Wotcha all, I'm on a well-trodden path to build a low-output amp, having previously built an excellent Gerhart Gilmore Jr. kit. I found the 1/2W output from this kit louder than I had hoped, even through an 8" Jensen P8R.
Wanting to avoid an expensive attenuation box, I'm next looking to build an adapted Bassman Micro design by Rob Robinette. Rob updated his design to incorporate an EF80 pp output stage, but this will still be too loud for me.
My plan is to use a pair of 5672 subminiature pentodes as the output stage. If this also proves too loud, I'm thinking of adjusting screen voltages to suit. Should this fail, my next attempt will be a single ended 5672 output stage. Luckily there have been lots of great projects with this tube already, notably Thomas Hafemann's fantastic builds, so I have some knowledge to build upon. I am a novice at this, so writing the project up will hopefully help get it right.
Wanting to avoid an expensive attenuation box, I'm next looking to build an adapted Bassman Micro design by Rob Robinette. Rob updated his design to incorporate an EF80 pp output stage, but this will still be too loud for me.
My plan is to use a pair of 5672 subminiature pentodes as the output stage. If this also proves too loud, I'm thinking of adjusting screen voltages to suit. Should this fail, my next attempt will be a single ended 5672 output stage. Luckily there have been lots of great projects with this tube already, notably Thomas Hafemann's fantastic builds, so I have some knowledge to build upon. I am a novice at this, so writing the project up will hopefully help get it right.
Last edited:
The first step is working out the output stage. I started by looking at the charts for the 5672 tube. Going off the chart provided for the 5672 with screen voltage at 67.5V, I tried to estimate the output transformer I would require.
The 5672 has maximum operating plate voltage of 100V, so I chose 90V and drew the load line through a linear looking part of the curves to a plate current of 6ma. Dividing the plate voltage by plate current with this load line (90V/0.006a) I get 15,000 ohms. This is then doubled for the push-pull design to 30K.
To achieve this I am planning to use the Hammond 125B 4 Ohm secondary, 15K primary impedance. I will hook this up to an 8 Ohm speaker, reflecting a 30K Ohm load to the primary.
The 5672 has maximum operating plate voltage of 100V, so I chose 90V and drew the load line through a linear looking part of the curves to a plate current of 6ma. Dividing the plate voltage by plate current with this load line (90V/0.006a) I get 15,000 ohms. This is then doubled for the push-pull design to 30K.
To achieve this I am planning to use the Hammond 125B 4 Ohm secondary, 15K primary impedance. I will hook this up to an 8 Ohm speaker, reflecting a 30K Ohm load to the primary.
Hi,
If you are going to go Push-Pull do not forget to draw that load line and decide where the class A and class B curves cross. You can squeeze some extra power out of it.
Another point is the screen voltage. Someone over instructables sent me some mutrace load lines with the screen also at 90v.
I did a drawing with 90v and a 30k transformer with -13.5V bias and it would give 0.27W.(used paint_kip for it, so not 100% correct). You just need to find something to drive the tubes.
If you are going to go Push-Pull do not forget to draw that load line and decide where the class A and class B curves cross. You can squeeze some extra power out of it.
Another point is the screen voltage. Someone over instructables sent me some mutrace load lines with the screen also at 90v.
I did a drawing with 90v and a 30k transformer with -13.5V bias and it would give 0.27W.(used paint_kip for it, so not 100% correct). You just need to find something to drive the tubes.
And why are you totally focused on output power as loudness? If a half watt is too loud then you are on the wrong path. Turn it down. Clearly you are not after power stage overdrive, so just reduce what comes from the preamp.
@Printer2 - projects give me something to dream about, so yeah any excuse!
@Enzo, actually power stage overdrive verses loudness was a major consideration for the selection of these tubes. Perhaps I'm barking up the wrong tree using two of these in PP and expecting output power low enough to result in less appreciable volume?
I am also planning to find a less efficient speaker as part of the equation, perhaps the Jensen Mod 6-15.
Regarding power stage overdrive, perhaps I was too conservative with my initial load plate line? I'll move to 100V idle at the plate and redraw. I am aiming for such distortion, so do please let me know how best to address this!
@Thomasha, thank you - I'll move onto class A & B curves now. If you happen to have the 90V screen curves to hand at some point that would be great!
