Transformer secondary resistors?

Hi all,
Likely an easy question for some of you. In many schematics with input and output transformers, sometimes we see a resistor across the secondary windings. I somewhat understand their purpose in a zobel network but what do they really do when there is no capacitor involved?

For example, the original EQP1A schematic has a 600ohm resistor across the secondary of the input transformer before the passive filter stage. Jakobs schematic has a 10k.

I could be mistaken (and I usually am...), but I think the 10K across the output secondary on Jakob's G9 is to keep the polarity switch from going bump in the night when you switch it.

From the circuit description on www.gyraf.dk...
"The output of the transformer is taken to SW4, the phase reverse switch, that has 10K resistors across
it to control transformer load when switching. And then to the output XLR....."

It sets the impedance. The inductance of each coil in the transformer has a certain inductance / reactance that defines the characteristic impedance that it can work with before you get low frequency loss or distortion or high frequency peaking and such. Meaning a transformer doesn't have a specific impedance like a resistor. It just reflects the impedance that it "sees" on the other side. So there should be a well defined impedance such as in the form of a resistor on at least one side to set a well defined impedance of the network surrounding the transformer. If the characteristic impedance of the transformer secondary is 600 ohms, it is customary to have a 600 ohm resistor or maybe 680 ohms would be a standard resistor value in range. A 10K transformer would use 10K.

A Zobel network is somewhat unrelated. It is very common to get "peaking" at the high end of the frequency response when using transformers.  At least the good ones usually do. To correct for this, a low pass filter is applied in the form of a capacitor and resistor to bring the peaking down just enough so that the peaking is damped.

You see the occasional old schematic with multiple x2 loadings, like 1200Ω on each side of a 600:600, or even (4) 600Ω, one on each split winding of a center tapped 600:600.    Old Altec tube preamps are a rare US example of matched loading also, 82K secondaries with 82K resistors on them, as opposed to the usual open grid arrangement. 

daskew81 said:

In many schematics with input and output transformers, sometimes we see a resistor across the secondary windings.

Click to expand...

A transformer without a load is a variable impedance, that increases with frequency. Electronic ciorcuits that don't rely on heavy NFB are sensitive to the impedances they are connected to.
Tube circuits in particular, but also many SS need to see controlled impedances. The stage the xfmr is connected to may have very large gain variations, that may not only alter the frequency response, but also result in unstability/oscillation.

I somewhat understand their purpose in a zobel network but what do they really do when there is no capacitor involved?

Click to expand...

  But here are always capacitors involved!  Particularly in xfmrs. There is capacitance between layers of wire, that appear as in parallels with the windings, and there is capacitance between primary and secondary, and between windings and frame/core.
Thesecapacitance react with the leakage inductance to produce a resonant Low Pass filter

For example, the original EQP1A schematic has a 600ohm resistor across the secondary of the input transformer before the passive filter stage. Jakobs schematic has a 10k.

Click to expand...

The EQP1A is a passive EQ. Its correct operation implies perfectly controlled input and output impedances. If the impedances were not correct, the response would be skewed.


A Zobel circuit is not unrelated, on the contrary; it's a way of compensating the increase of impedance, by applying progressively more load (damping), without unnecessarily loading the circuit at low and mid frequencies, which results in improved distortion and headroom compared to simple loading.

probably  helps keep distortion levels more constant,

all transformers require a bit of current to get them operating,

this exciting current is non-sinusoidal.

the output waveform is a combination of the exciting current and the signal current, so you can put a resistor on there to draw more signal current to help offset the distortion caused by the magnetizing current. some circuits have enough of a load to do this without the resistor, the EQP must have a larger input Z than 600 ohms,

I have a related question to this. Do you have a zobel-network AND a terminating resistor in parallel. Or is the resistor in the zobel de facto the terminating resistor?
/
Emil

Studio Mollan said:

I have a related question to this. Do you have a zobel-network AND a terminating resistor in parallel. Or is the resistor in the zobel de facto the terminating resistor?
/
Emil

Click to expand...

Sometimes you see both.  Look at the Jensen documents. 

Studio Mollan said:

I have a related question to this. Do you have a zobel-network AND a terminating resistor in parallel. Or is the resistor in the zobel de facto the terminating resistor?

Click to expand...

The termination resistor and the Zobel have different purposes.
The termination resistor stabilizes the primary impedance, when the Zobel deals with the HF resonance.
However they both interact. When there is a termination resistance, the Zobel must be significantly altered. To the point where, sometimes, the Zobel has so little effect it can be dispensed with.

What would be typical value for say a 150:600 output transformer?

I'm asking because I used once a carnhill in 150:600 configuration and tgere was no ringing so I didn't use any zobel. Neither a load resistor. Is this wrong practice?

warpie said:

What would be typical value for say a 150:600 output transformer?

Click to expand...

very hard to say. It could be all over the place. Typically the resistor would be in the 500-2k ballapark and the capacitor a few nF.

