transformer impedance and cheese

GroupDIY Audio Forum

Help Support GroupDIY Audio Forum:

This site may earn a commission from merchant affiliate links, including eBay, Amazon, and others.

5v333

Well-known member
Joined
Jun 30, 2013
Messages
631
Location
Gothenburg
i am unsure how to see this.

if i have a 15K puspull output transformer, what impedance does it want to see coming from to triodes?

is it 15K or 2-5 times lower?


when the feedback loop in my circuit is open, the low freq roll off is about 12hz but when i close the loop it moves don to about 3hz. and this must be because output impedance is reduced.

if i add a local plate to grid feedback, i can rebalance to get my roll off back up to 12hz.

i asked lundahl about the properties of my transformer and he pointed out that the roll off is depending on the source impedance and the inductance which in this case should be around 200H.

R/(2xPIxL) = f

15K/(2xPIx200) = 11,9hz


and by the way, what did the cheese say in the morning in front of the mirror?
 
Transformer doesn't "want" anything.

The Designer works it out to do a specified thing in specified conditions.

For transformer in your hand, best bass extension comes with Zero source impedance.

For single pole rolloff, you would expect 4:1 NFB to shift cutoff from 12Hz to 3Hz. Just due to gain relationships. Whether this can be accounted to "lower source impedance" is a chicken/egg cause/effect question which can't be answered on my first cup of coffee.

Why do you "rebalance" to get back to 12Hz? Big iron-core parts are NOT good low-cut tools. Their working inductance varies with current. If you cut it fine, the iron distorts. Make your OT too-good and then put a 12Hz filter at a low-level stage, probably R-C because film caps are pretty solid parts.
 
5v333 said:
i am unsure how to see this.
when the feedback loop in my circuit is open, the low freq roll off is about 12hz but when i close the loop it moves don to about 3hz. and this must be because output impedance is reduced.
As PRR wrote, it's one way of looking at it, because the dominant hi-pass effect is indeed due to the xfmr inductance.

if i add a local plate to grid feedback, i can rebalance to get my roll off back up to 12hz.
This is a little counter-intuitive, since anode-to-grid NFB also reduces the output impedance (not all NFB schemes reduce output Z); I would venture the input impedance, that is drastically reduced by this topology, interacts with the coupling caps.
+1 for creating a distinct C-R high-pass.
 
hmm.. my approach with the hpass and output impedance was prob wrong then, it lookd very good when scoping the other side of the transformer, my squarewaves looked  a litle bit cleaner and hi freq bandwidth  improved aswell.

since 12hz and 200H(which should be fairly constant???) told me that the source impedance is around 15K and my transformer is rated for 15K sources, i thought that 3hz and 200H would be a proof that the output impedance had changed...

i had a look now in the other parts of the circuit, and yes, there  are some lo freq bumps that prob should not be there...

ill have some experimenting with the interstage poles..!


the cheese says HELLO ME!





 
i havent drawn how the circuit looks like yet...

with the  feedback loop open, i have a RC interstage between the two triode stages, with a pole of about 13hz to match the output transformer rolloff.

i then close the loop with a cap and resistor that has a pole about 0,7hz. now the roll off after the output transformer has moved down to about 3hz.

should i be worried that my poles inside the circuit no longer are proper set because of the  roll off change? for ex should i change my interstage pole to 4hz and feedback pole to 0,4hz?

when i look at the output after the transformer on scope, its very straight and rolls off nice without any obvious resonances. however inside the circuit i have some crazy low freq resonances.

what matters here? since im not experianced i tend to worry alot about stuff that i probobly shouldnt care about...