@Enzo, actually power stage overdrive verses loudness was a major consideration for the selection of these tubes. Perhaps I'm barking up the wrong tree using two of these in PP and expecting output power low enough to result in less appreciable volume?
I am also planning to find a less efficient speaker as part of the equation, perhaps the Jensen Mod 6-15.
Regarding power stage overdrive, perhaps I was too conservative with my initial load plate line? I'll move to 100V idle at the plate and redraw. I am aiming for such distortion, so do please let me know how best to address this!
@Thomasha, thank you - I'll move onto class A & B curves now. If you happen to have the 90V screen curves to hand at some point that would be great!
So after initial feedback, a quick redraw of the initial load line using 100V idle plate volts results in a required primary OT impedance of 33K Ohms. Achieving this with easily obtainable transformers and speakers looks a faff, so I'll stick with the 30K option here.
Sticking with a 30K Ohm primary load, with 100 idle plate volts, I need to aim for closer to 6.66ma plate current.
Moving on, the bias line is drawn. 100% bias as I'm choosing cathode bias. An idle point of 100V, 2.8ma idle plate current, around -7.2V bias, 0.28W. Push-pull, so idle current is 5.6ma.
Bearing in mind that I'm taking a lot from Rob's guide here, so please fully expect glaring lack of knowledge along the way on my part.
Moving on, the bias line is drawn. 100% bias as I'm choosing cathode bias. An idle point of 100V, 2.8ma idle plate current, around -7.2V bias, 0.28W. Push-pull, so idle current is 5.6ma.
Bearing in mind that I'm taking a lot from Rob's guide here, so please fully expect glaring lack of knowledge along the way on my part.
I did a 5672 PP amp somewhat along the lines of a 5D3 Deluxe a while back, with a Hammond 125B output transformer like you're planning. I would describe the volume with my 8" alnico speaker as quiet enough for apartment use even when cranked into power amp distortion. I still use it pretty regularly for that reason.
Building something like a Bassman is a nice idea, since the direct-heated 5672s have to be fixed-bias in my otherwise Deluxe-like circuit anyway.
I didn't know how to draw load lines at the time, so I did everything by trial and error. Maybe I'll learn something from your design and revisit mine.
Building something like a Bassman is a nice idea, since the direct-heated 5672s have to be fixed-bias in my otherwise Deluxe-like circuit anyway.
I didn't know how to draw load lines at the time, so I did everything by trial and error. Maybe I'll learn something from your design and revisit mine.
Sticking with a 30K Ohm primary load, with 100 idle plate volts, I need to aim for closer to 6.66ma plate current.
Moving on, the bias line is drawn. 100% bias as I'm choosing cathode bias. An idle point of 100V, 2.8ma idle plate current, around -7.2V bias, 0.28W.
Class A operation, both tubes sees 15KΩ. 100V/15,000Ω = 6.66ma allows me to draw temporary class A load line, which is then moved to intersect my idle point, giving the actual class A load line.
Class B operation, Rload is 7.5KΩ. 100V/7,500 = 13.3ma. This allows me to draw (estimate anyway!) the class B load line.
The Class B load line crosses the 0V grid curve at 7.55ma peak current, 44V minimum swing.
Max Class B power therefore = (100v - 44V) * .00755A/2 = 0.21W.
At the intersection of Class A and Class B we have 62V, 5.1ma.
Max Class A power therefore = (100V - 62V) * .0051A/2 = 0.097W
So there we have the load lines. Are they too far inside the max plate dissipation curve? Any changes I should consider bearing in mind I would like to achieve power amp distortion? Any feedback would be much appreciated.
Moving on, the bias line is drawn. 100% bias as I'm choosing cathode bias. An idle point of 100V, 2.8ma idle plate current, around -7.2V bias, 0.28W.
Class A operation, both tubes sees 15KΩ. 100V/15,000Ω = 6.66ma allows me to draw temporary class A load line, which is then moved to intersect my idle point, giving the actual class A load line.
Class B operation, Rload is 7.5KΩ. 100V/7,500 = 13.3ma. This allows me to draw (estimate anyway!) the class B load line.
The Class B load line crosses the 0V grid curve at 7.55ma peak current, 44V minimum swing.
Max Class B power therefore = (100v - 44V) * .00755A/2 = 0.21W.
At the intersection of Class A and Class B we have 62V, 5.1ma.
