Kofi Annan said:But now its up to eleven and not really that loud... and I have pretty sensitive speakers (96dB Fostex 206Es).
Don't forget that it is allowable to use the rest of the volume control's range. I generally consider that my control should be at 1 o'clock for normal listening, and that full volume shouldn't clip.
Glad you got the amplifier working and are pleased with it. Are you going to let us see internal pictures?
Glad your amplifier is now working 🙂
The drop in gain is due to the fact that the negative feedback is now working correctly. As Sy says, increasing the value of the feedback resistance will reduce the amount of feedback, and increase the gain.
The drop in gain is due to the fact that the negative feedback is now working correctly. As Sy says, increasing the value of the feedback resistance will reduce the amount of feedback, and increase the gain.
EC8010 said:
Don't forget that it is allowable to use the rest of the volume control's range. ...
I had a mental block on this for a while. Built my amps and preamp (with gain) and at normal listening the volume control was set about 8:30 (not much control range there!).
I later used the amps with a passive pre and normal listening level increased to about 11 o'clock. For quite a while it just didn't "feel right". It really did feel like I was getting less power from the amp and I actually put it on the scope.
No surprise that max output before clipping was 7.5 watts with the active preamp and ..... tada..... 7.5 watts with the passive pre. Just had to use a bunch more dial to get there which once I got used to it is actually a lot nicer with more control.
But now its up to eleven and not really that loud
I may be wrong here, but I believe this is a nod to the parody rock group "Spinal Tap" where the lead guitarist had all his amps fitted with knobs that went to 11 instead of 10 so that they would play louder

I believe what our Secretary General is saying is that full-up isn't loud enough.
-Casey
I may be wrong here, but I believe this is a nod to the parody rock group "Spinal Tap" where the lead guitarist had all his amps fitted with knobs that went to 11 instead of 10 so that they would play louder
I believe what our Secretary General is saying is that full-up isn't loud enough.
Correct. I mean, it's OK, but it felt more dynamic and puchy before with the NFB removed.
Kofi
Kofi Annan said:
Correct. I mean, it's OK, but it felt more dynamic and puchy before with the NFB removed.
Kofi
Well now that it's built and sorted, you get to do the fun stuff. Here's where it becomes Your amp. Follow SY's advice and play with the amount of NFB. You get to decide just how much you like. That's your reward for building it yourself, and knowing the design.
And I'm with EC on the volume control. Shouln't clip on full volume.
Sheldon
Kofi Annan said:Of course, there's a good possibility that I screwed this up again....
Kofi,
Shut up. 🙂 🙂 🙂
You have done a great deal right, whether with or without advice. Experience is a great teacher .... but second hand (learning from others) is cheaper! I am glad it works. A picture would still be great.
Just to echo the advice of others, don't fall into the trap that I call the "fast accelerator syndrome". This refers to some of the smaller motor cars, where the accelerator linkage is such that you hardly touch the pedal when it flies away. But soon afterwards it runs out of steam, because it is not a powerful car. The particular accelerator design just opened most of the throttle at an early stage instead of gradually.
The amplifier equivalent is a low power thing but with too much gain before the volume control or too large an input signal. The result is that the output increases smartly at an early control setting, but you are not halfway up the volume control before it overloads, and you never use the rest of the pot.
By definition a pre-amp input for say a phono cartridge might specify 2,5 mV for full output. That is a scientifically correct statement but unpractical. Ideally you should have enough watts left at full volume control to take care of say 10 dB peaks above the normal specification, otherwise you are stuck with a part of the volume control which you will never use. It is quite in order (and in fact an indication of a mature design) to occasionally have to use all of the volume control. I hope you follow this. (In superior pre-amplifier designs there is often a pre-set input control to adjust whatever input is used to control too high an input signal.
Tutorial over, the BV is not a very sensitive amplifier. I notice 2V input spec. The feedback appears to be a modest 10 dB approximately (that is about 3,3 times). Perhaps later you could invest in a simple pre-amplifier (eg. a ECC81 for both channels) on the same chassis in preference to lowering the feedback. I could post a design (as could many other members).
