• WARNING: Tube/Valve amplifiers use potentially LETHAL HIGH VOLTAGES.
    Building, troubleshooting and testing of these amplifiers should only be
    performed by someone who is thoroughly familiar with
    the safety precautions around high voltages.

Local feedback between grid-cathode

Thanks wavebourn, I switched the OT to a CFB UL one, things should have worked out. I used to have almost 0 second harmonics, now it is very high. I don't know what is wrong, I think the phase is alright as it didn't went into oscillation with the feedback.. so this should be alright, but something is very wrong, THD should be below 0.1, now it is in the 4 % outch...

I am thinking, maybe the UL is incompatible with the CFB... this could make sense.
 
In the case I disconnect the UL, I need some voltage ratio ref. from the B+, I need another cap and a dropping resistor, no regulator, just a big cap for both g2s.. its the kt120, I have no idea which voltage to choose. I sent an email to see what could cause the high second harmonics, for sure something is not working at all there. As if only one tube was working or something like this.
 
If it is a temporary solution then ok but if definitive then regulation is worth.

Last UL + CFB I did was with the 6GK6 (10GK6 in reality) in parallel PP and it worked very well. This tube is basically an up-rated EL84 due to a slightly different pin-out. At 330V plate and g2 voltage I got about 35W very clean with just 10% CFB and no loop fbk.
 
I can connect UL and CFB, no issue there... could I have reverse the phase and still do not cause oscillations because of the CFB?

I will short the CFB windings, if it is the drivers in overload it is going to restore the gain and 2nd thd should go down as usual.

The overloaded drivers theory makes 0 logic there, I could drive with the same source level to no problem, it means something else is going on.

The worst case is that one side of the windings is gone kapuut or that one tube is not amplifying, it is only conducting...
 
Did this yesterday out of pure boredom and was shocked at how easy you can make a KT88 to resemble a 300B.
 

Attachments

  • curves.JPG
    curves.JPG
    331.9 KB · Views: 250
  • Curves_LTSpice.JPG
    Curves_LTSpice.JPG
    96.1 KB · Views: 277
The worst case is that one side of the windings is gone kapuut or that one tube is not amplifying, it is only conducting...

Well, are you getting that on both channels? Try one faulty channel alone and see. If it is the same then you can rotate the tubes to see if it's due to one of them. Then the other channel if all tubes combinations give the same.
If you don't see the problem with one channel driven then a 3rd option might also be the PSU.
 
UL & CFB in One Amp

This project was published in AudioXpress magazine in 2005. It is based on Norman Crowhursts Twin Coupled Amplifiers which were published in the late 50's. In a way it is a poor man's McIntosh. But instead of the plate to cathode coupling taking place in the one very expensive OPT, Crowhurst used two OPTs, the coupling was done by paralleling their secondarys. He also used caps to be sure the higher frequencies were adequately connected.

For this project I went a couple of steps farther, both CFB & UL are used. The power tubes are 6LU8s, the included hi mu triode used as CFs to drive the 6LU8s power section to the limit at low distortion. All of the parts are 'off the shelf', no boutique at all.

The B+ is supplied by a Zener controlled FET follower system with slow startup. The -ve supply is full wave right off the PT, set to the voltages required by Zeners. And a Current Limiting resister is on the line side of the supply.

The experimental chassis can be switched from Crowhursts input scheme to a full 2-stage differential. The full loop NFB can be switched on or off.

At clipping the power is 37 Watts. But could be easily scaled up.:cool:
 

Attachments

  • Power Supply Twin Coupled Amp.jpg
    Power Supply Twin Coupled Amp.jpg
    124.2 KB · Views: 257
  • Twincoupled Amp B.jpg
    Twincoupled Amp B.jpg
    183.1 KB · Views: 214
  • Twin Coupled Amp Front for RHS Veiwing.jpg
    Twin Coupled Amp Front for RHS Veiwing.jpg
    215.8 KB · Views: 218
This project was published in AudioXpress magazine in 2005.

Very cool. I built a Unity-Coupled amp with the Plitron transformers which are very expensive. I've always been fascinated by the Twin-Coupled design and was just thinking about it recently.

