Grid Stopper Advice - Graphs Included!

[Magz]

Hello all,

I am running a Supratek Chenin preamp which I have modified to use a cascoded IXYS 10M45S CCS on the plate, Cree SiC Schottky Diode bias, and single 2C22 output tubes in a parafeed-type configuration, replacing direct-coupled 6SN7 output tubes (too much gain!). Output transformer is a Lundahl LL1674 and Parafeed cap is 0.56uF. The CCS has 1k gate and anode stoppers already installed.

Here's the problem: I am currently installing the Lundahl x-former and am measuring its effect using TrueRTA on my Laptop. When I ran a square wave through the preamp I saw pretty severe ringing on a the leading and trailing edges of a 1kHz square wave (first plot below). There was no grid stopper stock on the 6SN7, but the 2C22 tube has the grid as a top cap, so it needs a longer lead to reach the cap, which is probably exacerbating the problem with stray inductance compared to the 6SN7.

I have tried a series of carbon comp resistors soldered directly to the top cap; plots 2 and 3 below are a 55k and a 79k grid stopper, and in the next post plot 1 is a 100k and plot 2 is a 150k carbon comp - they do decrease the ringing quite a bit.

Calculating the -3dB point using the Miller capacitance of the 2C22 shows a -3dB point of 21kHz with a 100k grid stopper. Based on that I'd rather go with the 79kohm resistor, but there is still some ringing on the trailing edge of the square wave.

The question is: How much ringing is acceptable? I'd like to eliminate it altogether, but not at the expense of frequency response. Any advice is greatly appreciated!

[Magz]

Here are the plots for the 100k and 150k grid stoppers, respectively:

[astouffer]

Do you really mean 100,000 and 150,000 ohms for grid stoppers? They are meant to stop high frequency oscillations and should be 100 or 150 ohms. Ringing is usually caused by the output transformer. Try looking into snubbers instead.

[Magz]

I tried 1k carbon comps on the grid and they did nothing. I needed to get into the 10s of kohms to see any effect at all. Once in that range, they do seem to stop most of the ringing, and the frequency response still looks very good from 20-20kHz with up to a 100kohm stopper. If the issue is with the transformer, why do the high value grid stoppers seem to work?

I intend to look into lead dress as well - the Supratek preamps are a bit "disorganized" in terms of their point-to-point wiring. They sound very nice, however!

[Magz]

Hmmmm....

I found the reference linked below which states that:

"The use of a square wave to analyze an amplifier is shown in full in Tremaine’s Audio Cyclopedia, 2nd edition, Audio frequency measurements chapter, pg. 1521, as a means to test the transient response of an amplifier. If the amplifier has ringing at the top of the leading and falling edge, this shows an extremely good high frequency response from the transformer in use."

Am I correct in interpreting this to mean that graph#1 in the first post above is what I WANT the square wave output to look like? If so, then I guess I'm already there without any grid resistors!

Can someone please confirm that this is correct?

Electra-Print.com Tech Blog


.

[Wavebourn]

Did you load the transformer when measured?

[DualTriode]

Hello Magz,
I trust what Jack has to say. I have bought several output transformers from Jack. Correct? No I do not think so. In the link that you provided Jack’s entire point is do not use square waves to test audio devices. The fact that you see ringing under test is not a positive result. What was Jack’s analogy don’t blow backwards through a clarinet you may not like the sound.
DT
All just for fun!

[Arnulf]

A: Fc just above audio (at 21 KHz) seems absurd for audio and even more so for your test as it does exactly the opposite of what you're trying to achieve. Square wave contains odd harmonics many many levels above the first. This .gif should be animated, it displays square wave generated with first 15 harmonics. Yours, with 19, is just marginally better. Notice remaining ringing at the end of the animation sequence ?

B: Instead of relying on Cmiller to match your expectations (because it usually won't, at least not in real life) you should be using a fixed known value capacitor from grid to cathode.

C: For the sake of discussion, route signal output directly into input (bypassing your amplifier) and run your test again to see whether your sound card is capable of producing suitable square wave in the first place - chances are it might be adding some distortion of its own thanks to finite sample resolution.

[kenpeter]

Stubs are resonant pieces of transmission line. And if unshielded,
also act as resonant antennas. A grid by itself could be a stub,
and a receiving antenna, and a transmitting antenna, and an
amplifier. And even if shielded from the outside world, its sitting
in an electric field, and a stream of moving electrons modulated
by its own resonances. See where this could become a problem?
Moving electrons is also a Magnetic field. So go figure..

The end impedance of a Beverage longwire antenna might be 450
to 800 ohms. Midpoint impedance of a dipole antenna might be 25
to 50 ohms... If you want to terminate and capture energy from
any antenna into a dummy load, it would need to be somewhere
within this range.

If you make the stopper value too large, instead of de-Q-ing the
problem stub and removing resonant energy. You simply cut the
stub in two high Q pieces that each can resonate. Maybe with a
capacitor now bridged across the gap where you think you have
put a stopper...

If you need a large value resistance to limit a current or prevent
bias pumping, then you still also need small value physically near
the grid, to stop it from behaving as a high Q resonant stub.

Irrelevant trivia factoid: The characteristic impedance of free
space is 377 ohms. Not suggesting this makes ideal stopper.
In a wire, the value could translate 25 to 800. And different
for every frequency vs length. Standing waves are a mess.

[SY]

First and biggest problem- you're using the wrong tool to measure. Doesn't it strike you as odd that the transformer starts ringing before the impulse?

This is one where a good old fashioned analog scope is the right thing to use, with the source signal coming from a function generator, not the sound card.