Tube stage HF roll-off problem

[abbey road d enfer]

As far as I can tell now, it's the impedance mismatch between the reflected impedance to the secondary and input grid impedance.

Agreed, but the grid impedance is supposed to be nearly infinite, except when Miller effect sets in, but with Cag1 at 0.05pF, even with a gain of 100, it does not count.
You need to understand why the impedance is not correct. It seems it's not only capacitive, but also resistive, since there is attenuation at low and mid frequencies.
Can you drive the input from a known resistance, let's say 50k or 100k and graph the frequency response as well as the attenuation?

 

[ruffrecords]

Yes it could well be Ian, but Sowter offer this as a replacement for a v76 input trafo, and the v76 has an input impedance of 200 ohms I think. Looking at Ilya's inductance measurements at 100Hz (370mH each pri/1970H sec) then with primaries in series this will theoretically be 1.48H total, which then not considering the load on the secondary, this L presents an impedance of 465 ohms at 50Hz, or 232.5 ohms at 25Hz. To me this is Sowter cutting it as close as they can for a 200 ohm input Z and minimal secondary loading. With such a high ratio maybe they needed to so that the secondary wasn't a million turns.

The Sowter 1:10 transformers I use in my tube mic pres have a primary inductance in the region of 5H so the value you are reading seems to me to be more suitable for a 50 ohm mic than a 200 ohm one. It is worth remembering that a transformer's inductive reactance is usually expected to be about 5 to 10 times the source impedance at the lowest frequency. However, if it is too low, what you tend to get is LF roll off not the HF roll off you are seeing. Your problem is certainly a strange one.

Ceers

ian

 

[Ilya]

I've made some tests with the primary loaded with 56R and the response improved quite a bit.
Here's the primaries in series:

Here's the primaries in parallel:

In both graphs 56R terminated plot is the lower line

 

[Ilya]

Can you drive the input from a known resistance, let's say 50k or 100k and graph the frequency response as well as the attenuation?

Here's a series of curves that I get when I drive the grid directly (via 0R, 47K and 100K resistors). They are virtually identical besides the 100Hz hum pickup due to my crude test setup:

And this is the grid via 1M drive:

 

[ruffrecords]

The primaries in series plot with the 56R source looks reasonable now. It even has the a peak near 40KHz.

Cheers

Ian

 

[Ilya]

I played with the secondaries swapping and the shield. The best I could get is when I swap them and disconnect the static shield from ground. Not dramatic results, but still:

The primaries in series plot with the 56R source looks reasonable now

Yes, but that will load the mic and how would that affect the output of the mic circuit?

 

[emrr]

I agree tangentially with ruffrecords about apparent intended source Z ASSUMING (big assumption) that all is right with the transformer. Series primary appears the only remotely reasonable connection as experienced and it’s still a heavy load. As pointed out, DCR is very low, which seems true of many Sowter. Larger gauge wire than is optimal, certainly very different from old school designs.

Maybe this German replica works great with a bunch of modern Neumann/etc 50 ohm condensers, but then who needs the gain?

 

[abbey road d enfer]

Here's a series of curves that I get when I drive the grid directly (via 0R, 47K and 100K resistors). They are virtually identical besides the 100Hz hum pickup due to my crude test setup:
View attachment 97321
And this is the grid via 1M drive:
View attachment 97322

Assuming both graphs have been made with the same settings, it shows that the grid loading is negligible, however it's hugely detrimental to the response of the Sowter xfmr.
Can you measure the source impedance the grid sees?

 

[ruffrecords]

Yes, but that will load the mic and how would that affect the output of the mic circuit?

Series connection loads it less than parallel connection.

Cheers

Ian

 

[Ilya]

Can you measure the source impedance the grid sees?

If I read your test setup correctly, feeding -10dBu results in -50dBu across 1K resistor.

Series connection loads it less than parallel connection.

Do I understand this correctly that I can strap 56R across the primaries and be done with that?

 

[ruffrecords]

Do I understand this correctly that I can strap 56R across the primaries and be done with that?

No, I am saying your transformer works better with a 50 ohm source. So you either need to use a 50 ohm mic with it (back in the day 50 ohm mics were very common) or change the transformer. to something like a 1:10 Cinemag.

