amp Replacements. The things that jump out are the extremely high number for the Bandmaster, and the differences between the various options listed for Fender Twin. The 1750WA is noteworthy as it is so much higher than the others, and it is the one that is listed for '65 Twin Reverb replacement. The data sheets include frequency, phase, and THD plots too.

[R.G.]

Good chart, Martin!

My morning-coffee mind lept to computing the ratio of primary inductance to leakage inductance...

Most lower-cost guitar amp specific OTs are no longer interleaved either at all or only P-S-P at most. Hifi OTs once had as many as ten - or more! - layers.

[martin manning]

My morning-coffee mind lept to computing the ratio of primary inductance to leakage inductance...

Thanks, and there you go. Should have done that the first time.