I read the entire thread containing the postings by Gary Kaufman. Sounds like he replaced the single cathode resistor with four separate resistors, and he got an unfocused sound. But, notice there was no mention of a pot to set the bias current the same for each tube. If all he did was to replace the single original cathode resistor with four individual resistors without adjustment pots, and if his tubes were not well matched, or if there was something else going on that was not apparent in his post, then I guess it isn’t surprising that there were problems. I don’t know that we can directly compare Gary Kaufman’s experience to what Dave Gillespie discussed in his article.
But, it would be good to keep Gary’s example in mind. When we say “individual bias,” it could mean that you just broke-out the single cathode resistor into four, one per tube. OR, when we talk about individual bias, we might mean the setup Shannon Parks and Kevin Devaney use, having a small resistor network including a potentiometer for each EL84. You cannot match the bias current in each EL84 if you simply use four individual resistors off of each EL84’s cathode.
Now, here’s something really interesting I found. If you can, pick up a copy of the July/August edition of the absolute sound magazine, issue 234. On page 122, there is a review of Bob Carver’s VTA20S “Black Magic” power amp. The review includes quotations from Bob Carver talking about the circuit. It uses four EL84s and two 12AX7s. The 12AX7s are wired for voltage gain and then cathodyne phase inversion just like in the original Stereo 35, except Carver uses a 12AX7 instead of the 12DW7 used in the Stereo 35.
Carver then goes on to describe the common cathode bias scheme he uses in the amp – with a zener diode to clamp the bias voltage at the single cathode resistor. Sound familiar? Dave Gillespie acknowledged the zener as a method to achieve the same goal as his EFB, but Gillespie also went on to discuss the zener’s potential drawbacks, as well. The discussion in the absolute sound article reads like an abridged mirror of the discussion not only in the Gillespie article, but also of other discussion you can find in forums. For instance, regarding the plate dissipation limits of different current production EL84s in the Stereo 35, where you would want to adjust the B+ so you don’t prematurely kill the EL84s. You can find Jim McShane talking about this over at Audio Asylum. Bob Carver’s comments in the absolute sound article are uncannily similar to Gillespie’s article and McShane’s posts.
Talking about the zener bias regulation in his VTA20S “Black Magic” amp, Bob Carver stated the following: “The effect is quite significant, and flies in the face of what I have always been taught-that is to use separate cathode resistors. Not a good idea, actually.”
Maybe this is why he called his amp “Black Magic.” And, on the face of it, it looks like Carver, in implementing something akin to Gillespie’s EFB, also thinks that individual cathode resistors are not a good idea. But, notice here again, as in the case of the comments from Gary Kaufman in the AA post, Carver did not mention an individual bias adjustment pot for each EL84. Maybe he would say a pot wouldn’t make any difference, but we really can’t tell from the article. When he pointed-out that a number of different variations of cathode biasing schemes are in use in different amplifiers, Carver stated: “That tells me that there is no known consensus. I like my way the best, especially with the clamp.”
What I get from Dave Gillespie’s article is that the impetus for the EFB is to find a way to stabilize the EL84s’ operating points under load to compensate for B+ sag, especially with both channels driven. I interpret him also to say that aside from the EFB, you could accomplish this by increasing the energy storage in the power supply with larger capacity electrolytics. However, Gillespie wanted to find a simple, unobtrusive means of stabilizing the operating point – something simple, clever and cost-effective in the spirit of the original Dynaco philosophy. Increasing the PS capacity would be more complex and expensive than the EFB. (But, the cost-effectiveness doesn't seem to be there if you have to keep buying matched quads of EL84s with any regularity.)
Item 10 on page 16 of Dave Gillespie’s article talks about the adjustable individual bias resulting in worse performance than the original bias circuit (no EFB). What I have been wrestling with is exactly how he comes to this conclusion. Evidence he offers refers to a test done by Triode Electronics in Chicago comparing their Dynaclone OPT to some other replacement brands. The test was performed by dropping each OPT into a DIYTube Stereo 35 clone, which is known to use the adjustable individual bias. Although Triode did not specifically say so, Gillespie assumes that the test results were obtained with only one channel driven, with the resulting distortion measurements equal to (in the midrange) or worse than (at the frequency extremes) what Gillespie found driving both channels of an SCA-35 having the original single cathode resistor scheme (no EFB).
Per his discussion in the article, one channel of the SCA-35/Stereo 35 performs considerably better than both channels being driven together (due to B+ sag). So, if one channel driven in the Triode Electronics test results in equivalent or worse distortion than Gillespie found driving both channels of an SCA-35 with a single cathode resistor (without EFB), then this seems to be the reason why he thinks adjustable individual bias is “worse” than the original scheme.
Problem I have with this is that the Triode Electronics test Gillespie references may or may not have been driving one channel only – we really do not know for certain if this was the case. There may have also been other variations of test parameters that may not be obvious. Even if Gillespie referenced distortion measurements from the original Z-565, how do we know that there weren’t substantial differences even among different samples of original Z-565s? The point being that Gillespie did not conduct his tests with the same amplifier having the same set of Z-565 transformers that Triode Electronics had, and if Triode’s test showed anything it is that there can be variation among different transformers, never mind any other differences between the amplifiers. I know Triode highlighted differences between different brands of OPT, but it seems to me that there could be significant variations in performance among original Z-565s, especially if they were manufactured by different vendors over a span of many years.
Gillespie may be exactly right, but I’d like to see tests done under carefully controlled conditions rather than drawing inferences from another party’s test that was done for a different purpose, in Triode Electronics’ case, testing for differences between brands of OPTs. I’d like to see tests done with one amp having one particular pair of OPTs only, measuring distortion both with and without closely matched tubes for: 1) single cathode resistor without EFB, 2) individual bias adjust with potentiometer (no EFB), and 3) basic EFB without individual bias adjust.
Again, Gillespie could certainly be exactly right. But, even by his own words, there are other ways to deal with B+ sag. And, here I will come full circle. Hifiguy began this thread considering either the adjustable individual bias or EFB. With either choice he would have to replace his cap board and go back to the quad-section can. I don’t know what the total capacitance of HifiGuy’s cap board is, but keeping the cap board could possibly obviate the need for the EFB. Also, going from low ESR electrolytics rated for 85 C or 105 C on the cap board to a quad-section can rated for 55 C could introduce its own change to the sound that may or may not be desirable.
Gillespie’s basic EFB seems an efficient and effective solution with the caveat that you need to keep matched tubes in the amp. If you like to roll different kinds of tubes that may not be matched very well, or if your tubes do significantly drift apart over time and you want full use from them, then I think the individually adjustable bias may be the best way to go. Pending evidence from a more targeted and closely controlled test, I think the jury is out on whether the individual bias adjust is really worse than the original single resistor (without EFB, that is). Perhaps it is worse. But, again, when I installed my adjustable bias, I did not get an unfocused sound like Gary Kaufman reported. If anything, the focus improved. But, then again, I listen at moderate levels, and any “worseness” might not show up readily unless the amp were pushed hard...
Last edited by PeterCapo on Tue Jun 25, 2013 10:35 am; edited 18 times in total