Negative feedback pot is scratchy

[dscottguitars]

I am using a 10K pot on the negative feedback line to adjust for more or less grind.  The turning the knob is always scratchy.  Why is that?  I always thought a scratchy pot (good one) was due to DC voltage as if you didn't use a coupling capacitor.  But the signal coming from the secondary on the OT is AC, yes?

Usually I turn down the volume and don't hear it but with this recent build that does not work.  It is really loud.

Thanks,

Daniel

[sluckey]

That circuit has DC on the NFB pot.

[dscottguitars]

Is there any way to minimize the scratchiness?  I don't know why, but I thought the output was ac like the signal coming in...

[Geezer]

Usually I turn down the volume and don't hear it but with this recent build that does not work.  It is really loud.

You might have positive feedback (OT secondaries reversed). I know, folks say "it doesn't squeel like a stuck pig, so it can't be that", but that is not always the test or symptom.....the loud(er) scratchy(er) pot might be the indicator in this case(?)

Here is a test that will confirm the feedback is correct (negative) or wrong (postive)

*Disconnect the feedback wire and connect an alligator clip lead to it.
*With the wire disconnected, leave the guitar cord plugged into the amp, but unplugged from your guitar, and turn up the volume so you can hear the noise that's generated, then reconnect the wire (alligator clip) while listening closely to the noise

Does the amp get softer (less volume) or louder when the feedback wire is reconnected??

If softer, you have the OT wires correct/negative feedback is being generated.

If the amp gets louder when the FB wire is reconnected, then you have positive feedback, and the OT wires need to be reversed.

$0.02

G

[dscottguitars]

I will try that as I want to make sure it's connected correctly.

I am wondering if a capacitor between the dropping resistor and pot would work without acting like a tone control.  Anyone know for sure?

[HotBluePlates]

... the knob is always scratchy.  Why is that?  I always thought a scratchy pot (good one) was due to DC voltage ...

That circuit has DC on the NFB pot.

More explicitly:

From your inverter's 20kΩ tail resistor, the path continues through the feedback pot to 2 parallel paths to ground: the 470Ω resistor and the 47kΩ series feedback resistor to the OT secondary tap and to ground.

I am wondering if a capacitor between the dropping resistor and pot would work without acting like a tone control.  Anyone know for sure?

You'd either need a huge-valued cap to make it a low impedance to all audio frequencies. But even then it probably won't work, because a cap (like all reactive components) will cause a phase shift and will have a good chance of turning negative feedback into positive feedback at some very low frequency. It is a tricky bit of engineering to figure out if that will happen and how to correct it if it does.

The non-tricky way is to slap a large-valued cap in there (like several uF's or more) and see if the amp becomes unstable. If it does, you'll have to reduce the feedback by increasing the 47kΩ resistor until the instability stops. Depending on a number of factors, you may have to do away with feedback altogether.

A different thought is that while variable feedback seems like a useful thing, most people find full feedback and zero feedback is close enough to what they want (with maybe some 3rd intermediate amount). In that case, the pot could be replaced with a switch offering 2 (or maybe 3) set levels of feedback.

Such a switch might still pop when engaged with the amp running, so you'd want to manipulate it with the amp in standby.

[dscottguitars]

Thanks for that explanation...

I've been thinking about going to a rotary switch for different positions too.  One for very clean ~20%, intermediate for 'Bassman' clean ~12%, grind like Marshall/Orange ~5% and extra raw-less than 1%.

[dscottguitars]

G-

That test did not work.  I unhooked the feedback wire at the pot.  The guitar was very quiet with the guitar cord plugged in.  I hooked the line back up and it was definitely louder.  I switched the OT leads, but now I get huge feedback, telling me the OT wires are not correct.

Any thoughts?

[Geezer]

Don't know why the strange behavior, but if the amp gets louder when the feedback wire is connected, that's definitely positive feedback, not negative.

[HotBluePlates]

I've been thinking about going to a rotary switch for different positions too. 

You could use a rotary switch for different settings, but those switches are expensive and bulky. If you think 3 setting would be enough for you, and SPDT On-Off-On switch could be arranged to give 3 levels of feedback.

But you'd have to determine the needed resistance first.

Don't wire up the feedback level pot as you have it. Connect the 20kΩ tail resistor directly to the 470Ω resistor. Wire the Feedback pot as a variable resistor (use only the wiper and one outside leg), and connect that between the 47kΩ series feedback resistor and the junction of the 20kΩ/470Ω.

Twiddle until you find a sound you really like, then measure the total resistance (with the amp off) of the 47kΩ resistor and the pot. After writing down all 3 of your desired settings, it will be pretty easy to calculate the feedback switch setup. In the end, it will be a SPDT switch and 3 resistors.

[12AX7]

Mine does it too, but for me it's not nearly an issue. But i will say this...why are you using a 10k pot when you could use a much larger pot and get a ton more variance? I use 100k and before that i used a 250k ! The 250 at it's full resistance is like removing NFB completely, and that gives you max variance. 10k is barely noticeable. I now use 100k because i found that i rarely went much beyond that, but if it were easy to find a 150k pot i'd likely have used that instead. But i often have it fully open so i have the minimum NFB and it's often very desirable like that. I may go back to the 250k. But 10k? Man, you're missing a lot of possibilities, especially in tuning to a dead room.

