Op-amp/Transformer Power Amp

[davidallancole]

Hi all,

I was fiddling around with a small power transformer the other day.  I hooked it up to an op-amp with a bit of gain and a +-15V supply.  I used my Boss MT-2 pedal for some distortion plugged into it and it sounds pretty decent.  I was quite surprised with the volume as well.  I haven't optimized this setup yet, but I could get it loud enough that it would P.O. my wife in the other room.

I ran some simulations of this setup as well.  With 4 op-amps I could get ~1 watt of power out of it into an 8 ohm load.

[brett]

Hi
this is quite interesting.  As well as a micro-amp, it might give some interesting mild distortion and a mids boost.
What op-amp are you using?  Most have a current-limiting output resistor (e.g. TL07X have a 128 ohm resistor). 
I might mention a couple of trade-offs that you face:
To minimise power losses R123 should be much lower resistance than the impedance of the primary of TX7 (there's also an interal current-limiting resistor in many op-amps e.g. 128 ohms on TL07X).  So that limits your current in the primary to 9V/(R123+128+Zprimary).  (It's less than 9V, but these are ballpark fighures).  If Zprimary is 1k ohms, that's 8mA. 
The good thing about a high Zprimary is that the voltage in the primary is high, and the resistors consume less power.  In the case above, Vprimary is approximately 9*1000/1248.  That's over 7 volts.
To get a reasonable amount of voltage and power out of the secondary (e.g. 1V = 0.125W at 8 ohms), the impedance ratio in T7 will need to be a moderate step down. Around 1K:8ohms (or 120V:12V for a PT) should be good.  That'll give a voltage step down of around 11 times (sqrt of impedance ratio) and about and the voltage to 0.6V and 0.1W at 8 ohms.  With an efficient speaker 0.1W can sound quite loud (85dB).
just my 2c...

[davidallancole]

The 120 ohm resistor I threw in because it was the dc resistance of my transformer.  On the sim it seems to drop the voltage by 3dB.  I don't have it in my test circuit, the output of the op-amp goes right into the transformer.

The op-amps I have tried are the TL071 and the LM741.  They both sounded fine.  I tested the transformer with my function generator.  I put in 7.5Vrms and measured 0.3865Vrms across a 6 ohm load.  If I remember correctly, the current measure to a little over 5ma on the primary side.  So the transformer is not optimized for this situation, but it definitely makes something useful.

With a more optimized transformer based off of the load, you can achieve (from sims) ~650mWrms

The circuit I am simming is posted below.  The two op-amp setup doubles the current, so assuming you are using an op-amp capable of source/sinking 35ma, you can get ~70ma.

[Transmogrifox]

You can put pnp & npn push/pull stage on the output of the op amp as a current booster.  You can also parallel these to get more juice. 

I think the rating on a 3904/3906 BJT is around 800mA. 

If you drive a 16 ohm speaker at 9Vrms the result is about 500mA current, equating to 4.5W power.  A single pair of standard small signal BJT's could easily get you there without the power transformer....however it looks like part of the effect is having the power transformer in there for the resulting filtering and distortion.

It's amazing how loud 1 Watt can be with a high-efficiency speaker.

[davidallancole]

There is definitely better ways of getting power to a speaker, but I just wanted to try this for the fun of it.  It seems neat to me that you can get enough power out of a single op-amp to be heard.

I have played around and built a few circuits with 3904/3906 pairs and they work very well.  I usually find I have the volume turned way down when playing on these still, so I am basically trying to see how low you can go.

[Transmogrifox]

I see the point now.  It is surprising how much sound you can get from 200mW.  There are some guys at my church who used the line out to drive the monitor speaker from the mixer because they didn't know what they were doing.  At low levels it was surprisingly loud and clean (this is probably just an op amp output).  The vocalists came to me wondering how they could get it louder.  Of course it got really distorted when you tried to turn the volume up.

My response was, "Erm...you need a power amplifier..."
So I gutted the power amp section out of a guitar amp a guy gave me and stuck it in a box.  It works a bit better that way.

While you're on the topic of trying to see how much sound you can get out of a low power device, you may consider trying to build a Class D amp with a CD4049 chip.  You could probably use one for power and the other for signal modulation.  Take a look at the MXR envelope filter for ideas of how to modulate the audio.  I have poked at the idea of making a headphone amp out of one of these.  Self-oscillating is probably the way to go.

