Characteristics of Guitar Amplifier Output Transformers, and criteria for design

[Gregg]

But, in this particular case, I do want to design for (relatively) wide bandwidth, avoid saturating the core, and operate at full drive output without breaking a sweat.

Most of the time guitar OT's can go down to 50 Hz at full power without problems and saturating the core is not so easy. Concerning the top end 15kHz is more than enough and is not difficult to achieve. You can use these as a start. In calculations shoot for -3dB. For guitar OT-1dB is overkill.


Couple of examples from real guitar OT's (and what you'll actually get from calculations at -3dB):

1/ 50W - iron sizes EI96x32-40mm,
primary 1600-1800 turns/ 0.25-0.3mm
secondary 60-68 turns/0.8-1.2mm
2/ 100W - iron sizes EI96x45-60, EI108x40-50mm, EI115x35-45mm
primary 1000-1200 turns/ 0.35-0.45mm
secondary 52-57 turns/1.2-1.5mm
The wire thickness for the secondary is for a single 4 Ohm winding which means that if you have two secondaries in parallel you must divide that by square root of 2 etc. All wire thicknesses are without insulation/lacquer.

After you get your numbers it's time to choose the winding configuration and there are many. Then you have to see what is the actual wire thickness (including the insulation/lacquer) and do some calculations how many turns will fit in one layer on your transformer bobbin so they can spread equally in all layers. Some corrections may be necessary but if you get to that stage we'll help.

[Malcolm Irving]

In a band context, with a Bass player, I think it sounds better when the low frequencies of the guitars are attenuated. Especially when there are two guitars, their low frequencies just muddy the sound and 'get in the way' of the clarity of the real bass line. (Keith Richards was famous for 'inventing' the five-string guitar by throwing away his low E string. )

For a solo jazz player, in the style of Joe Pass, a bit of hi-fi low frequency could be good, though.

[J M Fahey]

(Although, I'm really surprised at the 700Hz figure for voxes. I wouldn't have guessed that.)

I was also amazed when I calculated it

See for yourself: original VOX AC30 from the 60's



Check Bright Channel volume control and its coupling cap, at low volume you have C1 500pF and VR2 500k .
Use this handy online calculator:RC pad corner frequency upper and lower cutoff frequency calculation filter calculate time constant tau RC voltage power calculator capacitance resistance - sengpielaudio Sengpiel Berlin

Using those old series nominal values you get some 639Hz cutoff (at mild 6dB/octave, so you *still* have some Bass and Low Mids, just quite attenuated) , with modern normalized values it would be 470pF and 470k = 720 Hz. In practice same thing.

And that is with volume set low; when set to "10" capacitor also sees R9 and R7 to ground (depending on Normal Channel Volume setting, but which can be assumed is set to 0 if unused) so cutoff frequency starts at some 1200/1400 Hz

I BET Chris Jennings did not fire up his slide rule to calculate a cutoff frequency, but most probably had somebody playing his prototype LOUD and tried different capacitors until he found one which cleaned the distorted sound a lot .... tried and true design technique

FWIW revered Trainwreck amps are *basically* "a Fender amp with an extra tube for gain/sustain and a VOX type strong Bass cutoff for clarity"



Notice 0.002 coupling cap and 180k grid resistor at the third triode, which is also a cold cathode clipper, cutting below 440Hz.
The rest of the circuit is again a basic Fender with tweaked (Marshallish) Tone Control values.

Like at Mc Donalds, where they can offer a couple dozen "different burgers" using basic 6 or 7 "components".
Actually less, because they *always* need to use the bun and at least 1 patty

[Helmholtz]

Most of the time guitar OT's can go down to 50 Hz at full power without problems and saturating the core is not so easy.

True, but an OT being down 3dB@50hz at full power may be down 3dB@500Hz at low power.
Saturation headroom for a given core increases with the number of primary turns.

[Gregg]

True, but an OT being down 3dB@50hz at full power may be down 3dB@500Hz at low power.

I didn't quite get that?

[Helmholtz]

I didn't quite get that?

What is the specific question? (Maybe you missed my post #17)

[J M Fahey]

I think he´s baffled by the claim that at low power the OT cuts off Bass below 500Hz:

True, but an OT being down 3dB@50hz at full power may be down 3dB@500Hz at low power.
Saturation headroom for a given core increases with the number of primary turns.

something we do not experience when actually playing ... or even measuring amp frequency response on the bench.

A real World amp is as flat at 1W as at, say, 50% to 80% of full power (simply to put saturation out of the question).
yet I don´t doubt your experimental results, just think that a signal generator with its tiny weak output, most are designed to drive >600 ohms and, say, 1V RMS which means some 1.5 mA current capability definitely can not meet any magnetizing current requirements .... but any real World amplifier can pre-magnetize (consider it some kind of magnetic bias) output iron simply with power tube current imbalance ... let alone if it´s a Class A output.

