Ground on OTL

[jevta]

Hi,
I `m finishing at last night and first time start on OTL on speaker on one channel my OTL Ciuffoly ver 8.2. I lost many time for ver 6 and after more one year i interrupted and take ver 8.2. I spent one night for change connections and resistance. Result is fascinated with detals, clearly and presence of instruments but I have too much noise( comparrated with mosfet SONY ESP 590). My PS consist of torroidal trafo, per channel have 2 Spragua of 3500 mF and 4 Philips of 1500 mF for +-B and for input tube choke 1,5 H with 35 Ohma with block capacitors 4x25mF. I stricly do on scheme Ciuffoly. Now, I suspiction on little value block capacitors and maybe low filtration or ground. Tell me, please did you use one point for ground or seperate input of putput stage and what happend with virtual ground. If you have experince abaut that plese answer me.

All you need`s music

[fdegrove]

Hi,

Is the noise you hear hiss or hum?

Regarding ground, I'd recommend a single star ground.

Here's the circuit diagram of the Ciuffoli OTL V 8.2.

Cheers,

[tubetvr]

Hi Jevta,

As Frank write it depends on if you have 50Hz hum or 100Hz. If it is 50Hz it could depend on the grounding scheme used, the grounding method I always use in apower amplifier regardless if is OTL or transformer couples is as follows:

All points that should be grounded except output, B+ capacitor ground and rectifier bridge alternatively transformer center tap is connected to one point, usually I use the point where the first stage cathode resistor is connected. From this common point one wire goes to the chassis groundpoint which in my case usually is one input connector ground tab. -on rectifier bridge or centertap on transformer is connected to the - of the B+ capacitor, from the - of the capacitor one wire also goes to the common point mentioned above and one wire to the output ground connector. I have never used DC for heaters in a power amp but never had any serious hum problem but what I usually do is to connect a 100ohm wirewound pot across the heater terminals and connect the slider of the pot to either the common ground point or some DC potential if that is appropiate for instance if I use a SRPP. The pot is adjusted for min hum level.

This is the method I have used in my OTL and I can not hear any 50Hz hum at all even if I put my ear close to the speaker, (I use Lowther horns at around 100dB sensitivity). However I can hear some 100Hz hum but only if I put my ear very close, a couple of centimeters from the speaker, I estimate that the hum level is below -80dB or even less. What I have found is important is to use high quality electrolytic capacitors with very low ESR, this will have a big impact on the noise level, currentlly I use 3300uF +2200uF Panasonic elyts especially made for OTL's, (at least that is what is says on the capacitors), earlier I used surplus capacitors of similar values that gave much worse result.

BTW using a choke for the driver tube power supply is not at all necessary and a waste of money, I use 100uF + resistor + 100uF for driver tubes and additional resistor + 220uF for the input tube, this gives absolutely no hum problem.

I hope any of this make sense and is of some help.

Hans

[arg_]

Hi Jevta,

I am glad you finally got it to work even if you had to change version. I totally agree with your comments regarding the amp's performance as I am also delighted with the result.
Now for the noise issue, I am sure it can be resolved as in my amplifier which is the same version I have only about 3mV AC at the output.
Frank and Hans' suggestions above are very considered and I too recommend that you first find out whether noise is 50 or 100Hz.

If I may add a few comments based on my experience on the same circuit:
I use a star grounding scheme where all grounds from drivers and output are returned to a single point between the output ps capacitors (the floating ground) and from there to the chassis. I suggest you try this first, after all it costs nothing
For PS filtering I think the capacitance is sufficient however I totally agree with Hans that capacitors quality play a key role and in my case when I changed from cheapo's to Panasonic TSUP series (hi ripple, low ESR) there was a dramatic improvement in noise, from 16 to 3mV! Please note that I use just 2x3300uF per channel.
For the input stage filtering my experience is somewhat different from Hans' suggestion, at least in this particular circuit. I initially tried to get away with CRC filtering but I had considerable ripple and I finally resorted to CLC (with 5H choke) and all was fine.

I am not sure if you have built the version depicted above or the output capacitorless, which I have built, but in the latter case have you measured the output DC offset? I have about 90mV in both channels and due to floating ground if I adjust the bias to minimize offset in one channel, the other one is increased. I guess it can only go down with more closely matched 6c33's but even at this level it is not a problem at all

Congrats again and keep us posted

[tubetvr]

Hi arg and others,

Just one more thing to consider, the 100Hz ripple on the output depends also on the grounding scheme, with totally 2x5500uF per channel I have 0.4mV Pk-Pk, but using other grounding schemes then what I described I had much more ripple using exactly the same components.

Regarding use of chokes or resistors for driver B+ it really depends on how much voltage you can afford to drop over a series resistor. Please play with PSUD and you can see that it is quite easy to get lower ripple then with a CLC filter.

Hans

[jevta]

Thank you for your suggest and advice. I appreciate your time and wish for help. It`s to important for me.
Arg, I have identical voltage on measure points which you sent me. I experimented with connections E182CC and resistor for negative reaction and decided put 23 KOhma. I made DC heater for input first stage E83CC with stabilized 12V 7812. Output tube for heating I connect 2x100Ohm fix and middle point conne4ct on central ground point !?But, ground formed on star with central point assume for hround of choke 1,5 H for PS input stage and all wires simply put in this point. Now, I see my Mistake ( sincerly, this is my first hand work althought age 43 ). Ripple have 4,5 mV on output stage and it`s constantly very strong noise of 50 Hz on distance of 2 m ( excuse me, 100 Hz thinking after diode bridge Hz ?). I `ll apply your suggest and hope that be OK.
Thank you, all

[fdegrove]

Hi,

Hmmmm...I just noticed that the schematic I posted doesn't correspond to the one I worked on while helping Arg.

Much to my surprise they carry both the same version numbers but use different output stages:

This is by far the better one IMHO and is a reworked version (by yours truly):

Cheers,

[arg_]

No problem Jevta, I've been through this before and have an idea of what's it like
Since your ps capacitance is certainly adequate, I am sure you'll be able to bring hum bellow 2mV by playing around with the grounding scheme. Moreover you are already using dc in drivers filaments which can help even further (I've not tried it)

Frank, nice rework! I've done exactly the same to make it more readable and to raise the morale; I cant stand greys
BTW I have already provided Jevta with the same schematic with check voltages at various nodes; it will be useful for debugging should anyone in the future decides to work on it

Cheers all

[fdegrove]

Hi,

Output tube for heating I connect 2x100Ohm fix and middle point conne4ct on central ground point !

Yes, that's the idea. It's probably not necessary though.

Have you tightly twisted all AC heater wires together and pushed them against the top of the chassis well away from any signal carrying wires?

I experimented with connections E182CC and resistor for negative reaction and decided put 23 KOhma.

You mean you're using a 23K resistor in the NFB loop, right?

BTW I have already provided Jevta with the same schematic with check voltages at various nodes; it will be useful for debugging should anyone in the future decides to work on it

Good on you.
Do you mind sending me a copy or you could even post it here for future reference?

I think the version of the circuit as it is now is by far the best of the Ciuffoli OTL.
Personally I'd replace the 12AU7A with a 12SN7 and the E182CC with a ECC99 unless there's an octal version that's close enough to it.

Cheers,

[jevta]

Frank,

Have you tightly twisted all AC heater wires together and pushed them against the top of the chassis well away from any signal carrying wires?

I try but I build on one chassis with 5 mm Aluminium.I made seperate transformator only for heating and use Zn-Al protect tube for lead tone signal wire.

You mean you're using a 23K resistor in the NFB loop, right?

Yes.
pardon what mean BTW?
Regard