Putting feedback around an SET is really not a good idea IMO.
The problem is bandwidth and harsh clipping. Because bandwidth is not that great, with just a simple resistor in the feedback loop, you will get a peak in the highs before it rolls off, which will be at about the same point as without the feedback, owing to the fact that the OPT is the bandwidth limit, not the rest of the circuit.
You might want to read this- one of the better articles I have read on applying loop negative feedback:
Feedback and fidelity part 1
The problem is bandwidth and harsh clipping. Because bandwidth is not that great, with just a simple resistor in the feedback loop, you will get a peak in the highs before it rolls off, which will be at about the same point as without the feedback, owing to the fact that the OPT is the bandwidth limit, not the rest of the circuit.
You might want to read this- one of the better articles I have read on applying loop negative feedback:
Feedback and fidelity part 1
Why is bandwidth any different to a PP?
@Rewind, if you really get 5W through 101dB/Watt speakers you'd be generating around 108dB which you would certainly notice.
Therefore I suspect either your amp is producing much less or your speakers are not as sensitive as they claim - or both. I got huge volume from 87dB/Watt speakers and about (est) 8W - which works out as a paltry 96dB which was certainly loud in my living room!!
Indeed.
I suspect Plitron OT's don't have a gap. So they can only take a minimal amount of DC current before the Hc field causes premature drop of permeability.
Most PP transformers with EI cores will work fine with 3-5 mA DC unbalance. One only needs to see if the available headroom is enough for the full signal at the desired low frequency.
That was my opinion as well since saturation will set in if I go much above 5mA even at low power levels (10 or 20 watts).
The point I was trying to make is that they want some offset. They work best with 1 or 2 mA, not zero. The need for imbalance is somewhat dependent on the set of tubes being used, but the polarity of the offset needed for lowest distortion does not reverse whet the two tubes are swapped, but does reverse when the OPT plate leads are swapped, indicating a transformer issue.
These transformers were from their surplus page about 8 years ago. They are rated for 400 watts at 20 Hz and were originally intended for bass guitar amps. I have measured something like 8 Hz to 73 KHz for the 3 db points at 150 watts.
I have a few dozen guitar amp quality P-P OPT's (rated for "80VA at 70 Hz")and they all need some offset to work best, but they are all different. These transformers perform the best when driven from a low impedance source like 300B's (no saturation at 25 Hz and 25 watts) or sweep tubes in pentode mode with local feedback around the output tube only (starts to saturate at 40 watts and 25 Hz). Running EL84's in pentode with GNFB is worse case as far as low frequency power without saturation. Saturation is beginning at 15 watts and 25Hz.
I did restack a pair of them with the E's and I's all lined up so I could play with a gap. Small gaps (1 layer of Kapton tape) help the EL84 amp a little, and make the others worse. Any larger gap makes all amps worse. I suspect this has a lot to do with lack of primary inductance to begin with.
Have been reading this thread with interest.
Current amplifier is an all pentode, phase splitterless, fully balanced push-pull.
It runs 12SJ7 pentode inputs, feeding 6CA7EH Electro-Harmonix output valves.
Screens on both input and output valves are regulated with VR tubes.
Feedback is plate-to-grid around the outputs and the amp runs class A/B fixed bias.
Over the years I have realised the important role the power supply plays in getting good bass from a valve amplifier.
On this amp, I use a CLC PSU with SS rectification. There is a 2200uF storage bank after the choke. Bass is tight, clean, controlled, extended and powerful with this arrangement, which has given the best sound from pp I have had to date. Just fabulous.
I think (in terms of bass performance) SE and PP need different approaches, especially if you run class A/B on the pp side. My whacking great reservoir might not be appropriate for SE (I've never tried it) but it certainly works in A/B,fixed bias, pp.
In the pic below, can be seen the big cap bank, next to the power transformer.
I've gained some useful info from this thread, for future development.
Current amplifier is an all pentode, phase splitterless, fully balanced push-pull.
It runs 12SJ7 pentode inputs, feeding 6CA7EH Electro-Harmonix output valves.
