need opinion pls - 1:10 transformer on mix bus?

Hi.
I'm thinking if it would be a good idea to use Lundahl1578XL, 1:10 transformer(20dB gain), to boost level from mix bus?
It would boost gain by 20dB, with no added noise, is it? Otherwise it is mic transformer. http://www.lundahl.se/pdfs/datash/1578_8xl.pdf
What do you think about this idea?
thnx
gnd

My first thought:
Look at the "Max Prim. Level. 1% dist, 50Hz " on the datasheet and compare to regular Line input transformers.
And it is not clear to me what you are realy planing to do.

With no added noise.

Click to expand...

This is naive thinking. It surely does add noise with at least one, probably two mechanisms (depending on the summing amplifier design). A comprehensive answer would be rather lengthy, so right now just a short version: it depends, likely it won't help much--in any case you'd need to post the schematic of the mixing bus.

Samuel

[quote author="DerEber"]My first thought:
Look at the "Max Prim. Level. 1% dist, 50Hz " on the datasheet and compare to regular Line input transformers.
And it is not clear to me what you are realy planing to do.[/quote]

1% is at +12dBu input, thats +32dBu output!
it has 0.2% at 0dBu input, +20dBu output!
I need +4dBu output, so I'm at -16dBu input = distortion will not be the problem. And it is flat from 6Hz to 50kHz at -10dBu.

What am I planning to do? 10 channels, mixed via 10 resistors, and fed to this 1:10 transformer for gain makeup. Simple. Would it work?

gnd

[quote author="Samuel Groner"]

With no added noise.

Click to expand...

This is naive thinking. It surely does add noise with at least one, probably two mechanisms (depending on the summing amplifier design). A comprehensive answer would be rather lengthy, so right now just a short version: it depends, likely it won't help much--in any case you'd need to post the schematic of the mixing bus.

Samuel[/quote]

Schematics? It is just 10 stereo preamps, balanced outputs, going to summing box with 10k resistors, and then transformer primary. Passive summing. Transformer secondary goes to balanced input of another unit, like compressor or recorder. Thats it.

Would this work?

...

Output impedance will be then 100k! (10k/10)*10*10. So you'll need at least a buffer-amp.

And LL1578 propably doesn't like that high source impedance. The result is you'll loose some high-end. Try smaller resistors (2k-5k) insted of 10k. Then add a buffer. Op-amp, source-follower, what ever...

[quote author="Tomba"]Output impedance will be then 100k! (10k/10)*10*10. So you'll need at least a buffer-amp.[/quote]

You mean buffer after transformer, is it? Thats unity gain buffer, is it? I can do that, I guess, that shouldn't be a problem.

But otherwise, is using 1:10 transformer like this a good idea? Would it work as expected, with 20dB gain, bringing 10 summed balanced preamps back to balanced line level? Any problems to be expected?

...

I'm doubtful the output driving section of your 2-buss would be particularly happy.

Assuming you loaded it with a typical 10K load, the desk would see 1/100 of that = 100 ohms. A lot of opamp outputs would have quite a high distortion level at this load.

If you use the transformer to boost a voltage gain and a buffer to boost a current gain you may make your buffer with a voltage gain = 10 eliminating one expensive and distorting part.

Completely different story would be if you have input resistance after the transformer equal to output impedance of a source multiplied by a resistive ratio of the transformer, for example to feed 600 Ohm signal to input of a tube amplifier.

Transformer on input is needed to isolate input from in-phase noises, it is the main purpose of it. Speaking of colorations, everything will add colors, either instrumental amp, or a transformer, but if to choose one evil from 2 some people choose the one that looks more natural, i'e' which distortions sound closer to distortions of a sound in the real (not electrical) world. It may be a transformer.

The second purpose of the transformer is to match impedances, i.e. to pass the signal with the less as possible power loss. I do not know where 600 Ohm standard come from; air lines have higher impedance, cables have lower impedance, but we have to live with standards and match the gear to each other according to them.

If input impedance of amplifier is higher the ratio of the transformer must be higher, it means more voltage / less current is needed from it's secondary winding. The higher is bigger until input capacitance start roll off highs too much, or the transformer becomes expensive out of the desired range.

In case of transistor inputs the story is a bit different; they start produce more noise when source impedance is higher than certain level, or lover than some other level. It is always an equilibrium to get the better ratio between un-noised and un-distorted signal (of course, noises and distortions always present, the difference is in their amount).

Output transformer again is needed for couple of purposes.

A first, an impedance match to save the power and get wider dynamic range and lower distortions.

And second, again to help the next peace of equipment and minimize noises on it's input from ground loops and electromagnetic waves on cables.
If output impedance of your amp is good without the transformer you do not need it. But if you are going to connect it to the symmetrical input of the next device it is better to make it symmetrical.

And again, you may use an active phase splitter, or a transformer. Both will add coloration, the question is which colorations do you prefer.

