45 Watts per channel push pull 801As

stephie's place

Moderator: sbench

Post Reply
sbench
Posts: 296
Joined: Fri Jan 28, 2005 3:45 pm

45 Watts per channel push pull 801As

Post by sbench »

There's no doubt about it, at 16 tubes, this is a complex amplifier. The good news: it sounds great, thanks in part to Dave's driver iron!

Specs:
.111V produces 1 watt (very sensitive).
Residual hum ~0.6 to 0.7mV, even with AC on all the filaments/heaters.
Frequency response -0.3dB at 20Hz to -0.8dB at 20k
Distortion at 45 watts per channel: about 3%
Distortion at 1 watt: less than 0.1% (and no feedback).
Class A - A2 - AB2. It is almost impossible to tell when it transitions between A and A2 and A2 and AB2.
Unusual bias regulator with harmonic equalizer increases power output and lowers odd order distortion. This thing sounds more like SE.

Attached are pictures, distortion curves and schematics.
Attachments
801he.JPG
801he.JPG (92.79 KiB) Viewed 27369 times
801pphe.gif
801pphe.gif (19.42 KiB) Viewed 27341 times
he801_cm.pdf
(17.1 KiB) Downloaded 1332 times
he801_ps.pdf
(23.69 KiB) Downloaded 1348 times
he801_am.pdf
(25.97 KiB) Downloaded 1967 times
dave slagle
Posts: 2085
Joined: Sat Jan 22, 2005 3:54 am
Location: NYC
Contact:

Post by dave slagle »

Hey steve,

Awesome... How about a bit more background on the whole "concept". I know the history, but others may not.

I'd offer up some output iron but the class B aspect changes what i want to explore with PP outputs since class B allegedly requires coupling between the halves. I know JJ has a pair of "test" OT's that eventually need to be replaced, but i think they are closer to 15K. (jj you remember?)

If you are game to give "decoupled" class AB2 a listen (using the same logic as the IT's) i'll get a pair of OT's to you to play with.

hell, i'd just like to discuss what would happen in theory for the various
sectional couplings for the various classes of operation, but i'm weird that way.

dave
dave slagle
Posts: 2085
Joined: Sat Jan 22, 2005 3:54 am
Location: NYC
Contact:

Post by dave slagle »

by the way, am i missing something when i see you are getting nearly 100% efficiency out of those 801's at 3% distortion?

the distortion from 0-45W seems beautiful, but what happens from 45 to 60W when you get into the perpetual motion machine? (is greater than 100% efficiency possible???)

BTW, i'm sure you saw that the ETF this year has a thoriated theme :-) I suspect it will not be too hard to convince someone (mahatma?) to safely transport.

dave
sbench
Posts: 296
Joined: Fri Jan 28, 2005 3:45 pm

Post by sbench »

Hi Dave and all,

On the power. That's a good point. What happens is that as the power output increases, the current increases so that the power input increases too. So that at 60 watts output, for instance, the power input is about 100 watts, and each tube is dissipating about 20 watts still.

On the "decoupled AB2". Sure thing! Actually, the class of service can be controlled by the impedance. With 8600 ohms plate to plate, the system goes into AB. At 14-15k plate to plate, the system stays in class A except for the highest power peaks.

On the overall concept. As everyone knows, push pull operation cancels even order distortion components through symmetry. Thus, you can push the tubes harder and achieve more power before distortion becomes objectionable. Well, those clever Western Electric folks came up with a "Harmonic Equalizer" concept in the WE86 amp. This exploits the fact that even order distortion is cancelled in PP to come up with a way to cancel the odd order products too. What they did was to take some of the even order products (developed at the common cathodes of the output stage) and feed them into the bias. This is not feedback: the injected frequency is the harmonic energy, not a copy of the desired signal. This allows these components to be mixed through intermodulation action producing odd order distortion components that just happen to be phased properly to cancel the odd order distortion produced. This lowers overall distortion even more. Note that this is a wideband effect. Distortion is lowered over a wide range.

