• WARNING: Tube/Valve amplifiers use potentially LETHAL HIGH VOLTAGES.
    Building, troubleshooting and testing of these amplifiers should only be
    performed by someone who is thoroughly familiar with
    the safety precautions around high voltages.

5842 Headphone Amp

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
I'm making a headphone amp using parts I have lying around, which includes 5842s for the signal and possibly an 80 (4-pin 5Y3) rectifier. The 80 has a maximum input capacitor rating of 10uF. I wanted to split the power supply at the cathode of the 80, with one CLCRC filter *per channel* as follows:

10uF -- 10H/408R (Hammond 157H) -- 50uF -- 1.3K -- 50uF

My question is: would two 10uF caps on the 80 cathode exceed the capacitance rating? It would see that as 20uF, right? So I'd need a different rectifier, 5uF input caps, or a single rail PS, right?
 
Just food for thought, considering you want to build a headphone amp and the 5842 (417a) doesn't need a lot of voltage, you might want to consider an LCLC power supply instead.

You wont have two power supplies per channel, and you will have considerably lower apparent voltage, but the voltage you would have would be cleaner (no clipped wave form).
 

PRR

Member
Joined 2003
Paid Member
'80 is a 4-pin 5Y3 which sure can use more than 10uFd.

There is NOT a "maximum capacitance". There is a max peak current. This is mostly limited by transformer resistance.

If you use a preamp size transformer (not some 7-pound lump), I would not hesitate to use 40uFd.
 
Tjj226: Thanks for the suggestion. As you note, two choke stages would provide much better PSRR. However, since a goal of the project is to keep costs down, two $10 Hammond 157H are substantially cheaper than, say, 4 x $15 Hammond 156R (1.5H). If I really like this circuit and decide to invest more in it, then I'll definitely add more choke filtration.

Roundabout way of responding to PRR:

OK, here's what I'm working with. All resistances measured with a Klein MM400 DMM.

PT -- Hammond 270BX
* Pri 120V/60Hz -- DCR = 6.6R
* Sec HT 275-0-275 -- DCR = 419R
* Sec 5V 2A -- DCR = 0.3R
* Sec 6.3V 2A -- DCR = 0.5R

Choke -- Hammond 157H -- 10H, 414R (higher than mfr. claimed 408R)

OPT -- Raphaelite OP5.5K5A -- 5.5K, 16R+16R [for 8 or 32R], 5W
Using 16+16 in series for 32R to match my 32R Grado headphones
* Pri DCR = 351R
* Sec DCR = 4.8R for *each* 16R tap and its associated 0R connection

5842 x 2 -- NOS Raytheon JAN


80 x 1 -- NOS Tungsol
* Max plate supply AC voltage = 350V
* Max peak inverse voltage = 1440V
* Max transient peak plate current = 2.2A
* Tube voltage drop = 60V @125mA

I've made a template to follow for PS design, following the process layed out on this page at DIY Audio Projects: Vacuum Tube Power Supply Design

For this iteration, I want 140V on the 5842 plates and a current of 18mA.

To figure out B+, I did V = .018 * 351R = 6.316V. That's the voltage drop across the OPT primary, so...

Desired B+ = 146.3V

Peak plate current = 72mA

Maximum peak inverse voltage = 353.2

Now I'm calculating the peak and average plate resistances. Peak plate resistance requires forward voltage drop as part of the equation. The 80 datasheet indicates this to be 60V at the maximum DC output current of 125mA.

Is voltage drop dependent upon current? If so, my peak plate current is 72mA, meaning my voltage drop would 34.56V. Is that correct?

60/.125 = x/.072
408 = x/.072
.072 * 408 = 34.56
 
Last edited:
You could add a couple of small value series resistors between the power transformer secondary leads and the plates of your 80 rectifier tube. This has several detrimental effects (increased power supply impedance, lower B+ voltage) but would allow you to safely use a slightly higher value of reservoir capacitor with the 80 rectifier tube.

Shown as "Rs" in this schematic:
figure05small.jpg


A 10H 414R choke is probably not going to be optimal for a choke input filter with continuous 72mA current draw. Its spec sheet shows 10H rated at 50mA. But if you had two of those, you could do a separate LC filter for each channel. I think that's okay, right? You'd still be able to use your 80 rectifier with no worries, and each 10H 414R choke would be well within its limitations.

Maybe?
--
 
Last edited:
Thanks, rongon. I figured the 50mA rating was RMS. Maybe I'm mistaken about that? I have to bring the voltage way down, anyway, so judiciously adding resistance before the rectifier plates or in series with the cathode in order to allow more input capacitance is not a bad idea, provided that supply impedance is not raised too high.

The 72mA is *peak* plate current per plate.

Idle current draw based on tube setup is 18ma per channel or 36mA total. Assuming I can make a split rail PS with the 80, with one 157H per channel, that would be 18mA quiescent current going through each 50mA-rated choke, which I assumed was more than enough headroom. But again I could be wrong about that too. I guess we'll see if these things get really hot.

