• Disclaimer: This Vendor's Forum is a paid-for commercial area. Unlike the rest of diyAudio, the Vendor has complete control of what may or may not be posted in this forum. If you wish to discuss technical matters outside the bounds of what is permitted by the Vendor, please use the non-commercial areas of diyAudio to do so.

GK-2 cutoff frequency for SS section

Hi Hugh,

When I created the SS output on my GK-2, I used a 1 uF teflon cap. When we discussed this prior to my implementation, I remember you said it should just be between 1 and 10 uF.

Just out of curiosity - what would the cutoff frequency (- 3 dB point) for the SS output be with the 1 uF cap?

It is looking into the Ground Sound filter's 20 k input impedance.

I do not feel there is a problem, but since I have already implemented a very steep (48 dB/octave) sub filter at 10 Hz in the Ground Sounds, there is no need for additional sub filtering, and if that turned out to be the case I would simply change the SS output cap value to ensure I do not "get too much of a good thing" ;)

Cheers,

Jens
 
Simple: the -3dB roll off frequency is when the reactance of the capacitor (1/(2 * pi *frequency * capacitance)) equals load resistance. In this frequency, phase angle between voltages becomes 40 degrees and voltage across loads becomes .7071 of the input voltage.

If you get a calculator, then F 0 1/(2 pi capacitance [microfarads] * resistance[mega-ohms]) will give frequency in hertz.

Fortunately, this calculae is independent of capacitor brands and personal tastes. Is as is.
Good luck.
 
Simple: the -3dB roll off frequency is when the reactance of the capacitor (1/(2 * pi *frequency * capacitance)) equals load resistance. In this frequency, phase angle between voltages becomes 40 degrees and voltage across loads becomes .7071 of the input voltage.

If you get a calculator, then F 0 1/(2 pi capacitance [microfarads] * resistance[mega-ohms]) will give frequency in hertz.

Fortunately, this calculae is independent of capacitor brands and personal tastes. Is as is.
Good luck.

Errr, well, yes ...

Unfortunately, I'm not much of a mathematician and have never used a scientific calculator, so for me this is not simple.

So what is "F 0 1"? And what "resistance" are we talking about?

Cheers,

Jens
 
Errr, well, yes ...

Unfortunately, I'm not much of a mathematician and have never used a scientific calculator, so for me this is not simple.

So what is "F 0 1"? And what "resistance" are we talking about?

Cheers,

Jens
OK, it was a typo I would write f= 1/ 2 * pi * Capacitance * resistance, and it gives you frequency. Resistance must be considered the equivalent resistance to any ground before the capacitor in the way of signal flow. Example:
lets consider a 1 microfarad ( µF, or 10E-6) farad and a 20Kohms load.

So the "corner" frequency, say, the frequency at which the signal after the cap is -3dB (.7071 times) respect to the signal before the cap will be:

f = 1 / ( 2* pi * 20000 ohm * 10E-6F) = 15.91Hz.

Then, at about 16Hz, it is considered the lowest frequency your circuit will pass without "severe" attenuation and or distortion.

It is independent if the load resistor is a single 20000ohm or two 40000 in parallel, or or two 10K in series, or any other combination of resistances that gives you 20000ohms as result.

Aclareta: 1E-6 means .000001 in scientific notation, or 1/1 million or 1 millionth part of any measure.
 
OK, it was a typo I would write f= 1/ 2 * pi * Capacitance * resistance, and it gives you frequency. Resistance must be considered the equivalent resistance to any ground before the capacitor in the way of signal flow. Example:
lets consider a 1 microfarad ( µF, or 10E-6) farad and a 20Kohms load.

So the "corner" frequency, say, the frequency at which the signal after the cap is -3dB (.7071 times) respect to the signal before the cap will be:

f = 1 / ( 2* pi * 20000 ohm * 10E-6F) = 15.91Hz.

Then, at about 16Hz, it is considered the lowest frequency your circuit will pass without "severe" attenuation and or distortion.

