[peterpan188]

Update- Finally got to order from partsConnexion. Just the caps and resistor is going for ~$120. Ordered 6 per pair of resistors that will to be matched, and after discussing with the people who are paying for this together, we decided to just use Blackgate Std (which is still pretty damn expensive) excpet C7, if not, $120 will just be the cost of Blackgate

. Got some solen caps for C3 and C4 also.

 

[dsavitsk]

Quote:

Originally Posted by peterpan188 Update- Finally got to order from partsConnexion. Just the caps and resistor is going for ~$120. Ordered 6 per pair of resistors that will to be matched, and after discussing with the people who are paying for this together, we decided to just use Blackgate Std (which is still pretty damn expensive) excpet C7, if not, $120 will just be the cost of Blackgate . Got some solen caps for C3 and C4 also.

The cathode bypass C2 is just as important as the ouput cap.

Bypassing the output and cathode bypass (i.e. populating C3 and C4) with a small film cap (either solen or wima) will likely do more harm than good if you are using BG in those positions. BG N series sound better than solens anyway, which are a bit grainy. In fact, I think the Wimas are every bit as good a cap as the solens.

I'll not comment on the usefulness of BG on the PS positions with this amp other than to say that $120 gets you 80% of the way to a real tube amp.

 

[peterpan188]

Quote:

Originally Posted by dsavitsk The cathode bypass C2 is just as important as the ouput cap.

There is a real estate issue on C2. The only choices I have are Blackgate Std 220uF and Cerafine 220uF. How do you compare them?

Quote:

Bypassing the output and cathode bypass (i.e. populating C3 and C4) with a small film cap (either solen or wima) will likely do more harm than good if you are using BG in those positions. BG N series sound better than solens anyway, which are a bit grainy. In fact, I think the Wimas are every bit as good a cap as the solens.

Well, we have been talking about those bypass caps. The conclusion is that since they are not huge bucks to spend, (altogether around $20) I might as well just get them and swap between caps and no caps at all, I might end up finding what's best for the configuration.

Urhhhhhhhh, you got me thinking about power caps. Well, put it this way, the only places that stocks a good line of Cerafine, Muze and BG are Welborn labs, Handmade's, and partsConnnexion, respectively. If I chagne the power caps to anything other than BG, I will have to order from somewhere else. Counting the shipping in, I might as well just order BG std for those positions altogether from partsConnextion

Regard,
Peter

 

[NeilR]

I thought we had discussed C2; I guess not. It certainly should get he same attention as C7.

Although the C2 footprint is small, you can get away with a smaller voltage rating. I think Pete Millet spec'd 16V. In practice, I have only seen a couple volts across the trimpot, which is all the cap has to handle.

If you have a lot of courage and a driving need, you might even get away with 6.3V, although I'd be interested in hearing other opinions on that. I say that because with a constant current source running through the trimpot, the max voltage across the trim pot and C2 is 5000 (ohms) * 0.00056 (Amps) = about 2.5V. If something goes wrong, however, you will smoke an epensive cap or two....

 

[n_maher]

Quote:

Originally Posted by NeilR Although the C2 footprint is small, you can get away with a smaller voltage rating.

Another option is to mount the cap below the board, flopped on its side. That is what I'm thinking for my Millett, since the board will be mounted on standoffs anyway.

If I had to do it over again we'd have provided a bit more real estate in this area.

N

 

[peterpan188]

Neil,
We did talk about C2, playing on the safe side, I will just use 16V caps. I did it onces, the explosion of caps "sound" nice but smells just horrible.

Can't they make something that smells better?

Nate,
That sounds like a very good idea. I will add diamond buffer later, so space is gonna be tight around C2. I might or might not use standoff, if the case is going to be a Hammond. Going with the max out idea, it will be a NX 680uF, $30 bucks right here again

Peter

 

[dsavitsk]

Quote:

Originally Posted by NeilR I thought we had discussed C2; I guess not. It certainly should get he same attention as C7. Although the C2 footprint is small, you can get away with a smaller voltage rating. I think Pete Millet spec'd 16V. In practice, I have only seen a couple volts across the trimpot, which is all the cap has to handle. If you have a lot of courage and a driving need, you might even get away with 6.3V, although I'd be interested in hearing other opinions on that. I say that because with a constant current source running through the trimpot, the max voltage across the trim pot and C2 is 5000 (ohms) * 0.00056 (Amps) = about 2.5V. If something goes wrong, however, you will smoke an epensive cap or two....

Okay, a little math.

You need to be concerned about 2 things here. 1 is the capacitance required to set a decent 3dB point, the other is the voltage requirement, and, importantly, these influence each other.

First, for the voltage. The cap will see whatever the cathode is biased up to. Generally, as shown above, this is below 3V, so this is not much of a concern unless the CCS shorts. Even in that case, I doubt you'll be in too much danger. The trim pot is used to set this and as the resistance is decreased, the bias can only go down from the max of 2.8V.

