How do I determine if my NFB is optimal?

[sstudley]

Hello Folks,
I have a Chinese 5F6A kit amplifier all assembled and working. For the most part. However, it still sounds muddy to me and breaks up way too soon, imo. I really hate to sound like an idiot here, but I really don't trust my ears. Does anybody have a simple formula/technique for measuring and optimizing NFB?

I apologize for asking what are likely fundamental questions, but I really don't understand why switching primary OT plate leads made such a significant difference in audio quality?

I should mention (again) this kit came with no, I mean NO instructions of any kind. So just a simple understanding of transformer leads was left for me to figure out. Easy for the PT, not so much for the OT. Primary leads on OT red, brown and blue. Pretty sure I have those right. Secondary, black, white, green and yelo. For 8 ohm speaker load I selected first black & yelo, now I am using white and yelo and after making that change I had to swap the plate leads again to deal with what I believe is referred to"motor-boating"?

I really do appreciate the help!

Steve

[shooter]

guessing yours looks sorta like this

https://el34world.com/Hoffman/files/Hoffman_5F6A.pdf

motorboating can be anything from ps issues to parasitic oscillations

your OT secondary's are probably multi-tap speaker impedance, probably should web surf with whatever #'s you have on the OT.

did the bias set properly?
are the PS voltages inline with schematics you've looked at or values the kit says?

[sstudley]

Thanks Shooter!
yes, PS voltages all checked out, very accurate to the schematic. I have done some web surfing with resistance values and through elimination, came to the conclusion black or white was common. I'm not sure why I chose to go with black initially. I have several OT in my shop with white commons. Choosing yelo for 8 ohms was based on most common color to tap anecdotally.

Right now I have my scope connected to V3 grids trying to compare the two. I'm feeding a small 1kHz sine wave to the amp. I have to admit the schematic, albeit simple, is losing me when it comes to inverting the NFB signal. I am trying to verify the NFB is actually negative and if so, how I might tweak it to make for a better sounding amp.

as always, thank you for your help!

[sstudley]

Sorry, I neglected 2 of your questions. Firstly - yes, the pdf, schematic and layout is very close to my kit. Very very minor variation power supply filtering. Secondarily, The bias seemed to be in range to me. The numbers tossed around were between 50-60 VAC and my specific voltage was 54 VAC, if memory serves.

[2deaf]

If you have a dual trace 'scope, you can tell if the signal at the treble wiper is in phase with the speaker or out of phase.  You would want the two signals to be in phase.

But first, you need to get that OT figured out.  What are you using to generate the 1KHz signal?

[shooter]

"motor-boating"?

Glad you have a scope, invaluable for chasing that gremlin!

I've found it small enough that it will hide in the background noise floor, sometimes frequencies closer to tremolo 5-10hz, typically it will be everywhere 

happy hunting 

[tubeswell]

When you say 'muddy' and 'breaks up way too soon', does it have any cleans? (What number on the 'Volume' dial?).


Could be you haven't grounded the pre-amp filter cap.


Or it could be cold bias on the 6L6s.


Or a bad tube.

[sstudley]

But first, you need to get that OT figured out.  What are you using to generate the 1KHz signal?

I have rather runofthemill function generator. Very reliable "GWInstek GFG-8219A" my scope is a Rigol DS1102E, dual trace.

[sstudley]

When you say 'muddy' and 'breaks up way too soon', does it have any cleans? (What number on the 'Volume' dial?).


Could be you haven't grounded the pre-amp filter cap.


Or it could be cold bias on the 6L6s.


Or a bad tube.

I think we can safely rule bad tube out. I have put in 3 sets, 2 very mediocre and now running a very nice matched pair of NOS JAN RCA's, I think they black tops, if memory serves...

Cold bias? I will need to look that up and learn!

I will double-check the grounding of the preamp filter cap, but I am pretty certain it's grounded.

The quality isn't bad up to about 3-4, 4 and over is nasty. That is playing a Fender Strat "Eric Johnson" set to 7, single coil as I'm sure everyone knows...

The signal is the lowest I can set my GWInstek function generator to. 1kHz, sine wave, nothing fancy.

