What could account for severe loss of LF response of mic input xfmr?

[nevetslab]

While I'm nearly finished restoring this 45 year old Lombardi 8-Ch Input Mic Mixer, it having a Binson Echorec Magnetic Disc Delay System built in, I've discovered the source of the total lack of LF response on Ch 4, which was found prior to my pulling all the modules out to replace the 45 year old Ducati Electrolytic Capacitors throughout the mixer.

When I had the mic preamps removed, which was where I thought I'd find the problem, all eight preamps measured the same. When I finally re-installed them, I again found this lack of LF response on Ch 4. Clipping my Fluke 8060A DMM in RMS dB mode across the secondary of the input xfmr, and swept the response, I found it was the transformer itself having this fault.

The first item I found with regards to the transformers was this HF boost added in the preamp, but never found any de-emphasis network in any of the modules. I had been measuring near flat frequency response on the output. But after looking at the response of the input transformer in the HF range, I also found the transformers have a serious HF rolloff.

I just got collecting data on the Ch 4 input xfmr, as well as one of the other typical input xfmrs for comparison. Ch 4's HF response is also vastly different than the others. Mechanically, it looks the same, but measures way differently. I measured the input impedance of the xfmrs, using a 2 ohm source impedance, passing thru an ESI decade resistance box, with a Fluke DMM in AC RMS mode, dB mode, REL mode, and at each spot frequency, adjusted the decade box for a -6.0dB output, which would occur at the input impedance of the load....load being the input xfmr + the secondary load, which is more than 2 decades higher in resistance than the input impedance, so negligible.

What would account for this severe loss of LF response on Ch 4's Mic Input xfmr? I wondered if, due to the low overall system gain between mic input and mixer output being only 38dB, and plugging a typical dynamic mic into the mixer, such as a Shure SM57, I barely got any meter movement on the VU meter. I had to resort to plugging in a Symetrix SX-202 Mic Preamp ahead of the mixer, so I could give a listen with a typical condenser mic (SX-202 having Phantom Power). Could this mic xfmr have been severely overloaded by a similar set-up, and the problem not found right away, finally having such input level do damage to the transformer?

The lack of HF response on all of the transformers also is puzzling. I can't see any rational reason for selecting an input transformer for a professional mixer to have severe HF rolloff, unless that was a production mistake from the transformer mfgr, and the production fix was adding a HF shelving boost across the emitter resistors of Q1 input xfmr (C4 20nF across R7 2.2k + R6 47ohms).

I've attached the measurement data recorded today below, along with the Preamp schematic:

LOMBARDI MIC #4 INPUT TRANSFORMER DATA.pdf

LOMBARDI MIC INPUT TRANSFORMER DATA.pdf

Lombardi Mic Preamp Schematic-3A.pdf

[Helmholtz]

This a long text and a lot of info.
I have problems to sort it and understand how exactly you measured input impedance and frequency response
Could you post a little drawing showing how the equipment was connected?

Generally any (audio) transformer is a bandpass.
Bass response is determined by primary inductance, and treble response depends on leakage inductance and self-capacitance.
Frequency response changes with source and load impedance.
In the preamp the transformer may see something like a 100k load impedance (just a rough guess).
Input impedance should be measured with the designated load impedance, best connected to the preamp.

Poor bass response might be due to an unwanted/increased airgap (or a crack in a ferrite core).
Do you have an LCR meter?
What is the voltage transfer ratio @1kHz? Any difference between bad and good ones?

Quality wide-band audio transformers are expensive, so it makes sense they compensated the treble loss in the preamp.

Also not quite sure how you use dB.
Without any additional suffix db doesn't mean a level but rather a ratio of levels.
So 0 dB means a ratio of 1.

