Go Back   UG Community @ Ultimate-Guitar.Com > Instruments > Gear Building & Customizing
User Name  
Password
Search:

Reply
Old 12-06-2010, 12:52 AM   #1
CECamps
Schematic Junkie
 
CECamps's Avatar
 
Join Date: Feb 2010
Location: USA
Troubleshooting tone issues: a power amp analysis

*Warning: this post is long and is intended for people who are interested in technical details. If that doesn't interest you, skip this thread.*

One of the amplifiers I currently have in for repair is a silverface Fender Super Reverb. Not sure of the exact year, but this is a 45 watt master volume version. There were two separate complaints that the owner had about the amp. One of them dealt with tone and in my troubleshooting the issue I found the problem with this amp to be a great example to share and discuss here for anyone interested in designing or tweaking tube output sections, or just learning more about what's going on technically.

The complaint the owner had was that the amp's tone was somewhat offensive. It wasn't very warm and he was experiencing some really non-musical distortion that he hadn't experienced before. The distortion was occurring pretty low on the master volume.

My first inclination was that there was an issue with the power supply. My hunch was that the power supply wasn't delivering the voltage it should, thereby decreasing headroom all around and introducing harsh clipping. After pulling a schematic, the first thing I did was check voltages to see what was happening in the amp. What I found was that the opposite condition was happening. The voltages in the amp were considerably over spec, and the bias voltage on the power tube grids was 10v off from spec. This prompted an analysis of what was going on in the output section.

The Super Reverb has a dual 6L6 class AB output section. The OT has a 4k primary impedance. The schematic calls for 445v on the plates and 443v on the screens with a bias voltage of -48v. Right off the bat these numbers seemed like they were not exactly ideal for a 4k primary, but admittedly I feel this way about a lot of Fender designs I've seen. At any rate, one of the things I needed to do was sit down and analyze the output section with these parameters--which I did and will get to in a bit.

The actual voltages in the amp on my bench read 493v on the plates, 491v on the screens, and a bias voltage of -58v! Definite red flags shooting up and an almost certain case for terrible crossover distortion. So it was time to sit down and see exactly what was going on with this output section.

Analysis of actual parameters
The 6L6 datasheet only provides pentode plate characteristics graphs for screen grid voltages of 250v and 400v. So in order to analyze the actual output section in the amp I'm working on as well as Fender's specified output section, I created graphs at screen grid voltages of 491v (the actual) and 443v (Fender spec) using PSPICE. Displayed control grid voltages go from 0v to -60v in 5v increments. At the link below you'll see the graph for a screen grid voltage of 491v.

6L6 plate characteristics at Vg2 = 491v

The darker green dashed line is the class B load line, and the lighter green dashed line is the class A load line. The actual load line for our tubes is a combination of both as shown by the yellow line. The bias point of -58v is circled. The orange curve represents the 30w maximum plate dissipation for 6L6 tubes. The purple curve represents a portion of the 50w plate dissipation curve--well in excess of the maximum.

About the only thing good in this scenario is that the load line passes straight through the knee of the 0v control grid curve. Otherwise, we have some major issues. As you can see, our load line exceeds the 50w plate dissipation curve for a large portion of the signal swing! It's OK for a portion of a class AB load line to exceed max dissipation, but not by that much! In order to counteract this, the tubes have to be biased cold. Otherwise, redplating starts to occur during use. And the -58v bias point is quite cold indeed at 25mA of quiescent current. For our plate voltage of 493v, this equates to only 41% of max plate dissipation! But with the train wreck we have here, the bias point can't be brought much more positive without redplating issues under use. As a result of this, the owner is experiencing crossover distortion from the bias point being so close to class B.

So the solution is to reconfigure the power supply of the amp to deliver more appropriate voltages to the 6L6's in order that we may bias the tubes warmer and minimize the crossover distortion artifacts. But should we reconfigure to Fender's spec and call it a day? I decided to find out.

Analysis of Fender's parameters
Below is a link to my analysis of Fender's spec'd parameters straight from the schematic.

6L6 plate characteristics at Vg2 = 443v

At first glance this is looking more reasonable than our actuals. The load line still cuts through the 0v control grid knee, it no longer exceeds the 50w plate dissipation curve at any point (but still skirts it) and our bias voltage of -48v is more positive by 10v. But a closer look shows us that we haven't really increased our quiescent bias current by much. We're only at roughly 30mA now. At our plate voltage, that's only 44% of max plate dissipation! We're still really close to class B in terms of our bias as opposed to a nice happy medium between class A and class B. So in actuality, this isn't much of an improvement. And from the looks of it, the bias is probably cold for the same reasons as it was in our actual amp. We're still at risk for audible, non-musical crossover distortion in this configuration. There's got to be a better way to do this!

