Choking the Crack

[Mosez]

Ok, so I've read a bunch of choke related topics on the Bottlehead forum in an attempt to figure out the best power supply upgrade path for my Crack but I'm afraid I have more questions than answers at this point. One choke or two? Position one or two? And what to do with the capacitors - increase capacitance, decrease capacitance or leave as is?

I would really appreciate a breakdown of the options, pros and cons, and (basic) design rules when it comes to adding chokes to one's Crack. FWIW: I have 2 Triad C7-Xs on order from Mouser and I just got the Speedball upgrade in the mail, straight from the Bottlehead lair. Thank you very much in advance.

[Paul Birkeland]

Hello Daan,

For chokes in a power supply, position can depend a bit on whether or not you have a tube rectifier or solid state diodes, though in either case I generally put them first in the power supply.  Since you have two, that isn't as important of a consideration.

For figuring out the capacitor values, you can have a look at these sites:

http://www.ampbooks.com/home/amplifier-calculators/RC-ripple-filter/
http://www.ampbooks.com/home/amplifier-calculators/LC-ripple-filter/

In some ways, these are a gross oversimplification, but they are a good starting point (capacitor ESR will have a noticeable impact on the noise floor).  Using the stock values and ~75mA, we have -33dB of noise reduction.  With a C-7X and the stock caps, you get -62dB of noise reduction. 

This simulator would seem to suggest that 10uF of capacitance per stage with two C-7X's would be sufficient, but I think in practice that would be a bad idea.  What I would suggest is to keep two of the stock 220uF/250V caps in play, then put as big of a film cap as possible in the last position. 

-PB

[Mosez]

Hello Daan,

For chokes in a power supply, position can depend a bit on whether or not you have a tube rectifier or solid state diodes, though in either case I generally put them first in the power supply.  Since you have two, that isn't as important of a consideration.

For figuring out the capacitor values, you can have a look at these sites:

http://www.ampbooks.com/home/amplifier-calculators/RC-ripple-filter/
http://www.ampbooks.com/home/amplifier-calculators/LC-ripple-filter/

In some ways, these are a gross oversimplification, but they are a good starting point (capacitor ESR will have a noticeable impact on the noise floor).  Using the stock values and ~75mA, we have -33dB of noise reduction.  With a C-7X and the stock caps, you get -62dB of noise reduction. 

This simulator would seem to suggest that 10uF of capacitance per stage with two C-7X's would be sufficient, but I think in practice that would be a bad idea.  What I would suggest is to keep two of the stock 220uF/250V caps in play, then put as big of a film cap as possible in the last position. 

-PB

Thanks Paul. Very interesting simulator. A capacitance of 10uF vs. 220uF seems like a big difference, but what exactly would be the downside of lowering the capacitor value? (I'm wondering, not questioning).

Also, when you say "as big a cap as possible" are you referring to capacitance value or physical size? And what would the limiter be - the size of size of my wallet (allowance, really) - or something else?

Finally, does in this CLCLC configuration a bigger C in the final position mean you can go lower in the first and middle position, or is that not how it works? Trying to wrap my head around this. Thanks again for your help!

[Paul Birkeland]

By large, I meant as high in capacitance as possible without sucking away all your money (and, of course, it has to fit).

Lowering the cap value lowers the ripple reduction in the power supply, and at some point you will begin to hear noise from the output while the amp is operating. 

The C of the first position can affect the voltage of the power supply if the value shrinks enough, so that's not the place to get skimpy. 

The question about a CLCLC power supply depends on whether the chokes are all the same (which they are in your case).  Since the chokes are identical, I would put the largest (by large, one should always assume capacitance, the physical size of a capacitor is not considered until building time) capacitor closest to the bridge, then work smaller as you get out.  This will let you use higher quality capacitors with less capacitance in the signal current loop, which is ultimately the goal of this exercise. 

-PB

[Lar]

This simulator would seem to suggest that 10uF of capacitance per stage with two C-7X's would be sufficient, but I think in practice that would be a bad idea.  What I would suggest is to keep two of the stock 220uF/250V caps in play, then put as big of a film cap as possible in the last position. 

-PB
You mean like this? 

[Paul Birkeland]

Yeah, that's approaching soda can status.

[Grainger49]

Lar,

Being a confessed cap junkie this is just eye candy to me. 

Let me get this right, the big M-Cap is the last cap before the audio circuit, right? 

Those "Honker" output caps are just awesome!

[Mosez]

This simulator would seem to suggest that 10uF of capacitance per stage with two C-7X's would be sufficient, but I think in practice that would be a bad idea.  What I would suggest is to keep two of the stock 220uF/250V caps in play, then put as big of a film cap as possible in the last position. 