Regards.
Johan Potgieter said:Experience is a great teacher .... but second hand (learning from others) is cheaper!
Good post Johan. This particular quote reminds me of a slightly different version that I used to relate to my kids, hoping some of it would stick: "Experience is a great teacher, but a fool will learn at no other". I've forgotten its author.
Sheldon
Thanks, Sheldon.
Here I go again, philosophising .... but I am also reminded of another piece of good advice (also in audio!):
"He who asks is a fool for 5 minutes. He who does not ask is a fool for the rest of his life."
Wish I had thought of that.
Here I go again, philosophising .... but I am also reminded of another piece of good advice (also in audio!):
"He who asks is a fool for 5 minutes. He who does not ask is a fool for the rest of his life."
Wish I had thought of that.
Johan Potgieter said:
Kofi,
Shut up. 🙂 🙂 🙂
The amplifier equivalent is a low power thing but with too much gain before the volume control or too large an input signal. ...
By definition a pre-amp input for say a phono cartridge might specify 2,5 mV for full output....
Tutorial over, the BV is not a very sensitive amplifier. I notice 2V input spec. The feedback appears to be a modest 10 dB approximately (that is about 3,3 times).
Kofi, I hate to say this but I agree with Johan.
The volume pot should be such that you use atleast 70% of it. That means that you should reach max volume at about 12 o' clock (on a log pot). Just to make sure are you using a linear pot instead of a log one?
At 2V the amp is not really sensitive enough. I would consider a 500mV spec more in line.
OK-- so I've had some time to absorb the quality of sound eminating from the amp and I think I've decided that I like less NFB, not more. I disconnected the NFB entirely and I like the way that it sounds more freewheelin', but the distortion is objectionable.
So, I used a 3.5K resistor (or thereabouts) instead of the 2.45K that Jones recommends and that helped a bit, but I still feel like there's a bit of a veil over the sound-- like the "s"s are muddy.
I used just a basic metal film for the resistor-- nothing fancy, but I wanted to know if you'd recommend maybe a 5K job or a nice quality metal film or both for the job.
Also, I just used your basic electrolytics for cathode bypasses on the EL84s and I wonder if maybe Blackgates or another cap would serve me better there.
All in all, it's a nice amp, but I think it could be fantastic with some tweaking.
Advice?
Kofi
So, I used a 3.5K resistor (or thereabouts) instead of the 2.45K that Jones recommends and that helped a bit, but I still feel like there's a bit of a veil over the sound-- like the "s"s are muddy.
I used just a basic metal film for the resistor-- nothing fancy, but I wanted to know if you'd recommend maybe a 5K job or a nice quality metal film or both for the job.
Also, I just used your basic electrolytics for cathode bypasses on the EL84s and I wonder if maybe Blackgates or another cap would serve me better there.
All in all, it's a nice amp, but I think it could be fantastic with some tweaking.
Advice?
Kofi
Kofi,
In the Bevois Valley Amp the feedback resistor from the Output Transformer secondary is in parallel with the input stage cathode resistor. When you mess with the feedback resistor value you also change the bias on the input stage. That changes the input stage current and therefore the anode voltage (more or less voltage drop across the anode load resistor). Because the concertina splitter is direct coupled to the input stage you can also affect the current through and the voltage across the 2nd half of the ECC88 (6DJ8) used for the concertina. That is, when you change the feedback resistor you change the operating points of both the input stage and the concertina. Change them too far and additional distortion will result.
I assume that your regulator is set to give +285 volts. In the concertina you ideally want about 1/3 of this across each of the anode load, the tube and the cathode load - that means +95V at the concertina cathode. The ECC88 needs about -2.5 volts bias grid to cathode so the Concertina grid (and hence the input stage anode) need to be at +92.5 volts. Don't obsess over this - plus or minus 5 volts doesn't matter but plus or minus 10volts is getting to the point where you might want to trim it, plus or minus 15 or more volts and you definitely should trim it.
