Smoothing Choke - Marshall 50 Watt - Drake T 100

[CJ]

this is called a smoothing choke,

it is used in the power supply of many tube amps as part of the filtering circuit,

smaller amps will use a resistor in it's place, usually 6V6 and similar power tubes,

for bigger tubes, like the 6L6 and EL34, maybe a 6550 or KT120, you will see a choke in place of the resistor,

why use an expensive choke which takes up chassis space if you can get away with a 10 cent resistor?

good question, wish i knew the answer!  my guess is that the choke will have less of a voltage drop across it which will mean more screen voltage which will mean more power, which will mean more sales,

maybe it helps the stability of the amp, when a big tube starts working hard, there can be a big  AC voltage on the screen grids, this will want to move the power power supply B+ around a bit, some of this might get to the pre-amp stages,

a choke resists a change in current, so the power supply node feeding the screens will moving around on the supply side of the choke, but not on the downstream side, so you have less ripple on that second cap which will improve filtering and allow use of a smaller cap,

see that the choke and filter cap will form a resonant circuit, hopefully out of the audio range that we are designing the amp for,

we are going to pick this choke apart at the seems and see what makes it tick,

it is from a classic  79 Marshall 50 watt which features a Drake output which has been covered in a different thread,

 

[CJ]

attached is a Reactance graph for the choke, Reactance is just resistance,  to AC voltage that is, resistors might have inductance in the leads, especially the big 2 watt old school carbon comps, makes good for a grid stopper, eh?

but that is ho freq reactance because the inductance is so low,  you know, XL= 2 pi f L

but chokes with many Henries have a ton of reactance,

notice this choke craps out at about 6 grand, current has been steadily decreasing with frequency due to the above formula, but then it takes off again, what gives?

that dreaded capacitance, too many turns, but who cares, we are filtering dc ripple at 120 Hz,

but is that all? what about garbage from the output transformer or tubes, won't that choke filter that too?

a little bit, 6,000 Hz is fairly high for a guitar amp, so we get most of it,
that first filter cap should eat up any HF garbage, the higher the freq, the more they chew,
until a point, big lytics have inductance,
so some people bypass the caps with 0.1 caps, others use long B+  leads for chassis wiring capacitance,
note that a plate choke for a V72 must have better hi end specs, if 6,000 Hz and above were allowed to pass thru the choke due to capacitance, then you would lose your HF audio to the filter caps which can swallow hi freqs with ease,

note the steady increase with f, this is due to a constant core perm due to the air gap,

this graph has been corrected for the 107 ohms of DC resistance from the choke turns, this series resistance is only a factor at low freqs where the choke has less reactance due to low f,

keep getting spell check error for reactance, wtf, over? 

why is the uphill line flat and the downhill line curved? 

uphill is inductance, downhill is capacitance,

uphill is linear, 2pi f L,  downhill is of the curve family 1/x, as capacitor  reactance is 1 / 2 pi f C
so do your really have a bell shaped curve with a Pultec?
what is 1/x, a hyperbola or parabola? geometry book is in the garage and it is bite ass out there,  ;D

 

[CJ]

some info from a great resource, Aiken Amps, please check out their website,

after the blurb we have an inductance graph vs freq so you get the Henries, which is proportional to reactance,  expanded graph for lo freqs is shown, reactance for the UK (100 Hz ripple) and US (120 Hz ripple) is shown.

"Chokes Explained

General

    A "choke" is the common name given to an inductor that is used as a power supply filter element.  They are typically gapped iron core units, similar in appearance to a small transformer, but with only two leads exiting the housing.  The current in an inductor cannot change instantaneously; that is, inductors tend to resist any change in current flow.  This property makes them good for use as filter elements, since they tend to "smooth out" the ripples in the rectified voltage waveform.

Why use a choke? Why not just a big series resistor?

    A choke is used in place of a series resistor because the choke allows better filtering (less residual AC ripple on the supply, which means less hum in the output of the amp) and less voltage drop. An "ideal" inductor would have zero DC resistance. If you just used a larger resistor, you would quickly come to a point where the voltage drop would be too large, and, in addition, the supply "sag" would be too great, because the current difference between full power output and idle can be large, especially in a class AB amplifier.

Capacitor input or choke input filter?

    There are two common power supply configurations: capacitor input and choke input. The capacitor input filter doesn't necessarily have to have a choke, but it may have one for additional filtering. The choke input supply by definition must have a choke. Capacitor input filters are by far the most commonly used configuration in guitar amplifiers (in fact, I can't think of a production guitar amp that used a choke input filter).

    The capacitor input supply will have a filter capacitor immediately following the rectifier. It may or may not then have a second filter composed of a series resistor or choke followed by another capacitor. The "cap, inductor, cap" network is commonly called a "Pi filter" network. The advantage of the capacitor input filter is higher output voltage, but it has poorer voltage regulation than the choke input filter. The output voltage approaches sqrt(2)*Vrms of the AC voltage.