It is independent if the load resistor is a single 20000ohm or two 40000 in parallel, or or two 10K in series, or any other combination of resistances that gives you 20000ohms as result.

Aclareta: 1E-6 means .000001 in scientific notation, or 1/1 million or 1 millionth part of any measure.

Hmmm - I am afraid I still do not understand. If "1 /" means "1 divided by" then I cannot see how you get to the result shown. If I divide 1 by the result in the brackets, I get less than 1.

But I probably do not understand the notation ...

Anyway, since we do not know the resistance, I guess we can just as well wait for Hugh ;)

Cheers,

Jens
 
Hmmm - I am afraid I still do not understand. If "1 /" means "1 divided by" then I cannot see how you get to the result shown. If I divide 1 by the result in the brackets, I get less than 1.

But I probably do not understand the notation ...

Anyway, since we do not know the resistance, I guess we can just as well wait for Hugh ;)

Cheers,

Jens

Go to the windows calculator, and in the menu "view" click scientific, you will see the calculator expands.

Then, in the numeric right keyboard type (or in the calculator drawing press with the mouse the number) 2, then the asterisc that operates as a multiplier sign, with the mouse click the "pi" button, the * again, then type the value of the resistance load in ohms, the *, then the capacitance in farads (Again, 1µF is .000001F). Then, press equal button.

Once did it, press with the mouse, the 1/x button, and you'll get the value I post in the previous msg, (15.91xxxx) with lots of decimal numbers. In practice, only the first or two first digits are of interest.

But, if you don't know the load impedance form the output of the cap to ground, all this thread is meaningless.

Cordially, Osvaldo :headbash:.
 
Last edited:
Go to the windows calculator, and in the menu "view" click scientific, you will see the calculator expands.

Then, in the numeric right keyboard type (or in the calculator drawing press with the mouse the number) 2, then the asterisc that operates as a multiplier sign, with the mouse click the "pi" button, the * again, then type the value of the resistance load in ohms, the *, then the capacitance in farads (Again, 1µF is .000001F). Then, press equal button.

Once did it, press with the mouse, the 1/x button, and you'll get the value I post in the previous msg, (15.91xxxx) with lots of decimal numbers. In practice, only the first or two first digits are of interest.

But, if you don't know the load impedance form the output of the cap to ground, all this thread is meaningless.

Cordially, Osvaldo :headbash:.

Ah, that's how to do it! Thanks :D

Not all in vain - once Hugh tells us what the resistance is, I'll use your calculation to adjust my cap value, if necessary ;)

Thanks,

Jens
 
Hi Jens,

There's an easier equation to remember/use - although it's a slight approximation. This is: 160,000 / R * C (where C is in microFarads).

But, yes, this gives the -3dB roll-off of the R-C pair but you need to remember that the effects of the roll-off (in terms of phase) will extend up to 10x the -3dB frequency ... so in Oswaldo's example, this would be 159Hz.

Which is much too high.

The output cap in the GK-1/GK-2 is 1uF, so when combined with the 43K input impedance of my Lifeforce/Maya amps, produces approximately 4Hz - much better! :D

So, in your case - looking into the 20K input impedance of the Ground Sound filter - I would use 2.2uF as a minimum.


Regards,

Andy

PS: Don't forget that if you are using the "Sub out" facility to feed your bass drivers or subs, you also need to put an output coupling here (whose value depends on the Zin of the amp driving them).
 
Last edited:
Hi Jens,

There's an easier equation to remember/use - although it's a slight approximation. This is: 160,000 / R * C (where C is in microFarads).

But, yes, this gives the -3dB roll-off of the R-C pair but you need to remember that the effects of the roll-off (in terms of phase) will extend up to 10x the -3dB frequency ... so in Oswaldo's example, this would be 159Hz.

Which is much too high.

The output cap in the GK-1/GK-2 is 1uF, so when combined with the 43K input impedance of my Lifeforce/Maya amps, produces approximately 4Hz - much better! :D

So, in your case - looking into the 20K input impedance of the Ground Sound filter - I would use 2.2uF as a minimum.


