Tube DIY Asylum

Do It Yourself (DIY) paradise for tube and SET project builders.

Return to Tube DIY Asylum

Message Sort: Post Order or Asylum Reverse Threaded

Frequency Compensation

63.246.183.223
Posted on March 1, 2015 at 15:10:02
Triode_Kingdom
Audiophile

Posts: 10049
Location: Central Texas
Joined: September 24, 2006
 I decided to design a driver circuit for a new SET amp this weekend. In the course of tuning it up, I was reminded how difficult it can be to get all the tubes to play nice, even in simple circuits.

One issue I come across again and again is the loss of high end frequency response across multiple stages. For example, the 6SL7 stage I threw together works flawlessly when driving my scope probe. So does the 6SN7 it's intended to drive. However, connect the two together, and the 20kHz response falls off by a couple dB. The problem is that the 6SL7 can't drive the capacitance at the input of the 6SN7 without some loss of amplitude at higher frequencies.

Below is the method I use to compensate for this (RC at cathode). In fact, although the problem really lies with the 'SL7s output Z, I applied this technique to both stages. That way, no single stage has to provide all the necessary compensation. The 1dB response is now about 65kHz, and 3dB about 125kHz.

I use various methods to tune amplifiers, depending on the circuit and the type of problem it presents. I'm just wondering, how does everyone else handle this particular issue? I seldom hear it discussed.









 

Hide full thread outline!
    ...
RE: Frequency Compensation, posted on March 1, 2015 at 15:17:15
hennfarm
Audiophile

Posts: 535
Location: Oregon
Joined: October 8, 2008
 Low RP drivers. Parallel both sections ofa double triode or currently I triode rig a video pentode. 11hm7, 12hl7, 12gn7, 12en6, blah blah blah. Can you hear up to 20khz? JH

 

RE: Frequency Compensation, posted on March 1, 2015 at 15:22:08
Caucasian Blackplate
Industry Professional

Posts: 8313
Location: Seattle
Joined: June 18, 2004
 I deal with it by not using 6SL7's or 12AX7's.

Your circuit with no bypass cap has an output impedance of about 100K and gain of about 23dB, which is pretty rough. A full bypass of the cathode resistor cuts the output impedance in half and gives you 34dB of gain instead.

A 12AT7 with a CCS will give you around 30dB of gain and output Z under 15K. That's how I would deal with it.





 

RE: Frequency Compensation, posted on March 1, 2015 at 15:35:54
Triode_Kingdom
Audiophile

Posts: 10049
Location: Central Texas
Joined: September 24, 2006
 I've always found the 'SL7 and 'SN7 to provide excellent sound, and I believe they're more linear than most other octals and miniatures. Also, I'm building this particular amplifier with pre-WWII outputs, and I wanted to keep the late '30s look to the tube complement. That - plus the gain requirement - precludes the use of most other tubes for this.

FWIW, I did try paralleling both sections of the SL7 and bypassing the cathode. Neither technique produced flat response to 20K (I didn't try both at the same time).

I might try your suggestion for a CCS on some upcoming projects. Thanks for your input!


 

I use them all the time..., posted on March 1, 2015 at 15:40:54
Chip647
Audiophile

Posts: 2653
Location: The South
Joined: December 24, 2012
 but mainly in guitar amps. I have thought about it for hi-fi tone control by making it switchable.

 

RE: Frequency Compensation, posted on March 1, 2015 at 15:43:26
johnsonad
Audiophile

Posts: 759
Location: Central California
Joined: June 9, 2008
 Realistically, as there isn't much content at 20kHz and since 98% of people who can afford to build and listen to such a piece of gear can't hear anything up there anyway, why put so much effort into it?

 

RE: "20kHz response falls off by a couple dB.", posted on March 1, 2015 at 15:53:23
Ivan303
Audiophile

Posts: 48887
Location: Cadiere d'azur FRANCE - Santa Fe, NM
Joined: February 26, 2001
 Hook your ear up to it.

Fall-off likely more that a couple dB. ;-)


First they came for the dumb-asses
And I did not speak out
Because I was not a dumb-ass

 

RE: Frequency Compensation, posted on March 1, 2015 at 16:04:26
Paul Joppa
Industry Professional

Posts: 7296
Location: Seattle, WA
Joined: April 23, 2001
 6SN7 has about 30% more Miller capacitance than a 300B.

Driving a Miller capacitance has two components. Usually we just worry about the current, which is significant when shoving 70 peak volts into a 300B grid, but not important at 1-2 volts. But the frequency rolloff is not sensitive to voltage level.