    The choke input supply will have a choke immediately following the rectifier. The main advantage of a choke input supply is better voltage regulation, but at the expense of much lower output voltage. The output voltage approaches (2*sqrt(2)/Pi)*Vrms of the AC voltage. The choke input filter must have a certain minimum current drawn through it to maintain regulation.

    The voltage difference between the two filter types can be quite large. For example, assume you have a 300-0-300 tranny and a full-wave rectifier. If you use a capacitor input filter, you'll get a no-load max DC voltage of 424 volts, which will sag down to a voltage dependent on the load current and the resistance of the secondary windings. If you use the same transformer with a choke input filter, the peak output DC voltage will be 270V, and will be much more highly regulated than the capactor input filter (less variations in supply voltage with variations in load current).

How to select a choke:


Chokes are typically rated in terms of max DC current, DC resistance, inductance, and a voltage rating, which is the max safe voltage that can be applied between the coil and the frame (which is usually grounded).







        DC current



If you are using a choke-input filter (not likely, unless you are trying to convert a class AB amp to true class A and need the lower voltage, or if you are designing an amp from scratch and want better supply regulation), the choke must be capable of handling the entire current of the output tubes as well as the preamp section. Note that this doesn't mean just the bias current of the output tubes, but the peak current at full output. This usually requires a choke about the size of a standard 30W-50W output transformer, since the choke must have an air gap (just like a single-ended OT) to avoid core saturation due to the offset DC current flowing through it, and the choke also must have a low DC resistance, to avoid dropping too much voltage across it, which will lower the output voltage and worsen the load regulation. This combination of low DCR, air gap, and high inductance (more on that later...) usually results in a substantial sized choke. To calculate the required current rating, add up the full power output tube plate currents, screen currents, and the preamp supply currents, and add in a factor for margin. For a 50W amp, this may be 250mA or so.

If, on the other hand, you are selecting a choke for a capacitor input supply (such as the typical Marshall or Fender design), then the requirements are relaxed quite a bit. The purpose of the choke in these type supplies is not for filtering and voltage regulation, but just for filtering the DC supply to the screen grids of the output tubes and the preamp section. The screens typically take around 5-10mA each, and the preamp tubes draw about 1-2mA or so (for the typical 12AX7; 12AT7's are usually biased for around ten times that). This means that you can get by with a much smaller choke, and, in addition, the preamp supply current doesn't vary that much, so you can get by with a higher DC resistance, which means smaller wire can be used to wind the choke, which means higher inductance for a given size core. Just add up the current requirements of the screens and preamp tubes, and add a bit more for margin. For a 50W amp, a typical value might be 50-60mA.




    DC resistance



For a typical choke input supply, you need a choke with no more than 100-200 ohms or so DCR. A capacitor input supply typically might use a choke with a 250 ohm - 1K DCR. The higher the resistance, the more voltage drop and the poorer the regulation, but the cost will be lower.





    Inductance



As for the inductance value, this depends on how much filtering you want. The inductance, in conjunction with the filter capacitance, forms a lowpass filter. The larger the inductor, the lower the cutoff frequency of the filter, and the better the rejection of the 120Hz (if full wave rectified) or 60Hz (if half wave rectified) AC component of the rectified DC. In general, the larger the better, within reason (larger inductances at low DC resistances mean larger chokes, which cost more money). Typically, 5-20 Henries is a good choice with the standard 32-50uF electrolytic capacitors. The inductance and capacitance values also determine the transient response of the supply, which means the tendency for the supply to overshoot or "ring" with damped oscillations whenever a current transient is applied (such as at startup or on a heavy current surge, such as a hard "E" chord at full power!).





    Voltage rating



The voltage rating must be higher than the supply voltage, or the insulation on the wire may break down, shorting the supply to the frame.


I highly suggest going to Duncan Munro's website (http://www.duncanamps.com/ ) to download his power supply calculator program. It will allow you to experiment with different inductance and capacitance values and see the resulting residual AC ripple and transient response of the supply filter. Both capacitor input and inductor input filters can be simulated. It is a great educational tool."

 

[CJ]

here is a link for many Marshall schematics>

http://www.drtube.com/library/schematics/69-marshall-schemas#JMP-2203

here is the one for this choke>

http://www.drtube.com/schematics/marshall/2204u.gif

(OT - we like a 12AT7 phase inverter for stability and less distortion,  and we remove the 47 pf across the plates as it seems to make things worse)

here is an LC calculator>

http://www.1728.org/resfreq.htm

our choke and 50 uf cap forms a series LC  resonant circuit - rez freq = about 6.7 Hertz at 120 ripple

ok let's unwind this thing, find which lead goes to the B+ supply and which lead goes to the screens, start or finish, and we can get the core info and see what kind of gauss Drake is hitting the core with and also determine DC flux,

 

[CJ]

this varnish was pretty harsh, had to do the wake and bake to get it apart,

 

[CJ]

using regular lams as keeper lams,

Kraft Paper gap,

lucky we do not have to tear apart the lams with this gapped design. metric core, right in between 75 EI (3/4" tongue)and 87EI (7/8" tongue), no stock on hand for a re-lam, no bobbin either, so we save the parts for the rewind,

 