For the 3 dB point, you need to consider what resistance the anode of the cap sees to ground. This turns out to be the cathode bias resistor (the trim pot) in parallel with the plate load + the plate resistance of the tube divided by mu + 1. mu here is reasonably low, and the CCS is near to an infinite resistance (this is the whole point of a CCS -- see my page here http://www.ecp.cc/c4s.html), so effectively you can simply consider the trimmer. It will usually not be set at 5000 ohms, but it should certainly be set at more than 2500. So, to be safe, use 2500 ohms and figure the 3dB point. The formula is
F = 1/(2 * pi * C * R)
a back of the napkin analysis says to me that 100uF is plenty big. In fact, 47uF is probably just right, and you could get away with 22uF in a pinch.

-d

 

[NeilR]

Quote:

Originally Posted by n_maher If I had to do it over again we'd have provided a bit more real estate in this area.

Nate,

You know that your mission will never be completed until you rev that board with a 16mm C2

Regards,
Neil

 

[n_maher]

Quote:

Originally Posted by NeilR Nate, You know that your mission will never be completed until you rev that board with a 16mm C2 Regards, Neil

My days of revising the Millett PCB are officially over! The board is what it is at this point, if someone else wants to take over the cause they should feel free

.

And my C2 will be 16mm, it just won't be mounted in the allotted space

.

Nate

 

[NeilR]

dsavitsk,

You and I had a discussion here, where I think we came to the conclusion that 75-100uF was needed for a corner frequency of about 2HZ. I believe you picked that frequency to avoid phase shifts at frequencies several harmonics above the corner. That thread also linked to an interesting discussion by Steinchen regarding his calculations for the C7 value for various output impedances.

Thanks for the link to your ccs page. I am still trying to understand the fine points of current flowing though an infinite impedance

Regards,
Neil

 

[NeilR]

Quote:

Originally Posted by n_maher My days of revising the Millett PCB are officially over! The board is what it is at this point, if someone else wants to take over the cause they should feel free . And my C2 will be 16mm, it just won't be mounted in the allotted space . Nate

Yea yea yea... I've heard this before.

 

[dsavitsk]

Quote:

Originally Posted by NeilR dsavitsk, You and I had a discussion here, where I think we came to the conclusion that 75-100uF was needed for a corner frequency of about 2HZ.

Hmmm. I don't think I did the math there, or else I messed it up one of the times I did do it. Just to make sure this is correct, I'll show my work this time.

First, to calculate the impedence the cap sees, it is 2K5 (Rk) in parallel with

(RL + ra) / (mu + 1)

RL is the plate load, ra is the plate resistance of the tube, mu for these tubes is somewhere between 5 and 20 depending upon the tube chosen. ra is something I don't know, but larger than 1 (probably larger than 1000.) RL is big -- for convenience we'll call it infinite, but it is certainly larger than 1M (the bottlehead kit is about 3M for those interested.)

Anyway, 2K5 in parallel with anything large is going to be close to 2K5. Moreover, since 2K5 is an estimate, and in reality it is closer to 5K, this is a fine fudge.

So, using the cap equation, with a 47uF cap, you have

F = 1 / (2 * pi * .000047 * 2500)

which is a corner frequency of 1.35Hz. This is very low.

If it were me, and I were ordering from partsconnexion, I might try the 47uF os-cons. Mind you, I've not tried them, but I have heard from people who think they are very good for cathode bypass. Otherwise, there are a ton of different blackgates. Also, check out the elna selection at handmade.

-d

 

[dsavitsk]

Quote:

Originally Posted by NeilR Thanks for the link to your ccs page. I am still trying to understand the fine points of current flowing though an infinite impedance

Think of it as an infinite impedence to anything over the constant current. Sort of a brick wall filter than allows current with 0 resistance up to a certain point, but at an infinite impedence after that.

 

[NeilR]

dsavitsk,

I think the problem is the resistance chosen for your formula. In the thread I linked, I indicated that my trim pot was set at about 1K (12AU6A). My notes indicate precisely 1.17K. Changing your formula to 1K and 75uF, the corner frequency is slightly above 2HZ, which was the goal in that discussion.

I think the moral of the story here is that the computation is imprecise given the uncertainty of the trim pot setting, which varies greatly among the various tubes used. My 12FK6's bias at a much higher resistance, possibly as high as 3K, but I have no precise notes on that. I estimage 3K because I know that my 12AU6A's develop about 0.6V across R2, but my 12FK6's develop 1.6V. Maybe 1K should be used as a worst case.

Whatever the exact number is for any given configuration, your work does suggest that some value well under 220uF should provide acceptable results.

Regards,
Neil

 

[n_maher]

Quote:

Originally Posted by NeilR Yea yea yea... I've heard this before.

I didn't say I was done building them

, just done messing with those infernal PCBs! NO MORE GROUP BUYS

.

Truth be told I'm pretty much done building as well. I really just have to get mine started (like it'll ever be done, pfff) and that's it.