Regards,

Andy

PS: Don't forget that if you are using the "Sub out" facility to feed your bass drivers or subs, you also need to put an output coupling here (whose value depends on the Zin of the amp driving them).

Hi Andy,

This whole thread is in fact about the "Sub Out" (= SS output) ;)

The capacitors I am talking about are on the SS output, as recommended by Hugh, but they may need to be a somewhat higher value anyway.

Let's see what Hugh says!

Cheers,

Jens
 
Hi Jens,

Quick reply as I'm late for a appointment unfortunately.

Zout of the SS voltage stage in the GK2 is the reciprocal of the transconductance of the tube, so around 150R. So a 20K target impedance would be fine.

1uF is not enough, however. Impedance of a 1uF cap of 1uF at 10Hz is 16K, nearly the 20K of the source, so this would drop voltage by almost half, that is, 6dB.

I would increase this cap to 5uF, so that at 10Hz the impedance would be 3k2, so that all your very low frequency will give through. 5uF would also give you an impedance of 2k13 at 15Hz, MUCH lower than your 20K target and therefore about 0.5dB down, which is more than good enough bass.

Impedance for a cap is 160,000/uFxHz. Roughly, but get enough. You could use a good electrolytic here, since it has about 50V across it and with large polarising DC across it the sound quality will be outstanding.

Hope this helps,

Hugh

PS: My thanks to Osvaldo and Andy, good descriptions.
 
Hi Andy,

This whole thread is in fact about the "Sub Out" (= SS output) ;)

The capacitors I am talking about are on the SS output, as recommended by Hugh, but they may need to be a somewhat higher value anyway.

Let's see what Hugh says!

Cheers,

Jens

Aah, I was confused by your use of "SS", Jens.

For Sub out, I use a pair of 4.7uF BG bipolar caps in what they call the "Super-E" configuration ... so 9.4uF, giving a -3dB frequency of 0.4Hz (with 43K Zin). So the phase error will run only up to 4Hz. However, those caps no longer seem to be available.


Regards,

Andy
 
Hi Jens,

Quick reply as I'm late for a appointment unfortunately.

Zout of the SS voltage stage in the GK2 is the reciprocal of the transconductance of the tube, so around 150R. So a 20K target impedance would be fine.

1uF is not enough, however. Impedance of a 1uF cap of 1uF at 10Hz is 16K, nearly the 20K of the source, so this would drop voltage by almost half, that is, 6dB.

I would increase this cap to 5uF, so that at 10Hz the impedance would be 3k2, so that all your very low frequency will give through. 5uF would also give you an impedance of 2k13 at 15Hz, MUCH lower than your 20K target and therefore about 0.5dB down, which is more than good enough bass.

Impedance for a cap is 160,000/uFxHz. Roughly, but get enough. You could use a good electrolytic here, since it has about 50V across it and with large polarising DC across it the sound quality will be outstanding.

Hope this helps,

Hugh

PS: My thanks to Osvaldo and Andy, good descriptions.


Hi Hugh,

Thank you very much for this info.

Yes, 1 uF is obviously not enough.

Fortunately, I think I have a couple of Mundorf Silver/Oil 8.2 uF caps that I think will do very nicely for this job, so these will go in as soon as I have the time ;)

I am certain that the Mundorfs will be very good in this position, and I might consider moving the two 1 uF teflon caps to the tube output instead - but I won't do this at the same time, as I want to make sure to be able to listen to each change in depth before I do the next one :)

Osvaldo and Andy: Thank you very much for your good input and patience!

Cheers,

Jens
 
Jens,

My sincere pleasure!

With the advent of SMPSs in the last 20 years, there has been a steadily improvement in electrolytics, which are routinely now specced to 100KHz.