Screens on both input and output valves are regulated with VR tubes.
Feedback is plate-to-grid around the outputs and the amp runs class A/B fixed bias.
Over the years I have realised the important role the power supply plays in getting good bass from a valve amplifier.
On this amp, I use a CLC PSU with SS rectification. There is a 2200uF storage bank after the choke. Bass is tight, clean, controlled, extended and powerful with this arrangement, which has given the best sound from pp I have had to date. Just fabulous.
I think (in terms of bass performance) SE and PP need different approaches, especially if you run class A/B on the pp side. My whacking great reservoir might not be appropriate for SE (I've never tried it) but it certainly works in A/B,fixed bias, pp.
In the pic below, can be seen the big cap bank, next to the power transformer.
An externally hosted image should be here but it was not working when we last tested it.
I've gained some useful info from this thread, for future development.
Last edited:
Yes,
When I first started building valve amps about ten years ago, I simply followed existing schematics with their low capacitance power supplies. Although all my builds had tube magic, in the mids and highs, it was the bass that was always a problem.
It was when I read the late, great Harvey Rosenberg's musings on "pentodosity" and his "for the ten-millionth time" reiterations on the importance of the power supply a few years ago that it finally clicked.
Last edited:
Then they might have a so called "distributed gap" which is often found in toroidals but I still doubt they are cut.
Yes George, I understood your point and am with you. In general if there is no gap whatsoever the residual magnetization can be quite high, especially if they are of the higher quality of higher permeability type! If they have a tiny distributed gap it helps together with minimal unbalance. Possibly in your specific case these cannot considered truly symmetric for some reason. I don't have practical experience in winding toroidals. Cannot tell for sure.
With guitar OT's you might be able to apply more offset as these generally have less turns in comparison to Hifi relatives of the same size. Often these have pretty simple geometry but, if made properly, still work very well with low Rp tubes out to 40-50 KHz (-3dB). This also could explain why you get low inductance with a small air-gap already. However one should know the ratio between the thickness of the air gap and the length of the magnetic path to know precisely. The gap size alone will not tell much. The transformer size matters as well.
Last edited:
And that may have something to do with my SE amps having good bass. I've been using switching power supply bulk storage caps as my final filter caps.
Even the little 6P1P-EV SE amps which only put out 3W use 220uF caps.
I think the 6P41S SE amp uses 470uF caps, one for each channel.
The power supply filter is CLC-2(DRC). Since it uses SS rectification the first caps are 150uF, 10H choke, 470uf x 2, then the two 100R resistors and blocking diodes to isolate the final 470uF caps for each channel.
P.S.
@George, if you re-stack those small PP guitar OT's to use them for small SE amps you might try the intermediate solution for the air-gap that might give you a better balance between inductance and the necessary headroom at low frequency. Instead of alternating the E's and I's in single layers you can alternate them in small blocks of 3x3, 4x4 or 5x5, depending on the DC current you want to apply. This will result in a smaller gap than any practical spacer you might find.
If you have for example a 1" square core, the average path length will be about 15 cm. If the transformer has got 2000 turns and you only need a gap for 30-35 mA DC then even a 0.05 mm spacer is too big! That will cause higher DC capability than you actually need but lower inductance (should be some 11H with small signal and about 16H a full throttle but still away from saturation). If the nominal primary impedance has to be, lets say, 5K that is clearly not enough to go down to 20-30Hz properly. The effective primary load at 30Hz would be already about 2.5K. Or if you want to see it quickly in another way XL=2*pi*f*L has to be at least 8-10 times bigger, in the best cases, than Rp//5K to get low distortion. 5-6 times can still be acceptable (and actually is rather common with higher Rp tubes). So if Rp is 1.5K then Rp//5K = 1.15K. XL= 3K at 30Hz for 16H. The ratio is not even 3....
@George, if you re-stack those small PP guitar OT's to use them for small SE amps you might try the intermediate solution for the air-gap that might give you a better balance between inductance and the necessary headroom at low frequency. Instead of alternating the E's and I's in single layers you can alternate them in small blocks of 3x3, 4x4 or 5x5, depending on the DC current you want to apply. This will result in a smaller gap than any practical spacer you might find.
