Continued sidetrack discussion from "Tiniest Portable amp I can build" - nikongod microtransformer-based impedance step-down box

This thread continues a side discussion started in:
 
Tiniest Portable amp I can build
 
So, I was out running errands, drove by Radio Shack and figured I would pick up the parts to make nikongod's micro-transformer impedance step-down box. $12USD and change later:
 

 
I went with the DE-luxe phone jacks, otherwise, I could have saved three bucks and gotten the cheap versions they stock and this could have been attempted for under $10USD. This is one of the simplest builds ever. No additional parts are required as the transformer wires are plenty long enough. Even the ends are tinned out of the box. Ten minutes later and we are in business:
 

 
So how does it sound? Not too bad. The micro-transformers are pretty much bandwidth limited (rated: 300-10,000Hz +/- 3db) and roll off on the top end pretty hard, but, hey, what can you expect from a $3 part this freaking small? The mids are very nice and the noise floor is contained. If you have bright phones, this will certainly tame the top end a bit.
 
But the comments on the sound miss the point. Through my Image 5's, I was able to pretty much max out an iPod touch in volume and still have PLENTY of output. At full, they were far louder than I would normally listen. All of this goes directly to nikongod's main point, most modern devices have PLENTY of gain inherent in their design. Using this technique you are able to have ZERO digital attenuation and ZERO analog attenuation and have an acceptable listening level (at least for most folks and most IEM's). The entire signal is allowed to come though unabated, nothing is thrown away prior to hitting the transformers. The source sees what it likes to see and the phones see what they like to see. The trade off made in voltage gain is distributed nicely to current gain and everything maintains a very nice level of control. Very cool. No wonder active unity gain buffers are so popular for all but the most demanding of headphones. You get the signature of the design without loosing any precious signal. Certainly a path I am going to explore a bit more. I might even pick up a couple of better transformers and see how far we can take this implementation.
 
Anyway, thanks to nikongod for the concept, execution and low-cost/high-return nature of the project. It was fun and quite informative. I have to wait until I finish my Altoids and then I will box this puppy up.  
 
 
 
 

Transformer specs for those interested:
 

glad you like it
 
As a point to (hopefully) save face, the transformer bandwidth is measured a rated power - these little guys are toooooo small to have a hope with 1W . At the tiny power levels used by headphones they have MUCH better bandwidth that advertized. I'l measure them again tomorrow. 

Oh fun! Now I actually miss the day when there is a radioshack near the house.

So I remeasured these just a second ago. 
 
All measurements taken on the "full" primary (blue to green) and with a 1vp-p input signal 
 
 
Its worth repeating, these are the -1db points (-10%) NOT -3db. 
NOTE: on 32r, the transformers DO reach 20hz just before the -3db point, but they are clearly saturating and looking ugly. With a smaller input signal (I tried at 200mvp-p) they reach 20hz clean, although still about 3db down.I didnt bother checking with the other 2 impedances.
 
About the only thing that goes below 60hz is organ music. IME most bass in pop music is in the high 50/low 60's. 

Quote:

jdkjake said:

But the comments on the sound miss the point. Through my Image 5's, I was able to pretty much max out an iPod touch in volume and still have PLENTY of output. At full, they were far louder than I would normally listen.
 

Click to expand...

Is that the Klipsch Image X5?  I can't imagine having the iPod at full volume with these.  I've occasionally plugged mine into the iPod when the volume was set at 50% and that was way too loud.
 
 
Quote:

nikongod said:

So I remeasured these just a second ago. 
 
All measurements taken on the "full" primary (blue to green) and with a 1vp-p input signal 
 

1vp-p input	32ohm	12.4ohm	8ohm
-1db high frequency	250Khz	55Khz	45Khz
-1db, low frequency	50hz	50hz	45hz

Little radio shack transformers

 
Its worth repeating, these are the -1db points (-10%) NOT -3db. 
NOTE: on 32r, the transformers DO reach 20hz just before the -3db point, but they are clearly saturating and looking ugly. With a smaller input signal (I tried at 200mvp-p) they reach 20hz clean, although still about 3db down.I didnt bother checking with the other 2 impedances.
 
About the only thing that goes below 60hz is organ music. IME most bass in pop music is in the high 50/low 60's. 

Click to expand...

Isn't the line out voltage of an iPod 1V?  From using test tones, I know that none of my headphones accurately, or in most cases audibly, produce 20Hz at my listening levels. 
 
I know from experience that the iPod classic doesn't like music that has a lot of deep lows.  I have a solo organ CD that clips on the iPod.  It plays fine on my computer.
 