 
5v333 said:
i havent drawn how the circuit looks like yet...

with the  feedback loop open, i have a RC interstage between the two triode stages, with a pole of about 13hz to match the output transformer rolloff.

i then close the loop with a cap and resistor that has a pole about 0,7hz. now the roll off after the output transformer has moved down to about 3hz.

should i be worried that my poles inside the circuit no longer are proper set because of the  roll off change? for ex should i change my interstage pole to 4hz and feedback pole to 0,4hz?

when i look at the output after the transformer on scope, its very straight and rolls off nice without any obvious resonances. however inside the circuit i have some crazy low freq resonances.

what matters here? since im not experianced i tend to worry alot about stuff that i probobly shouldnt care about...
Roll-off should not be created inside the loop. The BW should be more extended than needed, and the filters that define the actual BW should be placed before.
 
i have read some where that one pole should be set just above the  OT´s roll off... and the rest should be set at 1hz or lower...

but in that case i should try setting the poles as low as i can.


the cheese that can hide a horse: MASC AR PONY
 
5v333 said:
i have read some where that one pole should be set just above the  OT´s roll off...
That would be more or less correct in absence of NFB. In fact, since the LF saturation is a 2nd-order function, the high-pass should be 2nd-order (12dB/octave). Very often a 1st-order (6dB/octave) high-pass is used, and its characteristic frequency shifted about an octave higher.
Applying global NFB produces an apparent downward shift of roll-off, but in fact, the high-pass should be set as if there was no NFB. In other words, the high-pass filter must be set in accordance with the open-loop response.
 
abbey road d enfer said:
That would be more or less correct in absence of NFB. In fact, since the LF saturation is a 2nd-order function, the high-pass should be 2nd-order (12dB/octave). Very often a 1st-order (6dB/octave) high-pass is used, and its characteristic frequency shifted about an octave higher.
Applying global NFB produces an apparent downward shift of roll-off, but in fact, the high-pass should be set as if there was no NFB. In other words, the high-pass filter must be set in accordance with the open-loop response.

hmmm... that sounds like my plan is correct then, since the open-loop response is dominated by: the inductance of the pushpull transformer (200H) and the output impedance of the 6sn7 plates (about 15K).  which yields  -3db at 12hz. fc = R/(2xPIxL)
and there for the interstage RC between the prestage and the 6sn7 is set to just above 12hz. and then i close the loop with a RC pole that is 1hz or below.



i wonder what suprise is waiting for me when i get the time to do a sweep with DC set on my scope..?

 
5v333 said:
hmmm... that sounds like my plan is correct then, since the open-loop response is dominated by: the inductance of the pushpull transformer (200H) and the output impedance of the 6sn7 plates (about 15K).  which yields  -3db at 12hz. fc = R/(2xPIxL)
and there for the interstage RC between the prestage and the 6sn7 is set to just above 12hz. and then i close the loop with a RC pole that is 1hz or below.
I would really set the interstage RC at a much lower value, which would insure the dominant pole is only the one created by the Ra/Zb combination and no additional phase-shift is created there, playing havoc with NFB. I would establish a dominant pole at about 16Hz passively by adding a capacitor before the grid of the 1st stage.
 
Ian also pointed out that the two poles (interstage&feedback) should have a decade between them atleast.
i wonder if it matters which one is the higher and which is the lower pole.
 
before the first grid is an input transformer that connects each end with the grids of a 12ay7. center tap connected to ground.

the rolloff from the input transformer is about 2,5hz
 
5v333 said:
Ian also pointed out that the two poles (interstage&feedback) should have a decade between them atleast.
i wonder if it matters which one is the higher and which is the lower pole.
Yes, it matters; the pole governed by Ra/Zp is difficult to improve (lots of copper/iron), the interstage RC is much easier to optimize. The main reason for tayloring the LF response is avoiding saturation effects in iron, capacitors don't have such issues.
 
5v333 said:
before the first grid is an input transformer that connects each end with the grids of a 12ay7. center tap connected to ground.

the rolloff from the input transformer is about 2,5hz
That is an issue; if there is nothing to roll-off LF before in the chain, it is hard to protect the OT against saturation.
 
why is that a problem?

perhaps the 2,5hz comes from an earlier stage by the way (before the input transformer is my passive eq with another input transformer as the front end of everything ).

 
Back
Top