Just guessing but find it a plausible explanation for the fact that actual Guitar amps do not show that huge Bass loss you mention.

[nickb]

Lp values this high are quite realistic in guitar amplifier OTs. But only around rated output power and at low frequenies (max. ac flux). I measured Lp of around 40 different OTs at 650Vpp and 50Hz and found Lp to lie between 60H and over 300H. But when measuring with an LCR meter @ 1kHz, the values were lower by a factor of 5 to over 10. (The reason for the different values is that the LCR meter measures at very low flux amplitudes where the ac µ is close to the initial permeability. The ac µ rises strongly with flux amplitude up to the onset of saturation.)

The results seem to indicate that the major design concern was to avoid excessive power loss from high magnetizing current at max. output, while the bass response at low output power may suffer. Of course NFB can take care of this.

I think the variation of permeability is something that is easily overlooked. Well, by me at least...

I did find a nice plot of permeability vs flux density here fig 27, which incidentally is a great introductory text on transformers (start here). Do take note that the x axis is a log scale so you only see the biggest lowering of permeability at very low flux densities. I doubt it has any audible effect in practice but does serve to illustrate that the simple transformer models need to be applied carefully.

[jmaf]

I didn't quite get that?

Iron cores are slow. As you increase frequency the core will be unable to react in due time and eventually power will get lost more and more easily.

3 dB loss means losing "half the power". So what he said is as you raise the frequency the point where you lose half your power to core saturation is lower and lower.

[Helmholtz]

I think he´s baffled by the claim that at low power the OT cuts off Bass below 500Hz:
True, but an OT being down 3dB@50hz at full power may be down 3dB@500Hz at low power.
Saturation headroom for a given core increases with the number of primary turns.
something we do not experience when actually playing ... or even measuring amp frequency response on the bench.

I admit that the example I gave is extreme and rather pessimistic. It was meant to sensitize to not only measure frequency response (and primary L) at rated power in cases where strong bass response matters. Of course the OT's bass response only directly shows without NFB. The magnetizing L of the OT decreases at lower currents and as the -3dB point depends on the ratio of reflected load to Lmag (Raa/Lmag), it will shift to higher frequencies at lower power. I think an increase of the OT's corner frequency by a factor of 3 at reasonably low levels is not an unrealistic assumption. NFB can compensate.

Imbalance means DC offset/bias and this further decreases primary/magnetizing inductance and results in a higher corner frequency.

[R.G.]

Just like the current gain of a transistor cannot be expressed as one number, neither can the primary inductance of a transformer or the permeatility of a loop of iron. The BH curve for a magnetic material is just that - a curve - and the "inductivity" of the material is the slope of that line at any given point to small signals. To large signals, the signal gets distorted by this constantly-changing loading.

The size of the primary inductance does indeed change with the size of the signal applied, and its frequency (that's a side effect I didn't get into in the preceding polemics ) and with any "operating condition" such as a smaller signal riding on a much lower frequency.

[Mike Sulzer]

The BH curve for a magnetic material is just that - a curve - and the "inductivity" of the material is the slope of that line at any given point to small signals.

It is not that simple. The observed effect is larger inductance at larger flux levels (higher transformer power). This requires an analysis that takes into account the movement around the hysteresis loop. I do not think that it is so easy to get a good intuitive understanding of why the inductance is higher at higher levels.

[Malcolm Irving]

I have on old textbook: 'Transformers for Electronic Circuits' by Nathan R. Grossner, 2nd Edition, 1983.

On page 8 it shows a family of B-H curves. For small values of B and H, the 'principal axis' (if I can call it that) of the B-H loop is rotated clockwise.

Also, on page 181 he states that: 'Initial permeability … at the instep of the curve is physically associated with domain growth and reversible boundary displacement'.

[SoulFetish]

on page 181 he states that: 'Initial permeability … at the instep of the curve is physically associated with domain growth and reversible boundary displacement'.

- goes without saying I'm sure.

But there's still one thing that confuses me in this whole conversation. What the hell does "'teaching my grandmother to suck eggs" mean?? Sounds very British, and with the last name Churchill, I feel like I should know this!

[SoulFetish]

I think the variation of permeability is something that is easily overlooked. Well, by me at least...

I did find a nice plot of permeability vs flux density here fig 27, which incidentally is a great introductory text on transformers (start here). Do take note that the x axis is a log scale so you only see the biggest lowering of permeability at very low flux densities. I doubt it has any audible effect in practice but does serve to illustrate that the simple transformer models need to be applied carefully.

[ATTACH=CONFIG]48899[/ATTACH]

Nice work, Nick! I just bookmarked those and am looking forward to the read.