This is a nice thread of @jhstewart9 about this kind of feedback:
https://www.diyaudio.com/community/...ut-transformer-secondary-free-for-all.396112/
https://www.diyaudio.com/community/...ut-transformer-secondary-free-for-all.396112/
I do something like that shown in posts 21 and 22 except that the cathode resistor is connected directly to ground. This squelches the concerns voiced by a few high efficiency speaker owners that the few millivolts of DC dropped across the DCR of the OPT secondary would fry their speakers or cause them to exhibit unwanted behavior.
@Tubelab_com Nice variation I’ve never seen before!
@Francois G You can try different amount of feedback by changing the tap you get the signal from.
@Francois G You can try different amount of feedback by changing the tap you get the signal from.
zintolo,
Great Question!
Not you zingolo, but somewhere along the way, the train jumped the tracks, from Hi Fi amps to Guitar amps.
First of all, I would not put Hi Fi Stereo Amplifiers, and Rock Guitar Amplifiers in the same category.
A ripe Philippine Yellow Mango, versus a Raw Steak.
Most methods, using 1. local negative feedback, 2. global negative feedback, 3. Combinations of negative feedback, or 4. no negative feedback;
They all can work very well . . .
If the circuit topology, parts quality, etc. is correct for the chosen method of 1, 2, 3, or 4. above.
I remember designing a single ended amp, using a 5k Opt, 16 Ohm tap for negative feedback, and a Triode Wired Pentode (g2 to g1 u = 11).
Gain before applying negative feedback, of course, was less than u.
With the output transformer voltage reduction of Root (5k/16) = 17.7, I think the negative feedback I measured was about 3 dB.
It did Increase the damping factor, Increase the lower and upper bandwidth, and it Reduced distortion . . . Good!
(all of those only slightly, with just 3dB of negative feedback).
I tried single ended using Pentodes and Beam Power tubes in their native mode. I knew I Must use negative feedback for the output stage, no matter which feedback topology was to be used.
I will no longer design that way; but am happy to listen to good SE and PP commercial, and good SE and PP diy designs.
I also used UL in single ended, and push pull.
For SE and PP, I first used DHT; then Triode Wired Pentodes and Triode Wired Beam Power; and now I use Ultra Linear Beam Power tubes, and Triode wired Beam Power tubes.
We all have our preferences.
They key is proper design, quality parts, and correct implementation within the topology we pick to use.
I often wonder how many designers, and on-lookers remember that negative feedback that includes the output transformer does Not fix saturation, it makes it worse, as soon as you reach the signal level that causes saturation.
Those of you who use simulation, that is great, it helps verify the "viability-keep going", or "toss this one" nature of the design.
Asking me to use software, is like asking me to put my hand in a box of snakes.
This is a Great Thread, so much to remember, learn, and try!
Great Question!
Not you zingolo, but somewhere along the way, the train jumped the tracks, from Hi Fi amps to Guitar amps.
First of all, I would not put Hi Fi Stereo Amplifiers, and Rock Guitar Amplifiers in the same category.
A ripe Philippine Yellow Mango, versus a Raw Steak.
Most methods, using 1. local negative feedback, 2. global negative feedback, 3. Combinations of negative feedback, or 4. no negative feedback;
They all can work very well . . .
If the circuit topology, parts quality, etc. is correct for the chosen method of 1, 2, 3, or 4. above.
I remember designing a single ended amp, using a 5k Opt, 16 Ohm tap for negative feedback, and a Triode Wired Pentode (g2 to g1 u = 11).
Gain before applying negative feedback, of course, was less than u.
With the output transformer voltage reduction of Root (5k/16) = 17.7, I think the negative feedback I measured was about 3 dB.
It did Increase the damping factor, Increase the lower and upper bandwidth, and it Reduced distortion . . . Good!
(all of those only slightly, with just 3dB of negative feedback).
I tried single ended using Pentodes and Beam Power tubes in their native mode. I knew I Must use negative feedback for the output stage, no matter which feedback topology was to be used.
