Overfiltering?

[Satamax]

Hi everybody!

Well, i want to make a simple deluxish like circuit. But i'm not too bothered to buy the caps, since i have all the other parts. Could i just use a 2X200µf multisection cap? 600V rating, so that part should be no problem! Or would it be pushing it?I also have tons of old caps from recaps over the years, how could i test dem without a tel-ohmike thing?

Thanks a lot.

Max.

[stokes]

If the multi section cap is the can type that has a shared ground,I would get 2 more 200uf caps and put them in parallel with the 2 in the can to get two 100uf's and use them for the main and the screen.I have 100uf's in two of my 5E3's,one has a 5v4 rectifier and the other is a 5R4.It actually sounds nice and tight with the extra filtering.As for using old caps,personally I wouldnt,but if you must I would just put them in the circuit and see how they work,check for excessive ac volts and keep your fingers crossed.Old caps that test good will still be liable to sudden failure,after all,they were taken out of an amp because they were old to begin with.

[Arthur B.]

Other than the load they might present to the rectifier, I see no with using such a large value. You can never have too much filtering n your power supply.

[Satamax]

Well guys! Thanks a lot.

Stokes, i think i'll have no tube recto! No 5V winding on the salvage trany.

Arthur, i was asking the question because i've heard that too much capacitance in a single stage could cause ghost notes. But as i discover every day, i'm quite prone to pick up wrong ideas!

Thanks and bye.

Max.

[Chris / CMW amps]

Please use fresh/new e-caps.

Regarding overfiltering in a tweed Deluxe; it makes the amp sound and feel dead. Just use 3 x 16-22u for the powersupply-nodes.

[Arthur B.]

Arthur, i was asking the question because i've heard that too much capacitance in a single stage could cause ghost notes. But as i discover every day, i'm quite prone to pick up wrong ideas!

Thanks and bye.

Max.

Ghost noting is attributed to too little capacitance. Capacitors are supposed to replicate ideal voltage sources, which act like shorts to AC. The worse the approximation in your power supply, the more issues with noise you'll have.

[Le Basseur]

Regarding overfiltering in a tweed Deluxe; it makes the amp sound and feel dead.

That's exactly what I noticed on several different amps when I tried the "steroid-capping" formula (no trace of ghost notes,though) instead of getting rid of the hum by more "layout-oriented" measures.The amps in question became dull,bassy/muffled and had weak dynamic response (subjectively).
I'm not sure why's that,maybe it's linked to ESR or something else.Maybe someone should share some knowledge about it.
Thanks!

Regards,
Le Basseur

[Arthur B.]

That's exactly what I noticed on several different amps when I tried the "steroid-capping" formula (no trace of ghost notes,though) instead of getting rid of the hum by more "layout-oriented" measures.The amps in question became dull,bassy/muffled and had weak dynamic response (subjectively).
I'm not sure why's that,maybe it's linked to ESR or something else.Maybe someone should share some knowledge about it.
Thanks!

Regards,
Le Basseur

My guess is we're seeing the circuits flaws, namely that it's bass heavy and lacking dynamic range. It's sort of like drinking a soda when it's flat; you're tasting the soda without the fizz covering up its nastier aspects of its taste.

[Le Basseur]

Arthur,
Nice comparison but I think we have something else here.
Saying that a certain circuit has flaws in terms of "bass heavy and lack of dynamic range" sounds like saying "..I don't like how this Ferrari runs because I suspect they're using too much lead in their red paint".
I built some "Ferrari"s in the past 8-10 years (hi-end amps involving 211,845,300B or 2A3,great OPTs and exquisite component quality) and it happened the same when over-built the HT with electrolytics.There were two possible directions to follow:first,small-to-medium value caps around a chunky piece of iron (a choke sized like,say,a Twin Reverb's OPT),or a small choke (2-5H) with plenty mFs around it.Remember,all this in the context of achieving a PS's desired -69 to -75 noise floor (horn speakers are merciless).The first solution proved to be always superior,in fact,the only one acceptable in terms of sound,fast response and whatever.The only thing that one cannot suspect in such simple or sometimes minimalistic designs is that the circuit has "flaws"...like we'd suspect in a guitar amp's config.
Regards,