[shooter]

4 and over is nasty

should be able to chase your 1k through the gain stages and see if it's clipping bad.

jack your strat into your oscope, strum n strum n strum.... to get a "reference" amplitude for the 1khz, set the sig-gen for that and follow it through the amp, stay outta the TS, things get wonky, I avoid the PA plates and pre-amp grids.

[2deaf]

I have rather runofthemill function generator. Very reliable "GWInstek GFG-8219A" my scope is a Rigol DS1102E, dual trace.

That'll work.  With amp off and capacitors drained, disconnect the OT RED CT from the power supply and disconnect all secondary leads.  Connect function generator to BRN and BLU primary leads (they don't need to be disconnected from the sockets).  Post the AC voltages between:

BLK-GRN
BLK-YEL
BLK-WHT
GRN-YEL
GRN-WHT
YEL-WHT

[PRR]

Disconnect NFB from OT. It should play "OK", but brash or boomy.

Connect NFB. Does gain go down or up (or howl). It should go down, in this case to about half. That's right.

Moving NFB from 4 to 8 to 16 made Jim Marshall rich. Try all options. It really depends a LOT on the speaker and the style.

[sstudley]

I have rather runofthemill function generator. Very reliable "GWInstek GFG-8219A" my scope is a Rigol DS1102E, dual trace.

That'll work.  With amp off and capacitors drained, disconnect the OT RED CT from the power supply and disconnect all secondary leads.  Connect function generator to BRN and BLU primary leads (they don't need to be disconnected from the sockets).  Post the AC voltages between:

BLK-GRN  = 2.0mVac
BLK-YEL   = 0.5 mVac
BLK-WHT = 1.2 mVac
GRN-YEL  = 1.6 mVac
GRN-WHT= 2.5 mVac
YEL-WHT = 0.7 mVac

These voltages would start out higher, 2.0-3.0 mVac and then as I would wait a few seconds, I gave it a consistent 30 seconds it would drop. I go the feeling these numbers could have gone a little lower. But I'm thinking you are looking for relative values here anyway?

[sstudley]

Disconnect NFB from OT. It should play "OK", but brash or boomy.

Connect NFB. Does gain go down or up (or howl). It should go down, in this case to about half. That's right.

Moving NFB from 4 to 8 to 16 made Jim Marshall rich. Try all options. It really depends a LOT on the speaker and the style.

The speakers I'm using are (2) Celestion 12-65 15 ohms, unused but NOS, supposedly from the 80's-90's. Obviously not the usual (4) x 10" Jenson Special 8 ohms.

[2deaf]

These voltages would start out higher, 2.0-3.0 mVac and then as I would wait a few seconds, I gave it a consistent 30 seconds it would drop. I go the feeling these numbers could have gone a little lower. But I'm thinking you are looking for relative values here anyway?

Crank that function generator to full output at 1KHz and try again.

[sstudley]

These voltages would start out higher, 2.0-3.0 mVac and then as I would wait a few seconds, I gave it a consistent 30 seconds it would drop. I go the feeling these numbers could have gone a little lower. But I'm thinking you are looking for relative values here anyway?

Crank that function generator to full output at 1KHz and try again.

BLK - GRN  = 206.1 mV
BLK - YELO = 84.6  mV
BLK - WHT = 206.7 mV
GRN - YELO = 290.8 mV
GRN - WHT = 412.8 mV
YELO - WHT = 122.4 mV

[2deaf]

Green is common
Black is four Ohms
Yellow is eight Ohms
White is sixteen Ohms

[sstudley]

Green is common
Black is four Ohms
Yellow is eight Ohms
White is sixteen Ohms

Thank you 2Deaf,
Green is common, does that sound strange to you?
Would the fact that I had been using black & white as common, stand to reason for the lousy sound quality?

[shooter]

lousy sound quality?

will definitely change things, it's "designed" to operate into a specific load, you've halved that load casing the PA tubes to operate "outside" it's designed parameters.

will that fix it, try it n see 

[sstudley]

will that fix it, try it n see 

Well, those changes (getting the OT taps sorted out) have definitely helped the raunchy distortion, no question about it! I turned it up to 5, with no significant distortion! Before it would begin to sound terrible just nearing 4!

sigh, it still sounds a little muddy and flat. But better, no doubt. I’ll keep at it.
As always, thank you muchly for your help!