Analysis of modified parameters
What needed to be done on this amp was a better output section design. So I set off to find the right voltages in order to achieve a more ideal scenario. I had to take into consideration the specified voltages of the rest of the amp and I was also relegated to the 4k OT primary. With that, I set out to play with curves at varying screen grid voltages an decided on a plate voltage of 400v, screen voltage of 400v, and a bias voltage of roughly -39v. Below is a link to my analysis at those parameters.

6L6 plate characteristics at Vg2 = 400v

In this scenario, the load line still passes through the knee of the 0v control grid curve albeit slightly above center. This is OK though as our screen stoppers force the load line in that direction under operating conditions anyway. We're also well below that 50w dissipation curve now and we don't even exceed the 30w curve by much anymore so our tubes are running cooler. Now we can bias them warmer!

The bias point here is right at 21w dissipation which is 70% of max. That corresponds to 53mA of quiescent current which is over double our problematic scenario. We've significantly expanded our range of class A operation and thereby reduced crossover distortion to an acceptable level. And best of all, we've improved the tone of the amp!

I just thought that this was a great example of how significantly a tube amp's operation can be affected by voltage disparities. It's also a great example of how stock configurations aren't always necessarily ideal and with a little bit of digging we may be able to find an improved configuration. Hope you enjoyed!
__________________
Associations:
CEC Amplification
ASG Amps

Howdy, my name's Craig.

Member of the GB&C and GG&A "Who to Listen To" lists......listen at your own risk.
CECamps is offline   Reply With Quote
Old 12-06-2010, 12:55 AM   #2
oneblackened
Future Breed Machine
 
oneblackened's Avatar
 
Join Date: Oct 2007
Location: Asgaard
imma sticky this and go over it later... this looks interesting even though I know next to nothing about amps.
__________________
Current Gear:
LTD MH-400 (SD Distortion/59)
Ibanez S970 (Dimarzio Evos)
Charvel San Dimas Style 1 (MIJ) (JB/59)
Peavey 5150 II head (JJ tubes)
oneblackened is offline   Reply With Quote
Old 12-06-2010, 01:08 AM   #3
supersac
Registered User
 
supersac's Avatar
 
Join Date: Aug 2009
i love reading this stuff
specially since im thinking of building a couple small amps soon
any info is helpful
supersac is offline   Reply With Quote
Old 12-06-2010, 03:54 AM   #4
Invader Jim
Registered User
 
Invader Jim's Avatar
 
Join Date: Aug 2005
we need more of this kind of stuff in the GB&C. excellent thread craig
__________________
The Wiring Thread
Invader Jim is offline   Reply With Quote
Old 12-06-2010, 03:59 AM   #5
LeviMan_2001
Registered User
 
Join Date: Mar 2008
YES! A thread that's not about "I want to refiniz and put new pickupz for meh brotalz oh and some modzzz"
__________________
Danelectro Longhorn guitar build: http://www.ultimate-guitar.com/foru...d.php?t=1564392
LeviMan_2001 is offline   Reply With Quote
Old 12-06-2010, 04:03 AM   #6
Invader Jim
Registered User
 
Invader Jim's Avatar
 
Join Date: Aug 2005
or those lame god-forsaken build planning threads...
__________________
The Wiring Thread
Invader Jim is offline   Reply With Quote
Old 12-06-2010, 09:34 AM   #7
CECamps
Schematic Junkie
 
CECamps's Avatar
 
Join Date: Feb 2010
Location: USA
Glad you dug it. I think that there's a school of thought out there that only takes into consideration % of max plate dissipation with bias. Multiply max dissipation by .7, divide by plate voltage, and dial it in.

This is a good example of how important it is to take into consideration OT primary impedance and screen voltage--two really critical pieces of the puzzle. It also illustrates that factory design configurations can be less than ideal and should be questioned.
__________________
Associations:
CEC Amplification
ASG Amps

Howdy, my name's Craig.

Member of the GB&C and GG&A "Who to Listen To" lists......listen at your own risk.
CECamps is offline   Reply With Quote
Old 12-06-2010, 07:31 PM   #8
salgala2000
GB&C's Noob of the Year
 
Join Date: Feb 2010
Thanks Craig. Stickied, I'll read over it later.

Jim, I think it would help if you'd add this to the HUB or the OP's preamp or amp building threads so it's easier to find.
__________________
GB&C HUB. Call me Sal
Wiring thread
Pickup thread
Pedal building thread

UG's Bokononist!
salgala2000 is offline   Reply With Quote
Old 12-06-2010, 07:34 PM   #9
Invader Jim
Registered User
 
Invader Jim's Avatar
 
Join Date: Aug 2005
i'll give it a shot.
__________________
The Wiring Thread
Invader Jim is offline   Reply With Quote
Old 12-06-2010, 07:58 PM   #10
jof1029
GO HOKIES!
 
jof1029's Avatar
 
Join Date: Nov 2003
Location: Atl, GA
Hmm, thats really interesting. Was the amp built out of spec like that, or was there a drift to such high voltages over time?

