suitable transformer for an audio isolation transformer?

[Rodgre]

I have been using an Ebtech hum eliminator to break a ground loop in the headphone send system at the studio. I would like to build a DIY version of the Ebtech box and wondered if anyone had any recommendations for a decent 1:1 audio transformer.

Roger

[R.G.]

They're expensive. A good, wide range transformer for use in a purely passive circuit is $50 and up, and will have loading issues running directly from a guitar.

The active/transformer isolator circuit in the geofex hum-free isolator is about the biggest bang for the buck I've found, if I do say so myself. The transformer is under $3.00, always in stock at Mouser, and the rest of the circuit only costs a buck or two for parts.

Mouser has recently started carrying some Triad audio transformers for about $4-$5 that look to have better specs than the Xicons, and will drop right into the Geo circuit.

[moosapotamus]

I've read that Edcor traffos give pretty good bang for buck... http://www.edcorusa.com

~ Charlie

[R.G.]

Hey, Moos! Thanks for mentioning that.

I gave the Edcor site a look, and they do have some pretty good values. Like the PCB industry, the transformer industry is getting more in tune with small quantities and agile distribution. I hadn't looked at their stuff in quite a while. What's new to me is that the transformers cost about half what I remember from the last time I looked, and they sell them with no minimums on line. Neat stuff!

For an isolator, I'd probably go with the PC10K/10K, which claims 20-20kHz with less than 1db variation. That's strong stuff. A guitar would still need a buffer to drive it to be absolutely sure it didn't lose treble, but this transformer doesn't need the low end boost. It's 1.3" by 1.64" (that's 33.8x41.6 mm to the sane world), so it's a little bigger footprint, but the specs look great and it sells for $6.37.

In fact, if it pans out to nearly the specs it claims, it should be a dandy replacement for all guitar isolators.

I was so intrigued that I decided to buy a couple of the PC10k/10ks, but it was two minutes after they closed ... 

I'll order a couple of them tomorrow for a trial.

[Rodgre]

Thanks so much for the tips!

Does it matter that I am trying to isolate a line level signal, not a guitar level?

Roger

[R.G.]

Only a little. The Edcor stuff is intended for line level, and if you are driving it with something that can drive a 10K load by itself, you could probably use just the isolator transformer. I'm pretty amazed at the specs. They have 600:600 transformers as well if you're running low impedance lines.

[Rodgre]

You won't believe this, but I whipped up a quickie test with a pair of the cheapo tiny Radio Shack audio isolation transformers and it worked. Not only that but the tone was not anywhere near as bad as I expected with their stated 300hz low frequency cutoff. There was plenty of bottom end.

I don't see these on their web site currently, so maybe they don't sell them anymore.

Roger

[Rodgre]

Spoke too soon: here they are for $3.99:



http://www.radioshack.com/product/index.jsp?productId=2103994

[R.G.]

I ordered a couple of Edcor's PC10K/10K. They specify them at 20Hz-20kHz +/- 1db, which is amazing if true. $6.36 each, and they happily sell direct and sell in one at a time. They also use the USPS fixed rate priority boxes, so the shipping costs are low on numbers under about 25 to 50 of each.

Good to know the RS transformers work for you. I finally wrote Radio Shack off entirely. I consider them to have nothing I want, so it's good if they can surprise use now and again.

[Paul Marossy]

I ordered a couple of Edcor's PC10K/10K. They specify them at 20Hz-20kHz +/- 1db, which is amazing if true. $6.36 each, and they happily sell direct and sell in one at a time. They also use the USPS fixed rate priority boxes, so the shipping costs are low on numbers under about 25 to 50 of each.

Good to know the RS transformers work for you. I finally wrote Radio Shack off entirely. I consider them to have nothing I want, so it's good if they can surprise use now and again.

Those are pretty good specs. Jensen makes some that have similar specs, but they want like $100 for just one. $6.36 is way cheaper!