If you have for example a 1" square core, the average path length will be about 15 cm. If the transformer has got 2000 turns and you only need a gap for 30-35 mA DC then even a 0.05 mm spacer is too big! That will cause higher DC capability than you actually need but lower inductance (should be some 11H with small signal and about 16H a full throttle but still away from saturation). If the nominal primary impedance has to be, lets say, 5K that is clearly not enough to go down to 20-30Hz properly. The effective primary load at 30Hz would be already about 2.5K. Or if you want to see it quickly in another way XL=2*pi*f*L has to be at least 8-10 times bigger, in the best cases, than Rp//5K to get low distortion. 5-6 times can still be acceptable (and actually is rather common with higher Rp tubes). So if Rp is 1.5K then Rp//5K = 1.15K. XL= 3K at 30Hz for 16H. The ratio is not even 3....
Last edited:
The guitar amp OPT's in question are very low cost open frame 6600 ohm CT transformers that weigh about 2.5Kg each. They were made by Schumaker, who also makes car battery chargers with similar looking transformers. I bought 200 of them nearly 20 years ago to make guitar amps. I got them for $16 each. I still have some.
I have dissected one to find no interleaving, just one half primary, then the entire 0-4-8-16 ohm secondary, then the other half primary, wound with thicker wire to even out the DCR between the two sides.
Maybe 10 years ago I breadboarded a 300B push pull amp with junk box components and a pair of these OPT's. It was a test design to see what worked, but it worked so well that I still have it. I tried rebuilding it with good parts, but the magic wasn't there, so I put it back the way I built it.
The "gap" experiments were done for some purposely unbalanced guitar amps I did several years ago. I was using dissimilar output tubes in a guitar amp for "tone color". One amp was a 6V6GT at about 30 mA paired with a 6L6GC at 50 mA. The amp is clean at low volumes, but makes sweet distortion when cranked. The lowest note on a standard guitar is around 80Hz and I purposely used smallish coupling caps, so saturation was not an issue.
I have used these OPT's as is in an SE amp, but I was in the 20 to 30 mA region using a 45 or triode 6V6GT. The $29 Edcor works far better in this application.
I have dissected one to find no interleaving, just one half primary, then the entire 0-4-8-16 ohm secondary, then the other half primary, wound with thicker wire to even out the DCR between the two sides.
Maybe 10 years ago I breadboarded a 300B push pull amp with junk box components and a pair of these OPT's. It was a test design to see what worked, but it worked so well that I still have it. I tried rebuilding it with good parts, but the magic wasn't there, so I put it back the way I built it.
The "gap" experiments were done for some purposely unbalanced guitar amps I did several years ago. I was using dissimilar output tubes in a guitar amp for "tone color". One amp was a 6V6GT at about 30 mA paired with a 6L6GC at 50 mA. The amp is clean at low volumes, but makes sweet distortion when cranked. The lowest note on a standard guitar is around 80Hz and I purposely used smallish coupling caps, so saturation was not an issue.
I have used these OPT's as is in an SE amp, but I was in the 20 to 30 mA region using a 45 or triode 6V6GT. The $29 Edcor works far better in this application.
SET transformers usually have DC current involved, which can lead to saturation problems. This makes them tricky to design; quite often the practical limit for full bandwidth (20Hz-20KHz) is about 7 watts (even then true 20Hz response is rare), and in many cases its obvious that the smaller 2A3 and 45-based amps have more bandwidth, part of the reason they sound better.
With push-pull the output transformers usually don't have the saturation issue as the DC currents of the power tubes are (theoretically) balanced and thus cancel. So you can build a much larger OPT and usually get more bandwidth too (please understand that I am glossing over a lot of transformer theory for the sake of brevity).
Quite often you don't notice the loss of bandwidth in an SET because the speakers used to good effect with an SET are high efficiency, which generally don't have bandwidth to 20Hz. So you don't hear the loss of the bottom octave. This frees the designer to optimize the unit for better HF response.
