Choke placement?

[jordan86]

In my continual effort to optimize this little SE marshall build, I have been reading up on Chokes. Read an article by Aiken amps and a few threads here.  I've never installed a choke, but a $15 experiment seems worth the fun to me. A few questions I have...

1) Placement. Marked some placements by number for ease. My understanding is that #4 would be best? I don't think it supposed to come before the B+1 node that feeds the OT CT.

2) Are chokes helpful in making SE amps less....well....Single ended :)  I'd love to tighten the lows, kill some hum, sweeten the overdrive, and get a little more punch like a Push pull amp. Will a choke do those?

3) Models. With two preamp tubes and one power tube, I think 50ma should be more than enough. Looking at the attached models. I think my voltages are a tad higher that the schematic, but still just barely under 400v. I'm thinking 50ma+, 5H should be more than enough, lower R value the better? Any thoughts?

FWIW. The preamp in pretty 1987 Super Lead-ish now. Paralleled the first two stages instead of cascading. That's more my style, and sort of what I am going for but in a smaller package.

Thanks in advance!!!!

[sluckey]

You would get the most benefit from a choke by replacing R2 with the choke. You would need a 100mA choke because that KT88 is pulling 70mA right by itself. I suggest a Hammond 158M or 159P.

[DummyLoad]

hammond 158Q would be a good fit + upside also, apart from much, much cleaner DC, is that it has the ≈ same DC resistance as R2 - 100Ω it's replacing. 

--pete

[jordan86]

Sounds like 158Q is a good fit. I just measured the B+ coming off that first node. 401v. Think I am  ok with a 400V choke? Most of the Hammond data online says the 158Q is rated for max 400V...but this one data sheet says it's 500V.
https://www.hammfg.com/files/parts/pdf/158Q.pdf

My gut says I am ok.

[jordan86]

I suggest a Hammond 158M or 159P.

Is 10H going to be much better than 5H in this placement? FWIW I believe I bias the KT88 to roughly 110ma when I use it. I've been liking a kt66 more lately though, which I do bias around 65ma.

[66Strat]

I would rely on the data sheet specific to the choke, i.e 500 volts. IMO the higher current capacity of the 158Q is preferable. You may even want to consider the 159R, 6H at 200 ma. The B+ voltage would drop and additional 4 or 5 volts. But I doubt that anyone could hear the difference between 400 volts and 395 volts.

Edit to add link.

https://www.hammfg.com/files/parts/pdf/159R.pdf

[HotBluePlates]

Is 10H going to be much better than 5H in this placement? ...

The reactance of the 47µF at 120Hz is 28.2Ω.  This forms a voltage divider with R2 (or the choke) that reduces the ripple present at C1 by 28Ω / (100Ω + 28Ω) = ~0.22 = ~1/5th.

A 10H choke has ~7540Ω of reactance at 120Hz, while a 5H choke has half that at 3770Ω.

     The 5H choke forms a voltage divider with C2 as 28Ω / (3770Ω + 28Ω) = 0.0074 = ~1/136

     The 10H choke forms a voltage divider with C2 as 28Ω / (7540Ω + 28Ω) = 0.0037 = ~1/270

So the 10H choke is twice as effective at filtering, and will cut ripple by half compared to the 5H.  You'll get the same effect by doubling the filter cap value to 94µF (which you'd really round up to 100µF).

... I have been reading up on Chokes. ...

2) Are chokes helpful in making SE amps less....well....Single ended :)  I'd love to tighten the lows, kill some hum ... Will a choke do those? ...

The #1 benefit of a choke is it is not a resistor.

Your schematic says there is 10v dropped across R2 of 100Ω.  You'd need a 7.5kΩ resistor to get the filtering effect of the 10H choke, which will drop 0.1A x 7500Ω = 750v (!) = you don't even have enough supply voltage to use it!

So you get the filtering power, but not the voltage drop that comes with using a resistor with the same opposition to ripple current (resistance & reactance are different, but both are opposition to current).

At the screen power supply node you get filtering that doesn't drop the supply voltage dynamically, the way a resistor would cause screen voltage to sag.  Sagging screen voltage will constrict peak plate current, constrict maximum power output, and introduce compression.  It's an artistic choice what is good/bad here; however, it is likely why big Class AB amps have chokes feeding their screen power supply node.

As for "sweetening"...     Less power supply ripple means less chance the amp-audio will add intermodulation distortion where the desired audio generates products with the ripple.  Some people really dislike that IM distortion, but perfect power supply filtering fights the squishy feel other players want.

[thetragichero]

however, it is likely why big Class AB amps have chokes feeding their screen power supply node.

I'd think another reason would be economy: feeding the screen supply rather than the plates/whole b+ rail requires significantly less current capability and one less filter capacitor

[jordan86]

Thanks all for the explanations and insights. I’m learning so much on here.

