Ideas for amp rebuild...

[hifiZen]

OK, here's the schematic, with all warts in plain view. The amp has sat idle for most of the past three years, due to reliability issues and a lack of time to commit to a re-build. Lately, I've been powering it up with a variac in order to enjoy it's sound without having to worry as much about it's health.

Suffice it to say, I really do not like the power supply much (even though I designed it all ). The zener ladders have proven quite unreliable and frustrating - they've got to go in favour of something which doesn't periodically self-destruct. You can see my humble attempt at a slow-start circuit for the B+... an RC time constant which slowly raises the gate voltage of the main mosfets until the zener ladder takes over. A major problem with this turned out to be the protection of the mosfet gates from overvoltage... I ended up bodging it with a large series resistor and some zeners. Yuk. This too must go. Finally, I had for some unknown reason, neglected to add a balance control for the output bias.

Still, for all it's flaws, it sounds very very nice when the power supply isn't acting up.

 

[Eli Duttman]

Well, by the sound of it, to avoid 'blocking', the 12AX7's going to have to go. I'm guessing a 6CG7, 12BH7 or 12AT7 (or even a 6SN7 !?) sounds like a better alternative that I could wring more current from. Though if I convert the first stage pentode to triode-wired mode as well, I may loose too much gain overall.

Majestic,

JJ's current production ECC99 is a good sounding tube with excellent drive characteristics; think 5687, but has a mu in the low 20s. The 12AV7 family, including the 5965 that HFZ uses, are also good; think 12AT7, but more linear and mu in the 40s. A 3rd possibility for the splitter is the Russian 6n1p; it has good drive and a mu in the 30s.

If you CCS load a triode, you get the "full" mu. In addition to giving the gain you need/want, CCS triodes also are VERY linear. That linearity allows the NFB loop to be around only the splitter and the "finals". Shorter NFB loops tend to work better.

 

[Majestic]

Majestic:
By all means, feel free to bring your amp into the discussion...

Hey, thanks Bro!

OK, here's the schematic

Whoa! That's some PSU all right

Not sure I'll be able to offer much in the way of helpful suggestions beyond what I have over the page, but I'm sure I'll learn a bit from looking it over.

JJ's current production ECC99 is a good sounding tube with excellent drive characteristics; ... The 12AV7 family, including the 5965 that HFZ uses, are also good; think 12AT7, but more linear and mu in the 40s. A 3rd possibility for the splitter is the Russian 6n1p; it has good drive and a mu in the 30s.

If you CCS load a triode, you get the "full" mu. In addition to giving the gain you need/want, CCS triodes also are VERY linear. That linearity allows the NFB loop to be around only the splitter and the "finals". Shorter NFB loops tend to work better.

I was just looking up all about the ECC99 before hitting this thread, so yeah, a definate contender. Not much experience with it myself yet, nor the 12AV7, or the 6N1P, though I've heard good things about them all.

My only "gripe" with the ECC99 (if it is one) is that the opportunities for adjusting the tonality of my resulting system via tube-rolling; decrease to nearly zero, since the 6BR7 is likewise pretty well only made by one company (Brimar). I'm not ruling out still using it, though.

And since I have a redundant electrolytic capacitor poking up through my chassis, removing it gives me an extra socket space for a third noval (or even octal!) tube; so a CCS could definately be on the cards.


Ok, here is a copy of the schematic of my own amps:

Amp | PSU (approx 300k each)

I hope it's readable (it's my first schematic posting here.)


Things to note:

• The circuit is 90% of a Mullard 5-20 so probobly not much new for the initiated; still, I have to start somewhere
• I've labeled my own companents by the original Mullard designations (ie R1, R2, ... etc.) for ease of discussion
• When thinking about what I wanted to do when I retire, I decided that building tube amps for my friends sounded fun. So I came up with a company name which appears on the schematic mastehad. The company is about 20 years from actually producing anything, since I first have to teach myself what I'm doing! (It also bears no relation to the originator of this thread!)
• Becuase I'm teaching myself, I probobly draw schematics funny to everyone else. On the other hand, the major circuit functions are usually clear to me (which is why I do them this way)
• The major differences between mine and a genuine Mullard 5-20 are: the values of input; R1, R2, C1, R5, & the presence of "R0"; grid stoppers: R16 & R19; feedback: C8 & R15; and the absence of C4 in the PSU. Naturally there is the low(er) noise 6BR7 pentode instead of the more usual EF86. Which could explain some of the above altered values.
• I wasn't 100% clear about the exact capacitance value of the 6BR7's anode compensation capacitors (C1) or the feeback capacitor (C8) because the polyester's were just labeled "22J" which, to me, means "22pF +/- 5%". Yet "220pF" seems more in the ball park of the value called for byt the circuit. So... <shrugs>

So clearly the previous (original?) owner had altered some things before me. ...Whether they had a good reason for doing so is something I guess we'll work out.