Compare this to a P-P unit (I like the H/K Citation 2 for example) wherein the OPTs are good to 10Hz no worries, and well over 50KHz on the top end.
SETs often struggle to go over 20KHz; if you add feedback to this you will experience a peak below 20KHz and the possibility of oscillation exists. Plus the clipping characteristic will not be as graceful, which is why most SET designers run zero feedback.
Total energy storage of the PS is key, I believe. My 833C monos each have an LCLC output supply (35H-22uF-8H-126uF) with a final cap bank (7 1000uF T-HA in series, actual measurement = 126uF) running at 2300V B+ that translates to 333J of energy storage (1/2squaredV*capacitance). Add in the 58J from the first cap bank and the energy stored by the chokes and that's quite a bit of juice!
Transients are very dynamic.
.
Last edited:
Can You please further explain this?
I have been wondering how I'm getting decent bass (recording of a bass Guitar) from 10nF Silver mica and Polystyrene Coupling caps. Great mids and highs.
I get smoother, but less detailed results from a 1.5uF motor run (as expected)...same smoothness but more detail with a 0.411uF motor run. The 10nF Mica's and Polystyrene sound best to my ears.
My amp is 1.4W SE and uses 6v6 30mA, or 6L6 50mA. Triode modes. Even the 6Y6's sound decent.
I'm using a filament transformer for my OPT- 115v pri.-6.3v sec. Center Tap...I'm using 1/2 the sec. by using the center tap which = 3.15v sec.
Last edited:
Don't know if you are recording direct, or MIC'ing the amps speaker.
Many guitar speakers have a resonant peak in the range of the low E string. Often the cabinets themselves also have a resonance in the bass region. Driving this type of speaker with an amp having a flat frequency response can result in a "boomy" or "muddy" sound. Often the amp response purposefully has a gentle roll off below about 150 Hz to reduce the boom. Smaller coupling caps are the easiest way to get this.
Some guitar amps use an OPT that is far too small for the power rating of the amp. Look at the OPT in any old Fender. You can stuff far more power through any OPT if you reduce or eliminate the low frequencies. The OPT's I have are rated for 80VA at 70Hz. If I use them in a HiFi application and test from 25 Hz to 20 KHz I see the onset of saturation anywhere from 15 to 40 watts, depending on the tube used. If I run them in the sweep tube amp, which can hit 130 watts on Edcor 100 watt OPT's, but test from 82 Hz to 20 KHz, I see about 100 watts before clipping sets in. All testing was done with an 8 ohm 500 watt resistor for a load.
If I take the same amp, reduce the coupling caps from .1uF to .022 uF, connect up a speaker cabinet and plug in my guitar, I can play that thing at any level from quiet to an overdriven fully clipped scream, and not see any hints of saturation. Cranking an open low E string (82 Hz) to the edge of clipping puts over 100 volts peak to peak across the speaker. That's almost 160 watts, but the amp clips at 130 watts. Why don't I see clipping or saturation?????
Look at the impedance curve of a typical guitar speaker, It is probably around 20 to 50 ohms at 82 Hz, depending on size. The amp is delivering far less than 100 watts and is limited only by its B+ voltage at this frequency.
So by rolling off the low frequencies, won't we lose all the bass? Well if the system was perfectly flat, and had zero distortion, maybe....but we are talking about a guitar amp, do guitar amps distort? A known phenomenon of psychoacoustics says that bass can sound fuller and louder if augmented with a good bit of lower order harmonic energy. Even a little bit of transformer saturation can boost the bass, but too much really sounds bad.....
and blows stuff up. Back in the 60's there used to be a saying...."Never play bass through a Bandmaster." One of my friends proved this one true. I got to fix the Bandmaster.
Magnetic transformer saturation happens when the magnetic core can not accept any more magnetism.....more energy into the transformer does not produce any more energy out of the transformer. Unfortunately, this isn't all. When the core saturates, the effective inductance of the primary drops....quickly. This causes the tube current to increase. An inductors effective impedance goes down with decreasing frequency, so the tube current goes up quickly as the frequency goes lower. The increase in tube current just serves to further saturate the transformer.