I was looking at Lundahl transformers which have four equal primary sections, the ends of which are all brought out to individual terminals. It seems like you could do something in between the Twin-Coupled and a true bifilar Unity-Coupled transformer with one of those. You could have the external capacitors to increase the capacitive coupling between the windings but have everything on one core and the transformer would be easy to source.
 
I disconnected the CFB, looks like everything went normal... very strange...

I am running the OT in UL now, CFB bypassed... see what this does.

Edit: I plug back KT88 and the band. is now from 10hz to 32.3khz at 10 watt so a little more extension as predicted by the excellent transformer design, however, at 10hz I experienced heavy harmonics, I will plug back the kt120 in UL mode only and compare the THD results.

I got better readings than with the hammond 1650R big time, with even the special driver tube removed and installed a basic tung sol there, the results are super good.

Thd are super clean, etc unbelievable, so, what could be wrong with the CFB?

I installed a 1.5 R current sense for bias at the end of the CFB, maybe it is generating 2n harmonics? it is very strange.

mmm big problem with music the bass is going insanely distorted, the high frequency gets very problematic and missing,, I THINK I CONNECTED THE TRANSFOR IN REVERSE LOL

weird, it should oscillate right away.... it sounds like it wants to go into hard oscillation but something stops it...

wow, just in UL this is a game changer, the sound is revealing all recordings imperfections, i hear everything, the bass is dry, the high pitch sound full power, everything holds together, if only it would work with the CFB wires connected.

the oscillation were only recording artefacts from cuts and voicing changes, it is disturbing...

No one resolved the enigma yet? (i did , a cookie for the person who finds it first)
 
ah, polarity was alright, the problem was elsewhere. it would have oscillated, I did with all the edcors because they have a + polarity, their windings are reversed hehe.

The problem: I had high second harmonics. Tubes were fine. I had the 13 db GNF connected, the CBF dedicated PP windings of 8 db and the UL on my KT120, it did the same problem with my kt88. It is simple ;)

fix bias was done also properly, balanced, (btw I biased to 65ma, it is measuring better now, thanks)
 
Last edited:
ah, polarity was alright, the problem was elsewhere. it would have oscillated, I did with all the edcors because they have a + polarity, their windings are reversed hehe.

Who can guess what went wrong then? :D
I once was pretty tired and made a sad mistake, swapped pins 4 and 5. It played, with pretty high intermods, however after control grids were already fried. :D
 
Think what could cause 2nd harmonics...

anything that is not working in tandem with the tubes, it could be an unbalanced driver,

it could be one tube going weak,

it could be one tube input disconnected

it could be not be a missing winding, or the tube wouldn't bias...

the PS was ok.

--- that is a bad mistake Wavebourn, I only did one destroying mistake, and it is a horrible one.... truly tragic like a divorce.

I had bought for a bargain almost NOS 6sn7 smoked, and they sounded nice, until i heard cracking sounds, I though the tubes had gone bad.

So, I threw 1 or 2 of them in the garbage... ok... those tubes bought 20 years ago are now selling for half a thousand now, this is not the sad part.

I bought the amp at a party and we had to cancel the amp because the cracking sound was terrifying, vibrations made the amp goes nut...

Back at home, I found the plate B+ wire was lousse, never been soldered since construction.

I now have the CFB and the UL, and it is not as good as the UL alone, I believe that the 6sn7 LTP is generating the higher THD now because the output has less gain, it doesn't explains it all...

The only advantage I had in the THD readings are with mix bass tones, the floor level was quite low at -50 db, i cannot recall seeing a tube amp with such a low noise floor when driven with 20 watts at 20 hz + 120 hz... so this is to keep watching.

Otherwise the THD figures with CFB + UL are worse than the hammond by a fair margin, it still sound a lot better than the hammond... which is another proof that measurements are not relevant of sound quality.

The second advantage in this particular setup is that for 16 khz tones etc the THD is lower under 20 watts, but this is not very relevant to music..

Bandwidth with UL at 10 watts is 10hz (between 8 and 11) to 33khz... with UL+CFB it is 31.77khz...