[Enzo]

Just my instant reaction, and considering my old friend Occam, any chance a shorted turn is responsible? Remotely in the sense a seized up speaker can still make sound, but all the bottom will be gone

[J M Fahey]

You´ll hate me for this,but can I ask you to remeasure?
In may favour, a somewhat simpler setup

1) You already measured transformer input impedance, like others suspected a shorted turn but the data does not support it.
A shorted turn murders inductance which is in parallel with input, so it should show lower impedance .... but suspect Xfmr shows it much higher , as in

* good one: 800-715 ohm @ 1k-100 Hz
* bad one: 1830-2250 ohm @ 1k-100 Hz

which makes me suspect the opposite: OPEN secondary because frequency response is (just taking 2 or 3 significant points) :

* good one: 100 - 1k - 10k Hz ..... -0.26 0 -4.77 dB

* bad one: 100 - 1k - 10k Hz ...... -13.8 0 +1.5 dB

so there is no *magnetic* coupling between them but *capacitive* coupling (tightly wound wires side by side have strong mutual capacitance) which murders LF response but keeps or even enhances HF .... what data shows.

2) that said, I would repeat measurements in a more realistic but simpler setup:

"To do what the mic does" I suggest you drive low impedance primary from a 200 ohm source (a very common DCR for a nominal 600 ohm microphone), do not attempt to match Xfmr input impedance (the mic does not do that either), AND load secondary with fixed 10k (a common value used when testing them), drive it with constant 50mV (again, what a mic does, it has no clue about compensating anything) and measure voltage (or dB) at secondary, across said 10 k

Also: use as dB value for BOTH Xfmrs whatever is 0dB@1kHz for the GOOD one, so we not only measure relative frequency response at each one, but also loss of performance at the bad one.

Meaning: suppose good one gets 50mV@1kHz at its primary (through 200 ohm and we measure them before the 200 ohm resistor, not across primary terminals) and secondary shows, say, 400mV across 10k load, so we call THAT 0dB, so if bad Xfmr at 1kHz already shows , say,200mV with same input,we know it not only has flawed frequency response but also it also has high loss or very poor coupling.

3) so in a nutshell I suggest this simplified retest:

* measure at only 3 frequencies :100 - 1000 - 10k Hz

* no attempt to measure or match impedance, just feed it from a 200 ohm source

* use fixed 50mV straight at generator terminals

* load with 10k

* take 1 kHz output from good one as 0dB as general reference for both

this will give us "efficiency" differences (for lack of a better word) and at 3 significant frequencies ,specially considering it´s a "voice frequencies" Mic transformer.

later we may check raw preamp on its own.

NOTE: I seem to detect an error in schematic: I expect C8 to be between 2n2 and 4n7 for a proper treble EQ boost value, not coupling cap class 4u7

in any case, thanks for taking the trouble of lifting that schematic up.

Here in Argentina we were quite used (at least I was) to work with Italian PA equipment, very popular here because ... ummmm .... ITALIAN Music was very popular, annual San Remo Festival was a MAJOR event (then Eurovision took its place) , Italian singers toured the Country all the time, records were always in our Top 40 , etc. , so "an Italian mixer with Tape Echo" was always found at shows .
Think EKO, Lombardi, FBT, Meazzi, and the undisputed King: Semprini.

Same era typical American "singer" PA (used by Frank Sinatra) was Shure Vocal Master, not even reaching the knees of the Italian ones.
Or German Dynacord.

Shure only used cheesy spring Reverb and their columns had no tweeters. .... worlds apart.

Even the Mamas and the Papas used European PA equipment (usually Dynacord).

[Helmholtz]

After rereading, it doesn't look like low inductance nor like shorted turns.
Both would cause low input impedance especially at bass frequencies.
But you measured higher input Z with the bad transformer.

An open winding can be found with an Ohmmeter.

I understand all measurments were done with the transformers connected to the preamp.

Is the input coupling cap C1 (4.7µ) of channel 4 good?
If it's low capacitance, load and input impedance at low to medium frequencies would be increased.
The input impedance of the transformer at mid frequencies is essentially the reflected load impedance.

[nevetslab]

This a long text and a lot of info.
I have problems to sort it and understand how exactly you measured input impedance and frequency response
Could you post a little drawing showing how the equipment was connected?

Generally any (audio) transformer is a bandpass.
Bass response is determined by primary inductance, and treble response depends on leakage inductance and self-capacitance.
Frequency response changes with source and load impedance.
In the preamp the transformer may see something like a 100k load impedance (just a rough guess).
Input impedance should be measured with the designated load impedance, best connected to the preamp.