Further, in tube amps, we all find higher voltages, up to 140V in the GK2. When caps are polarised, even electros, with DC voltage, the sound quality comes a huge rise in subjective listening. Some very large film caps can introduced EMI/RFI because they have large areas, much more than electros, so I would invite you to have a few good quality electros (Sanyo Oscon, Nichicons, Panasonics) and compare them. You will be very surprised, particularly as you are working at 100Hz and less, where resolution and the best possible quality is not noticed by most listeners.

Cheers,

Hugh
 
You could use a good electrolytic here, since it has about 50V across it and with large polarising DC across it, the sound quality will be outstanding.

Hi Hugh,

This raises an interesting issue, IMO ... where is the better place to put a coupling cap:

* on the output of the GK-2?

* or the input of the following component - in my case, an active XO?

Because you don't need them in both positions!

You say there is 50v across the output cap in the GK-2 ... which sounds great (as this large polarising voltage will certainly optimise the SQ of the output cap).

Whereas my active XO is running on +/-15v. So it would seem that (if I install the caps on the input of my active XO) there must be less polarising voltage ... so it would be much better to use them on the output of the GK-1?


Regards,

Andy
 
There is a fact that some people ignore or don't take into account until its too late and the circuits must be re-designed.

When using too big coupling caps together with a high impedance stage (output and input respectively), the cap will take more time to be the voltage across it stabilized when turning on and off the devices. So, in the very few instants, the audio is very distorted and/or noisier and or lower than in normal operation and POP's in the speakers. Also, the power supply must be severely filtered and or regulated so the ripple in the circuits at so low frequencies must be very low, and in case of unregulated ones, variations in line voltage will cause circuitry voltages an currents to vary, and this will be coupled to the audio chain causing woofer cones to move slowly at this frequencies causing non-linear distortion in them plus overheating in the coil because large DC offsets in it.

Good luck and greetings for all.
 
Andy, Osvaldo,

The output from the bass cap on the GK2 is at the collector of a CFP, which has very low Zout, around around 80R. A 5uF cap here would be fine. The midrange and top end output of the GK2 is passed through a cathode follower.

The power supply is very carefully regulated, and the switch on is also controlled.

This request to slightly mod the GK2 is has been inquired by a couple of customers, but it is an after market change, something I normally would suggest as the bass quality of this preamp is carefully sorted in the original design. However, some always strive for purist, and yes, with minor change, this can be accommodated.

You cover the issues, thank you for your inside suggestion,

Cheers,

Hugh
 
Last edited:
Just a small update on the implementation of new caps for the SS (Sub Out) output of the GK-2.

I have now succesfully replaced the two 1 uF teflon caps with two 8.2 uF Mundorf Silver/Oil caps. I have no problems whatsoever with any kind of noise pickup, despite the fact that these caps are huge!

It should be mentioned that I have shielding walls in place inside the GK-2 enclosure (they have been there all along), and the caps are placed within these shielding walls.

Changing the cap value from 1 uF to 8.2 uF certainly improved bass extension, and although a bit rough when first fired up, the caps are now bedding in very nicely and becoming more and more tuneful.

While I had the GK-2 on the workbench, I also tried out another change, as I had a 10K Alps motorpot lying around. Hugh specifies 20K for the GK-2, but we thought that 10K would be fine instead of the original 50K Panasonic motor pot supplied with the GK-1R kit.

The 10K pot is fine, and works nicely. However, because my speakers are very efficient, regulation for low level listening is difficult.

I have now decided to go all out on the pot side and have ordered a TKD 2CP-2500MC motor pot. I have found a 25K version (at Michael Percy), which I think is fine in my case because of speaker efficiency, even though Hugh specifies 20K as standard.

The TKD is rated by many as the best motor pot available, and I have even heard of people saying that changing from the Alps to the TKD gave an improvement similar to a major equipment upgrade. Whether that will be true or not remains to be seen - I will forgo the hype and let my ears be the judge of that. At any rate, the TKD is the most precise motor pot around, so I will be looking forward to getting the precision and the right value into my GK-2(R) ;)

Cheers,

Jens
 
Just a small update on the implementation of new caps for the SS (Sub Out) output of the GK-2.

