Transformer Distortion Project

[PRR]

> If ....you know the DC resistance ...and the inductance

Both given on the datasheet I linked.

DCR given explicitly. Inductance has to be teased-out from other specs. The response with 470r source is +/-1dB @ 600Hz, +/-4dB @ 300Hz. I say the "+dB" spec is only to cover the high frequency resonance, which will be above the guitar band. And the extra dB in "-4dB" is to cover general slop. If we *assume* -1dB @ 60Hz, -3dB @ 300Hz, with 470r source, we get just around 0.2 Henry.

Actually we should include the 108r Pri DCR. It is really tested with 578r source and Henries must be a bit bigger.

The inductance will vary with drive level (part of why it distorts).

If we drive with a zero-Ohm source (loudspeaker amp), we expect the "-3dB @ 300Hz" point to shift down roughly as 108r/578r or around 56Hz. Similar principles led R.G. to his transformer ground-lift giving good bass using transformers only somewhat better than modem lumps.

A low source impedance also makes inductance variation relatively moot, decreasing distortion..

> Digging into the Whitlock

Much of what Whitlock teaches is right for hi-fi but wrong for "transformer distortion project". Where he likes low-low source impedance and dead-zero DC, you want significant impedance and play with some DC.

[Phoenix]

One of the characteristics of transformer saturation is that it acts like a dynamic high-pass filter. PRR alluded to this already.
As the core becomes saturated, the inductance falls, so bass response drops as you get louder. Normally the way to get to this saturation point is to drive the transformer harder. But, as PRR also mentioned, you could get the same sorts of results with (unbalanced) DC current.
So, if you wanted to get really fancy, you could put together an envelope follower like found in compressors, noise gates etc. and use that to inject a dynamic DC current into the transformer primary. This way you could also alter the gain on the envelope to control the level at which saturation begins. Keep in mind though that this saturation high-pass is pretty subtle, you'd have to really crank that control voltage to get very pronounced effects.

[TejfolvonDanone]

Why don't you try to fiddle with a transformer coupled class A amp?
This idea crossed my mind before but (as with a lot of other ideas) it didn't even made to the breadboard.

[Rob Strand]

One of the characteristics of transformer saturation is that it acts like a dynamic high-pass filter. PRR alluded to this already.

A transformer also acts as a low-pass filter.   

I honestly think good a sounding transformer distortion relies on this to remove the objectionable high frequency "junk".   How much "junk" present in the first place depends on how soft the clipping or saturation is in the first place.

From an experimental point of view it is hard to find a "magic" transformer that work just right but from an engineering point of view it is possible to manipulate and experiment with the cut-off frequency.

[Phoenix]

One of the characteristics of transformer saturation is that it acts like a dynamic high-pass filter. PRR alluded to this already.

A transformer also acts as a low-pass filter.   

I honestly think good a sounding transformer distortion relies on this to remove the objectionable high frequency "junk".   How much "junk" present in the first place depends on how soft the clipping or saturation is in the first place.

Yes, that can be true (but not always - we often see conjunctive filters or Zobel networks added to transformer circuits to address unwanted high frequency response), but it is not a (very) dynamic effect - generally it is effectively a fixed low-pass. It may shift some with the varying output impedance of the driving stage, but this usually is well above the frequencies of interest, it does not shift nearly so much as a high-pass response can in response to saturation.

[PRR]

> As the core becomes saturated, the inductance falls

Inductance also falls at low signal level.

Transformers are very complicated beatsts.

[Phoenix]

Inductance also falls at low signal level.

Transformers are very complicated beatsts.

And now you're getting into hysteresis curves, that's cheating!

[Rob Strand]

Yes, that can be true (but not always - we often see conjunctive filters or Zobel networks added to transformer circuits to address unwanted high frequency response), but it is not a (very) dynamic effect - generally it is effectively a fixed low-pass.

Yes definitely!   I guess the point is if (for some circuits) the "magic" of using a transformer is only the low and high pass filtering then you are better off using electronic filters.  The transformer isn't adding any real magic and it far less hit and miss.  The low pass is generally 2nd order but in some case it might as well be 1st order as the second pole (time constant) is way above audio.