[Bruce / Mission Amps]

I don't like over filtered 5E3 amps either but I don't notice much of anything different (tone wise) untill I hit around 33uF to 47uF-450v values caps, and while I agree that over filtering can change the sound and feel of the amp, my experience has always been that this mostly noticable in the preamp stages.
The stiffer the power tube filtering is, usually the better the power stage works.... however, if that under filtered sound is what turns your crank, then by all means, use the minimal amount of capacitive filtering that keeps objectionable ripple off the OT and the screens.
I think that is totally 100% justifiable

And adding what La Basseur was saying ... IMHO,
the classic "Brute Force" PI filter works the best...
two smaller valued caps and a big choke in between is a wondeful B+ supply filter.
A killer filter section is:
B+ from the rectifier cathodes connected to a smaller 16uF cap, then a big honking choke rated at 5Hy or more and 200-300ma, followed by another smallish 16uF cap... then the B+ for every stage in the amp, including the OT is taken from that spot.
Then all the filter caps from that point on can be much smaller then what youd see in say, a big Engliosh amp with 50uF cap across the board.
The smaller, decoupling caps will work just like they are supposed and the amp will sound great.

[stokes]

I have found that if you put larger filters on the preamp supply you get these negative results as you guys describe.Using 100uf caps on my mains and filters just made it tighter,especially in the bass.I have to add that I use .02 coupling caps as well,the .1's are too muddy for my liking,and over filtering with the .1's would likely make it worse.

[Bruce / Mission Amps]

And I just reread my post and it as a bit nebulous... I ment to say that the B+ nodes start after the brute force filter... not every node is taking from that spot. You would still have the dropping resistors and their fitler caps but those caps can be 10uF to 16uF and you'll still haev a very very clean B+ rail, even with the little caps... pplus the preamp is not over filtered... if that is even a word.

[Matt T.]

What is the effect of having a lot of filtering on the plates but not a lot on the screens? I was looking at some of the older Marshall schematics and it looks like some of them had big filters on the plates but then had 2 each 32uF in series for only 16uF for the screens...and then the PI node had a lot of filtering. I recently rebuilt a homebrew smallbox and sort of went that direction; 100uF plates, 20uF screens, 40uF PI, then 22uF for each of the two preamp nodes. It sounds fantastic but I changed the PT (and voltages) at the same time.

[bob p]

i think that in many respects, what constitutes a good design for hifi amps isn't necessarily a good design for guitar amps, but i don't think that the two objectives are mutually exclusive. i guess i'd have to agree with some of the previous comments and disagree with a few others.

i would disagree with the suggestion that "over filtration" in the initial stages of a PSU is bad for a high end hifi amp. where you put the filtration, and how much you use in specific locations, is really important.

the whole purpose of the PSU is to provide a Pure DC, Zero AC, low impedance supply for the gain stages of the amp. most of us seem to have a pretty good handle on the Pure DC / Zero AC concept, but the low impedance aspect of the power supply seems a bit harder to get a grasp on. more on that later.

looking at the problem simplistically, the input filter of the PSU has two jobs: hum reduction / elimination of ripple voltage and delivery of a constant DC voltage whose amplitude is independent of current demands. to the extent that you have more of both, you're going to have a better sounding HiFi amplifier. (i'm going to sidestep the issue of "sag" in a guitar amp, as i think most of us have a pretty good handle on that).

as it turns out, i'm working on a HiFi "monoblocks" project for an old friend, so he visited last weekend to work on his amp project. i wanted to get a feel for what kind of performance / response / tone he was after, so we made appointments to spend the day auditioning components at one of the obscene HiFi shops in Chicago. (stokes, your local equivalent is probably Singer in Manhattan.) we had a chance to listen to some outstanding tube HiFi amps, that included amps like the Hovland Sapphire and the Audio Research reference series. one thing that all of these amps had in common was what we as guitar amp people would consider "excessive" input filtration. suffice it to say that we didn't observe "excessive filtration" compromising the sound of a $10,000 amp going through $20,000 Sonus Faber Homage speakers. the amps were exceptionally quiet, and had remarkable transient response. they were very transparent, musical and accurate. with that said, i'm sure that those setups would have not been desirable for guitar, for obvious reasons.