So, smaller coupling caps can help squeeze a bit more power through a given OPT, and can help keep a speaker cone from losing control near resonance, by reducing the power in the region of resonance, and boosting the harmonics (relative to the reduced fundamental) to make up some of the difference in sound.
Like the solid chest thumping sound of multiple 15 or 18 inch drivers.....use big caps, or maybe even a class D amp. Like the bright sound of a bass through eight 10 inch speakers, try smaller caps.
Many guitar speakers have a resonant peak in the range of the low E string. Often the cabinets themselves also have a resonance in the bass region. Driving this type of speaker with an amp having a flat frequency response can result in a "boomy" or "muddy" sound. Often the amp response purposefully has a gentle roll off below about 150 Hz to reduce the boom. Smaller coupling caps are the easiest way to get this.
Some guitar amps use an OPT that is far too small for the power rating of the amp. Look at the OPT in any old Fender. You can stuff far more power through any OPT if you reduce or eliminate the low frequencies. The OPT's I have are rated for 80VA at 70Hz. If I use them in a HiFi application and test from 25 Hz to 20 KHz I see the onset of saturation anywhere from 15 to 40 watts, depending on the tube used. If I run them in the sweep tube amp, which can hit 130 watts on Edcor 100 watt OPT's, but test from 82 Hz to 20 KHz, I see about 100 watts before clipping sets in. All testing was done with an 8 ohm 500 watt resistor for a load.
If I take the same amp, reduce the coupling caps from .1uF to .022 uF, connect up a speaker cabinet and plug in my guitar, I can play that thing at any level from quiet to an overdriven fully clipped scream, and not see any hints of saturation. Cranking an open low E string (82 Hz) to the edge of clipping puts over 100 volts peak to peak across the speaker. That's almost 160 watts, but the amp clips at 130 watts. Why don't I see clipping or saturation?????
Look at the impedance curve of a typical guitar speaker, It is probably around 20 to 50 ohms at 82 Hz, depending on size. The amp is delivering far less than 100 watts and is limited only by its B+ voltage at this frequency.
So by rolling off the low frequencies, won't we lose all the bass? Well if the system was perfectly flat, and had zero distortion, maybe....but we are talking about a guitar amp, do guitar amps distort? A known phenomenon of psychoacoustics says that bass can sound fuller and louder if augmented with a good bit of lower order harmonic energy. Even a little bit of transformer saturation can boost the bass, but too much really sounds bad.....
and blows stuff up. Back in the 60's there used to be a saying...."Never play bass through a Bandmaster." One of my friends proved this one true. I got to fix the Bandmaster.
Magnetic transformer saturation happens when the magnetic core can not accept any more magnetism.....more energy into the transformer does not produce any more energy out of the transformer. Unfortunately, this isn't all. When the core saturates, the effective inductance of the primary drops....quickly. This causes the tube current to increase. An inductors effective impedance goes down with decreasing frequency, so the tube current goes up quickly as the frequency goes lower. The increase in tube current just serves to further saturate the transformer.
So, smaller coupling caps can help squeeze a bit more power through a given OPT, and can help keep a speaker cone from losing control near resonance, by reducing the power in the region of resonance, and boosting the harmonics (relative to the reduced fundamental) to make up some of the difference in sound.
Like the solid chest thumping sound of multiple 15 or 18 inch drivers.....use big caps, or maybe even a class D amp. Like the bright sound of a bass through eight 10 inch speakers, try smaller caps.
Ah, I see exactly what you mean about the bass re. saturation - this sounds like it could be cured with a bigger transformer though?
Treble I also see to a limited extent - but most feedback in tube amps includes the transformer which is a crazy idea that makes a bad situation worse - like you describe. With the correct _local_ feedback operating so the transformer is driven properly in the first place;- for a low impedance source I think the treble should pretty much compete with a PP transformer (of comparable size!).
Have you tried a SEP or SET amp with local feedback only from the output tube anode to the driver cathode? I'd be interested to hear your experiences of you do/did, it made a night vs day different for my SEP.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.