If the input stage anode volts is too high that means we need some more current through it (more voltage drop across the anode load resistor), so reduce the bias by dropping the cathode resistor to the next preferred value down.
If the input stage anode is too low that means we have too much current (too much voltage drop across the anode load resistor)and we need more bias, so increase the cathode resistor to the next preferred value up.
Mr Jones gave a mathematical method to calculate all of these values but using the above you can happily do it experimentally.
I was VERY happy with my Bevois Valley and like you I liked it better with less feedback. I did do all those calcs, unfortunately I can't remember what resistor values I ended up with and it was in the days before I very sensibly started keeping a work book to record that sort of info.
Aside: A work book is just an exersize book in which I jot down what changes I've made and what I thought of them, my "back of the envelope" calculations, voltage measurements etc. These days I run a work book for each project - a practice I recommend to all of you. Print off the important posts and paste them into your workbook too.
I do remember that I calculated my feedback as being 5dB. Enough to nearly halve the output impedance and distortion but not enough to impact dynamics or imaging.
I have since rebuilt the amp to the "Baby Huey" circuit which I posted elsewhere on this forum so I can't even go back and check what values I ended up with - sorry. This is an option for you as well if you really can't get the Bevois Valley sounding the way you want.
I DID use 470uF/50V Blackgate Standard Electrolytics for the EL84 Cathode Bypasses and they were definitely an improvement over standard electrolytics. I also fitted 1uF Polyphenelene Sulphide (PPS) Caps across the 470uF and I think that added some top end. I also eventually replaced the cathode resistors with "Ring of Two" Current Sources (refer to the Baby Huey schematic for these if you want to try that).
After rebuilding the amp to the Baby Huey circuit I changed the bypass caps from the 470uF/50V Blackgate + 1uF PPS to 1000uF Blackgate N series. These took some time to settle in but were better again and probably would be in the standard Bevois Valley circuit too.
Cheers,
Ian
In the Bevois Valley Amp the feedback resistor from the Output Transformer secondary is in parallel with the input stage cathode resistor. When you mess with the feedback resistor value you also change the bias on the input stage. That changes the input stage current and therefore the anode voltage (more or less voltage drop across the anode load resistor). Because the concertina splitter is direct coupled to the input stage you can also affect the current through and the voltage across the 2nd half of the ECC88 (6DJ8) used for the concertina. That is, when you change the feedback resistor you change the operating points of both the input stage and the concertina. Change them too far and additional distortion will result.
I assume that your regulator is set to give +285 volts. In the concertina you ideally want about 1/3 of this across each of the anode load, the tube and the cathode load - that means +95V at the concertina cathode. The ECC88 needs about -2.5 volts bias grid to cathode so the Concertina grid (and hence the input stage anode) need to be at +92.5 volts. Don't obsess over this - plus or minus 5 volts doesn't matter but plus or minus 10volts is getting to the point where you might want to trim it, plus or minus 15 or more volts and you definitely should trim it.
If the input stage anode volts is too high that means we need some more current through it (more voltage drop across the anode load resistor), so reduce the bias by dropping the cathode resistor to the next preferred value down.
If the input stage anode is too low that means we have too much current (too much voltage drop across the anode load resistor)and we need more bias, so increase the cathode resistor to the next preferred value up.
Mr Jones gave a mathematical method to calculate all of these values but using the above you can happily do it experimentally.
I was VERY happy with my Bevois Valley and like you I liked it better with less feedback. I did do all those calcs, unfortunately I can't remember what resistor values I ended up with and it was in the days before I very sensibly started keeping a work book to record that sort of info.
Aside: A work book is just an exersize book in which I jot down what changes I've made and what I thought of them, my "back of the envelope" calculations, voltage measurements etc. These days I run a work book for each project - a practice I recommend to all of you. Print off the important posts and paste them into your workbook too.
I do remember that I calculated my feedback as being 5dB. Enough to nearly halve the output impedance and distortion but not enough to impact dynamics or imaging.
