5879 Princeton build

[firemedic]

I just finished (mostly) my 2nd build, loosely based on a SE Princeton. After hearing so many praises sung for the 5879 tube I had to try one.
I took an 18W Kustom Contender tube/SS hybrid, which sounded pretty awesome for $150, and tore everything out of the chassis. The idea was to build a straight up 5F2a but I just had to explore my options. So I settled on a 5879 input stage straight out of the HoSo56, to a Vibro-Champ AA764 T/B tone stack, to a 12AX7 parallelled triode, to a 250K master volume to replace the 220K grid load resistor, to 6V6GT SE power section. I have a second switched cathode resistor in parallel to enable the use of a 6L6GC. I did not use the princeton NFB loop.

I'm pleased to report that this little amp REALLY sings! Very rich tone! Wow!
Everything is jammed so close together inside, & I had to use terminal strips since I could never fit a board in. It's surprisingly quiet for how rough it looks.
This build took one week. My first build took 4 months.

Some tweaks:
- 16mF-choke-16mF-10mF-10mF for power supply. So there's more filtering & capacitance on the rail. The extra filter cap feeds the 5879.
- The HoSo56 5879 stage lost too much gain going straight into a tone stack, so I beefed it up w/ 100K plate & 1k cathode resistors. Originally 56k/470R. The 1000pF shunt resistor to B+ was not necessary in this build.
- A 5751 sounds pretty nice in the 2nd stage. Still rolling tubes....Sylvania 12AX7A was very noisy. 12AX7EH sounded good too.
- My PT may have a bit too high of a voltage (330-0-330) for a proper Class A build; a 5Y3 rect. gave me 365V plate to cathode. The biggest 10W resistors I had laying around were 560R, with those I got 23V bias voltage, 44mA current (6V6). Too much. Sounded good though.
A 5V4 rect. dropped the plate voltage to 350, bias current to 38mA, still a bit high but closer to what I guess is safe. The 5V4 takes a loooooooooong time to power up, good since there is no standby switch.
- The parallelled cathode resistor is also 560R since according to the Weber calculator the 6L6 types generally need about double the 6V6GT current.

Some issues:
- The bright switch 47p across the volume terminals seems to destabilize the signal somehow, lots of nasty distortion. Not really an issue, I'l just take it out. There's plenty of treble without it.
- The heater voltage is a little over 7VAC. I think this may cause problems in the future unless I am mistaken.
- One little cheap .1 mF tone cap had me running in circles for hours. It really gimped the sound, killed the 2nd stage plate voltage (I had 56v there & could not figure out why). I found the bad cap after basically dismantling the preamp. But I'm better now.

Thanks to everyone on the forum, I probably used the ideas of a hundred different people, plus the Library, to get this done. I wish I was good at the schem. program so I could post one but that will take a little while.
And Doug's customer service is really top notch, he ships fast & makes sure you're happy.
 

 
   
 

[PRR]

> plate voltage to 350, bias current to 38mA

Probably 320V plate-cathode, 12 Watts, entirely safe. Leave it alone.

> The heater voltage is a little over 7VAC. I think this may cause problems in the future unless I am mistaken.

When you are old and grey and still using the same tubes, yes, maybe.

Typical guitar abuse will kill the plate, screen, or cathode long before heaters die at 6.3V. 11% over-volt won't make a real difference.

If you obsess (and have room), put two 0.33 ohm 2+ Watt resistors in series with the winding. About 1 Ampere in 0.66 ohms, 0.6V-0.7V drop, spot-on. Personally I would not bother.

[Willabe]

Hi firemedic,

All right. I tip my hat to you sir. Any pics???

I've heard it said that the heaters are rated at, +/- 10%, so  6.3 > 10% > .63 = 5.67 to 6.93 (?) I've heard some guys say no big deal if your a little over, others say not a good thing. You can use ohms law to figure out a small value resistor to put in series with the heaters, will be less than 1 ohm (0.xx), flamproof type. Use ohms law again to figure out watts rating.

Also 5V4 should not drop more voltage than 5Y3, somethings not right with those voltages.