I will no longer design that way; but am happy to listen to good SE and PP commercial, and good SE and PP diy designs.
I also used UL in single ended, and push pull.
For SE and PP, I first used DHT; then Triode Wired Pentodes and Triode Wired Beam Power; and now I use Ultra Linear Beam Power tubes, and Triode wired Beam Power tubes.
We all have our preferences.
They key is proper design, quality parts, and correct implementation within the topology we pick to use.
I often wonder how many designers, and on-lookers remember that negative feedback that includes the output transformer does Not fix saturation, it makes it worse, as soon as you reach the signal level that causes saturation.
Those of you who use simulation, that is great, it helps verify the "viability-keep going", or "toss this one" nature of the design.
Asking me to use software, is like asking me to put my hand in a box of snakes.
This is a Great Thread, so much to remember, learn, and try!
Last edited:
I read it for the first time from you years ago on this forum. This and the phase issues is what has always made simpler for me to avoid nfb across opt in ho-fi. At the same time, I think that was this behaviour of distorted hi-fi amps that pushed towards early instrument amp design.
I would not go that far. Maybe a nice steak VS a cheeseburger. Since both the HiFi and guitar amp share the same set of bones underneath the skin, albeit with some different trimmings and the lineage gap widens with each subsequent generation. My first succesful DIY amp came somewhere around age 10. It was intended as a guitar amp but worked equally well as a mono HiFi amp. The original idea came from cracking open an old Fender Champ and tracing its schematic. We are talking 6SJ7 pentode driving a 6V6 pentode powered by a 5Y3 feeding a plywood box full of speakers robbed from old radios and TV sets.
Trying to wrap feedback around a cheap OPT, especially GNFB, can and often will make matters worse since core saturation causes the feedback loop to crank the closed loop gain once saturation is reached. A simple local feedback loop from the speaker winding to the output tube's cathode can help cheap OPT's that are a bit short in primary inductance, and maybe help with excess winding capacitance but nothing can fix core saturation.
I'm a big fan of using the secondary winding for CFB in SE amps.
Philosophically speaking it makes both the tube and the transformer slightly better just by moving one (1) wire: The feedback makes the tube act as if it has lower Mu and Rp and effectively wiring the OPT as an autoformer increases the turns ratio and the primary inductance a bit. The changes are usually small but in the right direction and it's all for free.
Yes, there will be some DC across the speakers (delicate fullrange drivers might have a problem with that, but I'm using 12" or bigger pro drivers) and the driver stage will have to deliver a bit more swing but it should have some margins of designed properly.
Philosophically speaking it makes both the tube and the transformer slightly better just by moving one (1) wire: The feedback makes the tube act as if it has lower Mu and Rp and effectively wiring the OPT as an autoformer increases the turns ratio and the primary inductance a bit. The changes are usually small but in the right direction and it's all for free.
Yes, there will be some DC across the speakers (delicate fullrange drivers might have a problem with that, but I'm using 12" or bigger pro drivers) and the driver stage will have to deliver a bit more swing but it should have some margins of designed properly.
Tubelab_com,
I should have stated the very large difference between:
A guitar amp playing Hard Rock . . .
and Dan Balmer's Jazz guitar playing in a quiet restaurant bar.
I guess that is like a Raw Steak, and Filet Minion respectively.
Marty Robbins 1960 recording, released 1961, "Don't Worry" ('Bout me), the baritone guitar mixing desk had a very bad distortion.
Marty did not like the sound, but the producer did.
Pretty soon, the industry was chasing after that fuzzy distorted sound.
Who knew that would happen?
I got a used KEF C25 at Goodwill. I think it is KEF's answer to the classic British small monitor, I can not remember that famous model number.
So many other manufacturers make similar small monitors. Anything famous and successful is bound to be cloned.
When I pulled it apart to check the crossover, I found out that it had a 1000uF Plastic Cap in series with the Woofer.
At 20Hz, 1000uF is 10 Ohms capacitive reactance, but that woofer does not work down there in that small cabinet.