I wrote off RadioShack about 3-4 years ago now. They have hardly anything in their stores anymore and what they do have is way overpriced. They are a handy source for little 12V transformers and small DC fans, though.

[head_spaz]

Have you had a chance to test those Edcore's yet R.G.?

Could you use a transformer like that to make a balanced output from a passive guitar?
And if so, would a higher turns ratio be better? Something like 15K/600 maybe?

THX

[R.G.]

Have you had a chance to test those Edcore's yet R.G.?

They shipped them USPS and they're not here yet. 

Could you use a transformer like that to make a balanced output from a passive guitar?
And if so, would a higher turns ratio be better? Something like 15K/600 maybe?

That's a problem. A guitar pickup has an internal impedance that runs up to about, very roughly, 100K at the top end because of the inductive nature of the pickups. Loading that down with a 100K load will lose half your treble in the 6kHz to 7KHz region. Lower impedances are worse, and this is the origin of "tone sucking". It is quite difficult to get both low frequency response and high impedance in a transformer. A passive transformer on a pickup will nearly always be a compromise. But you might find one that you like well enough. This line of reasoning is one of the reasons I like to buffer a pickup before transformer matching.

[head_spaz]

Well... you've just confirmed my reservations.
Buffer it is then.
What about the impedence ratio when using a buffer?
I've been pondering using a FET input stage -> NE5532 -> transformer.
Or... perhaps an OPA2132/4 -> transformer.
Or... how about an OPA2132 -> DRV134 and skip the transformer?
WWTD? (What Would Tesla Do?")

Thanks R.G. ... for all that you do.

[intripped]

sure I miss something...

but if the transformer is driven by a buffer, why couldn't it be a "line" transformer? (with 600 ohm input impedance)?

[R.G.]

but if the transformer is driven by a buffer, why couldn't it be a "line" transformer? (with 600 ohm input impedance)?

Maybe it could, if everything else was right. A lot depends on the hidden details.

Transformers are specified in terms of their target impedances, such as 600 to 600 ohms. They also have some other specifications on resistance of the windings, insertion loss, and so on, but the biggest one is probably the frequency response.

600 ohm line transfomers are often designed to a 300Hz to 3kHz frequency bandwidth. What that really means is that when loaded by 600 ohms resistive, they present approximately a 600 ohm load to whatever is driving their primary, and are reasonably flat in frequency response from at least 300Hz to 3kHz. The manufacturer usually says and guaratees nothing about what it does outside the specified frequency range.

But we can calculate what would happen if the transformer was **exactly** as good as the frequency range and other specifications say it is. Generally, the cheaper, smaller transformers are much better on the top end than they have to be. However, the low end is determined by the amount of iron used for the magnetic core, and the amount of copper and care in winding it that was used on the primary. The iron and copper are the primary cost materials for the transformer, so the manufacturer will not give you a whole lot more than the low frequency specification says he has to.

The thing that affects the low frequency response most is the primary inductance. The primary inductance has an impedance that decreases as frequency goes down. The transformer action makes the transformed secondary load appear across the primary, but the primary inductance is always there in parallel with the transformed secondary load. When the impedance of the primary inductance drops to equal the transformed secondary load, it is then "eating" as much signal current as is going across to the secondary load. If the transformer is being driven by the specified primary source impedance, then at this frequency, it is losing half its drive power to the primary inductance, and so this is the "half power" point which is normally taken as the low frequency cutoff. For a reasonably matched source driving the primary, at this frequency the secondary power response is half of what it is at the middle of the frequency band.

If you have a buffer available, you can keep the secondary output voltage from dropping off so quickly by driving the primary harder. If the buffer can deliver ten times as much current to the primary as a matched driver would, then it can keep the secondary voltage up to "flat" by just accepting that 9/10 of the primary current will be lost, but the remaining 10% of the primary current will drive the secondary OK.

