Advice wanted: Zero Field cancelling out 7.9K sec resistance of line in trafo

Hello All,

I've  been thru the various zero field frafo discussions over here but would like your insight in the following:

My EMT mixing console has Beyer line in trafos (BV3XX 001 081) with turns ratio 1:1 and a specified secondary DC resistance of 7.9KOhm.

These have a pad in front of them, but still exhibit distortion (mainly 3rd order) which is two orders of magnitude higher between 20 and 100Hz compared to 1KHz.

I have one unused opamp section left on the PCB; what would be the best basic schematic to try and cancel out the sec DC resistance?

I've read some claims that this high a DC resistance would be impossible to compensate, others say it should be possible.

Any help appreciated; I'll post distortion plots of results before and after.

Best Regards,

Rogy

Rogy said:

Hello All,

I've  been thru the various zero field frafo discussions over here but would like your insight in the following:

My EMT mixing console has Beyer line in trafos (BV3XX 001 081) with turns ratio 1:1 and a specified secondary DC resistance of 7.9KOhm.

These have a pad in front of them, but still exhibit distortion (mainly 3rd order) which is two orders of magnitude higher between 20 and 100Hz compared to 1KHz.

I have one unused opamp section left on the PCB; what would be the best basic schematic to try and cancel out the sec DC resistance?

I've read some claims that this high a DC resistance would be impossible to compensate, others say it should be possible.

Any help appreciated; I'll post distortion plots of results before and after.

Best Regards,

Rogy

Click to expand...

Secondary DCR is not a problem at all in regard to THD. Primary DCR is a problem in that respect. Zero-field technique is about using DCR instead of fighting it. The transformer then operates as a current-transformer, with very little voltage, thus very little magnetizing current (in fact the magnetic field created in the primary is almost completely cancelled by the current in the secondary.
Zero-field is generally pretty straightforward to implement, however, the best optimization happens with lower-turn windings. I have consistently used this technique using 1:1 mic xfmrs (600 ohms nominal). Same xfmrs rated at 10k:10k had noticeably inferior performance because the stray capacitance of the windings appears as a shunt to the input, which in turn requires large lead-compensation FB capacitor. Noise performance is degraded, as well as CMRR.

Thanks Abbey Road d Enfer,

The line in transfos are specified as having turns ratio 1:1 and 1KOhm:1KOhm so maybe well worth to try.

In your drawing, "C = A O T", what does this mean?

Best,

Rogy

Rogy said:

Thanks Abbey Road d Enfer,

The line in transfos are specified as having turns ratio 1:1 and 1KOhm:1KOhm so maybe well worth to try.

Click to expand...

Then it is not possible that the DCR is as high as 7.9k. Check specs. [/quote]
In your drawing, "C = A O T", what does this mean?
[/quote] Adjust On Test

Rogy said:

...Beyer line in trafos (BV3XX 001 081) with turns ratio 1:1 and a specified secondary DC resistance of 7.9KOhm.

Click to expand...

Is this the transformer?

Regards Andreas

Hello Andreas,

It's that transformer but without the center tap on the primary (3xx 101 081 has CT; mine is 3xx 001 081 without CT).

The specs on your pdf are the same as on my sheet; it reads primary DC resistance = 5.9K and secondary DC resistance is 7.9K

Do you have any experience with these?

Best Rogy

Hello Rogy,

yes, this Beyerdynamic transformers have many thousands of turns (7680 on primary an 7680 on secondary) on a small core - so they have a high DCR. At a high amplifier input impedance this plays no role.

Regards Andreas

This xfmr is specified for at least 50kohm nominal. You could use it; without input resistors, the FB resistor value should be equal to the sum of Pri. DCR + Sec. DCR. However, it would not really be a zero-field because the DCR of the secondary does not allow enough current to be reinjected back into the xfmr. You may use the alternative solution in the attached document.
TBD = To Be Determined.

Hi Abbey Road,

I have been playing around with both proposed circuits;

With the first one, I somehow have a high roll off which starts at about 2KHz and has its -3dB point at 10KHz. This happens regardless of the chosen feedback cap value.

The feedback resistor was implemented as a potentiometer so I could quickly see what it does. I started off with its value set to DCR pri + DCR sec (5K9+7K9).

There seemed to be no improvement in the distortion, and the feedback resistor potmeter seemed to act like a gain control, moving both the 50Hz testtone and all harmonics up and down in level, but at the same ratio. So no distortion improvement.