my only point in bringing this up is to point out that some truly exemplary tube hifi amps use what some of us might consider to be "excessive" filtration that would make an amp sound really bad. but in reality the amps sound really good. so it would seem that there IS a good way to do it -- we just have to figure out how they do it.

simplifying things a little and following Bruce's cue, beefing up the input filter is going to make an amp sound more HiFi (lower noise, lower ripple, lower current dependent voltage drop), and in a HiFi amp, more of that is better. (for a guitar, sag and "sonic bloom" are entirely different objectives.)

i think that one of the reasons that guitar amps tend to sound worse with higher filtration in the preamp sections is that most guitar amps have the filter nodes oriented in series, rather than coming off of the input filter in parallel. each node adds an RC circuit, and the result is that the series resistance of the PSU is cumulative. internal resistance of the PSU can become very high by the time that you reach the input stages. this phenomenon can impart extremely long time constants and significant internal resistance by the time that you get to the end of the rail. internal resistance is bad -- it causes voltages to drop in proportion to load, and the amplitude of the voltage drop is directly proportional to the amount of internal resistance in the PS rail. now that the damage has been done, the longer time constants impair the circuit's ability to recover from voltage transients. these add together to create a two-pronged problem: voltage drops and slow recovery from them.

to some extent, both of these problems are artifacts of PSU internal impedance. looking at this idealistically i think that impedance is more the enemy than capacitance.

(this post is getting a little long, so i'm going to break it up and continue in another)

[bob p]

A killer filter section is:
B+ from the rectifier cathodes connected to a smaller 16uF cap, then a big honking choke rated at 5Hy or more and 200-300ma, followed by another smallish 16uF cap... then the B+ for every stage in the amp, including the OT is taken from that spot.
Then all the filter caps from that point on can be much smaller then what youd see in say, a big Engliosh amp with 50uF cap across the board.
The smaller, decoupling caps will work just like they are supposed and the amp will sound great.

bruce, i like the ideas that you mentioned in another post about creating the brute force input filter and taking the other sections of the rail off of that radially. my preference for the 5E3 is to create an additional node in parallel to the last one on the fender schematic, and to use that node to supply the first gain stage (instead of supplying two gain stages off of the last node). in this approach works great to eliminate motorboating at high control settings when you run the amp with higher gain tubes like the 12ax7.

getting back to the idea of Pi filter design, i'd like to add a couple of comments from a hifi perspective. its really important to size your chokes appropriately for the current demands of your circuit, both in terms of inductance and rated current. depending upon whether the user is designing an amp that will have small or large current drain on the PSU, you might want to have a 5H choke that's rated at higher or lower current.

one of the things that can be confusing about chokes is that their inductance varies greatly with current, so spec'ing a choke can be difficult. optimally, you'd want a choke that has lots of inductance at lower currents (to store energy), that drops to a minimal value at the circuit's maximum current. this helps to prevent too big a choke from imparting "ringing" in the DC voltage rail in response to transients. if you've got too big a choke (in Henries), or a choke that is rated at too high a current for your circuit, it can impart really UGLY ringing artifacts on the DC rail. (Duncan's PSUD can demonstrate this well.)

picking out the right value of a swinging choke has always been conceptually difficult (at least for me) as nobody's ever developed a good mathematical model that completely explains all aspects of the problem. for this reason, choke sizing has always been a more of an art than a science, or something that tends to work out better from experience than from math. (stokes, note this occasion where i go on the record saying that math is no good! )

one reference that i've found particularly helpful is a nomogram for choke sizing that i found in one of Norman Crowhurst's books on hifi design. using the nomogram, you specifiy how much current your circuit draws at idle, and at what voltage, and you specify how much current your circuit draws under full load, and the voltage. the nomogram will tell you the optimum specifications for a swinging choke, both in Henries and the appropriate current rating. best of all, no math! (if anyone's interested in the reference, lemme know and i'll post a link).