I think I’m going to go with the 5H, 158Q, more so just due to physical size. That 10H is pretty massive. Excited to experiment though and see the tone differences. Also hoping for a quieter amp. Will report back.

[HotBluePlates]

however, it is likely why big Class AB amps have chokes feeding their screen power supply node.

I'd think another reason would be economy: feeding the screen supply rather than the plates/whole b+ rail requires significantly less current capability and one less filter capacitor

True, High-Current Many-H chokes are big and expensive.  Many have said over the years that once upon a time iron & copper were cheap, but many-µFs in a reasonably-small space were expensive.  Back then, iron instead of capacitors (or electronic regulators) was how filtering got done.

But chokes tend to choke or fight big changes of current.  While you can filter the entire B+ supply of a Class AB amp, you're gonna need many-many-µFs in the filter cap after the choke & at the OT/plate node.  Class AB amps have very large changes of plate current, which is part of what makes them AB (people focus too much on the biasing).

But Class AB is push-pull:  The voltage-output that would have occurred due to ripple fed to the center-tap of the push-pull OT is canceled in the windings, because of how they are wound in the same direction around the core but travel to opposite ends of the total primary winding.  So we can tolerate big-ripple at the OT/plate.

But output tube screens amplify from screen-to-plate.  Not as much as grid-to-plate, but an amplification factor on the order of 7-12.  As a result, hum at the screen is troublesome & louder than hum applied to a primary center-tap.

Single-ended doesn't enjoy the benefits of the push-pull OT, so hum matters at least some at the OT/plate feed.  But most amp-designs employing single-ended operation use a small output tube, to deliver lower output-power, which is less current-draw, which is a smaller & cheaper choke.

5v of ripple at the first filter cap might be typical of many amplifiers.  The 100Ω/47µF setup knocks that down to 1v, which is essentially a signal input across the OT primary winding.

The schematic shows 8Ω from White to Black, which will reflect 2500Ω according to the 125ESE data sheet.  The impedance ratio is 2500:8, or 312.5:1, and so the voltage ratio is 17.68:1.  1v/17.68 = ~56mV.

56mV across 8Ω is 0.4 milliwatts.  Pretty small.

But what if we have a speaker of 100dB SPL sensitivity, like a Celestion Blue Alnico?
     10 log (0.0004w / 1w) +100dB = 10 x -3.4dB +100dB = -34dB +100dB = 66dB.
     Hum at the speaker is loud enough to compete with a quiet/moderate TV, but not so bad if the amp is played loud.

Slug the ripple with a 10H choke, and the ripple applied to the OT drops to 18.5mV.  OT voltage ratio steps that down to ~1mV into 8Ω, for ~0.14 microwatts.  Hum output at the 100dB speaker falls to ~31dB SPL.

The 35dB difference is equivalent to a loudness-difference between a 1w amp and a 3,162w amp.

If anyone goes building themselves a tweed Champ or Princeton clone, they'd be wise in several ways to copy the 5B2-5E2 Princeton, or the 5C1 to 5E1 Champ.  Make sure to peep how they set up the power supply in those amps, then marvel at the folks who post on forums about not liking the hum in their new Fender 57 Custom Champ.

[thetragichero]

choke before ot is a MUST for single ended builds in my opinion, unless I'm doing something intentionally janky like an aa5/widowmaker (WITH isolation transformer)

[HotBluePlates]

... I believe I bias the KT88 to roughly 110ma when I use it. I've been liking a kt66 more lately though, which I do bias around 65ma.

Note that your 125ESE is rated for a max of 80mA d.c. in the primary.

[jordan86]

Note that your 125ESE is rated for a max of 80mA d.c. in the primary.

The friend who originally built this upgraded to the 20W Doberman OT by Heyboer. The one the AX84 guys custom make for this amp. Same 2.5k primary but it does handle 150ma+ of current.

[jordan86]

I'm also wondering about removing R1 that is paralleled across the first filter cap. What benefit would it be providing? I've only ever seen a resistor used across a filter cap, when combining two caps in series.

Is that called a bleeder resistor? Supposedly bleeds that caps upon powering the amp down?

[thetragichero]

yup leave it there. you'll thank yourself as you power the amp off to work on it. still measure the cap voltages before sticking your hands in there though!

[HotBluePlates]

I'm also wondering about removing R1 that is paralleled across the first filter cap. What benefit would it be providing? ...

I've been shocked by ~400vdc in an amp whose filter caps "remembered" a charge.  The caps were discharged, but then the amp sat for a few hours before I got a chance to get back inside it.  After I was zapped, I measured the 400vdc at the filter caps.

I've now got a Belt-n-Suspenders approach to safely discharging an amp before reaching inside.  As thetragichero says, you should thank yourself if you never feel what high voltage d.c. feels like (make 120vac funny by comparison).

[Mike_J]

I'm also wondering about removing R1 that is paralleled across the first filter cap. What benefit would it be providing? I've only ever seen a resistor used across a filter cap, when combining two caps in series.