So maybe again the driver is not doing the best...

So, I should reduce GNF and see... this transformer is built to work without any GNF.

The premature conclusion without changing feedback is that at 20db feedback total, the amp sounds like a solid state except it retains all musicality and is very enjoyable.

I prefer by far the sound of UL only so far, the best way I could describe the sound is like a triode amplifier, with unusually tight bass and detail, both things which are impossible with triodes... it is very contradictory, but I have this very full complex sound where you can hear the microphone coloration and the cuts, the words in chorale complex music is perfect, you would swear this is a super powerful 300B parallel push pull.

BTW wavebourn, I had one CFB winding in OPPOSITE phase, that is what generated 2nd harmonics LOL.

Now anyone want to see the amp THD without any GNF, only in CFB + UL ?
 
gabdx,

I just want to point out, in case you hadn't thought of it, that if you are tying the screens to a supply that is referenced to GND, then you are not running an experiment with pentode+CFB. Whenever the cathode voltage swings, the voltage between screen and cathode is changing, which is similar to the voltage changes that occur in UL. You would need a floating supply for each screen or a bootstrap cap.

How are you running the experiment? Are you keeping the voltage between screen and cathode constant (pentode operation) with a floating supply or bootstrap cap or is it varying as the cathode swings?
 
I am running the experiment in an old mono amplifier.

The amp has not the capability of running regulated G2, it has only basic UL/Triode.

What I added to the experiment is the CFB.

I was also able to get 0.5% thd at all conditions under 15 watts and a low output Z without any GNF.

It sounds OK but has the problems of non-GNF amps, it shouts at you, makes a lot of noise and have a deep perceived bass and too saturated sound. It has all this, however with low IMD, high bandwidth (10hz to 29khz at 10 watts, ref on 1khz), so you can still hear, there is not big curtain over the sound. The transformer UL is stellar, and with 8.5 db of (balanced) CFB it can drive any speakers.

So it proves that some transformers with UL+CFB can run with 0 feedback, nothing wrong, no XO-distortion from the CCS LTP and clean input stage, no hum etc.

The UL+CFB + GNF was extremely disappointing, because the builder told me the driving stage would be solicited, I now only build balanced CF with 6sn7 to drive tubes, unfortunately not in this amplifier...

If I could rate the performance, the Hammond UL+GNF is 3X better than the CFB+UL, which is 2x better than CFB+UL+GNF

The only advantage of CFB is in the 15khz and +, it has a very low THD, but other than that the IMD is higher in all other frequencies.

The only thing I can criticize on the UL+GNF is that the noise floor is high when playing multiple low tones, but it does sound better than the UL+CFB.

I tried to reduce GNF to 1.8 db with UL+CFB, and it is pointless, the CFB is very strong and I find that GNF does a way better job at keeping sound pure.

The only way I see a workable CFB+UL+GNF amp is if there is around 10 db of GNF, + 8 db of CFB, and around 50% UL,

this would require CF drivers, and two pre-gain stages, which builds up the complexity of the design, while a simple UL+13db GNF performs a lot better, but with a less precise bass (20hz - to 120hz) multiple tones at 20 watts have less than 0.15% THD, no IMD either), and less pristine 15khz + frequencies
 
you mean I should go pentode + CFB + GNF

It could work, but the kt120 pentode curves are no pretty.

gabdx,
How are you running the experiment? Are you keeping the voltage between screen and cathode constant (pentode operation) with a floating supply or bootstrap cap or is it varying as the cathode swings?

I will build a regulated supply, in spare time tomorrow maybe,(it is the same regulator from which I want to power the real amplifier that will drive the transformers, I just cannot figure out how to protect the supply from a tube disconnect or tube blowing up because the Regulator would then take the full dissipation and self destroy most likely) however I am tempted to run at the same time the input stage and LTP on the regulated supply... I don't see why not with a 4k or something dropper...

I should place a fast blow of around 50ma? in series with G2, I think I don' have that part, I have only 0.1ma slow blow which would not protect the G2s

In the same time, I can use the 50uf (of the drivers, input gain, sections) caps for the power tube, going from 100 to 150uf should help.