Incidentally, the Miller capacitance of the 6SL7 is even worse!

 

RE: Frequency Compensation, posted on March 1, 2015 at 17:21:17
Caucasian Blackplate
Industry Professional

Posts: 8313
Location: Seattle
Joined: June 18, 2004
 Try using a pentode input instead. If you can find something with decent transconductance, you can use a pretty low R plate load and drive that 6SN7 easily.

 

RE: Frequency Compensation, posted on March 1, 2015 at 19:17:25
Tre'
Industry Professional

Posts: 17306
Location: So. Cal.
Joined: February 9, 2002
 If it's 2db down at 20kHz then the -3db point can't be much above that.

The phase will be shifted a full decade below the -3db point, so a -3db point of 30kHz give phase shift all the way down to 3kHz.

Tre'
Have Fun and Enjoy the Music
"Still Working the Problem"

 

RE: Frequency Compensation, posted on March 1, 2015 at 20:42:14
johnsonad
Audiophile

Posts: 759
Location: Central California
Joined: June 9, 2008
 Thank you for the further understanding!

 

Thanks for the Comments, posted on March 1, 2015 at 21:22:07
Triode_Kingdom
Audiophile

Posts: 10049
Location: Central Texas
Joined: September 24, 2006
 Lots of good ideas thrown out on this. I'll try some of them in upcoming projects. FWIW, the two stages I built (6SL7/6SN7) are ruler flat. They have a combined voltage gain of about 170, and maximum output is greater than 200V p-p. That's more than I need in theory to drive the triode output stage beyond clipping. Hopefully, I can get one channel prototyped within a week or so and test it with this driver. Thanks to everyone for all the good ideas!

 

RE: Frequency Compensation, posted on March 2, 2015 at 02:51:27
Naz
Audiophile

Posts: 2184
Location: Sydney
Joined: September 2, 2005
 Also, there are many articles about life above 20kHz, why do we even bother with hi res digital if not. Why does vinyl sound so good? Ever tried Supertweeters, they really will convince you.

Naz

 

A late comment ..., posted on March 2, 2015 at 03:20:46
Naz
Audiophile

Posts: 2184
Location: Sydney
Joined: September 2, 2005
 FWIW, I think your compensation method is as good as any from a frequency response POV and I have used it myself in the past.

I do have a technical bias against it due to the loading ratio the following stage's input imposes on the the driving tube. Reason being that the mechanism causing the rolloff is essentially high OP impedance, (as you know) hence too much load on the driving tube at HF and I don't think that's a good thing for high order distortion (elliptical load line). The partial saving grace is that bypassing HF will reduce the OP impedance at HF but I prefer to use tubes with low enough OP impedance to drive the next stage.

As Paul pointed out it depends on the tube you are driving and for that matter, the type of circuit and even its gain where local NFB is utilised.

Naz

 

If you have $2..., posted on March 2, 2015 at 05:00:10
Chip647
Audiophile

Posts: 2653
Location: The South
Joined: December 24, 2012
 Really fun to play with is putting a MOSFET source follower between the driver and the output tube. Not really required in A1, but the MOSFET is really awesome in that role.

 

RE: Frequency Compensation, posted on March 2, 2015 at 07:03:06
TimFox
Audiophile

Posts: 697
Location: Chicago
Joined: January 14, 2006
 Another method, used in video amplifiers, is to add an inductor in series with the plate load resistor. See any of the 1940 to 1955 vacuum-tube textbooks for details on optimum value.
In general, it is difficult to obtain wide bandwidth at high impedance levels.

 

RE: Frequency Compensation, posted on March 2, 2015 at 09:32:37
rage
Audiophile

Posts: 793
Joined: December 17, 2010
 does this differ from what DF and JM are doing with multiple bypass caps? are they going to have a simliar effect when paralleled with multiple values?

 

RE: A late comment ..., posted on March 2, 2015 at 10:13:24
Triode_Kingdom
Audiophile

Posts: 10049
Location: Central Texas
Joined: September 24, 2006
 "too much load on the driving tube at HF and I don't think that's a good thing for high order distortion (elliptical load line)"

I'm hoping for a worse case being that this shows up only as increased harmonics of fundamentals about 10 kHz and above. That should render the resultant harmonic signature (20 kHz and up) reasonably inaudible. I'm planning to characterize THD and IM after the output stage is installed. If the overall design doesn't perform well, I might need to revisit this issue.