Poor bass response might be due to an unwanted/increased airgap (or a crack in a ferrite core).
Do you have an LCR meter?
What is the voltage transfer ratio @1kHz? Any difference between bad and good ones?

Quality wide-band audio transformers are expensive, so it makes sense they compensated the treble loss in the preamp.

Also not quite sure how you use dB.
Without any additional suffix db doesn't mean a level but rather a ratio of levels.
So 0 dB means a ratio of 1.

I'm NOT finding any open windings on the 'BAD' Xfmr Ch 4. The Inductance reading on it is VASTLY different. So is the voltage transfer ratio @ 1kHz. On the good ones Step-up ratio is typically 21.8dB, while that of this 'BAD' xfmr it's only 16.1dB at 1kHz. The Inductance reading (120Hz, Series) on typical ones is 9.22mH/Q0.03, though found that varied, a couple in the 10.2-10.5mH, another 12.3mH and highest was 14.1mH. The 'BAD' xfmr reads 2.36H/1.12Q. Secondary inductance readings seem odd. 4.8H on three, couldn't get ANY reading on the 'BAD' xfmr. Another read 5.43H, three others read from 6.8H to 7.7H. higher Q readings....0.09 typ 120Hz/Series circuit, all read on my GenRad 1658 DigiBridge





The connections on this DigiBridge jumper the two sides of the Kelvin Clips to a common Test Lead. I never picked up the adapter for this series to go to standard Kelvin Clip Cables which I have for my ESI 296 Bridge (in storage). I next removed the rear connector panel to access the xfmr terminal boards to try and get onto the Secondary side directly.



I sketched a couple test-set-ups used...one for measuring the Input Impedance, using in-line Decade Box between the Generator and the Mic Input Xfmr. Second one was that used for measuring the Step-Up Ratio of the Primary to Secondary. The Mic Preamps have new electrolytic caps, so yes....that 4.7uF cap IS brand new. 220k Secondary load in place.




The Test Setup for the Input Xfmr Impedance is shown below, to accompany the drawing and set-up notes in that 1st drawing. I used Ch 7's Input Xfmr in this photo:



The second DMM at the Mixer was set to 0dB Relative Mode, and used to show the Frequency Response for each Frequency after trimming the Decade Box to yield -6dB, reflecting that input impedance. I'll re-run those measurements on the Secondary side using the 200 ohm source, and trimming the Send Level at each spot frequency, since it will vary from the loading of the input impedance on the 200 ohm source. The B &K 1027 Sine/Random Generator has a 100 ohm source impedance, so I added a Pomona in-line adapter with a 100 ohm resistor to build out the source to 200 ohms. The 1st DMM is monitoring that output, and feeding the input cable to the Mic Preamp.

The next photos show the test setup & adjustments at the Source, the Secondary Step-Up Level and the Analyzer/Scope for monitoring the overall response thru the Mixer.

[Helmholtz]

Thanks, I am too tired now to read your full post.(2:am here). Will check tomorrow.

Please compare DCRs at primary and secondary.

With low inductances and low Q I would measure at a higher frquency (1kHz) to get better accuracy.
A Q of 0.1 means that impedance is mainly resistive. At higher frequency inductive impedance increases which makes it easier for the meter.

If No.4 transformer has no open secondary, could it be that primary and secondary are interchanged?
Wondering about the high primary inductance.

[nevetslab]

You´ll hate me for this,but can I ask you to remeasure?
In may favour, a somewhat simpler setup

1) You already measured transformer input impedance, like others suspected a shorted turn but the data does not support it.
A shorted turn murders inductance which is in parallel with input, so it should show lower impedance .... but suspect Xfmr shows it much higher , as in

* good one: 800-715 ohm @ 1k-100 Hz
* bad one: 1830-2250 ohm @ 1k-100 Hz

which makes me suspect the opposite: OPEN secondary because frequency response is (just taking 2 or 3 significant points) :

* good one: 100 - 1k - 10k Hz ..... -0.26 0 -4.77 dB

* bad one: 100 - 1k - 10k Hz ...... -13.8 0 +1.5 dB

so there is no *magnetic* coupling between them but *capacitive* coupling (tightly wound wires side by side have strong mutual capacitance) which murders LF response but keeps or even enhances HF .... what data shows.