To be honest, I am not quite sure what the original spec of the V76 input transformer is. Sowter make 'replicas' which are modern equivalents rather than exact copies. This means they use modern techniques and materials to ostensibly achieve the same function as the original but they do not always succeed.

Cheers

Ian

 

[abbey road d enfer]

If I read your test setup correctly, feeding -10dBu results in -50dBu across 1K resistor.

No. It maybe so at one frequency but it's important to see how it varies with frequency.

 

[Tubetec]

The split multi chamber formers found on many of the V series pre's transformers and inductors are long since gone , could a U-I lamination with dual split bobbins be used instead (4 chamber)? In any case how well does the original V76 input transformer suit impedences , levels and cable runs found in modern day gear ? music got a lot louder in the late 60's with the advent of the 100 watt tube amp and walls of speaker cabs . In the broadcast sector long balanced line drive and receive capabillity was hugely important ,now not so much . Is there really any point in sticking with a non optimal wind ratio in the pursuit of a memory of gone-by days ? I do understand thats theres an element of synergy going on with old gear , sometimes a short coming in the freq resp. of one wound component was compensated with boost in another so a more or less even response was the outcome , when you change one thing that whole balance is thrown out of wack and you end up chasing your tail performance wise .

 

[gyraf]

Still, V76 magic is in the iron - and who would we be if we gave up just because it's hard?

And yes, I believe the input transformer and the choke here to be very, very hard to get right - which is why there are no commercial clones of this gem. That sounds right, that is.

/Jakob E.

 

[CJ]

Try measuring the response after the second stage, that feedback wafer switch has a built in frequency compensator.

 

[vinyvamos]

Still, V76 magic is in the iron - and who would we be if we gave up just because it's hard?

And yes, I believe the input transformer and the choke here to be very, very hard to get right - which is why there are no commercial clones of this gem. That sounds right, that is.

/Jakob E.

Exactly, and it's not like we are trying to recreate some extinct vacuum tubes, blowing glass and playing with toxic materials in our basements ;-).
But yes, recreating these wound components accurately is a challenge and the trafo companies will never understand the obsessed madness of us sonic-nostalgia junkies. There are always "modern" ways to make something "work", but is it the same?

The only remake of the v76 that I know of is the Mercury M76 and they seem to be using all cinemag iron.

Even the solid state TAB stuff is full of very tricky coils. But if it was easy peazy then would be here on this thread?

FWIW I have had a similar issue with Lomo YP27 input trafos. They are designed for 50 ohm mics with a 1:14 ratio. If I don't strap the primary with a low R resistor (68R) then the Freq response is wild! First seen testing with a line out pad using 2x270R & 68R shunt. Alternatively I can load the secondary with an R that reflects low Z to the primary, but either way I am loading the mic too much...

 

[Ilya]

No. It maybe so at one frequency but it's important to see how it varies with frequency.

Here's what I get when I feed the 0dBu into the setup:
@10Hz = -41.6dBu
@100Hz = -44.2dBu
@1kHz = -44.6dBu
@10kHz = -43.7dBu
@20kHz = -41.7dBu
@50kHz = -35.8dBu
@100kHz = -34.8dBu

Try measuring the response after the second stage, that feedback wafer switch has a built in frequency compensator.

The HF response goes down there as well, the behavior is identical. What frequency compensator do you mean? series caps in the feedback loop?

The only remake of the v76 that I know of is the Mercury M76 and they seem to be using all cinemag iron.

These are V76m versions with 1:10 input transformer. Cinemag has some nice 1:10 species which I'm going to try in this unit.

That's a real pity that Sowter advertises this transformer as 200:180K while it's more like 50:45K. Even if we take the parallel primaries out of equation this still leaves the pad problem. It seems that the 2x3.16K in series + 200R shunt present too heavy load to the input stage, although as Abbey pointed out somewhere, it should have a huge impedance. My previous graph shows that feeding a signal via 1M starts to degrade HF response.
I'm at a loss now and don't see how this issue can be fixed besides swapping the input tx.

 

[CJ]

50uf cap gets closer to ground as you dial the gain sw