 

RE: Frequency Compensation, posted on March 2, 2015 at 12:15:04
Tre'
Industry Professional

Posts: 17306
Location: So. Cal.
Joined: February 9, 2002
 TK is only bypassing the cathode at the high frequencies. By my calculations, about 5kHz and up. That stops the current feedback for the high frequencies so the gain at those frequencies is higher.

What DF does is different. DF bypasses the cathode at all but the lowest frequencies.

Tre'
Have Fun and Enjoy the Music
"Still Working the Problem"

 

RE: Frequency Compensation, posted on March 2, 2015 at 12:16:48
rage
Audiophile

Posts: 793
Joined: December 17, 2010
 So essentially, if I understand correctly, each of the many bypass caps DF uses would have a simliar effect... (except bosting a multitude of frequency ranges)

 

RE: If you have $2..., posted on March 2, 2015 at 13:55:50
Eli Duttman
Audiophile

Posts: 10455
Location: Monroe Township, NJ
Joined: March 31, 2000
 I was thinking in terms of buffering the 6SL7. The little ZVN0545A exhibits tiny capacitances that even wimpy types, like the 6SL7, drive with aplomb.

Buffering the voltage gain triode with a MOSFET should eliminate the need for freq. compensation.


Eli D.

 

RE: Frequency Compensation, posted on March 2, 2015 at 14:15:16
Tre'
Industry Professional

Posts: 17306
Location: So. Cal.
Joined: February 9, 2002
 No, the total uf value of all of DF's caps added together is the cathode bypass capacitor value.

Using smaller "side caps" is a different thing.

Bypassing a tube's cathode is one thing, "side capping" the cap used to bypass a tube's cathode, in an attempt to change or "fix" the "deficiencies" of capacitors, is another.

We also see side capping used on power supply caps and coupling caps in a attempt to improve some attributes of the main cap.

Bypassing a cathode (to the lowest frequency of interest) is about preventing local current feedback and restoring the tube's plate resistance.

Bypassing a cathode to less than the lowest frequency of interest is "Frequency Compensation".

Again, side capping is a different thing.

Tre'


Have Fun and Enjoy the Music
"Still Working the Problem"

 

RE: Frequency Compensation, posted on March 2, 2015 at 16:03:29
rage
Audiophile

Posts: 793
Joined: December 17, 2010
 Thanks Tre'!

 

RE: Frequency Compensation, posted on March 3, 2015 at 19:43:01
91derlust
Audiophile

Posts: 1101
Joined: December 25, 2014
 Does that mean that the phase shift issues mentioned elsewhere occur higher up frequency, perhaps where the ear is more sensitive?

regards,
91
"Confusion of goals and perfection of means seems to characterise our age." Albert Einstein

 

RE: Frequency Compensation, posted on March 3, 2015 at 21:15:56
Tre'
Industry Professional

Posts: 17306
Location: So. Cal.
Joined: February 9, 2002
 Yes.

A high pass filter, like bypassing a cathode, will shift phase one decade above the -3db point. So if the cathode is bypassed to 2Hz the amplitude and phase will be correct at 20Hz. If the cathode is only bypassed to 7Hz then the phase is shifted all the way up to 70Hz.

A low pass filter, like the output impedance of a driver stage vs. the Miller capacitance of the output tube, will shift phase a decade below the -3db point. If the low pass (output impedance vs. Miller capacitance) has a -3db point of (let's say) 50kHz, the phase will be shifted all the way down to 5kHz.

If we want the amplitude to be flat and the phase to be not shifted, 20-20kHz, then all the low pass filters need to have their -3db points at about 2Hz and all the high pass filters need to have their -3db points at about 200kHz.

Of course there are other causes within a system that limit the frequency response and cause phase shifts. Like the output transformer, speakers, etc. But that's not an excuse to design our stages wrong.

I believe we should take care of the filters that we can take care of and not look for excuses not to.

Loss of amplitude and phase shift within a system will add. Minimizing them, where we can, will pay off.

Tre'
Have Fun and Enjoy the Music
"Still Working the Problem"

 

Thanks for the clear explanation Tre. (nt), posted on March 3, 2015 at 23:39:54
91derlust
Audiophile

Posts: 1101
Joined: December 25, 2014
 .
"Confusion of goals and perfection of means seems to characterise our age." Albert Einstein

 

Page processed in 0.034 seconds.

[ Top Top of Page ] [ Contact Us ] [ Support/Wish List ] [ Copyright © 2024, Audio Asylum, all rights reserved ]
[ General ] [ Speakers ] [ Tubes ] [ Vinyl ] [ Digital ] [ Hi-Rez ] [ Video Asylum ] [ Cables ] [ Tweaks/DIY ] [ Music ] [ Films ]