Filter choke design

[pullpush]

Hello everybody,

I'm trying to make a 5F6A Bassman clone. Since I don't live nowhere near the USA, importing transformers and chokes is a very expensive thing. Therefore I'm having transformers wound manually by a local manufacturer. The only thing remaining now is the choke, and I'm kindly asking if anybody could check these numbers before I send it out to be wound.

Since only screens and the preamp are choke-filtered. That should be in the < 50 mA ballpark.

I was thinking European EI-66 core, Sfe=4.84 mm2, 4600 turns of 0.3mm (~ awg 29) wire and 0.5 mm (~0.02 inch) air gap. Slightly thicker wire will, hopefully, keep the DC resistance low and 4600 turns with 4.84 mm2 surface area should be significant filtering.

Thanks in advance for any help on the matter.

[Gingertube]

Hi Push Pull,
I decided to reply since no one else had.
Hardly any of us design our own magnetic components (Chokes Transformers etc.) so it seems we are of little help.
The ONLY suggestion I have is to try some of the downloadable design software such as (I found this with a quick Goggle Search using "Choke Design Calculator"):
Metglas, Inc. - DC Choke Design Tool

For your application probably <25mA would do but designing for a bit more current is never a bad idea with chokes - their inductance actually decreases as you run more current through them.
Low DCR is not particularly important for screen supply and preamp supply duty, anything less than 1K would probably be fine. I base this on the fact that the Bassman runs 470 Ohms resistors in series with each of the screens anyway and one of the most performed mods to Fenders generally is to rip out the 470 Ohm 2W Carbon Screen resistors (hardly any of which are anything near 470 Ohms after a few years use) and replace them with 1K 5W wirewound "Rats Coffin" type resistors. This give additional protection to the tubes and it is claimed? increases the touch responsivity of the amp a little.

Good Luck with the Bassman build.

Cheers,
Ian

[Wombaticus]

My answer would be "why use a choke at all?". I would unhesitatingly replace the choke with a MOSFET capacitor multiplier, or perhaps a gyrator. It's literally a fraction of the cost, especially if you're contemplating a custom wind. And the performance is excellent. Moreover, it's relatively straightforward to tweak the level of filtering to get the desired results..

I've been using circuits similar to ones Merlin Blencowe presents in his power supply book with great results in my latest builds. I'm pretty sure I'm done with chokes for good once the couple I have kicking around are gone.



My answer would be "why use a choke at all?". I would unhesitatingly replace the choke with a MOSFET capacitor multiplier, or perhaps a gyrator. It's literally a fraction of the cost, especially if you're contemplating a custom wind. And the performance is excellent. Moreover, it's relatively straightforward to tweak the level of filtering to get the desired results..

I've been using circuits similar to ones Merlin Blencowe presents in his power supply book with great results in my latest builds. I'm pretty sure I'm done with chokes for good once the couple I have kicking around are gone.

[trobbins]

How does one tweak an RC or C filter to get an LC transient response to say a step (ie. fast attack) load change?

[Wombaticus]

How does one tweak an RC or C filter to get an LC transient response to say a step (ie. fast attack) load change?

It's an interesting question, and one I confess I hadn't considered. And, given you are who you are, I suspect that it's one for which you already have an answer, or at least a pretty darn good suspicion.

Can you tell us more about how & why these transient responses are likely to be fundamentally different? I think it could be an instructional thing for me to model.

In any case there's the separate consideration of whether or not any differences in transient response are readily audible. I like the sound of both LC and RC screen filters, and had always presumed that the differences were mostly attributable to the larger voltage differential between plate and screen that one typically has with an RC filter. But maybe there's more to it?

Either way, a cap multiplier can certainly give you choke-like voltage drops and with a comparable level of filtering. So I'd posit that this alone would make it sound much more choke-like in sound than RC-like given a constant level of filtering. In my experiments to date (with cap multipliers, haven't yet tried gyrators), the results have been tonally fully satisfactory. One caveat that I will mention is that I tend to prefer to a somewhat stiffer screen filter than some, because I find the ghosting under OD that often seems to accompany "classic" LC filter values rather distracting -- so perhaps I have a predilection towards the cap multiplier sound to begin with? Hard to say.

Cheers,

Paul

[R.G.]

How does one tweak an RC or C filter to get an LC transient response to say a step (ie. fast attack) load change?

I'll have to think about this some, but my first thought is that you can't. The Q is too low.

LC filters have their characteristic time response based on the cross-pouring of energy from the cap to the inductor and back at the natural frequency of the LC. Resistors can damp that, but can't make it higher. Setting aside active circuits to fake the energy storage in the L, I don't think any conclavation of Rs and Cs can cause the necessary cross storage and exchanges without also damping the energy storage too much. You can get two-pole rolloffs in the frequency domain from Rs and Cs, and you can get peaks and notches from Rs and Cs, but I don't think you can get the transient response to be as non-lossy as with an L.

In any case there's the separate consideration of whether or not any differences in transient response are readily audible. I like the sound of both LC and RC screen filters, and had always presumed that the differences were mostly attributable to the larger voltage differential between plate and screen that one typically has with an RC filter. But maybe there's more to it?

Screens are literally a second and lower gain input, and one that directly modulates the AC gain of a tube. If your screen supply rings, that would definitely be audible. But I think it would be difficult to get an L-C network on the screens with a Q high enough to ring, excited only by the change in screen current and in changes - if any! - in the preamp current draw.

Either way, a cap multiplier can certainly give you choke-like voltage drops and with a comparable level of filtering. So I'd posit that this alone would make it sound much more choke-like in sound than RC-like given a constant level of filtering. In my experiments to date (with cap multipliers, haven't yet tried gyrators), the results have been tonally fully satisfactory.

That was the exclusion I was talking about at first. If you have an active circuit, it can, perhaps unintentionally, have some of the transient response of an inductor. And if it's just a stiff screen supply that's being heard, well, a follower can do that too.

[trobbins]

Hi Paul - I didn't mean to stir the pot too much - but as the post was about cloning a classic, I thought it worth noting that sometimes changing a part value or type can have an impact.

I thought I had actually seen some assessment on the topic, and sure enough I just found the article again - check out:
http://www.jpbourgeois.org/guitar/pd...man%205F6A.pdf

Now the article doesn't go to trying to determine or isolate how noticeable a particular characteristic is, and I think that would be a pretty torrid job, but there certainly can be a 'noticeable' impact to waveforms. Maybe Kuehnel's book has some appropriate comments?

The damped ringing waveform that an LC can impart due to a step load change is not really emulatable with RC or C only (as they don't introduce the extra resonant ring frequency component).

But in practise, all clones are littered with compromises - and chokes are a definite pain in the neck unless you have one laying around and a chassis that can fit it.

Ciao, Tim