idea for hybrid tube amp?

Oct 9, 2005 at 9:37 PM Thread Starter Post #1 of 14

nikongod

DIY-ku
Joined
Jan 24, 2005
Posts
8,882
Likes
133
hello head fi.
the idea for this came to me sometime between buying my millett hybrid boards, and buying 2 very expensive electrolytic output caps.

i am curious if you think that it will/will not work. the "tube stage" is shown as a simple triode, in reality it will probably be a 5697 preamp the schematic for which i found on the web.



the painfully obvious "possible failure point" is the one of ground isolation. i think that the cap will block the high voltage dc in the tube stage from the ss buffer, and i plan to use 2X9V (in a grado ra1 configuration, no splitter device...) this of course avoids the problem of some funky power-line-ground loop. after that is avoided, are there any other glaring issues?
 
Oct 10, 2005 at 8:57 PM Post #2 of 14
Are you familiar with the YAHA amp? It's a similar tube/opamp hybrid but without the dual rail power supply you propose here.

The original YAHA design is described here:

http://www.fa-schmidt.de/YAHA/index.htm

I recently completed construction of a battery powered YAHA using 1J24B pentodes, as in my ugly adaptation of the original schematic:

http://rubix.areb.org/yaha-pentode.jpg

I'm currently using a single NiMH AA cell for the heater and two 9v alkalines for B+. My next step is to see if i can implement a simple "cockroft ladder" voltage multiplier about the shape of an AA battery holder, which would allow me to supply a higher B+ for better tube performance, and allow me to fit the whole deal into an altoids tin. After diode losses, 4x of 9v should still be within the maximum voltage of the opamp.

You're not the first person to suggest using a split rail for the output buffer, but to the best of my knowledge nobody has actually done this yet.

Really, nearly any opamp can run on a single voltage, and then the DC offset can be removed by an output cap.

But theoretically you can use a much smaller cap value between the tube stage and the output buffer, and thus theoretically it doesn't have to be an electrolytic. Try it out and tell us how it works :-)

I'd be interested in seeing that ck5697 preamp circuit.

Edit: Here is a picture of my liliputian tube hybrid, power supply not shown.

http://rubix.areb.org/gallery/20050917/img_0470
 
Oct 10, 2005 at 11:02 PM Post #3 of 14
yes indeed, the primary objective here is to eliminate electrolytics from the signal path. a secondary objective is to build a new tube amp for my k-1000's

oops, typing >me, as allways.
the tubes are 5687's my bad.

pat5687.gif


i think i will try this idea out on my millet (likely with lots of proto-boarding...) and then if it works, go full voltage on it
wink.gif
i think that MANY of the paralled caps can be eliminated, if the unit is to drive a dedicated buffer.
 
Apr 6, 2006 at 7:56 AM Post #6 of 14
WOOT

i tried the above in a millet hybrid.

all i have to say is DO IT!!! its awsome.

with a little "creative planning" a voltage divider can be added, to dump some gain, and make your millet even more grado-friendly.
 
Apr 28, 2006 at 4:29 AM Post #7 of 14
first off building this is dumb, taking advice from people on the internet that you dont know is dumb. you should probably not even build this what i describe below. but it sounds realllly good.


note: in the "schematic" above i say "omitted" regarding the caps spanning ground to v+, v-. i did use them, i jsut didnt draw them in the schematic.

nowthen:
the above is a drop-in replacement for the buffers on a millet hybrid amp, that uses no output cap. below i describe also how to use it without an "intermediate" coupling cap.

i built it on a piece of perfboard with 2 parallel strips on it (available at rat-shack, as described in tangents cmoy article) 2 dip-8 sockets can be hot-glued on, and the whole shebang will stand proudly, and drop in...

nowthen, assembely is straightforward. i kept the 2 voltages sperate on the board (left v+, right v+, left v-, and right v- from the millet) i also used 2 tle-2426's as this can drive k-1000's, and why not get the extra seperation from not sharing a grond chanel? for the purposes of driving "standard" headphones, i simply tied the 2 tle-2426's together. in the place of the 2 100k resistors, diferent combinations can be used. i employed this "voltage divider" as the lowest gain for the "drop in tubes" is about 7.5. this is simply too high for most headphones, so i droped it in half with those resistos. alternately the one closer tot he cap can be removed. be sure to keep the "apparent resistance" high, as the tubes have to drive these resistors as a load. the 12xx6 tubes dont like to drive anything anyways 200-250k ohm is about as low as i dared to go.

inital startup should go something like this: build everything. put the tubes you want to use in. put your buffers in the sockets ON THE MILELT BOARD. power up, check bias as normal. i HIGHLY recomend simply using a "bias voltage" as close to half of your rail voltage as possible. once staible, remove buffers, and install this board. install buffers (in todays buf-634 short scociety, who can spare 2 or 4 buf 634's to build this, its mostly parts you already have otherwise sockets are highly reccomended...) DO NOT HOOK UP THE HEADPHONES YET!! it would suck to blow your 'phones now from one weird conection. once the whole thing is all fired up and running with the board in place, check your output voltages. durr, you should do that with everything.