-PB
You mean like this? 

Impressive on so many levels. One of them being 'making-it-fit'.

[Mosez]

In all seriousness, I was thinking of going in THIS direction. Davew's CLCLC has 2 Hammond 158M chokes and 1 x 470uF electrolytic and 2x 20uF film capacitors (not sure what he ended up using as coupling caps - I'm thinking stock).

I noticed Paul commented on this design before - Paul, in your opinion, would 20uF for #2 and #3 caps be enough, or would suggest to up a bit?

[Paul Joppa]

I think you meant the other Paul but ...  the last cap in the power supply is in the output current loop. That means it is in series with the headphones, and with the output cap (100uF). So, at low frequencies, the voltage drop across the capacitor is inversely proportional to its capacitance. The cathode follower feedback tries to keep the voltage across the output cap /headphone constant. However:

1) the feedback is small because the gain of the 6080 is quite small, so it can't compensate for a small power supply cap very well. So a small cap in the power supply will still cause the bass cut-off frequency to be high.

2) when it does compensate, it imposes a larger signal on the power supply. If the feedback were very large, it would cause the power supply signal voltage to be 5 times the signal output voltage (!) if the PSU cap is 20uF and the output cap is 100uF.

Since modulation of the power supply voltage gets into the driver stage, I think it is to be avoided or minimized. I would never use less power supply capacitance (last cap) than in the output cap, and I'd prefer at least twice that - if it were my amp.

[Mosez]

I think you meant the other Paul but ...  the last cap in the power supply is in the output current loop. That means it is in series with the headphones, and with the output cap (100uF). So, at low frequencies, the voltage drop across the capacitor is inversely proportional to its capacitance. The cathode follower feedback tries to keep the voltage across the output cap /headphone constant. However:

1) the feedback is small because the gain of the 6080 is quite small, so it can't compensate for a small power supply cap very well. So a small cap in the power supply will still cause the bass cut-off frequency to be high.

2) when it does compensate, it imposes a larger signal on the power supply. If the feedback were very large, it would cause the power supply signal voltage to be 5 times the signal output voltage (!) if the PSU cap is 20uF and the output cap is 100uF.

Since modulation of the power supply voltage gets into the driver stage, I think it is to be avoided or minimized. I would never use less power supply capacitance (last cap) than in the output cap, and I'd prefer at least twice that - if it were my amp.

Thanks Paul (I know there are at least 2 Pauls with guru status here on the BH forum). I think I'm beginning to comprehend why there seems to be no single recipe for Crack power supply upgrade. If I understand what you are saying correctly, the PS design cannot be separated from the rest of the amplifier? So when PB says this:

[...] Since the chokes are identical, I would put the largest (by large, one should always assume capacitance, the physical size of a capacitor is not considered until building time) capacitor closest to the bridge, then work smaller as you get out.  This will let you use higher quality capacitors with less capacitance in the signal current loop, which is ultimately the goal of this exercise.  

-PB

does that mean that "smaller" should still not be less than twice the output cap? What if the output caps were smaller - say 70uF or 47uF? Would there be a penalty for doing that? Following your rule that would allow the last cap to drop to 140uF or 100uF respectively, right? And would I have to compensate for that drop somewhere else - for instance by increasing the value of the #1 and/or #2 cap? Thanks in advance for your patience - really trying hard here.

[Lar]

Lar,

Being a confessed cap junkie this is just eye candy to me. 

Let me get this right, the big M-Cap is the last cap before the audio circuit, right? 

Those "Honker" output caps are just awesome!

Yes thats correct Grainger, i knew that would get your attention.

[Paul Birkeland]

does that mean that "smaller" should still not be less than twice the output cap? What if the output caps were smaller - say 70uF or 47uF? Would there be a penalty for doing that? Following your rule that would allow the last cap to drop to 140uF or 100uF respectively, right? And would I have to compensate for that drop somewhere else - for instance by increasing the value of the #1 and/or #2 cap? Thanks in advance for your patience - really trying hard here.

Yes, this is a good rule of thumb to follow.  With high impedance headphones (300+ ohms), you can begin to decrease the value of the caps at the output a bit, and consequently the last cap in the power supply as well. 

You don't really need to compensate for that drop elsewhere, though there's nothing wrong with increasing the capacitance of cap 1 or 2 in the power supply (I put a 500uF cap for C1 in one Crack I built).

-PB

[Mosez]

Makes a lot of sense - thanks PB!

[daveottley]

Would it be ok to use a Hammond 157M in one position and a Hammond 158M in a second position? I have these chokes on hand and I'm not sure what the impact of different chokes will be on the power supply.