From mgarvi–(at)–anix.com Sat Jan 11 13:41:58 CST 1997
Article: 34038 of alt.guitar.amps
Path: geraldo.cc.utexas.edu!arlut.utexas.edu!news.io.com!fnord!news.thenet.net!news1.best.com!www.nntp.primenet.com!nntp.primenet.com!nntp.uio.no!news-feed.inet.tele.dk!news.algonet.se!hammer.uoregon.edu!news-xfer.netaxs.com!panix!news.panix.com!not-for-mail
From: mgarvi–(at)–anix.com (Mark Garvin)
Newsgroups: alt.guitar.amps
Subject: Re: My SF-BF experiment
Date: 11 Jan 1997 07:47:47 -0500
Organization: PANIX Public Access Internet and Unix, NYC
Lines: 279
Message-ID: <5b825j$lo--(at)--anix2.panix.com>
References: <32CFD4A4.667--(at)--hem.ucalgary.ca>
NNTP-Posting-Host: panix2.panix.com

> Scott Hinman writes:
>The experiment: stepwise conversion of my silverface to
> blackface specs.

Hi Scott,

Thanks for taking the time to do such a meticulous job and for posting
the results! And compliments to Kathy with the golden ears! (Really!
She could reliably tell the difference between grounded cathodes and
bypassed 150 ohm resistors? I’m impressed!)

I’ve spent lots of hours trying to answer tech questions on a.g.a.,
and once in a while it actually seems worthwhile. I’ll try to
explain some of my own thoughts on the Fender circuit. Maybe this
will help to put your test results in perspective.

>The amp: Early ’69 Pro-Reverb (all the pots are 68 except for one 69).
>
>The guitar: (Well, someone might think its relevant). 76 Gretsch
> superaxe. The pickups are Gibsons, don’t know what

Actually could be relevant to high frequency content. Good to know.

>Step two: Install three wire cord with ground to chassis.

Good! I should remember to suggest this more often.

>Step three: Added a second bias adjustment pot in parallel with
> existing one to allow individual bias adjustment of
> each power tube. (Not in BF circuits, but I thought
> it was a good idea anyway).

Another good addition. I’m surprised more boutique amps don’t
do this.

>Step four: Replace the plate resistors in the phase inverter circuit
> from the 47k / 47 k SF values to 82 k / 100 k BF values.
> Also changed the 100k bias feed resistors to the 220k
> BF values. Reinstall chassis in amp, and play.
>
>This is the least scientific of the tests I did, since I couldn’t
>quickly go back and forth between the BF and SF phase invertor
>circuits. The effect of the change was small, if any. At the time
>I wrote “I think its a little louder. It seems to be a bit cleaner,

Mathematically, this change should not make a huge difference in
gain with a 12at7, though I’ve heard a fairly pronounced difference
sometimes. I wrote a computer program to model the gain equation
for that circuit with various tubes and component values. Some of
the results are surprising.

Here’s something interesting about that circuit: Though the 12at7
has lower ‘Mu’ than a 12ax7, it can actually have higher overall gain
in Fenders with the 47k/68k circuit.

The Mu is sometimes thought of as being synonymous with gain, but
that’s not exactly true. Mu is transconductance times plate resistance.
Sorry that I can’t explain in depth, but when driving low impedances
(like the 47k/68k resistors in parallel), the lower plate res of the
12at7 works in its favor.

Tim Tube made some interesting comments regarding sound of 12at7′s
vs 12ax7′s as phase inverters recently. The general idea is that
12at7′s are better drivers, and the lower plate resistance can
help to avoid treble loss due to local capacitance. (more later)

>Step 5: Changed the resistor values in the power supply chain.
> In my amp, they went choke to 4.7k to 10 k, and in the
> BF circuits (and seemingly all the SF circuits except
> the AB 668) they go choke to 1k to 4.7k.
>
>This should change the plate voltages to both the preamp and phase
>invertor. If there was any diffence in sound from this, though,
>it was completely inaudible to me.

Agreed. May change the headroom a bit. May eliminate a bit of
supply sag, too.

>Step 6: Take a box with 8 SPST switches in it. (Just happened to
> have one). Rig it to be able to a) switch the existing
> 2000 pF grid capacitors (not present in BF circuits) either
> in or out of the circuit. b) switch the non-polarized cap
> (also a SF addition) coupling the two power tube cathodes
> in or out of the circuit, and c) switch two 220uF caps into
> the circuit from the power tube cathodes to ground
> – thus shunting the SF 150 ohm cathode resistors.
> (These caps aren’t BF entities, they were suggested
> by Mark Garvin – more on his reasoning later)
>
>This is where I’m going to get accused of having tin ears.