2) that said, I would repeat measurements in a more realistic but simpler setup:

"To do what the mic does" I suggest you drive low impedance primary from a 200 ohm source (a very common DCR for a nominal 600 ohm microphone), do not attempt to match Xfmr input impedance (the mic does not do that either), AND load secondary with fixed 10k (a common value used when testing them), drive it with constant 50mV (again, what a mic does, it has no clue about compensating anything) and measure voltage (or dB) at secondary, across said 10 k

Also: use as dB value for BOTH Xfmrs whatever is 0dB@1kHz for the GOOD one, so we not only measure relative frequency response at each one, but also loss of performance at the bad one.

Meaning: suppose good one gets 50mV@1kHz at its primary (through 200 ohm and we measure them before the 200 ohm resistor, not across primary terminals) and secondary shows, say, 400mV across 10k load, so we call THAT 0dB, so if bad Xfmr at 1kHz already shows , say,200mV with same input,we know it not only has flawed frequency response but also it also has high loss or very poor coupling.

3) so in a nutshell I suggest this simplified retest:

* measure at only 3 frequencies :100 - 1000 - 10k Hz

* no attempt to measure or match impedance, just feed it from a 200 ohm source

* use fixed 50mV straight at generator terminals

* load with 10k

* take 1 kHz output from good one as 0dB as general reference for both

this will give us "efficiency" differences (for lack of a better word) and at 3 significant frequencies ,specially considering it´s a "voice frequencies" Mic transformer.

later we may check raw preamp on its own.

NOTE: I seem to detect an error in schematic: I expect C8 to be between 2n2 and 4n7 for a proper treble EQ boost value, not coupling cap class 4u7

in any case, thanks for taking the trouble of lifting that schematic up.

Here in Argentina we were quite used (at least I was) to work with Italian PA equipment, very popular here because ... ummmm .... ITALIAN Music was very popular, annual San Remo Festival was a MAJOR event (then Eurovision took its place) , Italian singers toured the Country all the time, records were always in our Top 40 , etc. , so "an Italian mixer with Tape Echo" was always found at shows .
Think EKO, Lombardi, FBT, Meazzi, and the undisputed King: Semprini.

Same era typical American "singer" PA (used by Frank Sinatra) was Shure Vocal Master, not even reaching the knees of the Italian ones.
Or German Dynacord.

Shure only used cheesy spring Reverb and their columns had no tweeters. .... worlds apart.

Even the Mamas and the Papas used European PA equipment (usually Dynacord).

An OPEN secondary on Ch 4's 'BAD' Input xfmr.....YOU NAILED IT! THAT's why there NO Inductance reading on that xfmr! All the other xfmrs have a DCR around 4.88k, while this 'BAD' Xfmr only has the 220k termination resistor across the Secondary (bleeds off the DCV from the 4.7uF Input Cap C1.

When I was re-capping the preamp boards, and then traced out the schematic and drew it up in the program, I had removed C8 from one of the boards, and put it on the LCR Bridge. It measured roughly 4.7uF. Odd-looking part for a Dip-Tantalum. I agree with you that that value for the HF Shelving to work properly should be in the range you suggested....2.2nF to 4.7nF. All three of the other Film Caps are marked the same, pulled one which measured 20nF. I didn't verify the others. I also didn't think to order replacements for that part in the range you suggested. I do have Wima FKP2 Polypropylene 2.2nF & 4.7nF on hand, but only six of each (RATS!).

With finding the Secondary of Ch 4 Input Xfmr open, I think I'll have to try and unsolder those fragile tiny wires and see if it's the wires having a break in them, or if it's really open. Odd there'd even be output......though your explanation on the mutual coupling of tightly wound wires side-by side yield enough output to only be down by around 5dB or in the mid-band on up.

[nevetslab]

Thanks, I am too tired now to read your full post.(2:am here). Will check tomorrow.

Please compare DCRs at primary and secondary.

With low inductances and low Q I would measure at a higher frquency (1kHz) to get better accuracy.
A Q of 0.1 means that impedance is mainly resistive. At higher frequency inductive impedance increases which makes it easier for the meter.