Just a couple of thoughts.


         Brad      

PRR posted before I did. I did'nt see it, I'd listen to him.         


      

[jojokeo]

firemedic, you've really learned pretty fast. 

I'm wondering about a few things w/ the 5879 that you may or not have done w/ it. Did you use a bypass cap on the cathode resistor? Also what size screen resistor & cap? These will make gain differences on a pentode and you didn't mention them.

For the power tube did you use a grid leak bias resistor to go with the MV pot? Where did you put the 47p bright cap (on this control or in the preamp?).

[firemedic]

PRR- the 350V is plate-to-cathode. Next chance I get I'll pick up some higher-ohm 10W resistors across town. I use 10W to keep heat down, the bias bypass cap is pretty close to the resistor due to cramped space.
If I start blowing up tubes I'll install the series resistors in the heater circuit.

Willabe- thanks, and I got the 5V4 at a Hamfest for $4, maybe it's on the verge. I thought the same thing about the lower voltage, really. But it's so pretty!

jojokeo- I started with the HoSo56/56T 5879 stage. I changed the plate resistor to 100k, and the cathode resistor to 1k. The cathode bypass is 2.2uF. I did not change the screen resistor, 470k, or cap, 0.1uF.

I'm getting ready to tweak a little more. Separate cathode resistors for the parallel triode, maybe a 5.6k or 3.3k series resistor to the power tube (there's a lot of noise when MV is dimed) and perhaps change MV to 500k or 1M. As you can see I am not well versed in theory. Should I be using a grid leak bias resistor w/ MV? What is it?
Also, the schem. is almost done, will post this weekend.

[jojokeo]

Changing the screen resistor will change gain of the pentode as will the cathode bypass cap. The cathode resistor seems kinda low, you might like a 2.2k there. If you go too low, it's not the same as a 12a_7 and can actually give less performance and feel and maybe even like you're getting sag. It's worth a try to check it out, as is making the screen grid a 1M to see how that changes the gain.

The grid leak resistance I mentioned should really be called a grid-to-ground resistor & is important for a power tube's bias which is used to provide a DC ground reference for the grid circuit. Anyway, this resistor (or maybe pot w/ bright cap in your case?) should be constant & be no greater than 500k including a grid stopper for a 6V6/6L6 tube. Maybe it's best to wait for your schem before going further but this may be a source for some of your problems or instability?

[firemedic]

Well the bright cap was across the 1st volume pot, but I took it out. This circuit sounds plenty bright already. But it's pretty stable.
I will put in the grid leak resistor, 470k.
I made the 5879 cathode 1k because the 56t had 470R w/ 56k plate resistor, I figured since I doubled the plate I should double the cathode. Again, weak on theory. I'll try a 2.2k.

[jojokeo]

10-4 I wasn't sure about a couple of things & thought that I'd read something about instability. You were correct in raising the screen along with the plate load r, unless you want to change the gain structure of the tube. In one of my first builds using an EF86 I mistakenly used a bias switch to go from a 3k3 to a 1k like you'd do for a 12ax7, the tube performed great w/ the 3k3 but not nearly as much w/ the lower 1k - w/ everything else being the same. I normally use a 2k7 now w/ a 220k plate and 1M screen for that tube almost w/out thinking about it, or atleast as my starting point whenever/however I use it. The 5879 is close but not quite the gain of the EF86.
I never asked why Geezer & Tubenit used the values they came up with intially for the 5879. Maybe it was because of older Ampegs and such that used them had those values? But then, back in the day most amps weren't maximizing gain either. Same for the EF86 and they had higher screen r's to keep them cleaner and not sure that they used bypass caps on all of the designs? This is like Leo using 100k r's for his 12a_7circuits back in the day and today many that want higher gain use 220k's or more.