[dscottguitars]

HBP:  I have a few 5 position switches I got pretty cheap off ebay from Ukraine awhile back, really good quality too.  I put one in and it works nicely except it's a HUGE pop.  So, switching to standby is a must.  I'm not sure how this affects anything, speaker or components in the amp, but the pop is scary loud.  I'm not sure which is worse...

On ebay I see some others now, 4 position for less than $2 each. I would prefer 4 settings but may consider the on/off/on switch too.  Link:  http://www.ebay.com/itm/121072531917?ssPageName=STRK:MEWAX:IT&_trksid=p3984.m1423.l2648

12AX7: What size dropping resistor are you using before the pot?  I had read that much less than 1% NFB, or a ratio of more than 100 does not make much difference.  My setup goes from 22% down to 0.82%.

[sluckey]

I put one in and it works nicely except it's a HUGE pop.

Get a "make before break" switch.

I had read that much less than 1% NFB, or a ratio of more than 100 does not make much difference.  My setup goes from 22% down to 0.82%.

What are you basing those percentage numbers on and how are you calculating?

[dscottguitars]

I'll look into the make before break switch.  Is that the same as a shorting switch?

My calculations are:  dropping resistor(Rd)/load resistor(Rl)=ratio; 1/ratio*100=%.  Or easier to: Rl/Rd*100=%

Ex:  

All the way down-
Rd: 47K;  Rl=470+10K pot=10470;   10470/47000*100=22.4%

All the way up-
Rd:47K+10K pot=57K; Rl=470;  470/57000*100=0.82%

The pot is wired in reverse to add gain when turning knob up.

[dscottguitars]

The search engine in this forum really sucks.  The amp pops when I turn it off-unless I turn off the standby first.  Can that be fixed?

[12AX7]

12AX7: What size dropping resistor are you using before the pot?  I had read that much less than 1% NFB, or a ratio of more than 100 does not make much difference.  My setup goes from 22% down to 0.82%.

I have a 33k off the 16 ohm tap with a 100k pot in series. But i did have a 250k. The 100 doesn't take NFB out all the way. I'm not familiar with other designs like fenders, but my marshall build has a very useable tone from one end of the spectrum to the other, which with a 33k resistor and 250k pot equates to as much NFB as one would ever want to virtually none and everything in between. 10k would make almost no audible difference. 100k sure does, and 250k = virtually no NFB which can be nice for some things. Just try it. Take you a couple minutes.

[dscottguitars]

My setup is different and I can't figure out how yours calculates.  Do you have a cap on that circuit like a presence control?  If your pot is wired standard, all the way up would be 33K/100K and all the way down would be 133K/0.  However, at about two thirds the way up (linear pot) you have about 67K/67K which divides the voltage in half or 50%.  I get confused when the dropping resistor is small compared to the pot and the load.  With your setup you are getting quite a bit of NFB from your speaker with the pot all the way up and then a 50% cut on down to almost nothing.  But how can the amp handle so much NFB of more than 25-30 percent?  That seems too much to me.

Mine is set up as a voltage divider starting with 22% of NFB and cutting it to 0.8%.  That's a huge difference and sufficient.  I could try to use only 100 ohms on my load resistor to see if less than 0.8% has any effect.  That would equate to 0.002%  With your setup a 90% knob turn is still roughly 8% of the signal.  There's not much turn from 90% to all the way, so that's why I wonder if you have a cap in the circuit.

Also my amp is using the 4 ohm tap which is 4 times less voltage than the 16 ohm. 

[12AX7]

Ok, i think i see what you are talking about. We're talking about 2 different things here. Look at a typical marshall NFB loop. Theres usually a 47k to 100k resistor depending on what tap. Then theres a 4.7k to ground and those 2 form a voltage divider. I'm not talking about touching the 4.7k, just making the other resistor variable. I don't have to do the math because I know how it actually works and it works well to do exactly what I'm talking about which is tp go from about what the max amount of NFB (actually a bit  more due to the 33k off the 16 tap) to none, or in my current setup with a 100k pot, maybe 75% of the way there. So the 33k and pot are in series and the pot just acts to vary it to either extreme and the 33k is just there so you can't turn the pot to the point you have too much NFB. Instead of trying to figure all this out by calculating it, just try it. We're talking 2 minutes of work to tack in another pot just to test it. Calculations don't tell you anything much about what your ear will experience.

[dscottguitars]

I looked at the 1959T schematic that shows a 47K (100K) like you said and a 4.7K pot with 0.1uF cap from the wiper to ground.  That's a presence control.  Even if you don't have the pot and cap there you still have a 4.7K resistor as your load resistor.  That means with your dropping resistor plus the 100K pot you have a total of 133K divided by 4.7K for the voltage divider.  That works out to be a ratio of 1:27 or about 3.7% of NFB.  Before you had a 250K which gave you a ratio of 1:60 or 1.5%.  Why did you change?