And that's why a 600 ohm line transformer might not work. It starts with a small primary inductance. We're using a buffer to make up for the decreasing impedance of the primary inductance. It's easy to get into situations where you don't have a big enough buffer to make a 600 ohm transformer work. It can be done, perhaps by using a power amp chip, but you're well into diminishing returns. And heating both the transformer and buffer to try to make things come out right. So - maybe it would work; it depends on the transformer, the buffer, and how hot things get.  It's just easier to use a different transformer, given that many 600-600 ohm transformers are intended for modem duty and not all that good for normal audio to start with.

It's quite different if what you use is a wide bandwidth pro-audio specified 600:600 line transformer. That would  be much easier, as it already has a wide bandwidth. The trick is to make cheap semiconductors make up for a cheaper transformer. Good pro audio line transformers are both big and expensive. It's all in the details.

[trixdropd]

So what is the verdict on the edcor RG?

[intripped]

thanks R.G. for your answer!!!
Now it's clear that the core target is having a wide frequency bandwidth.
Looking at edcor products I find the PC series transformers: 20 to 20kHz freq. response, 1:1 ratio, 1/4W power, and different input & output impedances.

I see everyone is going for the 10k-10k impedance, but what if I choose the 600-600? isn't it the same, or better?

[R.G.]

So what is the verdict on the edcor RG?

They're in a box on the shelf in my workshop, waiting for me to get back to them. The delay in getting them here was long enough that I had to move on to other issues.   

I'll try to get back to them.

Now it's clear that the core target is having a wide frequency bandwidth.

Dead-on accurate.

Looking at edcor products I find the PC series transformers: 20 to 20kHz freq. response, 1:1 ratio, 1/4W power, and different input & output impedances. I see everyone is going for the 10k-10k impedance, but what if I choose the 600-600? isn't it the same, or better?

It then becomes a matter of power available to drive them and loading, and a number of other issues arise.

First, you have to consider what is driving the transformer and what the transformer in turn drives. For pedal/isolation use, the input which the transformer drives is usually another pedal or a guitar input with an input resistance of 100K or more, and often these days 1M or more. If the isolator is driving a true 600 ohm microphone input, that's another set of issues. The buffer output impedance will be quite low in all cases, usually much lower than even 600 ohms, so it can drive either a reflected 600 ohm load or anything larger - to *some* voltage level.   

In practice the buffer is usually an opamp, and the commonest opamps can only drive 600 ohm loads with small voltages, perhaps a volt or so. They don't have the internal power/current ability to drive a 600 ohm load to their full power specification. They **might** be able to do it with only a 9V battery as a power source, but it's a maybe. Some opamps like the NE5532 or LM833 do have the ability to drive 600 ohm loads well, so it gets into choices of opamps for buffers.

Because the buffer is driving the transformer, cable, connectors, etc. with a low impedance drive, it has the ability to drive even a 10K:10K transformer well enough to keep cable capacitance issues at bay. So the higher impedance transformers are good enough for the usual issues of driving normal pedal/guitar cable and amps, and may run into problem if the owner thinks "Hey, this is a 600 ohm output, I'm going to use it to drive a 600 ohm input."

So at a first order, 600:600 is OK-ish, and will work in some, perhaps many instances. It is getting into territory where it can be hard to drive and cause oddities with the buffer. My choice as a designer is to sidestep not only the obvious but also the secondary issues, so I like the higher impedance transformers better.

In my opinion, the 10k-10k is better for most pedal use because of the secondary issues. With a wide bandwidth 600-600, you could get good results depending on several issues. "Better" means you have to define what you're measuring.

[intripped]

thanks a lot R.G., I know something about audio transformers, finally!! 

[wavley]

I keep meaning to order the same transformers you did.  I'm not un-happy with the triad transformers in my hum free, but I'm don't really think of them as great either so I was gonna try these.  Plus I wanted to order some transformers to make a couple of DI boxes and I was thinking about a couple of those stereo to mono summing transformers to make a passive analog summing box for recording.