[Rob Strand]

And now you're getting into hysteresis curves, that's cheating!

The funny thing is series caps driving into clipping diodes have a similar variable high-pass cut-off effect.  See Aron's Shaka pedal.

[ashcat_lt]

> As the core becomes saturated, the inductance falls

Inductance also falls at low signal level.

Transformers are very complicated beatsts.

How long does the signal need to stay big (or small) for this to happen?  Out signals get big and small many times per second.  Is inductance changing at audio rate?  I can see where that might cause some distortion.

[PRR]

> Is inductance changing at audio rate?

Yes.

[ashcat_lt]

> Is inductance changing at audio rate?

Yes.

!!! 

And it follows something like a parabolic curve?

[D_Ex_Patria]

Hey everyone,

Just an quick update: I've breadboarded a new 386 and hooked everything up, and I'm experiencing some major squealing, way more than I experienced when just hooking it up to a soldered version. I also blew a cap by mixing up my power supply wires. I just finished putting everything away out of frustration, but when my blood simmers down, I'm going to get back at it.

I've heard that the 386 is really sensitive to any distance between things connected to it, so I may try soldering up a barebones model where I can swap out that key resistor/cap combo without leaving too much space between components.

I've also have some goodies coming in from Digikey, including a different brand 386 and another transformer, so I'll keep playing once they arrive.

[Rob Strand]

One think I forgot to mention is you can drive a transformer two ways:   current and voltage.

The method with a power amp and a cap is voltage drive.
The ITP preamp is current.
A third way is  power amp with a series cap and a resistor which is "in between"; in reality voltage drive has some resistance.

It will change the behaviour.

[D_Ex_Patria]

Progress!

I'm still poking around the LM386-LM380 setup to try to kill the weird noises, but in the meantime, I connected the transformer as a direct go-between from my my mp3 player to my audio interface.

This gave me a chance to put some white noise through the component and try out some RC values as we talked about, and I think I have a pretty good idea on where the transformer is sitting Henries-wise.



I was using the exact schematic from http://sim.okawa-denshi.jp/en/RLCtool.php with R=240Ohms and C=100nF. Playing around with the RLCtool, I'm going to guess that the transformer is sitting at about 0.16 Henries. Swapping out for a 10uF acted pretty much as expected, giving me a sharp roll-off south of about 125Hz.

I even played around with my bass guitar plugged in through the setup. It gives me a reasonable tone, though the bass starts to pick up hum from being so close to my computer.

Since I'm mostly messing about with a bass guitar, I'm going to pick out some bigger caps and see if I can't get an even lower bass response.

[snap]

Found this simple approach - http://www.diystompboxes.com/smfforum/index.php?topic=100258.0 - with seemingly only 2 basic components - http://www.diystompboxes.com/smfforum/index.php?topic=100267.0

[D_Ex_Patria]

Found this simple approach - http://www.diystompboxes.com/smfforum/index.php?topic=100258.0 - with seemingly only 2 basic components - http://www.diystompboxes.com/smfforum/index.php?topic=100267.0

It's definitely similar, and it's definitely simple. I'm hoping for a bit more magic with the distortion the 386 already adds. 

[amptramp]

Think of a transformer being able to sustain a certain maximum level of volts x seconds in the core.  Lower frequencies give more seconds, so the low frequencies will be affected first.  But like most clipping designs, once you get into saturation, the highs and the lows are both reduced because the highs are riding along on the low waveform and they both get squashed.  If you bias the transformer with DC current, you tend to approach saturation in one direction before the other which tends to favour the creation of even-order harmonics.  Without the DC, you tend to get odd-order harmonics which are not as desirable as even-order.

[ashcat_lt]

Think of a transformer being able to sustain a certain maximum level of volts x seconds in the core.  Lower frequencies give more seconds...

But how long is that window?  If it's long enough, everything adds to 0 anyway.

[amptramp]

Think of a transformer being able to sustain a certain maximum level of volts x seconds in the core.  Lower frequencies give more seconds...

But how long is that window?  If it's long enough, everything adds to 0 anyway.

You are counting the volt x seconds in one direction then in the other.