Ribbon Mics Loading Again

[Marik]

I remember an argument from last year about Chinese ribbon mics where I was claiming that loading the ribbon microphones affects mostly the low end and somebody else (was it Rossi?) was saying that it affects top end.
The last few days I was measuring a bunch of transformers, including Chinese stock, Cinemag, Lundahl, and my own wound toroidal transformer. The test revealed why we had the differences in our findings.
For the measurements I was using a "Vector Impedance" section of my HP 4274A LCRZ meter. The beauty is it measures the whole ribbon/transformer system up to 100KHz. It also measures the phase angle. The results were very interesting and highly enlightening, explaining many processes in ribbon microphones, as well as the reason the stock Chinese mics lose the high frequencies with loading.

For the test I used a medium ribbon motor structure (very similar to the Nady RSM4 and alike). With the stock ribbon of about 4-6um, and stock transformer the impedance between 100Hz and 500Hz drops from 868 Ohm to 396 Ohm, then stays there till 3KHz and then from 5KHz to 50KHz raises from 413 Ohm to 1.8KOhm, reaching 3.9KOhm (!!!) at 100KHz.

For comparison, with the same ribbon, but with my own wound toroidal transformer the impedance drops from 314 Ohm to 107 Ohm between 100Hz and 1KHz, then stays there until 10KHz, raising from 136 Ohm on 30KHz, to 290 Ohm on 100KHz.

Both Lundahl and Cinemag did not have that huge impedance spike on the top of the range, either. Cinemag was measuring little better than Lundahl, but then again, it is a lower ratio transformer. In fact, to be of the same quality transformer it should've been measuring MUCH BETTER. IMO, its very low permeability DU core is unusable for ribbon transformers and results in a very high Pri DCR (0.11 Ohm, whcih translates into high noise), and still low Pri. inductance.

The thinner ribbon (1.5um) with the same transformers measures much better both impedance-wise and in term of phase.
With my transformer it stays flat @224 Ohm from 1KHz to 10KHz, then just slightly raising to 295 Ohm at 100KHz.

For comparison, Lundahl LL2912 was flat @309 Ohm from 1KHz to 10KHz, and then raising to 455 Ohm at 100KHz. It had worse phase characteristics.

Hopefully, that was helpful.

Best, M

 

[emrr]

Excellent observations, thanks for sharing in such detail.  I had noticed Cascade only provides the impedance versus frequency chart with the mic, rather than in the advertising for all to see. 

The part I think about a lot is the following bit of mystery coupling; the mic preamp.  Especially when mating transformer to transformer, and often unclear info about loading, or lack thereof, on the secondary.  I usually feel unloaded transformer inputs sound better with ribbons than the standard modern resistively loaded non-transformer type.  Noise and phase angle are entirely different questions. 

 

[rodabod]

Thanks for that, Mark.

By the way, I just picked up an STC 4038 which is off for servicing at Coles just now. I am having the transformer replaced while it is there, so sould be able to experiment with the old model when it is returned.

Roddy

 

[Marik]

The part I think about a lot is the following bit of mystery coupling; the mic preamp.  Especially when mating transformer to transformer, and often unclear info about loading, or lack thereof, on the secondary.   I usually feel unloaded transformer inputs sound better with ribbons than the standard modern resistively loaded non-transformer type.  

I think the "mystery coupling" could be explained with the impedance curves, which show that different transformers will react differently. For me it was actually somewhat unexpecting to see such big differences as for how different transformers affect high end...
The interesting thing was my transformer is 1:39 versa Lundahl 1:37, but actually even though mine is (slightly) higher ratio, the system had lower impedance. When comparing to Cinemag (1:28) I used the 1:30 version of my transformer and on 1KHz mine had impedance of 54.30 Ohm (vs. Cinemag 157 Ohm).
Many differences (incl. noise) can be attributed to different copper resistances. The Lundahl Pri DCR=0.042 Ohm and Sec=59.38 Ohm. Cinemag Pri=0.11 Ohm (!), Sec=16.47 Ohm. Mine has Pri=under 0.001 Ohm, and Sec=4.6 Ohm (all the resistances are measured with 4-wire Kelvin probes on my HP3457A and then double checked with the GenRad1650A bridge).

Personally, instead of using two transformers in series, I much prefer the variable Z input pres, as they are much more predictable and stable in terms of uniformity of loading accross the band.
As those measurements show, the better quality transformers and thinner ribbons will be much more "immune" to the loading.

Generally, I feel there are many great ribbon microphones, but only very few have well designed transformers. MOST of even highly regarded "classics" could greatly benefit from swapping to even more generic "fit-it-all" but higher quality transformers.

By the way, I just picked up an STC 4038 which is off for servicing at Coles just now. I am having the transformer replaced while it is there, so sould be able to experiment with the old model when it is returned.

Cool! What transformer are you putting in? The 4038 has about 0.25 Ohm ribbon, so not many offshelf transformers will fit the bill...

Best, M

 

[PRR]

Your impedance curves (I wish I could see them plotted) make only slight sense.

The drop around 400Hz might be core eddy loss, the rise/peak at the top of the audio band is common resonance.

But if correctly proportioned, I'd expect core loss to hide behind loading, ribbon impedance.

And we may assume the ribbon is exactly 0.x ohms at "all" frequency, unless there are major mechanical or acoustic resonances. Which could be removed by taking the magnet away.

I'd also like to see the UN-loaded transformer impedance. It "should" be significantly higher than the reflected ribbon impedance. That can't be true to infinitely low frequency; the ~~20Hz impedance ought to matter but I don't see this fall-off in your remarks.

Feed say 10mV from perhaps 1K impedance into the secondary, with nothing on the primary. The voltage curve can be (tediously) computed to impedance. It is awkward because the few-mV level is small and imprecise, but ribbon outputs don't exceed dozens of mV and we don't want to get close to that for this measurement.

This curve is more informative if related to reflected ribbon resistance and nominal load (preamp) impedance.

 

[rodabod]

What transformer are you putting in? The 4038 has about 0.25 Ohm ribbon, so not many offshelf transformers will fit the bill...

Don't worry - I'm getting a 300-Ohm Coles toroidal model to replace the current 30-Ohm model. Should be pretty sweet...

 

[xmvlk]

For the test I used a medium ribbon motor structure (very similar to the Nady RSM4 and alike). With the stock ribbon of about 4-6um, and stock transformer the impedance between 100Hz and 500Hz drops from 868 Ohm to 396 Ohm, then stays there till 3KHz and then from 5KHz to 50KHz raises from 413 Ohm to 1.8KOhm, reaching 3.9KOhm (!!!) at 100KHz.

Hi, Marik

May be high frequency resonance of transformer. Where is maximum?
Normally high frequency res is damped with the ribbon resistance It may
enlight that HF resonance is flatter with thinner ribbon.

I do not have measured phase, only  modulus of ribbon impedance.
I had measured it in air and in vacuum on the side of ribbon
{i wanted to design transformer for the given motor, not to made motor for given transformer}
Here are some curves: http://radio.feld.cvut.cz/~vlk/thesis/kap7.pdf  do you like to read in czech ))  {personally, I hate to read in english...}

Some note: When in the chineese microphone HF resonance of the transformer
is used to flatten mic frequency characteristics, the motor is badly (acoustically) engineered.
I had designed ribbon mic transformer only for lowest noise and I had not problem with
resonance. I can not imagne what the hell this makes with transformer input (tube) pream.