What I did was to expand on this concept a little. The problem with the WE Harmonic Equalizer was that it essentially had to be AC coupled back into the compensated stage, so that the distortion products ended up causing small bias shifts as the signal dynamically changed in level. You could hear this effect. What I did was create bias regulators (the 12AX7s in the circuit) that provided the mixing action after the signal coupling, so that the bias shift effect went away. And, the signal added on top of the bias also happens to push the tube harder just when it needs a harder push, so that the available power increases. That's why I can get more power from a set of 801s at 500 volts than the "published" data shows even for class B at 600 volts.

Actually, I played with this concept in Spice "modeling space" for quite a while to see what the effect is likely to be for different conditions. Interestingly enough, the actual measured results follow the spice model VERY closely. (By the way, that's why I altered the 801 models a while back; I wanted an accurate model INcluding grid current.)

The 7 segment displays showing on the front panel are monitoring the 801 cathode currents, which are adjusted by the 4 knobs on top of the chassis. They are on little PC boards, controlled by a PIC processor. That's the he801_cm.pdf file. If it wasn't obvious, the amplifier schematic is the he801_am.pdf and the power supply is the he801_ps.pdf.

-Steve
dave slagle
Posts: 2085
Joined: Sat Jan 22, 2005 3:54 am
Location: NYC
Contact:

Post by dave slagle »

Very cool all around.

did you also use those pic controllers for regulation of the B+ too?

I'll start sketching out an output. What footprint do you have to work with and what are your ideal specs? (impedance and power).

finally, i'm guessing that the source impedance (801 Rp) varies quite a bit, do you have a guess at the range?

dave
DowdyLama
Posts: 104
Joined: Thu Jan 27, 2005 5:42 pm
Location: Blue Ridge Mtns, NC
Contact:

Post by DowdyLama »

Yes - uber-cool stuff!
Dave Cigna
Posts: 96
Joined: Sun Jan 23, 2005 10:58 pm

Post by Dave Cigna »

I agree, amazing as always; nothing boring from you!

I have a question that is a little boring; nothing to do with the HE function: did you have trouble balancing the direct coupled PP stages? It seems that even a slight mismatch in the input pair would result in a relatively gross imbalance in the second stage.

Direct coupled SE stages don't scare me too much, but whenever I've thought about doing it in PP I'd always felt that some method of adjustment would be needed. Maybe a balance pot under the cathodes of the input LTP...

-- Dave
sbench
Posts: 296
Joined: Fri Jan 28, 2005 3:45 pm

Post by sbench »

A few things to comment on.

The A431s currently installed as the OPT are about 3.75"x4.25" with 3"x3.5" mounting centers. There's room beyond that if needed.

Output impedance: The amp output Z as built measures about 7.5 ohms, so that puts the plate resistance of the 801s at 4 to 5k.

On the balancing: Actually I took the "simple" route, since 5687s aren't terribly expensive. With only that portion of the circuit operating, I simply plunked in left and right input stage and measured the plate voltage. Ended up getting about 50% yield from my stash of 5687s. Ditto pre driver stage. The cathode follower A2 driver didn't need matched since it's part of the output tube bias and that has a pot adjustment. So I used two of the ones that didn't match in the other positions. The 12AX7s matched close enough in the first 4 I plunked in. (The 12AX7s I used were -er- old old stock Sylvanias from a friend who worked in Emporium Pa. I think I was in high school at the time).
During design, I had toyed with balancing the input stage. I did not want to add degeneration in the cathode since I am depending on the tail for single ended to differential conversion. I almost decided to balance it with a pot in the plate circuit. (simply changing to 27k plate resistors and a 5k pot would do it). For those who don't want to deal with balancing the tubes first, this would probably be the thing to do.

On the H.E.: I've been listening to this thing for a while now with and without the H.E. I absolutely and unequivocally prefer the H.E. activated. There are two "regimes" where it shines. At low levels, below perhaps a couple watts, it turns the amplifier into what sounds like a beautiful 801 single ended stage. It has that same delicacy and sweetness that an 801 gives. The second regime where it shines is pumping out LOTS of sound. This thing has a really invisible clip. There is no obvious harshness at any level I care to torture my speakers with. I have a cut Christian sent of "Rat in the Kitchen" that simply BEGS to be played loud. And the amp seems to take it. I think the only other amp I have ever built that comes close to matching that volume was a push pull parallel 6146 beast I built in college. (and do not have any more). I think I was pushing slightly over a volt into the amp with "Rat" which calculates to something over 80 watts per channel, which is definitely well into clip.
JoshK
Posts: 13
Joined: Fri Dec 02, 2005 4:37 pm
Location: NJ/NYC