Im also interested in how the amp sounded at spec as opposed to your inproved redesign. I mean, was it so far off that it still sounded really harsh or just a bit off?

The graphs are too damn hard to read on my phone, I'm definately going over this later because it is really interesting. Guess its not something I would have thought to question.
__________________
I hope you meet someone your height
So you can see eye-to-eye
With someone as small as you

I have donated 3920 grains of rice!
jof1029 is offline   Reply With Quote
Old 12-06-2010, 08:07 PM   #11
XgamerGt04
Registered User
 
Join Date: Sep 2005
I don't know if you know this but you can make spice plot the maximum dissipation curves as well.

In LTSPICE if you add a new plot the equation would be as follows

(Watts to plot)/max(V(anode),100)

You have to set to use the max function at the bottom because at very high currents in this area spice can act a little funky.

Very good article otherwise though. A lot of amps from the factory are built less than ideally. Its all about cutting cost and saving money.
XgamerGt04 is offline   Reply With Quote
Old 12-06-2010, 08:57 PM   #12
MonkeyLink07
UG's Fire Gym Leader
 
MonkeyLink07's Avatar
 
Join Date: Dec 2008
Location: Vermont, USA
I got like 80%-85% of that, seems pretty cool, I guess that's why people will pay so much for "handwired" and "custom built" as they really are better designed and more reliable. I still have a lot to learn, but I'll get there eventually.
MonkeyLink07 is offline   Reply With Quote
Old 12-06-2010, 09:12 PM   #13
blandguitar
Retired Posting Machine
 
blandguitar's Avatar
 
Join Date: Feb 2009
Orrin, what don't you get?

Craig, great stuff, do you know what caused the voltages to be above spec by ~50v?
__________________
Quote:
Originally Posted by Invader Jim
thanks jared

Wah
Need Wiring Help?
Ultimate Threads

Rydw i'n dysgu Cymraeg.
blandguitar is offline   Reply With Quote
Old 12-06-2010, 09:18 PM   #14
XgamerGt04
Registered User
 
Join Date: Sep 2005
I'm willing to bet the wall voltage caused a lot of problems. Most older transformers were designed with a 110V primary, so now that wall voltages are more regulated and closer to 120 you have about a 10% higher voltage out of the secondary.

So 445V, 10% higher is roughly 490V. Give or take 5 or so volts just for differences in current draw etc.
XgamerGt04 is offline   Reply With Quote
Old 12-06-2010, 09:28 PM   #15
blandguitar
Retired Posting Machine
 
blandguitar's Avatar
 
Join Date: Feb 2009
That makes sense, can the change of the magnetic field of the laminations in the PT affect that as well? Or would the field be generally smaller and less likely to cause such an issue?
__________________
Quote:
Originally Posted by Invader Jim
thanks jared

Wah
Need Wiring Help?
Ultimate Threads

Rydw i'n dysgu Cymraeg.
blandguitar is offline   Reply With Quote
Old 12-06-2010, 09:34 PM   #16
XgamerGt04
Registered User
 
Join Date: Sep 2005
The laminations aging would probably lead to a little more voltage loss as the magnetic losses increase in the transformer, but nothing like having the higher primary voltage.

A 10% increase is pretty big when you look at a roughly 3.5-4:1 ratio for the secondary.
XgamerGt04 is offline   Reply With Quote
Old 12-06-2010, 10:05 PM   #17
CECamps
Schematic Junkie
 
CECamps's Avatar
 
Join Date: Feb 2010
Location: USA
Quote:
Originally Posted by jof1029
Hmm, thats really interesting. Was the amp built out of spec like that, or was there a drift to such high voltages over time?


The power supply caps have been replaced at some point, along with some of the resistors. But the values were all to spec. I cannot tell if the PT is original, but that's where the voltage disparity stems from. It's putting out more voltage than is called out.

Quote:
Originally Posted by jof1029
Im also interested in how the amp sounded at spec as opposed to your inproved redesign. I mean, was it so far off that it still sounded really harsh or just a bit off?


Well, this was the primary curiosity that led to me analyze the stock config. And as I mentioned, the difference between Vg2 = 491 with -58v bias and Vg2 = 443 with -48v bias was not significant. Both leaned very heavily toward the class B side with 41% and 44% plate dissipation respectively. It kind of leads one to wonder if a lot of the bad rep silverface amps get is due in part to PA designs that leave a lot of room for improvement.

Quote:
Originally Posted by jof1029
The graphs are too damn hard to read on my phone, I'm definately going over this later because it is really interesting. Guess its not something I would have thought to question.