[firemedic]

Okay, she's pretty much dialed in.
Grid leak resistor in, check.
I tried switching out the 1k cathode resistor with a 2.2, at the same time switching the 2.2 bypass cap w/ a 25uf. I know, too many variables.... but it sounded less lively & muddy so I switched back to 1k/2.2uf.
So far so good...
A 6V6GT seems to like my 5V4 rectifier w/ its lower voltage. A 6L6GC seems to like the higher voltage of my 5Y3. I don't understand why the 5V4 voltage is lower than the 5Y3. It doesn't look like a relabeled 5U4, it's an NOS Dumont 5V4GA, USA made. But, whatever, it's not arcing or anything, and power is steady.
The heater voltage has dropped a little, to 6.7VAC.
With a 6L6WXT+ I get a nice jangly sound I've only ever heard in bigger amps.
Jojokeo, what cathode bypass cap do you use with your 5879 circuit?
Stand by for schematic, I'm not on my home computer tonite.

[firemedic]

Okay, semi-final version, schematic posted below. Suggestions welcome, I may lower the 22k resistor on B+ to beef up preamp voltage some. With a 5751 parallel triode, Sovtek 6L6WXT+, and 5Y3, this is the best amp I've ever played thru. The harmonics are incredibly rich without getting muddy. Somehow I got lucky on this one.
I'd like to do a line out from the power section. It does have the 22k NFB loop (a la Princeton) installed, but it sounds better without it so I left it disconnected. Any ideas as to how to make that a line out instead, are welcome.

 Thanks again everyone for enabling me in this weird tinkering addiction I've developed.

[jojokeo]

>>what cathode bypass cap do you use with your 5879 circuit?
2.2uF like what you're using already

>>Any ideas as to how to make that a line out instead?
Line out w/ a vol control is recommended. As you turn the amp up or down, the signal strength will also do the same, w/ level control you make your line out independant of amp volume.

*I paralleled your triode by connecting the corresonding pins. If they are not actually connected together (which it appears by your schem) you can do so and use the corrected values in the drawing for the same operation and frequency response while lowering your part count.

[firemedic]

I can't open the file.

[firemedic]

Never mind, I opened it.
The parallel cathodes are separated, but identical. Grids & plates are connected.
On the line out issue I was thinking more of impedance matching with my little 4-track.
The line out on my 1st build works fine when connected to a slave amp, but sounded thin when recorded. I was recording at very low volume, though (in the garage with a sleeping toddler in the next room), which may be responsible for the thinning.
The matchless Hurricane has a 22k from the 8ohm tap, w/ 3.3k to ground. Maybe a that's a "catchall" type impedance which would work when slaving amps OR feeding a mixer....
I just don't have the room on this thing for another pot. Or I'm too lazy to drill another hole? My wife just rolls her eyes when I say "The amp's done", she's heard that so many times.

[jojokeo]

You can make it fixed values if you want to. Notice that the ratio for each set of values from mine and the other one are the same @ 6.6:1. This is the general divider amount + or - a tad. But, the gain on the recorder will have to be turned up more if you're playing quiet or turned down w/ the amp up. You may have to adjust ever so slight the ratio for you situation?

[mcrracer]

>>what cathode bypass cap do you use with your 5879 circuit?
2.2uF like what you're using already

>>Any ideas as to how to make that a line out instead?
Line out w/ a vol control is recommended. As you turn the amp up or down, the signal strength will also do the same, w/ level control you make your line out independant of amp volume.

*I paralleled your triode by connecting the corresonding pins. If they are not actually connected together (which it appears by your schem) you can do so and use the corrected values in the drawing for the same operation and frequency response while lowering your part count.

My machine will not open the schematic file . I could open the first one but not the second one. Can anyone convert it to another format or help me open it? Thanks

[firemedic]

Go to the schematics section & download the jschem software. Doesn't take long.
I went ahead & popped in the line out, works fine. That 6.6 ratio is what I was looking for, tho it should have been obvious....
Also tried replacing the 22k on the rail with a 5k, which brought the preamp voltages up nicely but I guess it allowed too much B+ interference for lack of a better word- didn't sound good.
So I'm done messing with it. 5V4/6V6 for a crunchy little blues box, 5Y3/6L6 for a nice clean tone. I may get froggy & try a GZ34. 
Not bad for a bald guy. Thanks again!

[mcrracer]

Still wont work. I give up. I guess I don't need to see it.