If No.4 transformer has no open secondary, could it be that primary and secondary are interchanged?
Wondering about the high primary inductance.

All eight xfmr's Primary Resistance measures around 98 ohms. The Secondaries measure around 4.88k, with the exception of Ch 4, which is open circuit (or a break in the tiny fragile wires).

[J M Fahey]

Glad you found it

I´m practicing hard to apply to Enzo´s Tech Team, I heard he doesn´t care much about "credentials" (what do I do now with my $70k Student Loan? ) but cripples something and puts it on the bench, to solve or die trying.

Oh well.

[Leo_Gnardo]

Odd there'd even be output.

Much like guitar pickups with broken coils - there's output but it's thin sounding (no bass) and significantly lower in volume. "Capacitively coupled" explains how signal passes though coil wire is discontinuous. Good hunting, you & Juan!

[Enzo]

An open secondary on a power transformer will still yield phantom AC readings. More than once I have encountered an open heater winding whose phantom voltage was right around 6v. Confused me the first time, and my junior tech when he first encountered it. "Hmmm, I have my 6v, but it goes away when I plug in any tubes..."

[Helmholtz]

"post #6"

Great, seems you know how to use your (sometimes fancy) equipment .
Taking frequency response plots is easy with the Bode plotter of my little Velleman PSCU200 USB scope

One thing I don't understand: What is the DMM on the secondary for when measuring input impedance ?

[nevetslab]

If my laptop hadn't died from it's Hard Drive failing, I'd have been running plots with my Velleman PSCU2000 and the Amber 3501a as a front end receiver, as I'd been doing for many years. My plotter pens have all dried up. Even the sealed containers for the older capillary ink-fed pen system on the Bruel & Kjaer 2308 XY Plotter ended up with no ink in them....SEALED containers. THAT really astounded me. So, at this moment, generating plots is a PITA. I did invest in plotter-style rapidograph pens, where I'd be able to buy colored ink and fill the small reservoir as I had done for years as a draftsman, doing pen and ink drawings. I'll have to return to that, and do likewise with the B & K capillary pen cartridges. Messy, but does the trick. I had invested in a large qty of K & E Vellum preprinted graph paper. With the B & K gear, the plotters drive the generator's frequency control.

The DMM shown across the secondary was there to show me relative frequency response (re 1kHz) while measuring the input impedance. I had thought about using the compressor loop in the B & K to produce a constant current source to drive the xfmr primary, but would have to scale the Vertical axis of the scope, driven by the Log Converter in one of the B & K measuring amps. The set-up to scale it led me to doing it all manually with pencil and paper. That always works when technology fails you.

This morning, having resigned myself to accept what I already knew...having a bad input xfmr, though not knowing where the fault was unitl you all helped me find it.....I checked the ninth transformer on that connector panel, which feeds a Male XLR connector thru a pot. Not sure where it's fed from....one of the three output busses, no doubt. I measured the DCR at the XLR Male connector, finding that around 4.9k....made me wonder if this was an input xfmr turned around. I removed the panel to see what the other side was. It was connected to the wiper of a pot. Pot measured a bit over 50 ohms, so I unsoldered the xfmr wire from the wiper. That winding measured open circuit. sigh..............so two bad transformers on this panel. So much for wondering if it was the same one, and if it measured like an input xfmr, I would have swapped it out with the Ch 4 xfmr.

[Helmholtz]

. I'd have been running plots with my Velleman PSCU2000 and the Amber 3501a as a front end receiver, as I'd been doing for many years.

What would the Amber be for when used with the Velleman?

My PCSU 200 is a complete measuring system putting out printable plot files. So no plotter required.

The DMM shown across the secondary was there to show me relative frequency response (re 1kHz) while measuring the input impedance..

Maybe I misinterpret, but this sounds like the source impedance was varied during frequency response measurement.
This would skew the response.at least somewhat.
.
A transformer's frequency response varies with source impedance.
So the response should be taken with a constant source impedance corresponding to the designated source, e.g. 200 Ohm for a dynamic mic.
In audio the input impedance of equipment should always be higher by a factor (of at least 5) than the source impedance to avoid loading the source.

Driving the primary from a current source would give wrong response.