I then moved on to the second circuit, which yielded an almost flat frequency response. No more high roll off, but rather a high boost towards 20KHz which was worse with higher feedback cap values; the flattest response was achieved without feedback cap.

But again no distortion improvement.

Above tests were done using the half of an NE5534 (tried LME49860 also); feedback cap values were varied between 1pF and 1.5nF.

Does it matter if a BJT or JFET opamp is used? I read somewhere that TedF suggested BJT for this circuit.

The output of this circuit is connected to the positive in of the next opamp stage ("trim").

Anything else I can try? I've gone over the circuit a number of times and I can't spot/measure any cabling errors.

Thanks again!

BR, Rogy

Hi Rogy,

I think I spotted your mistake - 1/2 of a NE5534 is just not going to cut it - single opamp.... ;-)

but joke apart, you might have a look at the Studer implementation, they had a patent a exactly that int the analog consoles. studer has a ftp with all the goodies, even though the manual is long and repetitive, but its all there. look for the 9xx series of mixing consoles, may be not the latest one.

ftp://ftp.studer.ch/Public/Products/Mixing_Analog/

cheers,

michael

Hi Michael,

Oops I meant NE5532 of course  :-[ shouldn't have drunk that last beer yesterday night

Thanks for the Studer link, will look it over!

Cheers Igor

Rogy said:

With the first one, I somehow have a high roll off which starts at about 2KHz and has its -3dB point at 10KHz. This happens regardless of the chosen feedback cap value.
The feedback resistor was implemented as a potentiometer so I could quickly see what it does. I started off with its value set to DCR pri + DCR sec (5K9+7K9).
There seemed to be no improvement in the distortion, and the feedback resistor potmeter seemed to act like a gain control,

Click to expand...

That's predictable.

moving both the 50Hz testtone and all harmonics up and down in level, but at the same ratio. So no distortion improvement.

Click to expand...

As I said earlier, the inductance is too high for Zero-Field. The voltage across the windings is the about the same as in voltage transformer, so no improvement is to be expected. If you want to reduce THD, you must add high-value resistors at the input (about 100k) and scale the FB resistor accordingly. then I suspect the frequency response will be worse and noise will increase. A bipolar opamp would be very noisy with 100k impedance; a TL0 maybe a better solution, although its noise voltage is higher than bipolar.

I then moved on to the second circuit, which yielded an almost flat frequency response. No more high roll off, but rather a high boost towards 20KHz which was worse with higher feedback cap values; the flattest response was achieved without feedback cap.
But again no distortion improvement. 

Click to expand...

  What do you get for THD? The improvement should be at VLF (50Hz and below). Important: remeber that the original uses an attenuator before the primary. This reduces the flux and thus THD. What is the value of the components constituting this attenuator? 

Does it matter if a BJT or JFET opamp is used? I read somewhere that TedF suggested BJT for this circuit.

Click to expand...

Yes it does. BJT opamps (5532 type) have better noise voltage than FET opamps, but they have significant noise current. Generally they want to see an impedance of about 5-10 kohms.

Anything else I can try? I've gone over the circuit a number of times and I can't spot/measure any cabling errors.

Click to expand...

These xfmrs are simply not suitable for this application. In fact they work, but there is not much improvement to expect over the standard voltage transformer operation. I warned you about this before: "however, the best optimization happens with lower-turn windings".
Simulation shows that the second method is best in terms of frequency response.
Last thing to check. You may find that frequency response is better by cross-wiring one of the windings; will flip phase though.

Hello,

Using the second circuit and after having removed the input pad, the THD at 30Hz is 0.045%. Without the zero field and using the original pad, I measure 0.026%.

The original pad is a balanced -6dB pad, made up of two 1k series resistors and a 5k6 resistor across the primary. Will it affect operation of the second circuit if I put it back in place?

I tried cross-wiring the secondary and it effectively yields a flatter frequency response.

Even if nothing is gained with my Beyer trafo's, I learned and had fun. I'll definately try this with a more suitable trafo one day; thanks for educating me and all others around here.

Best,

Rogy

Rogy said:

Hello,

Using the second circuit and after having removed the input pad, the THD at 30Hz is 0.045%. Without the zero field and using the original pad, I measure 0.026%. 

Click to expand...

That proves that it's difficult to expect any gain with this xfmr which is not adeaquate. 

The original pad is a balanced -6dB pad, made up of two 1k series resistors and a 5k6 resistor across the primary. Will it affect operation of the second circuit if I put it back in place? 

Click to expand...

Yes, but you need to adjust gain and noise will increase in the same ratio.