 

scompton said:

Is that the Klipsch Image X5?  I can't imagine having the iPod at full volume with these.  I've occasionally plugged mine into the iPod when the volume was set at 50% and that was way too loud.

Click to expand...

Yes, they are Klipsch Image X5's. Normally, when plugged straight into my iPod, I never get close to 50% volume. However, going through the transformers, there is enough of a reduction that the 100% setting is close to the 50% setting without the transformers. The X5's have an impedance of 50ohms, so, I am sure lower impedance IEM's will have a bit more travel. Regardless, it is plenty loud.

nikongod said:

So I remeasured these just a second ago. 
 
All measurements taken on the "full" primary (blue to green) and with a 1vp-p input signal 
 

1vp-p input	32ohm	12.4ohm	8ohm
-1db high frequency	250Khz	55Khz	45Khz
-1db, low frequency	50hz	50hz	45hz

Little radio shack transformers

 
Its worth repeating, these are the -1db points (-10%) NOT -3db. 
NOTE: on 32r, the transformers DO reach 20hz just before the -3db point, but they are clearly saturating and looking ugly. With a smaller input signal (I tried at 200mvp-p) they reach 20hz clean, although still about 3db down.I didnt bother checking with the other 2 impedances.
 
About the only thing that goes below 60hz is organ music. IME most bass in pop music is in the high 50/low 60's. 

Click to expand...

Interesting to see the transformers measure so well. Especially on the top end. The sound I am getting is definitely seems rolled-off on the top end when compared to the direct out of the iPod. Could be the transformer is just calming out some treble spikes? Perhaps the impedance curve of the X5's prefers something above 8 ohms to remain linear. I'll have to break out the Grados and see how they play as well. They generally sound pretty good right out of the iPod.

BTW, what did you use to measure the response? A frequency generator and scope or a computer based solution? Just curious.

Well it has to be asked: what's the next step up for better transformers? (in the same size?)

 
Quote:

scompton said:

Isn't the line out voltage of an iPod 1V?  From using test tones, I know that none of my headphones accurately, or in most cases audibly, produce 20Hz at my listening levels. 
 
I know from experience that the iPod classic doesn't like music that has a lot of deep lows.  I have a solo organ CD that clips on the iPod.  It plays fine on my computer.
 

Click to expand...

I think most portables have a 1v line out. The ipod does have coupling caps on the line out, and you cant adjust the line out volume, so Id probably go for the headphone out.  100mv into an IEM is still pretty loud.
 
I also have trouble getting "real" 20hz out of any of my headpohones. Once you pass about 50hz on speakers I find I feel it in my body more than hear it.
 
Quote:

jdkjake said:

Interesting to see the transformers measure so well. Especially on the top end. The sound I am getting is definitely seems rolled-off on the top end when compared to the direct out of the iPod. Could be the transformer is just calming out some treble spikes? Perhaps the impedance curve of the X5's prefers something above 8 ohms to remain linear. I'll have to break out the Grados and see how they play as well. They generally sound pretty good right out of the iPod.

BTW, what did you use to measure the response? A frequency generator and scope or a computer based solution? Just curious.
 

Click to expand...

It could be some funky distortion. I dunno. Im pretty sure they measure this way because they are being run so far below rated power. At 1w I doubt they measure this well. Seriously, I have no idea how anyone would even think to put 1w through these.
 
 

 
I used a signal generator & oscilloscope to test these. Im pretty bad with computer based test stuff.


Quote:

armaegis said:

Well it has to be asked: what's the next step up for better transformers? (in the same size?)

Click to expand...

Im not sure if there is one - at the same size
 
The worst thing about these is the low end frequency response, which is at least partially caused by the physical size of the transformers. 
 
I think things will be better there with larger transformers, but portability will certainly suffer. 

Thank you for doing this I was very interested when this was mentioned.
Also would this work better with a transformer rated for a wider frequency range.

On paper, of course
In reality, maybe only a little bit. 
 
The problem isnt building a better transformer; the problem is building a better SMALL transformer. 

So the next step is a DIY transformer...

Yepp. The guy I bought my oscilloscope from was into winding DIY transformers for a solid-state guitar amp he was working on. REALLY cool stuff, but a totally insane-awesome level of DIY.

looks like you could series-parallel 4x of the RS xfmr per channel and still fit the Altoid tin - cutting the excitation V by half should give big distortion benefits since saturation effects are very nonlinear
 
better core material could help but diy could be tough - the windings use insanely fine wire