I have since rebuilt the amp to the "Baby Huey" circuit which I posted elsewhere on this forum so I can't even go back and check what values I ended up with - sorry. This is an option for you as well if you really can't get the Bevois Valley sounding the way you want.
I DID use 470uF/50V Blackgate Standard Electrolytics for the EL84 Cathode Bypasses and they were definitely an improvement over standard electrolytics. I also fitted 1uF Polyphenelene Sulphide (PPS) Caps across the 470uF and I think that added some top end. I also eventually replaced the cathode resistors with "Ring of Two" Current Sources (refer to the Baby Huey schematic for these if you want to try that).
After rebuilding the amp to the Baby Huey circuit I changed the bypass caps from the 470uF/50V Blackgate + 1uF PPS to 1000uF Blackgate N series. These took some time to settle in but were better again and probably would be in the standard Bevois Valley circuit too.
Cheers,
Ian
Kofi Annan said:OK-- so I've had some time to absorb the quality of sound eminating from the amp and I think I've decided that I like less NFB, not more. I disconnected the NFB entirely and I like the way that it sounds more freewheelin', but the distortion is objectionable.
Yes, NFB does seem to have that effect, and there are always trade-offs. I prefer using NFB to clean up the residual sonic nastiness.
So, I used a 3.5K resistor (or thereabouts) instead of the 2.45K that Jones recommends and that helped a bit, but I still feel like there's a bit of a veil over the sound-- like the "s"s are muddy.
I used just a basic metal film for the resistor-- nothing fancy, but I wanted to know if you'd recommend maybe a 5K job or a nice quality metal film or both for the job.
Don't get ripped-off for special audiophool resistors. Any 0.5W metal film ressitor will work just as good at a fraction of the cost.
Also, I just used your basic electrolytics for cathode bypasses on the EL84s and I wonder if maybe Blackgates or another cap would serve me better there.
Fix the NFB first, then decide. It will probably sound quite good once that's done. Then if you want to spring for the Blackgates, that's fine. I don't need cathode bypasses since I prefer fixed bias anyway.
What sort of 0.1uF coupling capacitors did you use? The type ('propylene, 'carbonate, 'tetrafluoroethylene) makes a difference. Polypropylene gives a much cleaner but somewhat sharper treble than polyester. Polycarbonate gives a warm cuddly but slightly muddy sound (but they haven't been made for years). PTFE seems to give the best of both worlds to my ears - I use the 56n 500V Soviet capacitors. If you have room (they're big), you could fit 100n 500V PTFE.
Sorry for the delay in responding. Had some issues at work and me and Mrs. Annan may have an announcement to make shortly...
I used Auricaps, which I've used a before and I like. I'd be willing to hear arguments against that, though.
So, I've read and re-read all the recent responses and again, I'm so grateful for such a detailed reponse. I have tried to recalculate the feedback resistor and cathode bias resistor combo, but I'm confused about something. Well, I'm confused about a lot of things, but this one issue is at the forefront...
So, I believe the E88CC needs to see 618R from the cathode to ground to be properly biased. I thought I could just use the parallel resistor calculation (R1R2 / (R1+R2)) to come up with a configuration that would yield an appropriate resistor combination to appeal to the bias. However, Jones comes up with a calculation by using simultaneous equations that yields:
y=2.953x
...where y is the NFB resistor and x is the cathode bias resistor.
So, I'm confused about how to calculate this as the two ways I would conclude are acceptable to make this calculation may be at odds with one another.
Help.
Kofi
What sort of 0.1uF coupling capacitors did you use?
I used Auricaps, which I've used a before and I like. I'd be willing to hear arguments against that, though.
So, I've read and re-read all the recent responses and again, I'm so grateful for such a detailed reponse. I have tried to recalculate the feedback resistor and cathode bias resistor combo, but I'm confused about something. Well, I'm confused about a lot of things, but this one issue is at the forefront...
So, I believe the E88CC needs to see 618R from the cathode to ground to be properly biased. I thought I could just use the parallel resistor calculation (R1R2 / (R1+R2)) to come up with a configuration that would yield an appropriate resistor combination to appeal to the bias. However, Jones comes up with a calculation by using simultaneous equations that yields:
y=2.953x
...where y is the NFB resistor and x is the cathode bias resistor.
