Power Emitter Follower

[Marik]

I need to build a buffer to check ribbon transformers for bandwidth. I need a source impedance of at least 0.1 Ohm and even lower. The Pri. DCR of the transformer is 0.004 Ohm. My NAD amp did not like that load at all and was distorting like crazy :shock: .

I was not on the loop, so what are good power transistor choices out there for the job and (maybe) ideas for good topology for a very low output impedance EF?

Thanks, M

 

[Guest]

I would use some transformer connected to output of some power amp. For example, take a line distribution transformer and add several turns of thick wire, may be several such coils in parallel.

 

[PRR]

> I need a source impedance of at least 0.1 Ohm and even lower.

Do you mean "source impedance" (your NAD has that) or "load impedance" (your NAD will puke)?

At what level? 10V is hard. 1mV may be easy.

> The Pri. DCR of the transformer is 0.004 Ohm.

Taking some round ASSumptions: your primary impedance in the audio range is over 0.05R, your secondary impedance is 200R. Impedance ratio is 1:4,000 and voltage ratio is 1:63.

Also: you probably have 1mV to 50mV output. MAYbe 100mV. Ah, say 63mV is "very hot".

Then you need up to 1mV into 0.05R load, preferably from 0.005R source.

If you have an 8 ohm power amp, rig an 8R and 0.005R voltage divider. Oh, heck, round-off.... use 5R and 0.005R. That is 1000:1 ratio. To get 63mV at the 200 ohm output, you need 1mV at the 0.05R winding, which is 1V at the 5R end of the divider. A Sears Cardboard Bookshelf Stereo will put 1V in 5R (but maybe not to 20Hz; but your NAD will).

As Anatoliy says, you could do a transformer. "75 Watts" (24.5V across 8R) across a 24VAC power winding will give ~~about~~ 0.1V per turn. At 60Hz, that will be a heavy load on the amp because the iron is near saturation, so put say 20V across a 120V winding. Saturation will be toward 10Hz, expect ~~0.02V per turn. You really want more like 0.001V, so drive with 1V across "120VAC". Your "turn" needs to be thick copper ribbon/foil (or parallels) to get low resistance loss while fitting in the space around the coil. (You could de-wind the core, lay 600 turns primary, a copper-bar strap, and 600 more turns primary... more work than it is worth.)

BTW: it isn't "full turns", but "window passes". Bring a 120V E-I winding up on 120VAC. Shove a wire through one window. You get say 0.05V. Wrap it 9/10ths around the coil, you still get 0.05V. Bring it through the other window, boom, get 0.1V. At power and most audio frequencies, 99.9% of the flux is in the iron, meaning in the windows. So voltage is essentially the number of passes through the window, not number of full turns. With many-turn windings, don't care. When we speak of "one turn", we need to know which part of the turn matters. At power frequency, you might do one window-pass for least copper length. At high audio, you may need to "wrap" the main coil and capture the 0.1% leakage flux also.

A major problem with a transformer step-down is magnetic coupling from your driver transformer to your ribbon transformer. At 150R we can put them yards apart, but at 0.005R your interconnection must be fat and short. That favors the resistor divider which has small (not zero) magnetic field.

 

[bcarso]

Best would be to match roughly to the Z of your typical ribbon. At these low of impedances a little inductance could begin to be important.

Having said that, the point is to take advantage of the tiny power required, and work with reasonably small voltages, whether achieved with a resistive divider or other stepdown means.

 

[PRR]

> work with reasonably small voltages

I think he "must" work with small voltages.

If the tranny is normal mike-tranny size, the 30Hz saturation level at the "line" side may be 30mV to 1V (2V to 20Hz at low THD is a big chunk in my AKG 414s). Unless Marik's skill and the new/neo magnets reach some new high in ribbon output level, I'm guessing no musical source will put over 100mV on the 150/200 ohm side. He knows his ratio better than we do, but that's likely 1mV-2mV on the ribbon side.