Also, forgot to say before, another reason I liked the 157H is that the 408R (or rather 414R) series resistance would help to bring that B+ voltage down to where I want it. Really I should probably be using a smaller PT with something like a 150V or 160V HT secondary, but again -- for this iteration I'm working with what I have and trying to buy as few parts as possible.

I just want to listen to more music at night, when my kids are sleeping and wife is studying, and that means headphones.

I suppose it would help to share a schematic. I'm starting with DHTRob's "classic" circuit, but without the diode bridge, and with changes for the components I have:

Hoofdtelefoonversterker5842.jpg


And will eventually try his "modern" one en route to finding my own preferred version and op point:

Hoofdtelefoonversterker5842ultrapath.jpg


My concern of course is not to add too much capacitance lest the inrush current kill the rectifier.
 
Last edited:
Ah, silly me. The handy rectifier voltage drop chart in the DIY Audio Projects article corroborates my calculation of ~35V drop @72mA (using the orange-red line for 5Y3):


figure01.jpg



So peak diode resistance = 34.56 / .072 = 480R

Average diode resistance = 1.14 * 480 = 547.2R
 
Last edited:
My concern of course is not to add too much capacitance lest the inrush current kill the rectifier.

Yes, the Rs resistors in the power supply schematic in my previous post are meant to address that problem. The trade-off is loss of rectifier efficiency.

Why not use the diode bridge with the EZ80 (or a suitable damper diode) used in series as a soft-start mechanism? You could use UF4007 diodes to make your own bridge. They're cheap, and not bad.

I figured the 50mA rating was RMS. Maybe I'm mistaken about that?

I don't know for 100% sure, but I believe RMS.

The 72mA is *peak* plate current per plate.

Sounds to me like you're safe in this setup.

Idle current draw based on tube setup is 18ma per channel or 36mA total. Assuming I can make a split rail PS with the 80, with one 157H per channel, that would be 18mA quiescent current going through each 50mA-rated choke, which I assumed was more than enough headroom. But again I could be wrong about that too. I guess we'll see if these things get really hot.

I think you're totally safe with that. The question I have is whether there is any kind of hidden problem with 'splitting' the raw (high ripple) DC off the rectifier into two filter chokes. Those Hammonds are not actual swinging chokes, but I think they'd be de-rated enough in this application. But I'm no expert. Maybe someone with more expertise can chime in on this...

Also, forgot to say before, another reason I liked the 157H is that the 408R (or rather 414R) series resistance would help to bring that B+ voltage down to where I want it. Really I should probably be using a smaller PT with something like a 150V or 160V HT secondary, but again -- for this iteration I'm working with what I have and trying to buy as few parts as possible.

I hear you about working with what you got. I have too many parts collected. Must build, use the stuff! But 414 ohms won't drop much voltage at those currents.

I just want to listen to more music at night, when my kids are sleeping and wife is studying, and that means headphones.

I hear ya! Definitely. I never thought I'd enjoy listening in headphones as much as I do these days.

275V secondary into a choke input filter should yield about 245V from the rectifier. Then, 414R with 18mA will drop only 7.5V, so you're still way up at 237V for your B+. From the schematics, it looks like you want about 160V final B+ to each 5842.

What's wrong with adding a series string of RC filters after each choke input filter? You need to drop 237V down to about 160V, so that's 77V at 18mA you need to burn off. That's about 4.3k ohms. Hmmm... that's a lot. But if you divide that into four 1k-22uF sections in series, that could get the job done, and suppress ripple.

--
 
Re: diode bridge -- For purely irrational, emotional, nostalgic, sentimental, and human reasons, I want only a tube rectifier -- no sand, man! I will try an SS rectifier one day, but not today.



What's wrong with adding a series string of RC filters after each choke input filter? You need to drop 237V down to about 160V, so that's 77V at 18mA you need to burn off. That's about 4.3k ohms. Hmmm... that's a lot. But if you divide that into four 1k-22uF sections in series, that could get the job done, and suppress ripple.

--


Yes! Very good idea. However I'm just wary of adding all that capacitance and killing the rectifier. I guess what I don't understand is whether over-capacitance would come from the series addition or sum total of all the capacitor values or if, say, one excessively rated capacitor could do it.


I played around with this in PSUDII, and adding 100uF caps in positions 2 and 3 (I was wondering if I could use the JJ 100uF + 100uF can caps I have on hand) exceeded the maximum allowed current for the `80 (in that sim). I haven't done that math by hand yet, but there we are.
 
Yes, it's the current drawn by all power supply filters at startup that would overload the EZ80 plate dissipation capability. Very likely with a string of large-ish value like 100uF.

What does PSUD2 say if you use 4X 1k and 22uF instead?