My setup gives me only 0.8%; half of your amount you changed.  And I think it's definitely a good sound.

And you are wrong about the math. It tells you everything, it tells you what parameters are safe, and let me figure out what you have compared to mine. (and that doesn't take into consideration your tap is off the 16 ohm and mine is off the 4 ohm which is less NFB)  It took me a bit of trials to settle on this setup and figure out what I wanted to achieve; a sweep of not more than 25% (recommended as safe) to less than 1% (quite raw and almost no NFB). 

If you change that pot to a 500K you'll be at the point I am.  You could also change the 4.7K to a 1K and get the same result.

My idea of using a variable NFB loop was from understanding this page and how it works: http://www.aikenamps.com/NegativeFeedback.htm

[12AX7]

Ok, math aside i'm really not sure what you have there or what you are saying. I just know this. You say a 500k pot would give me what you have. That would be way past the point of no NFB. Are you saying you have no NFB with the 10k fully turned and that even at 50% of the pot's range theres still none?  Because I have no NFB with a 250k which makes me think your math has got to be off somehow. A 500k would do nothing past the 250k point or thereabouts, so whatever the math is you are using it doesn't seem to make sense in actual practice.

[HotBluePlates]

It's not just about a percentage, which we could rearrange and interpret (more conventionally for feedback examples) as dB of feedback.

If you want to try to calculate what you need to do regarding feedback before trying, then read all of Part 1 of High Fidelity Circuit Design by Crowhurst and Cooper on Millett's site.

May God be with you on that journey, as attempting to calculate dB feedback, phase and stability margin makes me dizzy. I don't think I've worked through a single complete amp in this way yet.

Or you can use existing known-good amps as an example, try and adjust-on-test. "AOT" may also require a scope and test signals, because crazy things can happen outside the audio range that are not there for you to hear, but exhibit themselves a fizzy/fuzzy sound or other odd behaviors.

[dscottguitars]

Are you saying you have no NFB with the 10k fully turned and that even at 50% of the pot's range theres still none?

Yes and no.  I have virtually no NFB with the pot all the way up.  At half way up, I have about 12% which is close to the Bassman specs.  At 3/4, I have about 5% which is what Marshall uses.  Your 3.7% is because you added a 33K resistor in the circuit.

Also, it does not matter the size of the pot you use.  It all depends on the ratios of resistance between the pots and resistors.  My math is correct, trust me.  You can achieve a certain percent of voltage reduction with a number of different pots and resistors:

Examples: from the resistor in front of your pot to the one going to ground-

50K, 100K pot and a 15K is the same as a 5K, 10K pot and a 1.5K or 500K, 1M pot and a 150K.  They all drop the voltage 90% or down to 10% of the original, depends on how you look at it.

Your setup is like this: 33K, 100K pot and a 4.7K give a ratio of  133K/4.7K=28.3  That means you have a voltage drop of 28.3:1.  Or 1/28.3=.035  To get a percent you multiply by 100 to get 3.5% ( I think I used 5K in my previous post to get 3.7%, rounding off)

I hope this makes sense...

[12AX7]

Well, math aside i'm simply describing how it works with my amp. Yours is apparently very different.

[HotBluePlates]

...  I have virtually no NFB with the pot all the way up.  At half way up, I have about 12% ...

You do realize that the way you have the NFB pot wired, as you turn it up for more negative feedback, you're also making the 470Ω resistor larger and increasing feedback 2 ways, right?

At the minimum setting, you have the full resistance of the 10k pot and the 47k resistor in series as one half of the feedback divider (the series resistance), with 470Ω to ground as the other half (the shunt resistance).

At maximum, you have 47k as the series resistance of the feedback divider, and 10k + 470Ω as the shunt part of the feedback divider. The 10,470Ω resistance to ground is also changing the resistance the long-tail is standing on, and may alter its operation somewhat.

So I hope you're not calculating assuming that the 470Ω to ground is a constant value, because it's not the way you have the schematic drawn (and presumably, wired).

You seem to be ignoring that the voltage present at the speaker terminals (which depends on output power and load impedance for the tap used) is the quantity being divided by the feedback resistor divider, and the resulting quantity of voltage determines the amount of feedback. Resistor ratios by themselves are meaningless without a knowledge of how much source voltage is being divided and fed back.

Anyway, all that said, if it works for you, then cool. Pick your best-sounding 3-4 arrangements, wire up your switch.

[dscottguitars]

I have the pot wired backwards so all the way up is less NFB.  And I do know that it is dividing the voltage off of the source, which is about 14-15 volts on the 4 ohm tap at about 55 watts.  I'm being very general about my description.  I would assume anyone discussing the NFB circuit knows what it's coming from.  But I didn't know about the change wrt to the tail resistor.  I'm not sure how that affects things, but the amount I am dividing seems to work and I read somewhere that more than 25% of the signal fed back is pushing things into oscillation.  That's why I keep it where it's at ~22%.  My understanding of all this comes from Aiken Amps' technical pages.

[dscottguitars]

I put in a make before break switch and that did not work.  It's still a loud pop when changing the positions even with the volume down but maybe not quite as loud as the regular switch, it's hard to tell.