> match roughly to the Z of your typical ribbon

Agree. You can create an artifically low bass cut-off with an artifically low source impedance. Also limited by winding resistance, but for good mike S/N the winding resistance will be lower than ribbon resistance. If tranny DCR is 0.004 and ribbon is 0.1, measuring tranny alone with a dead-zero source gives bass rolloff or saturation at 25 times lower frequency than he will reach in real life.

Taking a 0.1R lower resistor in this resistor divider, and 2mV max level, we actually need no more than 20mA. If he had a 50 ohm 1V source, 49R9:0R1 nails it. If he has 10V in 600R, then 599R9:0R1 reaches 1.66mV but the math is annoying. Given a wideband loudspeaker amp, 100R:0R1 gives a light amp load, an easy 1000:1 ratio, 1V-2V out of the amp gives ribbon-level, and momentary accidents to 10V won't smoke a 1/2W resistor.

There is the problem of finding 0R1 or 0R005 resistors. I assume anybody working with ribbon mikes can eyeball the milliOhms in a scrap of foil or an inch of wire. If this 0.005 DCR winding is self-wound, he can use that wire scaled to a shorter length.

We are far into the range of "4-wire resistors". If I had a 0.0X scrap of copper wire, I'd cut an inch extra so I could wrap the ends around short pieces of #12 house-wire and solder. Then the NAD common and 5R or 100R hot resistor go to one end of the #12s, the winding to be tested goes to the other ends of the #12s.

There is much logic to just using "the ribbon" as the lower end of the voltage divider. Main why-not is that it may be hard to get a 4-wire connection so NAD currents don't flow in transformer lead-in wire. To a first approximation it should not need a magnet. If it has a magnet you need to ignore ambient sound, but at 1mV on the ribbon this should not be difficult.

You don't need any exact or known divider ratio to know Frequency Response. You do need to know you are not driving at unlikely high level, but the level at the 150R side tells you that. Capacitance is a non-issue at these impedances. Ribbon and resistor inductance will be less than DC resistance and thus negligible far past the audio band. Or at least we can assume the resistor divider is far more perfect than the ribbon mike, which only works from bass resonance to where its physical dimensions get not-negligible. The NAD will have poles at 5Hz and 100KHz but these can easily be measured; likewise if he needs a preamp to get level in his ACVM.

 

[Marik]

Of course, with old school motor structure dimentions and using modern Neos I could easily go up to at very least 1Tesla in a gap, but for sake of both HF and LF response the field is much more modest, so we are rather in 1mv region.
I actually the most interested in measuring the extremes of the bandwidth and esp. the LF.

My main problem with transformer approach is not even trying to measure one nonlinear device with another nonlinear device, but the fact that even some small copper resistance of the "matching" transformer will through things off on LF.
Measuring HF can be also little challenging without knowing which transformer actually rings.

Resistive divider looks like a much better idea and making either 0R1 or even 0R005 resistors is not a problem. My old Genrad 1650A still measures in that range with ease.

Thank you Gentlemen!

 

[bcarso]

[quote author="Marik"]Resistive divider looks like a much better idea and making either 0R1 or even 0R005 resistors is not a problem. My old Genrad 1650A still measures in that range with ease.

Thank you Gentlemen![/quote]

I pulled a 250 amp meter shunt out of the wrong bin at All Electronics once, which I still have---I forget what they thought it was. A good buy at the time, I've not yet gotten the right bolts for the two big tapped holes on the ends.

 

[Marik]

Since the transformer primary inductance is very small, it seems in case of using resistive divider the source impedance as seen by transformer primary will be output impedance of the amp (say 2 Ohm)+8 Ohm (divider)=10 Ohm (the 0.005 Ohm part is small enough to disregard).

Say the primary L=1.7mH then the -1db point will be @ 1885 Hz :shock:

Am I missing something?

Best, M

 

[Guest]

If you use 10 Ohm and 0.005 Ohm divider your tranny will see 0.005 Ohm