I'm asking because I used once a carnhill in 150:600 configuration and tgere was no ringing so I didn't use any zobel. Neither a load resistor. Is this wrong practice?

Click to expand...

Although in theory all xfmrs could benefit from a Zobel, some xfmrs don't require these elements as long as they are driven and loaded with the right impedances. These low-Z xfmrs are ringing in the MHz region, so you may not see them.

abbey road d enfer said:

very hard to say. It could be all over the place. Typically the resistor would be in the 500-2k ballapark and the capacitor a few nF.

Click to expand...

Sorry if I wasn't clear. I know how to adjust the values on a zobel network. My question about a typical resistor value is about the termination resistor. From what I remember I have seen values of 620 Ohm as well as 1k2 Ohms and how can I know if a termination resistor is needed at all? What would it depend on?

warpie said:

My question about a typical resistor value is about the termination resistor. From what I remember I have seen values of 620 Ohm as well as 1k2 Ohms and how can I know if a termination resistor is needed at all? What would it depend on?

Click to expand...

Essentially, the response should be as flat as possible and witout ringing. It depends on the construction and on the circuit that drives the xfmr. Circuits with little or no NFB have a high drive impedance and may be sensitive to the primary-reflected impedance.
Some xfmrs have their response dependant on a specific load, like 600 ohms. In that case, if the are loaded with a higher Z (typically most modern inputs are 10-20k) they require an additional load resistor.
In the end it's the frequency response that counts.

thanks everyone for this...

and CJ, thats really interesting - i have measured that resistive secondary termination not only flattens response but also lowers THD, in particular 2nd order harmonics. nice to know what is actually happening to cause that. the current thru the winding is something i hadn't considered.

To the OP - this is easy to observe on the bench if you have a signal generator that will go above 200kHz.

Pretty much any of the common vintage or modern transformers will show some amount of resonance starting before that, unloaded vintage transformers can easily exhibit 15 - 20dB of peaking - if you have the ability to sweep thru the region you can see it.

empirically determining the value of the resistor termination to overcome this resonant peaking is as simple as subbing a 5k or 10k pot hooked up as a rheostat across the secondary and repeating the sweep while lowering the resistance of the pot/rheostat until the peak is gone and the response is flat thru the target bandwidth and sloping down gradually above it.

you can confirm this with a square wave test. The square wave should show little to no ringing but also not exhibit premature rolloff.

one thing to consider tho is that this termination is in parallel with the load - so if there is an instance were you will have a hard 600Ω load, this corrective termination will not only change the bandwidth/response of the transformer, but also lower the overall load as seen by the amplifier driving the transformer.

This may or may not be a problem but should be kept in mind...

ok!
T.

mutterd said:

i have measured that resistive secondary termination not only flattens response but also lowers THD, in particular 2nd order harmonics. nice to know what is actually happening to cause that. the current thru the winding is something i hadn't considered.

Click to expand...

The explanation is quite simple: when the secondary is loaded, the primary voltage decreases and the reflected impedance appears in parallels with the primary. The improvement in THD is the result of less voltage and a lower equivalent drive source impedance. The cost is the transfer loss.

abbey road d enfer said:

The explanation is quite simple: when the secondary is loaded, the primary voltage decreases and the reflected impedance appears in parallels with the primary. The improvement in THD is the result of less voltage and a lower equivalent drive source impedance. The cost is the transfer loss.

Click to expand...

Also trying to wrap my head around this. Lets say I want to build a DIY passive transformer color box as an I/O. My interface has a 100Ω output and a 10kΩ input. So when using a 600:600 transformer and a 620Ω resistor loading the secondary you will improve THD.
Can you compensate the THD "loss"/ increase the saturation with more gain?
In this particular case what would happen if you "switch off" the resistor?

Not sure why you would want to improve THD in a color box? 600:600's don't sound like much in the first place, usually.

Not actually looking to. It was my question due to abbey road d enfer's comment:

abbey road d enfer said:

The improvement in THD is the result of less voltage and a lower equivalent drive source impedance

Click to expand...

And I'm really interested in the answer to this two questions

eazyhiro said:

Can you compensate the THD "loss"/ increase the saturation with more gain?
In this particular case what would happen if you "switch off" the resistor?

Click to expand...

Depends on the transformer what happens when you toggle the resistor. There’s a broad range of ways that can go from nearly nothing to something, practically speaking.

eazyhiro said:

Also trying to wrap my head around this. Lets say I want to build a DIY passive transformer color box as an I/O. My interface has a 100Ω output and a 10kΩ input. So when using a 600:600 transformer and a 620Ω resistor loading the secondary you will improve THD.

Click to expand...

Yes, but imperceptibly.

eazyhiro said:

Can you compensate the THD "loss"/ increase the saturation with more gain?

Click to expand...

Yes.

eazyhiro said:

In this particular case what would happen if you "switch off" the resistor?

Click to expand...

It would slightly increase THD AND level.