[brett]

Hi

My response was, "Erm...you need a power amplifier..."

Classic.  You must have been choking on a laugh or two.
 
I would guess that the line out was from an opamp with +/-15V (maybe more), so the available power was a bit more than with 9V.

Two opamps (e.g. a TL072) in push-pull configuration with a 1K:8 ohm transformer would be cool.
cheers

[davidallancole]

Thanks for the responses guys.

I kind of ran out of time to experiment much for the next few weeks.  The wife had a baby Monday night.

I will have to check into the pulse width modulation.  I haven't studied it much yet.

I did have time for a quick sim.  Using +- 15V supplies, the TL072 in a balanced driver configuration with a 1k:8 transformer will give ~300mWrms and if you use the Radio Shack 120-12.6 transformer you get ~370mWrms.  I thinks that pretty damn cool.

[Transmogrifox]

Congrats on the baby.  I like the little critters.

I didn't look at the TL072 datsheet, but the classic LM741 output limiting current is 25mA.  With +/- 15V supply this is V*I = 3750mW peak power, which makes for about 250mW RMS for an undistorted sine wave.  This is single ended.

With the balanced driver config, it doubles, so you can actually get 1/2 Watt out of a dual op amp chip using the right transformer.  Assuming 25mA, you would need a 1.2k:8 (150:1)  transformer to optimize power transfer to the load....so your 1k:8 transformer is about right.

[davidallancole]

Thanks Transmogrifox.  I guess I have more time for this stuff than I thought.  Maybe some day in the future he will be perusing these boards as well.

I use PSPICE for my circuit simulations.  I insert a resistance in series with the transformer so that PSPICE doesn't get an error.  I make the resistance the same as the DC resistance of the transformer so this causes a voltage divider effect to happen, thus the power out the other side will be lower.

There are some easily available op-amps that have higher output current handling as well, like the NE5532/4.  They are rated to 38mA.  There is definitely a lot of options to play around with to get more and more power.

I was thinking about the pulse width modulation a bit.  For a few more parts, a pwm could be setup with an op-amp running into the transformer.  It should make for a more efficient amp.  I also though that the transformer might work as the filter eliminating the carrier frequency.  As I get some free time in the future, I will give it a try.

[Transmogrifox]

I recommend using higher speed chips like CMOS inverters and/or high speed comparators for class D.  To keep distortion at a minimum you don't want either the rising edge or falling edge to be more than about 50 nano seconds.  I tested a little low power class D amp using 2N700 MOSFET + 2N587 pnp BJT to drive my guitar amp speaker at approximately 1 Watt.   This was acceptably clean sounding.  Switching frequency was about 80kHz.  Most Hi-Fi amps switch in excess of 300kHz.  Many switch at 500kHz.  The rule of thumb is your switching frequency should be about 10 times the highest frequency of the input signal. 

For electric guitar, the pickups start to roll off the signal at around 5kHz.  An anti-aliasing filter at 8kHz is quite reasonable for nice sounding clean electric guitar.   There are higher frequency components in distorted guitar, but I have messed around with EQ's using Audacity to make a low pass brick wall at 6 or 7KHz on distorted guitar w/ effects.  The difference is too subtle for my loud-guitar-torn-ears.

That's a long way of explaining why I determined 80kHz was an acceptable frequency for electric guitar.

This low power Class D amp I tested on my breadboard was actually the Power MOSFET gate driver of a real high power Class D amp I'm making out of an old UPS Power output board.  It takes a respectable amount instantaneous high magnitude current to drive power MOSFET gates.

If I get some time I might do a simulation with a CD4049.

I know you're exploring possibilities with regular IC op amps.  The reason I discourage you from using them for Class D amps are the following:
1) They're too slow.  At 16V/us slew rate (for a fast one), your minimum rise time is about 1us (remember 50ns requirement).  Simply due to slow switching speeds you sacrifice sound quality AND efficiency since a respectable amount of time is spent switching from high to low and low to high during each switching cycle.
2) They're probably not more efficient.  Class D efficiency relies on MOSFETs for low ON resistance.  An op amp output is not really designed to have a low saturated resistance like this.   In fact, because of the way they're designed internally, they're probably less efficient when driven rail-to-rail.  Some rail-to-rail op amps may help you out here....but still, they're too slow.

You're already on the right track for powering speakers with regular op amps.

[davidallancole]

I would be interested in seeing a circuit using the CD4049 for a Class D amp.