Is that called a bleeder resistor? Supposedly bleeds that caps upon powering the amp down?

From what I recall reading, in addition to it bleeding voltage on power down I believe it helps spread the voltage to both caps so for instance you have 350VDC caps in series giving you 700VDC. Without the 220K resistors the first 350VDC cap can be hit by the voltage before the second at start up. Since the main purpose of having 700VDC is to protect from a voltage spike at start up it is important to have the resistors across the caps.
Not so sure about a single cap. You are probably right that it is a bleeder cap.

I made a 5E3 tweed deluxe clone. It hummed so I took a chance and threw a choke in it. It cured the hum problem and tightened up the bass a lot. However, since the purpose of having a tweed deluxe is not to have real tight bass I had to figure out what was causing the hum and fix it without the choke. It is quiet now and I enjoy playing it quite a bit. Like everything it is what you are trying to get out of your particular amp.

[HotBluePlates]

From what I recall reading, in addition to it bleeding voltage on power down I believe it helps spread the voltage to both caps ...

Two resistors in series from a voltage source to ground = Voltage Divider.

You're probably aware of an adjustable voltage divider: a Volume Control.

Two capacitors in series are also a voltage divider, though usually just for a.c.  Normally, the division is covered by the capacitance in each cap, just as it is with resistance of each resistor.

     But electrolytic caps can't be counted on to be a precise value, so divison is uncertain.

     And the series-caps will tend divide the voltage according to their leakage current.

     Between the two, the series caps from 500v to ground could wind up with 50v on one cap and 450v on the other.  Or really any-possible split of the 500v.

So the two resistors across series-caps force equal-d.c. volts across each cap.  And when you turn the amp off, they provide a resistive path between the cap's terminals, discharging the stored voltage.  So it's at least 2 different jobs performed by one set of parts.

[66Strat]

According to the schematic, there is one 220K resistor wired parallel to ground across a single 47uF capacitor.

[sluckey]

in addition to it bleeding voltage on power down I believe it helps spread the voltage to both caps so for instance you have 350VDC caps in series giving you 700VDC.

This is not applicable to Jordan's question. He has one cap and one resistor. The sole purpose of the resistor is to bleed the cap charge when power is removed. Nothing else.

[Mike_J]

in addition to it bleeding voltage on power down I believe it helps spread the voltage to both caps so for instance you have 350VDC caps in series giving you 700VDC.

This is not applicable to Jordan's question. He has one cap and one resistor. The sole purpose of the resistor is to bleed the cap charge when power is removed. Nothing else.

I was a little late to the party in reading Jordan's question thoroughly but I did modify my post to say exactly what you just did.

[jordan86]

Thanks all for the insights. Got the choke installed. Easy straight up swap for the first filtering resistor.

While the tone results were not as drastic as I hoped I would say it is an improvement in every area. Low end holds together better. Less like a small amp trying to keep up. It’s a tad stiffer but not in a bad way. Overdrive seems a bit smoother. Again, all subtle. If I need a touch more sag I can run it in on the variac.

I was surprised it did not improvement anything noise-wise. I still have this 60hz hum. Very subtle but happens even with the amp only in standby (not full on) and without any tubes installed. Added an elevated heater / virtual CT onto the kt88 cathode. No change. In a good spot though. Thanks again all.

[Williamblake]

All the good advice already givven i would like to remind you that good choke placement is about placing it 90 degrees to the OT and PT to not catch hum.

[brewdude]

Hum with no tubes?
Does this imply the OT is coupling to the PT?
Can you adjust the physical position of the OT?

[jordan86]

Yea, I think moving the OT is the best choice. Possible but not terribly convenient so will live with it as is for now.

[HotBluePlates]

... Got the choke installed. ... I was surprised it did not improvement anything noise-wise. I still have this 60hz hum. ...

Where is the choke installed in-the-circuit?

[jordan86]

Where is the choke installed in-the-circuit?

Did a straight swap for R2, right after the first filter.

[HotBluePlates]

Where is the choke installed in-the-circuit?

Did a straight swap for R2, right after the first filter.

Move your output transformer's Brown wire to the other side of the choke, at C2.

I'll bet the cost of the choke your hum drops dramatically, for the reasons I already explained.

[sluckey]

Move your output transformer's Brown wire to the other side of the choke, at C2.

It's always been connected to C2.

[HotBluePlates]

Move your output transformer's Brown wire to the other side of the choke, at C2.

It's always been connected to C2.

Is it?  I only have his schematic to go by, which shows it connected to C1.

[sluckey]

His schematic clearly shows the brown wire connected to C2.

[HotBluePlates]

His schematic clearly shows the brown wire connected to C2.

Thanks!  I misinterpreted the "1" in the green circle to be "B+1."   

[jordan86]

Yeah that’s where it is. The amp is pretty quiet in action. The hum that I can’t kick I think may be transformer placement. Faint enough to live with for now.