[CJ]

got the coil off, got the KP wrapper off, warped bobbin to to varnish bake,

look at all the room left on that bobbin, maybe we can play some tricks with this coil to up the freq response, just for the heck of it, see if we notice anything different about the amplifier, sound, waveform,

after all, a choke can be considered as the primary of a transformer, it responds to the same techniques used to improve HF response, coil sectioning, turns count, wire size, inter-layer insulation,

 

[CJ]

random wound at about 100 turns per layer,

unwind on the machine so we can use the turns counter,

 

[CJ]

first layer distorted due to a tight wind,

 

[CJ]

ok so here is what we did to improve HF response,

added some 15 mil KP between layers, until we noticed that we would not have room for all 2199 turns of #30,

so we finished off with thin mylar tape in between layers, wound every layer flat, and ended up with about 1871 turns of #30,

DCR dropped to from 107 to 80 ohms,

 

[CJ]

since we have less turns, and inductance is a function of turns squared, we can use a smaller gap to compensate by raising the permeability of the core,  that way we don't lose too many Henries,

but what about the DC, won't it saturate the core? well, since we have less turns, we have less DC flux, as DC flux is a function of amp-turns,  so maybe we had 20 ma thru the coil for screen current and preamp tubes,  amp turns will be .02 x 2200 = 44, now we have .02 x 1871 = 37.4, or a drop of about 15 percent,

what about AC flux, fewer turns will mean more AC flux, well, in this case the AC voltage is pretty low due to the first filter cap dropping the ripple down pretty far, and AC flux is usually a small percentage of the flux,  the amp turns contribute more to flux than ripple, usually by a large margin, we can calculate all this with a spread sheet,

so how much improvement in hi freq response did we get?

here is the new chart for reactance on our rewind job, before we rolled off at 6 KHz>

 

[CJ]

so now we roll off at 20K, this should isolate the OPT B+ node from the Screen node a bit better at the high end, we have eliminated some capacitance by putting more space between layers,

what about inductance?

here is the before and after inductance charts,

bake and dip this thing and install it in the amp, plug and play, fun fun fun...

 

[CJ]

as expected, our AC flux for this choke is very low, we used 5 volts as the ripple level, it might go up a bit when the amp is really crankin,

here are the two levels of flux with the stock 2200 turns and the new 1871 turns,

 

[PRR]

The usual point of a Screen choke is (as I think you said) you can have 5K of impedance at buzz-frequency with only 100 Ohms of DC drop.

20mA at 100r = 2V drop
20mA at 5K = 100V drop

The screens, driver, and maybe the preamp will benefit from a B+ 98V higher.

This also explains the un-full winding. If the thin wire gives 2V drop, what would you gain from fatter wire? 1.6V drop? Is 0.4V more really meaningful at 400V? So you (the factory designer hoping to get paid) have to think: can I use a thinner gauge, save several pennies each, and defer bankruptcy that much longer? OK, heat goes up, but lifetime is still likely to be 20-30 years, by that time I'll be gone and probably nobody will be playing these things any more. So thin wire, half-empty bobbin, all is well.

There may be a second-order effect in HARD-driven guitar amps. On heavy hit, the screen voltage will dip and then recover. (Somewhat as your 6.7Hz resonance.) This may modify the overdrive attack sound.

I don't see any point in "extended HF response" here. The 16uFd-40uFd cap on the end sucks all that up.

There is an additional use for choke short of a full Choke-Input design. Some Sunns ran a small first cap, a choke, and another small cap. This reduced cap-cost for these very high voltage amps. It also limited rectifier abuse for the early models with hollow rectifiers. (Even the early sand-rectifier Sunns push their rects hard.) This was probably designed from Radio Amateur handbooks, which often covered the topic clearly.

 

[CJ]

good thing about choke circuits is that you can take them out and put in a resistor to drop screen volts, since there is a trend now days to run JJ 6L6 tubes at 510 volts B+ with 509 on the screens,

yes this HF choke biz is just an experiment,

a side note, that Drake OPT that we rewound with more paper between layers now runs flat to 100K with no peaks, the old one had a peak at 60K and then dropped off, so it will be interesting to see how this hi-fi git amp sounds, there was some slight HF osc seen on the output waveform, probably from the inverter, going to see if this gets better or worse with the new iron,

here is a comparison of DC flux, nothing to worry about as far as a reduced gap, as in both cases, the DC flux is very low, why did they use such a big core? to get the inductance with lower turns, and thus lower DCR than Fender, 

notice the big drop in effective perm with the gap,  .003 is about as good as you can do with a plastic hammer on an EI 87 with no KP, there is always some air in there,

we are guessing at an initial perm of 8000 for 29gaM6 or even barn roof,

so our max combined flux will be about 2700 + 700 = 3.4KG, well below the 18 KG rating for steel lams,

 

[CJ]

as an added bonus for anyone who endured this boring choke thread, we have a pic of Rein Narma and Les Paul thrown in as a bonus, 

 

[dmp]

What frequency is the inductance of Chokes typically spec'd at?
I think Fender blackface amps had 4H chokes. I'm curious how that compares to this Marshall.
Great info in this thread.