Not by me, though my own test results may have been a bit different.

>First, neither Kathy nor I could hear any difference whatsoever
>on taking the 2000pF grid caps out of the circuit. We went
>back and forth several times – and notta. You can even switch
>while playing a sustained chord, and there ain’t no difference.

Recently both Tim Tube and Fat Willie agreed (Eek!) that there
would be a big difference when doing this. Nostradomas warned
that the alignment of opinions of Tim and Willie were one of the
signs of the end of the world, but we’ll ignore that for now.

The aforementioned 12at7 should have a plate resistance of
around 10k or so. The theoretical corner frequency (point
where highs start rolling off by -3db) can be calc’d as:

f = 159000 / (C * R) where C is in microfarads.

So the ‘RC’ (resistor/Capacitor) filter in this case has a
corner at:

f = 159000 / (.002 * 10k) the 10k is the plate resistance.

so….

f == 8000hz Highs start rolling off at 8Khz…pretty high!

Note that a 12ax7 has a plate res around 62K ohms, so changing the
phase splitter tube could make a big difference, though the plate
load res’s would also start figuring into the equation (they would
be paralleled with the plate resistance), so the value would be
around 35k to 40k.

With hifi/stereo gear, the 8Khz range is indeed audible, but most
guitar speakers are not putting out much sound up there. They start
to roll off sharply around 5Khz. Another thing to keep in mind is
that fuzz boxes and overdrive preamps can sound really harsh if the
high end is over-accentuated.

Given that the equations above are pure paper (in this case
computer screen), they are not always to be believed, but my
own listening tests confirm your opinions. Yes, once in a
while I actually disagree with Tim! But that’s what keeps things
interesting.

You also mentioned that you were testing with humbuckers. This
could also have some bearing, and the particular speakers in your
amp, etc.

>Next, switch the cathode coupling cap out of the circuit, and the
>two 220uF shunt caps into the circuit. BIG DIFFERENCE here.
>Kathy’s words “It just sounds more focussed”. At lower volumes
>the highs stood out more, and the bass seemed less muddy. Definitely

I have to admit that I’m surprised that you noticed that BIG a
difference here, though I expected some change. And I expected it to
be more noticeable at high volumes. However, I’ve always thought that
the cap bypass method (rather than the resistors themselves) was the
main problem with this circuit.

Here’s why…a bit of background first:

Class A cathode biased amps usually hook the two cathodes together and
provide a single resistor from their junction to ground. The shared
resistor provides a set amount of current to both output tubes. Since
the tubes operate in push-pull, when one starts to conduct more, the
other starts to conduct less: a seesaw effect. There is no need to
bypass the resistor because the current thru it does not change.

This works fine until one of the tubes turns off completely (this
means that the output circuit has gone into class AB). Kinda like
one end of the seesaw hitting the ground–no more seesaw effect. The
other tube is trying to pull more current, but now it must pull it
thru the cathode resistor cause there is no more ‘give’ from the other
tube. So from the point where one of the tubes turns off, the cathode
resistor will ‘enter the circuit’ and will limit current thru the
other tube. Bypassing the cathode resistor with a large cap solves
the problem, as it provides a low impedance AC path around the cathode
resistor.

That’s why class A amps with shared cathode resistors do not need
cathode bypass caps. Class AB amps do.

Now for the Fender circuit:

Consider what the non-polarized (np) cap is doing in the normal Fender
circuit: it provides a virtual short between the cathodes, but *just
for ac signals* (caps block DC). So imagine what the circuit would
look like if you shorted the two cathodes together: Just like the
class A circuit described above!

The problem, of course, is when the output circuit goes into class
AB (when the volume is cranked). Same thing happens as described with
the class A amp above.

Another problem is that the value of the cap that Fender used is
too small to bypass low freq’s effectively. So at low frequencies,
the 150 ohm resistors will limit current.

And that’s why I suggested using the two separate, large-value bypass
caps in your experiment.