Now I’m off to a build some effects pedal!    I bought 2 kits from from guitar pedal parts. and get this, I'm not even kidding, the first kit has NO schematic or instructions! I am on a roll! I bought both a digital reverb and fuzz kits. Maybe they put the instructions for both in one, I only opened one so far.

{EDIT- untangled quote --PRR}

[2deaf]

Green is common, does that sound strange to you?
Would the fact that I had been using black & white as common, stand to reason for the lousy sound quality?

I've seen Marshall's with violet and even brown for common.  Sovtek's with yellow for common.  Green doesn't seem all that strange to me, especially since green has been associated with ground throughout colored-wire history.

That amp is probably meant to have a 4K primary.  WHT-YEL or BLK-YEL was probably reflecting 22K to 47K with an eight Ohm load.  I would guess that those extreme impedances would result in lousy sound quality.

Are you interested in how I decided which color was what?



 

[2deaf]

Getting back to the NFB question.

Inject a signal into one of the inputs like you did before.  If you don't have a dummy load for the speaker output, you get to listen to the test signal. 

Turn the presence control to the lowest setting.

Measure the AC voltage at the end of the feedback resistor (57K? 27K?) that goes to the presence control.

Measure the AC voltage at the treble control wiper.

These two voltages should be in phase.  The voltage at the wiper will be bigger than the voltage at the NFB resistor.  The ratio of the two voltages gives you a good estimate of how much NFB you have.  If the voltage at the NFB resistor is 50% of the voltage at the wiper, the output at the speaker will be around half the output without NFB.  If the voltage at the NFB resistor is 80% of the voltage at the wiper, the output at the speaker will be around 20% of the output without NFB.

How much NFB is optimal is a matter of personal taste.       

[shooter]

it still sounds a little muddy and flat

Look at the link schematic I posted,
Is your V1 close?  2 inputs?
If so you should hear a difference, which one is closer to your liking
(when deciding, make sure the "other volume" is 0)

[jjasilli]

https://www.aikenamps.com/index.php/what-is-negative-feedback

[sstudley]

Green is common, does that sound strange to you?
Would the fact that I had been using black & white as common, stand to reason for the lousy sound quality?

Are you interested in how I decided which color was what?

Absolutely!

[2deaf]

The largest voltage is 413mV which is between GRN and WHT.  The largest voltage is between the ends of the entire secondary winding, so we know that either GRN or WHT is Common.  The secondary taps vary by a factor of 1.414, i.e. the voltage from Common to the 8r tap is 1.414 times the voltage from Common to the 4r tap and the voltage from Common to the 16r tap is 1.414 times the voltage from Common to the 8r tap.

The 16r tap is at 413mV, so the 8r tap is going to be around 413mV divided by 1.414 which is 292mV.  Now we look for a voltage like that between GRN or WHT and the other wires because one of them is Common and the other is 16r.  GRN to YEL is 291mV and WHT doesn't have a voltage to anything that is close to 292mV.  So now we know that GRN is common and YEL is 8r.

By the process of elimination, BLK would have to be the 4r tap if this is a 4, 8, 16 secondary.  We can test this by taking the 8r voltage and dividing it by 1.414 to get a 4r voltage of 206mV.  GRN to BLK is indeed 206mV.

1.414 is the square root of two and it pops up frequently in electronics.  When we went from the 16r voltage to the 4r voltage, we divided the 16r voltage by 1.414 then divided that by 1.414.  This is the same as dividing the 16r voltage by 1.414 squared.  1.414 squared is the square root of two squared, which is just plain ordinary two.  So the 4r voltage is half the 16r voltage.

The drawing says uV instead of mV, but the principle is the same.  A couple more beers and it might have been pV.       

[sstudley]

Wow, thank you 2Deaf! I really appreciate you taking the time to enlighten me. I have a ton of OT in my shop I can practice on. Many I do not know the secondary tap configuration. So I can practice and provide myself a little useful information as the same time.

ok, now I'm going to get back to the issue of NFB and overall sound quality.

thanks again! Very useful information and something I have needed to understand for quite some time!

Steve