I suppose that cap was there to deal with a solid state amp's DC offset, and the few tube amps with cathode feed back that also have a DC offset.
Near equal voice coil depth and magnetic pole piece depths do not deal very well with DC offsets, and the same for overhung voice coils.
I should have stated the very large difference between:
A guitar amp playing Hard Rock . . .
and Dan Balmer's Jazz guitar playing in a quiet restaurant bar.
I guess that is like a Raw Steak, and Filet Minion respectively.
Marty Robbins 1960 recording, released 1961, "Don't Worry" ('Bout me), the baritone guitar mixing desk had a very bad distortion.
Marty did not like the sound, but the producer did.
Pretty soon, the industry was chasing after that fuzzy distorted sound.
Who knew that would happen?
I got a used KEF C25 at Goodwill. I think it is KEF's answer to the classic British small monitor, I can not remember that famous model number.
So many other manufacturers make similar small monitors. Anything famous and successful is bound to be cloned.
When I pulled it apart to check the crossover, I found out that it had a 1000uF Plastic Cap in series with the Woofer.
At 20Hz, 1000uF is 10 Ohms capacitive reactance, but that woofer does not work down there in that small cabinet.
I suppose that cap was there to deal with a solid state amp's DC offset, and the few tube amps with cathode feed back that also have a DC offset.
Near equal voice coil depth and magnetic pole piece depths do not deal very well with DC offsets, and the same for overhung voice coils.
Last edited:
Expressing this cathode feedback in volt percentages like for UL, for a 5k:8 transformer it would be 1/(sqrt(5k:8))= 4%
I normally see around 33% UL (a to g2) and 10% cathode feedback.
Expressing the same feedback in dB, with EG an EL84 swinging from 50 to 450V at B+ of 250v with an input signal of 24Vpp, I would get 16Vpp on the cathode, so 20xlog((12+16)/12) = 7,4 dB
Concerning core saturation, I’m sure there are many solutions and I’m reiventing the weel, but the simplest one (with the right tubes and split secondaries like some cheap Hammond have) would be heaters supplied with a CCS through a dedicated winding having a ratio with the primary proportional to the heater/primary-idle currents.
Back to the previous example, an EL84 at 250V B+ and 90% dissipation will have 43 mA through it, while an EL84 and EG a 12AT7 will have 1.060 A, so a ratio of 1/25, that is exactly another 8 ohm secondary. OPT with split secondary (like some Hammonds) can be used for this purpose.
The same 16Vpp will be on this secondary as well, so it is needed to supply the CCS at least 8V (+4 for the CCS) above the 6,3V of the heaters. What you get from the OPT, you loose on efficiency (20 W vs 6,3 W for heaters).
Thanks for your experience. George solution, splitting the AC path from the DC, is indeed a very nice idea to solve this issue.
A clever solution indeed, one that I would take a closer look at if I was using more fragile speakers!
I can see two small problems with it though: If I visualize things correctly, the cathode resistor effectively ends up in parallel with the output signal. Probably not much of a problem. The second thing is that it requires a cathode resistor, so fixed bias is out of the game.
Off Topic: I think most guitar amps have Zout within 10 (American oriented) and 30 Ohm (British oriented). Except some modern high-gain amps that have tighter power amps and Zout highly dependant on frequency to use the impedance of the speaket to equalize the sound. The two amps above should have a Zout around 20 Ohm.
The triode/pentode control on the EF86 is left without cap intentionally, like the scratchy presence control of early Marshalls.
I agree with you: we are always at least one order of magnitude higher with the cathode resistor, often way more than that, so it won’t waste power.
Yes, it is indeed.
@Tubelab_com
What if we reference the voltage divider of the UNSET to anode and secondary, instead of anode and ground? Can it be good to include the OPT dc coupled to the nfb this way?
@TonyTecson have you applied Schade nfb to the driver as well? It is a very simple yet effective kind of feedback, the drawback I see on the driver stage is the low input impedance of the amp.