Post by JoshK »

I'm trying to get my head around this circuit. I am not sure I see how you are accomplishing this
What they did was to take some of the even order products (developed at the common cathodes of the output stage) and feed them into the bias.
I just don't see where/how that is done. From my naive eyes, it looks like you are imbalancing the bias on the 5687 so that it predistorts the signal a little that then cancels in the output stage. Kinda like the way some like to use the 12AT7 as gain in a PP amp so that it creates the appropriate HD profile.

Do I have this all wrong? If not, then how do you determine the appropriate amount of pre-distortion to add? Is it based on measure the FFT of the total stage?

I think this is the coolest concept I've seen, but I am trying to understand how it works so I can potentially add it to an arsenal of things to try.
----
Josh
JoshK
Posts: 13
Joined: Fri Dec 02, 2005 4:37 pm
Location: NJ/NYC

Post by JoshK »

Maybe Steve doesn't come around here much. Does anyone know how this works and could help enlighten me? I wasn't questioning whether it works, just trying to understand. Its a very cool concept and worth exploring.
----
Josh
sbench
Posts: 296
Joined: Fri Jan 28, 2005 3:45 pm

Post by sbench »

You're right. Steve doesn't get around here much. He doesn't exist any more. On the other hand, you can never tell about Stephie. She might just occasionally show up.

The bias point is being altered by injecting a second harmonic signal (sort of like a "push-push" (sic) frequency doubler. This is phased so that the IM distortion present in the amplifier then produces an odd harmonic component (f + 2f = 3f) precisely aligned to remove the third harmonic also produced by the same amplifier. Thus, the odd order components are lowered or removed. As it turns out, the phasing is such that the "clip point" is slightly extended, raising the maximum available power output.

-Stephie
JeffreyJ
Posts: 320
Joined: Fri Jan 28, 2005 2:53 am
Location: stuck in the USA for now...
Contact:

Post by JeffreyJ »

YEAH!!! Welcome Stephie!!!

I am looking forward to more tube madness...

woo hoo!!!!

Peace,
Jeffrey
JoshK
Posts: 13
Joined: Fri Dec 02, 2005 4:37 pm
Location: NJ/NYC

Post by JoshK »

Welcome Stephie and thanks for clarifying. I still have to chew on this some more but I think I am closer to understanding.
----
Josh
dave slagle
Posts: 2085
Joined: Sat Jan 22, 2005 3:54 am
Location: NYC
Contact:

Post by dave slagle »

Hey Stephie,

I have a customer asking for OT iron for this circuit and I see you used 4300R P-P with an 8R load on the 4R taps for an effective P-P of 8400R.

The request is for some 12K P-P OT's and I just wanted to check with you to see if this changes anything in the design.

Also how much of the output power relies on Class AB?

thanks

dave
Get Your Fix
www.hifiheroin.com
sbench
Posts: 296
Joined: Fri Jan 28, 2005 3:45 pm

Post by sbench »

Hey Dave and all,

Got busy, so haven't been on the forum for a bit. This girl has several things happening....

On the higher impedance... The amp will stay in Class A / A2 over a longer range, and the power output will be just a tad lower, but will work well. If you look at the distortion plot way at the beginning of this thread, you'll see where the distortion curves kink somewhat. That's the transitions between A1 A2 and AB. IIRC, at the higher impedances, you'll get 30 or more watts before it hits AB.

Warmest Regards,
Stephie <3
sbench
Posts: 296
Joined: Fri Jan 28, 2005 3:45 pm

Post by sbench »

Just redid some of the sims with higher impedance OPT. For 80+80H to 200mH (12.8k to 8 ohm) the circuit remains essentially A1 or A2 never hitting class B.

The clip point is slightly over 25 watts. At 25 watts the distortion is 3%, at 30 watts (into clip) it is 6.5%.

Transition between A1 and A2 is almost 4 watts. At 1 watt, distortion is .1%

The harmonic equalizer values I used are 300k and 27k for the voltage divider at the 12AX7's.

Stephie <3
Post Reply