>Step 7: Rig the switches to short the power tube cathodes directly
> to ground, shunting the now parallel combinations of 150
> ohm resistors and 220uF caps.
[and correcting bias]
>
>We spent a long time evaluating this one. I honestly couldn’t
>tell any difference between shorting the cathode resistors directly
>to ground, or shorting them with the 220 uF caps. At low volumes,
>Kathy said she thought it sounded better with the 220 uF caps handling
>the AC, as opposed to the direct to ground connection. While she
>couldn’t explain to me what “better” meant, she did seem to be able
>to pick out the same configuration as better, even when I was running
>the switches and she couldn’t see them. Then, when I cranked it up,
>she identified the _other_ configuration, with direct shorts from
>cathode to ground, as “better”, and seemed to be able again tell the
>difference even when she couldn’t see the switches. To me, there
>was no difference. (Again, grid caps made no difference in or out).

I would expect very little difference there. Good for Kathy if
she could tell! There will be a bit of difference in operation,
esp. as the circuit is played louder.

Some do like fixed bias better than cathode bias. This circuit
(with switches open) would be partially cathode biased, so maybe
that’s what Kathy was picking up on.

>The rational between using the 220 uF caps and leaving the cathode
> resistors in is from Mark Garvin, who wrote in aga not long ago:

>>I may have mentioned this in the past, but cathode resistors help to
>>relax the grid-dc-res-to-ground spec of output tubes. They help to
>>keep tubes from ‘going gassy’ due to ionization when the grid resistors
>>are high (which they are, at 220k). So…it would be great if the
>>Webernoia regarding the cathode resistors could be confirmed (as
>>opposed to sonic diff’s due to phase splitter and bias res values).

>>I’d still love to hear a logical explanation for why mixed cath/fixed
>>bias is a bad thing. Either bias method sounds fine alone, right?

Your experiments and my own would seem to confirm this, Scott.
I’m not sure exactly how much the low-value cathode resistors would
relax the constraint on the DC-grid-to-ground spec, so you may not
be gaining much by leaving the resistors and large caps in there.

>(P.S. gonna put a sticker in the speaker cab,
> -BF’d with a little help from aga-)

Thanks again for following through and posting such detailed
results, Scott (and Kathy)! I didn’t want to post any lengthy
explanations until you finished your test. I hope this post
helps to make sense of what you were hearing.

I’d still like to resolve why you heard such a big difference
between the standard Fender circuit and the version with large
bypass caps at higher frequencies and lower volumes. The highs
should have been bypassed even by the low value np cap. And in
fact, I didn’t hear a *pronounced* difference when I tried this.

It would also be great to find out what the difference is with
Tim’s and FW’s amps/speakers/guitars that would make the .002′s
so audible.

Over the years I’ve been trying to resolve situations where
listening tests do not agree with mathematical results. It
has always turned out to be an oversight in the mathematical
model. Usually an educational experience.

Mark Garvin

From timtub–(at)–ol.com Sat Jan 11 13:42:15 CST 1997
Article: 34069 of alt.guitar.amps
Path: geraldo.cc.utexas.edu!cs.utexas.edu!howland.erols.net!feed1.news.erols.com!news.bbnplanet.com!cpk-news-hub1.bbnplanet.com!portc02.blue.aol.com!audrey01.news.aol.com!not-for-mail
From: timtub–(at)–ol.com
Newsgroups: alt.guitar.amps
Subject: Re: My SF-BF experiment
Date: 11 Jan 1997 19:14:10 GMT
Organization: AOL http://www.aol.com
Lines: 23
Message-ID: <19970111191201.OAA2455--(at)--adder01.news.aol.com>
References: <5b825j$lo--(at)--anix2.panix.com>
NNTP-Posting-Host: ladder01.news.aol.com
X-Admin: new–(at)–ol.com
X-Newsreader: AOL Offline Reader

In article <5b825j$lo--(at)--anix2.panix.com>, mgarvi–(at)–anix.com (Mark Garvin)
writes:

>
>It would also be great to find out what the difference is with
>Tim’s and FW’s amps/speakers/guitars that would make the .002′s
>so audible.
>
>
Thanks for the all the leg work and consise reporting Scott. Thanks for
the review Mark.

Generally the first thing I do when I see those .002 caps is snip them. To
me, it is just like turning up the tone control on the guitar a notch. If
you turn up the amp and just listen to the air noise, you should certainly
be able to hear a difference when you A/B test.

Tim

A great amp can make a lousy guitar sound great.
A lousy amp will make a great guitar sound lousy.