• The original Mullard circuit specified a value of 8.2k for the feedback resistor R15. This apparently provided for 26dB of feedback for a 15ohm load. Now that this current circuit has a 10k resistor in it's place, so how much feedback is that overall now?
• I'm a little unclear, believe it or not, about what the current output trnasformer taps are set for. I believe it's 7.5ohm. I'm hoping it's not 15ohm!

ok, I think I've written too much already, so I'll leave it here and come back with my proposed modifications list later.

 

[Eli Duttman]

Majestic,

Do some research into Gary Pimm's CCS designs. A CCS loaded triode voltage gain block will give you plenty of "room" to roll tubes.

 

[hifiZen]

Majestic:

I can't seem to download your schematics... is the link broken for other people too, or is this just a problem with my ISP?

Anyway, you were asking about open-loop gain of the amp with a triode input and a medium-mu phase splitter... I just finished running SPICE simulations of my amp circuit, pretty much exactly as shown in my posted schematic, but with ideal power supplies instead of the regulator arrangement. Open-loop gain measures 38.5dB, and the low and high feedback settings apply 9dB and 12.5dB feedback, respectively. So closed loop gain in my preferred low-feedback setting is about 29.5dB, or a voltage gain of about 30.

The SPICE results correlate very well with my real-life measurements, though I'm not too sure of my OPT model... I had a hard time measuring the secondary inductance and the shunt capacitances are just guesswork. Still, the midband gain figures will be accurate regardless of what's happening in the frequency extremes. As I mentioned before, I find the gain is a little high, even with my unity-gain preamp. A CCS-loaded triode with mu around 40 would be pretty much a direct drop-in replacement for the 12AT7 in my circuit. Or, a lower-mu triode could be used in combination with a different feedback network in order to reduce the amp gain while preserving the amount of feedback applied.

 

[fdegrove]

Hi,

or is this just a problem with my ISP?

Other than AMP and PSU pointing both to the PSU I don't have any problem.

Cheers,

 

[Majestic]

oops. ($*#%! stoopid late-night postings!)

hey, we all know that an amplifier is just a modulator for the power supply, so I decided that the purest sounding circuit must just be the PSU alone....

umm, seriously, changing to lower case "...amp..." should do the trick. apologies for that.

http://members.optusnet.com.au/zenaudio/A&R_4001_amp.jpg

Sorry if my ISP is playing up for you, but it seems to be working fine for me. Let me know if you're stilll having troubles.


(Oh, and before I forget; for anyone who's interested:

• An externally hosted image should be here but it was not working when we last tested it.
  6BR7 DataSheet
  from Brimar tube manual #10. PDF hosted by Keith Snook.
  (note: Adobe Acrobat Reader required)

 

[planet10]

Ok, here is a copy of the schematic of my own amps:

Amp | PSU (approx 300k each)

Saving those as gifs instead of jpg will keep them cleaner and a lot smaller (63 & 53 k respectively for the ones i just converted)

dave

PS i fixed the link

 

[hifiZen]

Hrm. Still not working for me. Perhaps you could post them as an attachment to the thread, or just email them to me? My email starts with my diyaudio username at yahoo.com.

 

[hifiZen]

Ok - update... so yesterday I bought TCJ's P-P calculator and fiddled around a bit with the EL34 and 7027 (in lieu of the 'super' 6BG6). I'm a little disappointed that the software doesn't do pentodes or UL, but the triode mode curves for the 7027 definitely look nicer than the EL34 curves. I'm pretty happy so far with what I see.

I've decided to do the rebuild in stages, starting with a new power supply and the 6BG6 output tubes, but leaving the input and phase splitter as is. Then, I'll start incorporating 'upgrades' one by one... CCS-loaded input tube (maybe 5687, ECC99 or 6SN7/7N7), then convert the phase splitter to Schmidt w/ CCS from neg. rail and bring the feedback to this point instead of the input. Then maybe go for the DC filament supply. This way, I can hear the change each mod makes instead of doing it all at once.

I'm debating over the possibility of adding another dual triode to handle cathode-follower duty for AB2 output. It could be a pain to add this later if the hole isn't already in the chassis. Any advice on AB2? It's completely new territory for me, but I'm intrigued by the prospect of getting the extra power...

 

[hifiZen]

Majestic:

This article may be of interest to you. It's a reasonably thorough breakdown of the Mullard type circuit, including lots of useful technical information ranging from quite basic to moderately advanced. The article explains why input sensitivity was so high, and suggests triode-strapping the EF86 pentode as an improvement. Among other things, there is also a graph showing that for a pair of EL34s in ultralinear, a 6.6k p-p OPT is just about perfect in terms of output power and distortion.