I have now succesfully replaced the two 1 uF teflon caps with two 8.2 uF Mundorf Silver/Oil caps. I have no problems whatsoever with any kind of noise pickup, despite the fact that these caps are huge!

Changing the cap value from 1 uF to 8.2 uF certainly improved bass extension, and although a bit rough when first fired up, the caps are now bedding in very nicely and becoming more and more tuneful.

Cheers,

Jens

Thank you for your update - I am now inspired to put some larger caps on my "Sub Out". :)

However, I suspect that Mundorfs are an overkill for a signal output that doesn't really go above 100Hz? IOW, you could've used a different (ie. much cheaper) 8.2uF cap without ny sonic penalty in the bass region. Anyway, I will try something different.

Hugh specifies 20K for the GK-2, but we thought that 10K would be fine instead of the original 50K Panasonic motor pot supplied with the GK-1R kit.

Mmmm, in the schematic Hugh sent me 10 years ago, he specified a 25K pot.

The TKD is rated by many as the best motor pot available, and I have even heard of people saying that changing from the Alps to the TKD gave an improvement similar to a major equipment upgrade. Whether that will be true or not remains to be seen - I will forgo the hype and let my ears be the judge of that. At any rate, the TKD is the most precise motor pot around, so I will be looking forward to getting the precision and the right value into my GK-2(R) ;)

When I replaced the stock motorised pot in my GK-1, I used a (manual) 25Kohm TKD stepped attenuator - as per Bob Crump's usage in his "Blowtorch" preamp. This is bracketed by some Vishay resistors - which Bob found improved the sound even further.

So I suggest if you want the convenience of remote volume ... you will be sacrificing some SQ. ;)

Regards,

Andy
 
Thank you for your update - I am now inspired to put some larger caps on my "Sub Out". :)

However, I suspect that Mundorfs are an overkill for a signal output that doesn't really go above 100Hz? IOW, you could've used a different (ie. much cheaper) 8.2uF cap without ny sonic penalty in the bass region. Anyway, I will try something different.



Mmmm, in the schematic Hugh sent me 10 years ago, he specified a 25K pot.



When I replaced the stock motorised pot in my GK-1, I used a (manual) 25Kohm TKD stepped attenuator - as per Bob Crump's usage in his "Blowtorch" preamp. This is bracketed by some Vishay resistors - which Bob found improved the sound even further.

So I suggest if you want the convenience of remote volume ... you will be sacrificing some SQ. ;)

Regards,

Andy

Hi Andy,

Yes, the Mundorfs are indeed overkill. I only chose them because I had them lying around idle. They may or may not make a difference in this position, but the thing is I know they are darn good, so after putting these in there I will have no doubts about this component any more. Hugh also suggested trying something else. I believe I would have chosen some Panasonics if it had been necessary for me to buy new ones ;)

I believe the schematic with the 25K pot you are referring to is for the GK-1. However, the GK-1R was supplied with a 50K Panasonic motor pot, as you probably know. I have no idea how the 25K would perform in the GK-1, as I haven't tried it. The circuit and placement of the pot in the GK-2 are very different, but after the testing done with the 10K, I am sure that 25K is a good value in the GK-2.

Whether or not there is a sound quality penalty in using a pot compared to a stepped attenuator would probably depend on the attenuator and the pot. Some people claim that the TKD pots are better than some of the widely used attenuators like the DACT. I have no idea whether this could be true or not, but the TKD people certainly use some very sophisticated techniques in their pots ...

I also have no idea how the TKD pots compare to the TKD attenuators ...

Anyway, I do not want to loose my remote capabilities, so for my use, the best choice is the TKD motor pot :)

My pot was shipped from Michael Percy yesterday, so with a little luck and not too much customs hazzle, I should have it some time next week :D

Once installed, I will try to make an assessment if it makes a difference in sound quality. I am not expecting a lot, but even small improvements are nice!

Cheers,

Jens