Buffer/splitter advice needed

[BigBoy]

Hey, y'all, new member here.

I come seeking your advice, since I'm good at building tube amps, but haven't built a solid state device since my earliest work for the Army back in the mid 1980's!

I'm looking to build a buffered one-into-three splitter so a friend can run his signal into three amps.

So I looked at the design for the AMZ one-into-three splitter, shown here: http://www.muzique.com/lab/splitter.htm (I can't seem to figure out how to add pictures.)

That looks pretty straightforward. However, it seems to me (correct me if I'm wrong) that if this is run into three amps, there may be ground loop issues. If this is correct, could I just isolate each output by using the 1:1 little audio transformers from Mouser on each output? I see that in the design for an A/B/Y box at geofex.com= Improved Hum Free A/B/Y

I'd really appreciate any guidance y'all can give me. Like I said, I'm a tube amp guy, not a SS builder (yet--I want to try my hand at some stompboxes this winter).

Thanks a bunch
Rich

[grapefruit]

Hi Rich,

There's a chance you'll get hum with the AMZ JFET splitter. I'm not sure if you could just add a transformer to the output. It would depend on the transformer you use. Personally I would use an op amp to drive the transformer as the one at geofex does. You don't have to use a split rail supply, but if using a single rail supply I would go for 18 volts to get enough headroom.

Cheers,
Stewart.

[Artie]

You can go to Jensen Transformers, and look to the left side, under "Applications", click on "Schematics". Scroll down pretty far to the "Musical Instrument Application Schematics" catagory, and look at AS013 and AS014. Especially AS014. Those use the more expensive Jensen transformers, but there shouldn't be any reason you couldn't substitute any transformer you want, (asuming you keep the specs close). You may take a performance hit, but at least it shows the basic idea.

[R.G.]

I'm looking to build a buffered one-into-three splitter so a friend can run his signal into three amps.
So I looked at the design for the AMZ one-into-three splitter, shown here: http://www.muzique.com/lab/splitter.htm
That looks pretty straightforward. However, it seems to me (correct me if I'm wrong) that if this is run into three amps, there may be ground loop issues. If this is correct, could I just isolate each output by using the 1:1 little audio transformers from Mouser on each output? I see that in the design for an A/B/Y box at geofex.com= Improved Hum Free A/B/Y

Hi, and welcome.
Your instincts are good. The AMZ splitter will almost certainly give you hum problems with amps. At best it will do an OK job splitting signals within a pedalboard, but not to amps. If the amps are well constructed, and have properly set up third-wire grounds, you can probably make do with the opamp-only version of the splitter. That will be fine for hum voltages of only a volt or two, and well-grounded amps will usually be in that range.

If one or more of the amps is a favorite old two-wire death... er, sorry, relic  then the hum will sometimes, in some settings, be big enough to need one of the transformer coupled setups. The transformer setups can get rid of tens of volts of AC hum.

...
Those use the more expensive Jensen transformers, but there shouldn't be any reason you couldn't substitute any transformer you want, (asuming you keep the specs close). You may take a performance hit, but at least it shows the basic idea.

Tee hee! Yes, that's absolutely true - just like a new operating system release is a simple matter of programming. 

The Jensen page was the thing that kicked off a lot of the thinking that went into my transformer isolated article. The reasons the article didn't just say "just go to the Jensen page and do that," was that:
(a) the Jensen transformers are pretty expensive. You might want to price them. I did. Gulp.
(b) 'any transformer you want, (assuming you keep the specs close)' is a bit of an understatement. There don't seem to be many transformers with near-equal specs to the solid-platinum Jensens that don't also cost like them.

The thrust of the GEO article was to get good-enough results with a $3.00 transformer and a buck or so in parts for the DIY minded.

[Artie]

'any transformer you want, (assuming you keep the specs close)' is a bit of an understatement.

True, true. I've just recently acquired a Tamura transformer (MET-09) with similar I/O specs as the Jensen JT-11P-1 that Digi_key has for around $14. Here's the spec sheet in PDF format from Tamura's webpage: http://www.tamuracorp.com/clientuploads/pdfs/pg30.pdf

Note, that I use the term "similar", very loosely. I'm going to be doing some listening tests with it this week. I'll post back results.

RG; Do you have a link to your transformer article? I'd love to read it.

Thanks;
Artie

[R.G.]

True, true. I've just recently acquired a Tamura transformer (MET-09) with similar I/O specs as the Jensen JT-11P-1 that Digi_key has for around $14. Here's the spec sheet in PDF format from Tamura's webpage: http://www.tamuracorp.com/clientuploads/pdfs/pg30.pdf

Note, that I use the term "similar", very loosely. I'm going to be doing some listening tests with it this week. I'll post back results.

I have to say that getting anything even close to the Jensens for $14 each is a great deal indeed. Last time I looked the bar stood at about $40.

RG; Do you have a link to your transformer article? I'd love to read it.

Hmmm... as it turns out, me too. I did a look for it and the last iteration of updates seems to have destroyed either the link or the article itself.

I'm pretty sure I typed in the gist of it here at least once.

But in short:
There are four things you have to watch for in audio isolation transformers:
1. low frequency response, which tends to be isolated in the primary inductance
2. high frequency response, which tends to be isolated in the leakage inductance and self capacitances
3. core distortion and saturation from the use of ferromagnetic materials
4. miscellaneous effects: hum and noise pickup, extraneous resonances, bridging capacitances, longitudinal balance, insertion losses, etc.