after verifying that there is not excessive voltage offset, plug in and listen. betcha a dollar that if you used a decent cap (i used a pair of orange drops i had in the parts bin) or better you will be quite pleased.

nowthen, in another thread (which i will link to here in it) it was asked how to use NO cap in the signal path.

the purpose of the cap is NOT to get rid of dc, its to get rid of DC OFFSET. BIG diference.
simply: you dont get rid of the plate voltage, you make it not have any offset compared to yoru new virtual ground.
how to do that is as follows:
but first, i will repeet: i moded what i built above to be what i describe below, and unmoded it. sound improved, but i didnt think it was worth the risk. if you think its worth the risk, and want to try it, that is yoru choice.

what you need to do is set the plate voltage to EXACTLY half (where a tle-2426 railsplitter sets to) of the rail voltage. its a very delicagte opperation, and sensative to how warm your tubes are. after doing this for a few days, i decided that i was too afraid to listen adequately: aka it wasnt worth it, and put the caps back in. what actually caused this was a bit of tube distortion/a driver crackling on a recording that i had not previously heard (the distortion) i was listening on my k-1000's and i heard this. my first gut wrenching thought was "crap, i toasted a driver" it was very real, and right there.... i realised how close to "bad stuff" i got, and took a step backwards.

build the above, and bypass the caps with either a bit of wire or a resistor.

follow general install instructions as above.
after that, my general method was to turn the amp on for a few miniutes (no headphones on it) hook my dmm upto the outputs of the amp, and then turn the bias pots untill i got a couple mv of offset. then do the other side, and repeat. it is VERY hard to get a good low offset on the output with the tinny pots that the millet uses as standard parts.

some form of active "bias driver" based on output dc offset would be nice, but i lack the skills to design one, or the motivation to steal a design. the help of others is always welcome here
smily_headphones1.gif
i have a couple ideas for building a "self biasing" circuit for the millet, but eeh.

the amb "dc and delay device" would probably add some piece of mind for this mod
smily_headphones1.gif
another bit of gear i lack the motivation to buy and build... it took me this long to get this from the paper napkin in the first post of this thread to a bit of perfboard...

if you are afraid to be the second person to build this, dont build it.
 
Apr 28, 2006 at 4:40 AM Post #8 of 14
ooh, i forgott to say, if you do build the above, the output must be kept off of the millet board.

run wires off of the daughterboard to the output jack. be sure that the outptu jack is isolated from v- (millet board ground) as failure to do so will result in a little smoke, possibly fire, and a bad tle-2426.
 
Apr 28, 2006 at 5:34 AM Post #9 of 14
Just curious, but has anyone thought of using a MOSFET driver (a la Szekeres) instead of an opamp in the YAHA design?

Aditya
 
Apr 28, 2006 at 6:22 AM Post #10 of 14
Quote:

Originally Posted by a1rocketpilot
Just curious, but has anyone thought of using a MOSFET driver (a la Szekeres) instead of an opamp in the YAHA design?

Aditya




I did

but not YAHA, SOHA instead

then I put in dual voltage and also DC servo, as such I dont need any output caps. only some small 1uF input cap to the szekeres part
however, for 12AU7, apparently I cant put the mosfet directly from the tubes. I still need the op-amp as intermediate buffer.

I'm now using SOHA as pre-amp, the szekeres as a power amp, able to drive my pair of bookshelves.
very very negligible DC offset. the DC offset servo schematic can be read from headwize's szekeres article's addendum
 
Apr 28, 2006 at 6:31 AM Post #11 of 14
Quote:

Originally Posted by a1rocketpilot
Just curious, but has anyone thought of using a MOSFET driver (a la Szekeres) instead of an opamp in the YAHA design?

Aditya



not a YAHA, but sijosae did just that in a MHHA (Multi Hybrid Head Amp). i'll see if i could dig a schematics.
 
Apr 28, 2006 at 6:37 AM Post #12 of 14
Apr 28, 2006 at 7:25 AM Post #13 of 14
Also, is there any possibility of tuberolling with the YAHA (something other than the 6922 that is)? If not directly compatible, is it possible to tweak the circuit to allow for other tubes?

Aditya
 
Apr 28, 2006 at 8:48 AM Post #14 of 14
Quote:

Originally Posted by a1rocketpilot
Also, is there any possibility of tuberolling with the YAHA (something other than the 6922 that is)? If not directly compatible, is it possible to tweak the circuit to allow for other tubes?

Aditya



i guess so. just need to tweak the heater current to supply the correct amount to the tube used. well maybe there's the Mu and gain of the tube to be considered. i reckon 12AU7/12AX7 could also be used in the YAHA.
 

Users who are viewing this thread

Back
Top