I think the majority of folks would not question a big name manufacturer's design. Sometimes they nail it and sometimes they don't. Only way to find out for sure is to question what they've done. Glad this piqued your interest!

Quote:
Originally Posted by XgamerGt04
I don't know if you know this but you can make spice plot the maximum dissipation curves as well.


Yeah, I know it can be done. Unfortunately my implementation in PSPICE errors out on me. The non-programmer in me shrugs it off and takes an extra minute to hand draw it.

Quote:
Originally Posted by MonkeyLink07
I got like 80%-85% of that, seems pretty cool, I guess that's why people will pay so much for "handwired" and "custom built" as they really are better designed and more reliable. I still have a lot to learn, but I'll get there eventually.


Well, the silverface Super Reverb is actually handwired from the factory. And there are boutique builders out in the amp world who build handwired and "custom" clones of classic amps right to the manufacturer spec. So the method of construction doesn't exactly delineate the design quality of the circuit.

Quote:
Originally Posted by blandguitar
Craig, great stuff, do you know what caused the voltages to be above spec by ~50v?


Thanks, and no I don't know for sure. Like XgamerGt04 stated, the PT is the likely culprit. But it may not necessarily be due to the intended mains voltage (although that could very well be playing a role). I frequently use PT's new from Hammond that put out over spec voltages when the proper mains voltage is applied. In fact, a PT in an amp I just recently built was delivering somewhere around 35 volts over what it should have!
__________________
Associations:
CEC Amplification
ASG Amps

Howdy, my name's Craig.

Member of the GB&C and GG&A "Who to Listen To" lists......listen at your own risk.
CECamps is offline   Reply With Quote
Old 12-06-2010, 10:12 PM   #18
XgamerGt04
Registered User
 
Join Date: Sep 2005
Quote:
Originally Posted by CECamps
Thanks, and no I don't know for sure. Like XgamerGt04 stated, the PT is the likely culprit. But it may not necessarily be due to the intended mains voltage (although that could very well be playing a role). I frequently use PT's new from Hammond that put out over spec voltages when the proper mains voltage is applied. In fact, a PT in an amp I just recently built was delivering somewhere around 35 volts over what it should have!


Manufacturing standard for transformers is usually either +/- 5% or +/- 10% as far as turns and other things go so its probably the tolerance of the PT there as well. Hammond also tends to be a bit odd about their designs. They put 120V on some transformers, when 115V will get you the actual output. In any case, the larger voltages usually aren't a huge issue and nothing that you can't find a way to fix.

As far as the maximum dissipation curves, it took me a while to figure it out. SPICE gets really bitchy when you have very high currents compared to voltages so it freaks out. That is why I usually just start the curve at 100V.
XgamerGt04 is offline   Reply With Quote
Old 12-06-2010, 10:22 PM   #19
CECamps
Schematic Junkie
 
CECamps's Avatar
 
Join Date: Feb 2010
Location: USA
Quote:
Originally Posted by XgamerGt04
Manufacturing standard for transformers is usually either +/- 5% or +/- 10% as far as turns and other things go so its probably the tolerance of the PT there as well. Hammond also tends to be a bit odd about their designs. They put 120V on some transformers, when 115V will get you the actual output. In any case, the larger voltages usually aren't a huge issue and nothing that you can't find a way to fix.


Yes, most of the time it is a non-issue.

Quote:
Originally Posted by XgamerGt04
As far as the maximum dissipation curves, it took me a while to figure it out. SPICE gets really bitchy when you have very high currents compared to voltages so it freaks out. That is why I usually just start the curve at 100V.


You know, I just went into PSPICE to try once again adding this trace. All the times I was trying before, I was using the wrong output variable for plate voltage in my function! Haha, I just got it to work.

Guess it just wanted to be finicky.
__________________
Associations:
CEC Amplification
ASG Amps

Howdy, my name's Craig.

Member of the GB&C and GG&A "Who to Listen To" lists......listen at your own risk.
CECamps is offline   Reply With Quote
Old 12-06-2010, 10:49 PM   #20
XgamerGt04
Registered User
 
Join Date: Sep 2005
Yeah spice is very picky about things. If you get it a bad input you get bad output, but if you know what your doing SPICE is about accurate as actually building most circuits. Since it was written for IC development it has to be extremely accurate with good inputs.
XgamerGt04 is offline   Reply With Quote
Reply


Thread Tools Rate This Thread
Rate This Thread:

Posting Rules
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

vB code is On
Smilies are On
[IMG] code is On
HTML code is Off
Forum Jump



All times are GMT -4. The time now is 09:22 PM.

Forum Archives / About / Terms of Use / Advertise / Contact / Ultimate-Guitar.Com © 2014
Powered by: vBulletin Version 3.0.9
Copyright ©2000 - 2014, Jelsoft Enterprises Ltd.