Anyone used Hawksford kind of feedback like @smoking-amp suggested here?
https://www.diyaudio.com/community/threads/hawksford-style-error-correcting-output.35001/
https://www.diyaudio.com/community/threads/hawksford-style-error-correcting-output.35001/
Zintolo,
You always ask good questions.
As a kid I built 2 Knight Kit 12 Watt mono amplifiers; 12AX7 mag phono, 12AU7 gain and concertina, and EL84 push pull on a very small output xfmr. Simple, but the sound impressed me . . . on my Electro Voice Wolverines, and Stephens Free Cones drivers.
After the US Navy, I built a Harmon Kardon stereo amp, more controls, but with output tubes similar to 6V6, and not much bigger output transformers.
Then about 1968, I changed to solid state amps . . . the SWTP little tiger (the crossover distortion champion) and a Dyna SCA80.
I did not come back to tubes until about 1995. A Dyna Stereo 70, became an experimenter's platform. Remove the output transformer laminations, and re-stack them with an air gap; single ended and a single EL34 per channel, cathode negative feedback from the output transformer secondary.
Then a stereo 300B with lots of 9V batteries in series for fixed bias, with Real SE output transformers was built on the same Dyna Stereo 70 chassis.
I found AC filaments for DHT to have too many disadvantages, fix the hum with a hum pot, but still have Hum IMD on every musical note (I wrote about that effect too many times); and hum modulation from the filament's unequal distance to the two sides of the steel plate . . . have to select tubes until you find ones that have filaments and plates that are more carefully located versus each other.
To solve the above Three problems for DHT on AC filaments, it was time to use DC filaments on 300B tubes.
Eventually, I did not want to use all the the extra parts, increase complexity, Schotky bridges, and the need for higher current rated filament windings required for AC to DC circuits (peak and rms current of capacitor input filters for the filaments). Then I used DC filaments for my later 300B amplifiers.
I built DC filament power for SE 300B and PP 300B.
Then AC filaments on 45s and 2A3s. If I ever do another 45 or 2A3, it will have DC filaments.
Some people get to sleep by counting sheep.
In a similar manner, I "saw" indirect heated output tubes "passing" in front of me.
Aha . . . no DC required for the filaments.
6L6GC, EL34, KT66, KT88, KT77, and 5881. I can Drive my Car over to Eurotubes.com, and pick them up (All JJ products).
There were even some NOS 6CK4 triodes too (with Real cathodes).
For me, EL34 and EL84 are the only pentodes I would use (I would get JJ from Eurotubes).
Early on, I discoveed that Native pentode and Native Beam Power mode were not for me. Very non-linear without lots of negative feedback.
So I tried UL.
I did not have a Real single ended UL transformer, so I re-used the Dyna ST70 Air gapped one, and the center tap became a 50% UL tap;
20% and 80% if you use the screen taps.
I purchased PP output transformers with UL taps. CCS and cathode coupling is my phase inverter of choice; I did try paraphase and concertina phase splitters, but was not a fan of them.
I built some self inverting output stages, with triode, and than triode wired beam power tubes.
I liked the distortion, damping factor, and simplicity of Triode Wired Beam Power tubes.
I finally built a balanced amplifier, triode wired beam power tubes, to use with my XLR outputs CD player.
I have lots of different speakers; but I use the ProAc Tablette Signature 2000 with my modification of the crossover.
I use a JBL powered subwoofer whenever I want to.
A lot of my amplifier designs along the way were not very good, so they either got modified, or were re-cycled into completely different topologies.
I did listen to my various amplifiers on many different speaker models; most often I would use either the Left channel to drive a mono-block, or the Right channel.
With the sound coming from only one speaker, it can reduce the complexity of the sound, and allow for easier analysis of the sound; especially for those who only have one working ear, such as me.
When I first lost one ear, I would walk across the room in front of the stereo pair of speakers.
Boy! The comb filter effect almost tore my head off.
So many recordings on an XY plot look like a 45 degree line, very little width, and with an upward slope. That creates a comb filter, as the same sound coming from the two speakers adds or cancels as the distance of each frequency from each of the two speakers changes the relative phase.