Not to be picky, but some argue it's inappropriate to describe transformer turns ratios as dB of gain, since deciBels were historically used as power references (0dBm= 1 mW) and dB was used to describe power ratios. Transformers if anything exhibit slight power loss, and accomplish voltage ratios with a commensurate impedance transform.

It is perhaps viable to use a step up transformer at the back end of a passive summing network, if the combining network drops down to a low source impedance similar to a microphone (<200 ohms). In this case the transformer would be exactly like the transformer in front of a mic preamp. Nothing free per se, but impedance transform up to something more compatible with opamp gain stages.

I am not a big fan of transformers in audio paths so don't personally see much merit in this, but it is pretty much the same thing as hanging a mic pre on the output of a passive combining network, so whatever floats your boat.

I recall trying to close the feedback loop around a transformer input stage, before we could buy such quiet stuff off the shelf and never came up with a workable solution. I figured I could use a smaller, cheap, transformer since it was inside the feedback loop and would be linearized by the overall negative feedback. Apparently another of my design hallucinations since I never came up with anything clever or workable, before low noise solid state solutions came along in late '70s, early '80s.

JR

Not to be picky, but opamp stage after 10K fader loaded by 100K summing resistor is believed to give a "gain" despite there actually is a great power loss. Not to mention huge power losses inside of that opamp itself...

[quote author="Wavebourn"]Not to be picky, but opamp stage after 10K fader loaded by 100K summing resistor is believed to give a "gain" despite there actually is a great power loss. Not to mention huge power losses inside of that opamp itself... [/quote]

Well since you are indeed being picky or argumentative, I'll ruin your day and agree... Mr. Bell is long dead and logarithmic scale is far to convenient to limit to only power ratios. I use decibels myself routinely when characterizing cascaded (voltage) gain stages. The old school types even tolerated that if you made the somewhat arbitrary qualification that input and output was terminated with the same impedance.:roll:

Modern interfaces focus on voltage rather than power transfer. Not only that but I have even seen modern transformer companies speccing out their turns ratios in decibels (probably because customers think that way and want to buy them based on that). I find the use of decibels for voltage ratios useful myself and only share the caution should somebody have to interface with a strict old school engineer, perhaps a boss, and don't want to appear uninformed.

It also seemed appropriate to point out that the voltage ratio of a transformer is not free, but has a cost in terms of impedance and even losses.


JR

10 channels, mixed via 10 resistors, and fed to this 1:10 transformer for gain makeup. Simple. Would it work?

Click to expand...

I second the opinions that say a transformer is an expensive and rather unhappy solution for the design at hand.

An active summing amplifier is the much better solution here. There are many ways to skin the cat--without knowing much further about the rest of your project I suggest you'd use a pair of NE5534, perhaps followed by a NE5532, configured as instrumentation amplifier with balanced output. Alternatively you could use an integrated mic preamp such as the THAT 1512--in this case you should scale the summing resistors to about 3k ohm (say 2k if the driving amplifier is happy with that) for best noise performance. You might have a look at the summing amp design in my SGA-SOA-1 documentation (SGA-SOA-1_documentation.pdf) as well.

Samuel

I think the transformer could work to your advantage. I agree with what the others have said, but look who is asking the question... Ive seen enough from analag to know he loves transformers. and remeber that one of the most beloved mix busses EVER uses this topology (Neve 80-series). in their case they used a 1:2 transformer feeding a BJT input current feedback amp. the bus impedance was made about the same as the input impedance of the TF, as terminated into the makeup amp. so you loose 6 more dB but get good power transfer. the transformer does 2 things, matches impedance and isolates ground. in a physically large console that isolation could help alot, but the passive voltage summing topology gets noisy at high channel counts anyway. I would maybe try a lower-Z buss so that the transformer has a better situation. you will loose LOW end if you drive a TF with too high of an impedance. but in neebler's case it sounded good. a higher ratio transformer will work, but you will need a different makeup amp, like a tube based one.

if you are hell bent on this approach, you could use high current drivers in each channel and low value bus resistors. shoot for as low of a bus impedance as possible, like under 100 ohms. then with a modest step-up transformer you might be able to get away with no make-up amp. but the impedance will still be "high" if you use a big enough ratio to get enough level step-up. but if you were feeding a A/D chip right off the bus you would want a lower level balanced signal, and the TF secondary center tap could be connected to the A/D reference voltage...

Big THNX to everybody for opinions.

There was a pair of Lundahls on eBay, so I was planning to for it. But after reading your opinions, I decided to let them go to somebody else.... They went for 69eur pair, which is a nice purchase, i guess....

Anyway, as usually, I was looking for siple solution to solve all problems, but things get complicated quickly.... :wink:

Maybe I should have grabbed those Lundahls anyway, for mic pre.... Whatever..., maybe next time.

Thnx again.
gnd