So, I'm confused about how to calculate this as the two ways I would conclude are acceptable to make this calculation may be at odds with one another.
Help.
Kofi
Here's a good cheat that will save you some calculation: to turn off the feedback without upsetting operating point, disconnect the output transformer end of the feedback resistor and parallel it (the resistor) with the existing cathode resistor.
Kofi Annan said:Sorry for the delay in responding. Had some issues at work and me and Mrs. Annan may have an announcement to make shortly...
I used Auricaps, which I've used a before and I like. I'd be willing to hear arguments against that, though.
So, I've read and re-read all the recent responses and again, I'm so grateful for such a detailed reponse. I have tried to recalculate the feedback resistor and cathode bias resistor combo, but I'm confused about something. Well, I'm confused about a lot of things, but this one issue is at the forefront...
So, I believe the E88CC needs to see 618R from the cathode to ground to be properly biased. I thought I could just use the parallel resistor calculation (R1R2 / (R1+R2)) to come up with a configuration that would yield an appropriate resistor combination to appeal to the bias. However, Jones comes up with a calculation by using simultaneous equations that yields:
y=2.953x
...where y is the NFB resistor and x is the cathode bias resistor.
So, I'm confused about how to calculate this as the two ways I would conclude are acceptable to make this calculation may be at odds with one another.
Help.
Kofi
If I look at the schematic in the book I see 1.2k paralleled with 2.7k. If we take 1.2*2.7/3.9, the result is 0.830kohms, or simply 830 ohms. But in order to set the proper DC conditions, we are looking for a cathode resistance to ground of 618 ohms. The trick here is that the feedback is applied to the cathode. Look at where that feedback resistor goes. It goes to the speaker, which is only 8ohms from ground. So we can ignore the resistance of the speaker. The feedback resistor is 5.1*4.7/9.8., or 2.44k. And what you really have is the cathode resistor in parallel with the feedback resistor. So, 0.83*2.44/3.27 = 0,619k - close enough. If x=830, then x*2.953= 2.45K. So that works fine.
If you want to reduce the feedback, increase the value of the feedback resistor. In order to maintain the same DC conditions, however, you have to decrease the cathode resistor so that the parallel combination of cathode resistor and feedback resistor remain at about 620ohms. Try a 5 or 10k pot. Start at 2.5k and work your way up. Note that at 10k feedback resistor, your cathode resistor will be about 660 ohms. With no feedback (infinite feedback resistor) your cathode resistor will be 618 ohms. Pick a value about halfway between 830and 618 at 5k (close enough). You may want to reduce the compensation cap in proportion to keep the same frequency roll off. A couple of steps should do, say 500pf at 5k and 220 at 10k (don't need exact values here).
Sheldon
edit: ah, just saw SY's post, what he said if you simply want to eliminate the feedback.
SY said:Here's a good cheat that will save you some calculation: to turn off the feedback without upsetting operating point, disconnect the output transformer end of the feedback resistor and parallel it (the resistor) with the existing cathode resistor.
Another great tip to put in "SY's Book of Great Tube Amp Tips"!
Sherman said:
Another great tip to put in "SY's Book of Great Tube Amp Tips"!
And to take it one step further: Kofi, forget the pot for now. If you have used the parallel resistors as per the schematic, you can do as SY suggested. But you can go halfway too. Take one of the feedback resistors and attach the speaker end to ground. Then add another 900pf or so cap across the remaining feedback resistor to keep the same high frequency roll off. Presto, about half the feedback with no change in operating point or compensation. (Actually, you can take any combination of the four resistors and have some go from cathode to ground and some go from cathode to speaker +. You just have to recalc the compensation cap so that the product of cap * feedback resistor stays about the same as the original).
Sheldon
- Status
- Not open for further replies.
- Home
- Amplifiers
- Tubes / Valves
- Kofi Annan in: "Push and Pull with Me"