For low frequency response, the trick is to get the primary inductance large enough to get the low frequency rolloff you want. That translates to making the primary inductance be equal to the reflected secondary impedance at the -6db (voltage) point for your chosen frequency response. There is a lot of specsmanship involved, and some transformers are specified at -3db, -1db, etc. It all comes down to specifying the primary inductance, and that sets the amount of iron and copper needed for the core and primary windings. This is THE primary thing to do in a transformer design first. Most of the rest depend on it.

For high frequency response, you have to control the leakage between primary and secondary windings and the self capacitances. The big one is leakage inductance. You do this by interleaving and/or sectionalizing the primary and secondary windings.

Core distortion is a result of the nonlinearity of the magnetic properties of the core material itself. All ferromagnetic materials are more or less nonlinear. You minimize this by (a) using small flux density excursions (which makes the core large, needs more copper, and drives expenses up),  (b) driving the primary from a low impedance, and (c) picking the most linear core materials, which also tends to be expensive.

Finally there is the miscellaney. Hum and noise pickup you cure by shielding and isolation; extraneous resonances are usually above audio, but you need to test for them; balance is critically important in balanced line audio work, but for our single-ended stuff, not even on the radar. Insertion loss and so on are fine points which are not of any critical importance to effects work.

What I did was to note that Jensen had some good ideas for people who could afford them, and look for cheaper alternatives. I did the obvious and ordered a selection of cheap transformers and measured what they would do. The Xicon cheapos from Mouser were specified for response (which usually means -6db point if they don't say anything else) were noted to be 300Hz to 3.4khz. They actually did mid 200's to well over 20kHz, above which there was a resonant peak and then a big falloff. The midrange was flat, as you'd expect.

Fortunately, in a low power application, I could afford to spend power to get range extension. The low end response was with matched drive impedances.  I mismatched the drive impedance by drivng it from an opamp output. This let me pour enough current into the primary to satisfy the sagging impedance of the primary inductance and still have the voltage be constant with frequency all the way down to 60-something Hz, under guitar frequencies even for drop-D tuning. At the top end, I got flat response out to 20kHz with that hump above audio. I decided not to mess with trying to tame that, as they might easily change that on different runs of the transformer.

So for the cost of a $0.30 opamp and a sub-$3.00 you get all the response range you need plus isolation. And to me, that was good enough for rock-n-roll.

[Artie]

Thanks R.G. I appreciate you taking the time to type that all out. Even though I've spent the last 40-something years involved in electronics, transformer theory still eludes me. Voltage in/voltage out, ditto for power, is easy enough to understand, but reflected impedances and "stuff" are still a gray area. If you look at that PDF document that I linked to, would you say that the MET-01, with a 200k/1k I/O impedance might be better for guitar work?

I may just buy one to experiment with.

[R.G.]

Thanks R.G. I appreciate you taking the time to type that all out. Even though I've spent the last 40-something years involved in electronics, transformer theory still eludes me. Voltage in/voltage out, ditto for power, is easy enough to understand, but reflected impedances and "stuff" are still a gray area.

See if you can find a copy of Nathan Grossner's "Transformers for Electronic Circuits". It's out of print, but it's a good general text on why transformers act like they do.

If you look at that PDF document that I linked to, would you say that the MET-01, with a 200k/1k I/O impedance might be better for guitar work?

I think you'd be disappointed. Tamura only guarantees 300hz low end response on those, so you're back and needing an active driver to drive the bass response down. They do call for 100kHz, which probably means the ultrasonic flatness is good; problem is, my ears only get out to 13kHz these days, and I don't think I ever got over 20KHz, so the high end doesn't do you much good.

The MET-01 with a 200K to 1K impedance ratio is a 14:1 voltage step down, a 23db signal loss. If I was going to pick one of those to play with, I'd get a MET-09, 10K:10K impedance and 1:1 voltage ratios. Also 40mW versus 10mW for the -01. 40mW into 10K is a voltage on the primary of 20Vrms, good enough for what we play with.

On this one, the opamp driver should work well; you'll still need it for bass extension.

[BigBoy]

Hey, y'all, thanks for all your input. I guess the short answer is: yes, I can put the transformers on the outputs of the jfets for the AMZ design. When I get it going, I'll let y'all know how it's going.

Cheers,
Rich

[Artie]

Hey, y'all, thanks for all your input. I guess the short answer is: yes, I can put the transformers on the outputs of the jfets for the AMZ design. When I get it going, I'll let y'all know how it's going.

Cheers,
Rich

Cool. Curious to see how this turns out.

I think you'd be disappointed. Tamura only guarantees 300hz low end response on those, so you're back and needing an active driver to drive the bass response down.

I did a small test, of sorts, in the lab. (I work in a Navy calibration lab.) I ran a signal generator into the input, set for 350mv's rms, (which is similar to a guitar output), and ran the transformers output into a 'scope. The 'scope has a 1M input impedance, so its close to a typical guitar amps input Z. I swept the signal from 20khz down to 0. The amplitude didn't start to roll off until I got down to around 30hz, and even then it wasn't much. Maybe 1 or 2 db. It didn't roll off steeply 'til about 10hz.

Luckily, it is the MET-09 that I bought, so I'll continue my tests with it. 

[R.G.]

I used to have access to well stocked labs... sigh...

The trick here is to find out what your signal generator driving impedance is. If it's a good, low impedance source, it's doing the opamp trick for you. Stick a 10K resistor and a 2H inductor in series with the signal generator to fake a humbucker and try that again.

'Course, it's always possible that the Tamura transformers are just way better than they specify them at, too.

[Artie]

I used to have access to well stocked labs... sigh...

And I get payed to play there.   

The signal gen has a 50-ohm output. (Most of out test equipment does.) I'll try the trick with the inductor and resistor. Thanks.