[VMS]

Open .sch file with ExpressSCH program

and .jsch file with JSchem program:

http://dhost.info/jschem/

[jojokeo]

Sorry, haven't seen the thread until now. Here's the file as a .jpg

[mcrracer]

Thank you

[firemedic]

I must need therapy or something. Why can't I stop messing with my amps?
I added a series 100k resistor to the signal just before the MV since there was some oscillation when I had the thing dimed. I made a couple other changes too; changed the 5879 plate resistor back to 51k, cathode back to 470r, like the HoSo56. Raised the preamp voltage a little. I'll post the new schem. tomorrow in case anyone wants to see it.
I'm gearing up to convert my 1st build, the Dan-O-Lux, to a TOS. It's time.
I'm also going to make a Pro Reverb, this year I guess, Normal channel a straight 5879 stage. That is one awesome tube.

[firemedic]

Here's the updated schem.

[Fresh_Start]

Just noticed one detail which might not mean anything.  Your "grid return" (or "grid leak") resistor for the power tube is an effective 323K [(1 meg || 470K) + 3.3K], but the classis Fender circuits all use 220K grid return resistors.  6v6 and 6L6 spec sheets both show a max value of 100K for fixed bias but 500K for cathode bias, so you're within max specs.  Still, a lower grid return may affect the gain and tone of the power tube. 

If you replaced the 470K with a 270K resistor, you'd be close to 220K effective.

I'm also curious about the arrangement of the master volume pot.  I can see how it works with the 100K resistor in series with the wiper.  However, the frequency shaping of the 0.02uf coupling cap becomes a function of where the master volume is set.  If you eliminated the (potentially) extra parts and had a 250K master volume pot wired in a typical fashion (IOW wiper going to power tube grid), the combination of the 0.02uf cap and the 250K pot would be like a BF Champ.  OTOH you could use a 0.01uf cap and a 500K pot to get the same frequency response.  Either way, the master volume pot would become a variable grid return resistor.  That's not necessarily a bad thing since a larger grid return resistor results in more gain from the power tube (at least I think that's what happens).

I'm sure you had good reasons for the master volume circuit design.  I just don't "get" them.

Hope this helps,

Chip

[firemedic]

Ummmm, I guess there's not anything to get. I started with a simple circuit & added stuff until it sounded right to me. I initially was trying for a variable grid return resistor & originally did have a 250k pot there, no 3.3k or 100k in series. It sounded fine.
But I HAD TO MESS WITH IT. I figured correctly that a 1M pot would give me even more gain. I liked the sound even better, real brassy. But it made some oscillation problems, so I fixed those w/ the series Rs. I do not know what I'm doing, I just fix things as they come up.
To be honest, the more I try to design things according to my peabrain understanding of the RDH, the more I just resort to trial & error. 5879SE, exhibit B. The DanoLux, my first build, is exhibit A.

[firemedic]

Okay, I switched it back to the original 250k pot. Now I'm really really done. Thanks FreshStart for the reality check, & jojokeo for your attention to this matter.

[Fresh_Start]

FM - how does it sound now? 

I assume that firemedic's master volume pot attenuates the volume, but how?  When there was a resistor between the preceding stage and the pot's wiper, I could see the voltage divider.  Now I don't see one.  Does it work because it's a variable load for the preceing stage?  IOW the portion of the volume pot below the wiper in parallel with the 100K load resistor for that stage?

Cheers,

Chip

[jojokeo]

FS - I think it's just drawn wrong?

[firemedic]

No, it's drawn right. Maybe designed wrong.....I started out with a SE 5f2 Princeton layout, which has no series resistance between the last PA stage & the power tube and a 220k grid leak resistor. I figured the 220k could be replaced with a 250k pot. I left the 3.3k grid resistor in because it isn't hurting anything being there.
It sounds really good. Everything they say about the 5879 is true. If I make any more mods I'll probably just replace the master volume or whatever it needs to be called, with a fatness pot, affecting the parallel triode on the 2nd stage. Probably soon since I keep the MV turned almost all the way up mostly anyway.