One solution would be to just buy a 115V:115V isolation transformer like the Triad N51X. That would give you about 150V for your B+. Then get a separate transformer for the 5842 heaters, maybe like this 12.6VCT 1A trannie.

If you don't mind more complexity, you could burn off that extra voltage using a voltage regulator. That's what I've been wrestling with lately... +150V DC Vacuum Tube Regulated Power Supply?
 
Well, maybe I'm not using PSUD2 correctly, but consulting the 5842 datasheet shows me that the plate resistance is about 1.7K at 150V (https://frank.pocnet.net/sheets/049/5/5842.pdf). So made the value of the end resistor 1.7K (see attached screenshot).



That results in 77V on the final resistor -- about half what I need -- but the ripple is incredibly low at 1.9uV.
 

Attachments

  • Screen Shot 2018-06-13 at 2.53.26 PM.png
    Screen Shot 2018-06-13 at 2.53.26 PM.png
    161 KB · Views: 319
Last edited:
That 1.7k rp is with 150V plate voltage, 25mA plate current, with a 60R cathode resistor, so -1.5V grid bias.

According to the plate curves, plate voltage of 150V with plate current of 18mA should mean grid bias of about -1.9V, so cathode resistor of about 105R. Transconductance would be a bit lower at that operating point, so I'd figure the rp would be closer to 2.5k.

http://www.westernelectric.com/spec_sheets/417A.pdf

If you're sure your 5842 will be drawing 18mA plate current, try using an 18mA current sink as the load in PSUD2.

--
 
OK, posting results. After going through the first calculations, it turned out that my peak plate current was actually more like 5x the average current. So I recalculated the other key parameters.

Here's what I originally got:

B+ (Vl) -- 146.3V
(Il) -- 36mA
Vm -- 120V
Fm -- 60Hz
Average plate current -- 18mA
Peak plate current (assume 4x average) -- 72mA
Max peak inverse voltage -- 353.2V
Peak diode resistance -- 480R
Average diode resistance -- 547.2R
Effective PT secondary resistance (Rs) -- 453.66R
Voltage step-up ratio (N) -- 2.29167
Effective peak transformer resistance -- 933.66R
Effective average transformer resistance -- 1000.86R
Equivalent load resistance (Rl) -- 7.639K
Frequency time constant (wRC) -- 28.78
Peak source resistance to load ratio -- 0.1222
Average source resistance to load ratio -- 0.1310
Check of peak diode current in Reich/Schade's plots reveals it to be 5x

Recalculating key parameters with the new peak current reveals the following:

Peak plate current -- 90mA
Peak diode resistance -- 444.4R
Average diode resistance -- 506.6R
Effective peak transformer resistance -- 898.1R
Effective average transformer resistance -- 960.33R
Peak source resistance to load ratio -- 0.117
Average source resistance to load ratio -- 0.1257
Check of Reich/Schade plot reveals no significant departure from original plot

With the new information, some more calculations:

Rectifier efficiency (Er) -- 0.725 (looks good to the right of the knee in the chart)
PT peak secondary voltage -- 379.3V
PT RMS secondary voltage required -- 268.25V (very close to my PT's 275V per section)

The `80 datasheet says max. peak steady state plate current is 400mA, so the 90mA is still well within spec (unless I'm mistaken or have botched any of these calculations).

If you're sure your 5842 will be drawing 18mA plate current, try using an 18mA current sink as the load in PSUD2.

Making the load an 18mA CCS results in 243V B+ and 877nV ripple. (attached)
 

Attachments

  • Screen Shot 2018-06-13 at 3.14.27 PM.png
    Screen Shot 2018-06-13 at 3.14.27 PM.png
    146.8 KB · Views: 233
Making the load an 18mA CCS results in 243V B+ and 877nV ripple. (attached)

Try making C1 0.1uF (in other words, make it a choke input psu).

The `80 datasheet says max. peak steady state plate current is 400mA

Confusing. "peak steady state plate current"? 400mA?? That's a ton of current for any tube rectifier. I don't think that's right.

According to this data sheet (Philips EZ80) the max continuous output current (Io max) with 250V per plate is 90mA.

--
 
According to this data sheet (Philips EZ80) the max continuous output current (Io max) with 250V per plate is 90mA.
--
I'm using the `80, not the EZ80. http://www.r-type.org/pdfs/80.pdf

And I think what we're looking at is plate current, not output current.

According to the linked datasheet chart, though, if it's 90mA output then that would seem to exceed the limits of the tube.
 
Some more calculations following the procedure document.

Ripple factor (Rf) = 0.022
Output voltage (Edc) = 289.1V
Ripple voltage (Vr) = 6.2V (2.2%)

Now I need to figure out the smoothing factor and -dBs of all this, and the document seems to assume I know how to do that already (I don't).

For any noobs like me who are reading this, I'm looking at this document for audio db calculations now: http://eaw.com/docs/6_Technical_Information/StudyHall_and_TechNotes/dB_calculations.pdf
 
Last edited:
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.