Max Class A power therefore = (100V - 62V) * .0051A/2 = 0.097W
So there we have the load lines. Are they too far inside the max plate dissipation curve? Any changes I should consider bearing in mind I would like to achieve power amp distortion? Any feedback would be much appreciated.
@sahazel, cheers for the feedback. Looks like I'll have to work out how fixed bias effects the bias and load lines, back to the drawing board! The attraction for me to the Bassman design was also that It would only involve a change of a preamp tube to move in a Marshall direction as well.
Last edited:
Here you have the load lines for a screen voltage of 90v.
and here my attempt with load lines
An externally hosted image should be here but it was not working when we last tested it.
and here my attempt with load lines
An externally hosted image should be here but it was not working when we last tested it.
Last edited:
If you want high distortion at low power, why draw "optimal-like" loadlines? Go low voltage and low impedance. Like 67V and 8kCT? Gets you well into milliWatt territory, which still may be "loud" for the home.
Fostex FE83 3" Full Range is not cheap, and surely too good for guitar, but about 10dB less sensitive than "good" guitar speakers. SB Acoustics SB10???? 3" Full Range is similar at 1/3rd the price. Put a steep 4kHz high-cut in the system so aharmonics don't bite your ears.
Fostex FE83 3" Full Range is not cheap, and surely too good for guitar, but about 10dB less sensitive than "good" guitar speakers. SB Acoustics SB10???? 3" Full Range is similar at 1/3rd the price. Put a steep 4kHz high-cut in the system so aharmonics don't bite your ears.
Clearly I need to learn a bit more about loadlines in general, no point in drawing them if I'm not sure why I am! You're right, I shouldn't try to extract maximum power for my purposes. Thanks for the feedback.
Last edited:
I think everybody has a different opinion on the subject.
IMHO the goal is always to take the most out of what you have on hand. The bigger the challenge the better.
More new stuff to learn and keep the mind occupied.
IMHO the goal is always to take the most out of what you have on hand. The bigger the challenge the better.
More new stuff to learn and keep the mind occupied.
I would be tempted to approach this by designing for "optimal" output wattage, then use something like a power scaling circuit on the output tube, so you can drop the plate/screen volts on the power amp section. You are working within 1/2 or 1 watt at full output power so knocking that down would not require a huge mosfet/heatsink to regulate the 90 volts down to say 45 volts or so.
I think you will have to bias it colder. You want the load lines (class A and B) to contour the max. dissipation line, without crossing it.
No to make too sharp a turn but have you considered this: Instead of trying to overdrive an output stage so weak as to be hardly audible, Make your 1-2-4 watt output stage, whatever power, and overdrive it into a dummy load. DO this right inside the amp, then sample the now overdriven signal from the dummy load, and then feed THAT to a small amplifier you can adjust to the loudness you want to actually drive the speaker. A basic little single ended stage would serve for that.
Lest you think I am nuts, that is exactly what Fender did inside their Princeton Recording Amp.
Lest you think I am nuts, that is exactly what Fender did inside their Princeton Recording Amp.
Thanks for the suggestions.
@Enzo, it is a tempting solution, but I'm hoping/thinking that 0.1W should still be plenty loud?
@Shanx, that is a very interesting solution, I might pop it in the build.
@Thomasha, I think that is the direction I'm moving towards. Going low OT impedance, & low voltages, the load lines didn't look right to me. I have lowered B+ voltage to 70V and gone back to high impedance, 30K ohms. As the 5672 appears to be simpler to implement with fixed bias, that is what I'm going forwards with, with a 75% idle bias to start with.
The load lines are still crossing the max. dissipation line here, so idle bias and B+ may have to come down further.
Can I ask, am I drawing the the idle bias line and class A/B load lines correctly here for a fixed bias solution?
@Enzo, it is a tempting solution, but I'm hoping/thinking that 0.1W should still be plenty loud?
@Shanx, that is a very interesting solution, I might pop it in the build.
@Thomasha, I think that is the direction I'm moving towards. Going low OT impedance, & low voltages, the load lines didn't look right to me. I have lowered B+ voltage to 70V and gone back to high impedance, 30K ohms. As the 5672 appears to be simpler to implement with fixed bias, that is what I'm going forwards with, with a 75% idle bias to start with.
The load lines are still crossing the max. dissipation line here, so idle bias and B+ may have to come down further.
Can I ask, am I drawing the the idle bias line and class A/B load lines correctly here for a fixed bias solution?