Improved SCA-35/ST-35 Performance

for the DIY ST35, the Dynakit and every other PP EL84

Postby SilverT » Mon Jan 10, 2011 11:09 pm

Thanks for your reply Dave. Your explanation gave me the final piece to complete the EFB™ in my SCA-35!

Well, perhaps "complete" is bit premature. I seem to be experiencing adjustment issues. The closest I can get to the 0.27 vdc mentioned in the article is 0.29 and 0.296 respectively, and that's with the adjustment pot at its stop. Turning it in the other direction increases the voltage.

I read your reply to Marshman earlier in this thread, when he was having bias issues. I will have to use the advice you gave him to troubleshoot my amp. But I will do that in the morning. I've been hunched over a hot soldering iron all day, I'm tired, and my shoulders are sore.

Jeez, can you actually get too old to solder? I hope not, but it sure feels like it at the moment! Maybe I'll just have to cut down on the all day marathons, but you know how it is when you want something to be finished.
SilverT
 
Posts: 5
Joined: Mon Dec 27, 2010 12:25 pm

Postby dcgillespie » Tue Jan 11, 2011 7:16 am

T -- If you're confident in the value of the divider resistors used (the 360K and 15K resistors) and the value of the pot installed, then the current readings you are getting show it is likely that the tubes you are using simply require more bias voltage than the available range provides.

The range of the bias pot as designed covers a bias voltage range of 5 vdc, which I have found adequate for any tubes I have used to date. But it is possible that some tubes may need more. The easiest way to achieve this is to replace the 15K resistor with a value of 18K. This will have the effect of shifting the range of the bias control up by ~ 3 vdc. While the range of the control itself will still be about the same, it will now provide this range over a voltage that is ~ 3 volts greater.

This should take care of the issue. Good luck and keep us posted on your progress!

Dave
dcgillespie
KT88
 
Posts: 399
Joined: Sun Mar 08, 2009 5:34 am
Location: Ball Ground, GA

Postby SilverT » Tue Jan 11, 2011 10:17 pm

The problem has been solved, and I wanted other potential users of the EFB™ to know what the issue was.

It turns out the nut behind my soldering iron was loose.

Dave Gillespie was gracious enough to correspond with me via email to sort out the problem. I made a simple wiring error to the bias adjustment pot by misinterpreting the schematic and the picture included with the article.

I'd like to publicly thank Dave, certainly for sharing his EFB™, and writing such a great article. But also for his kindness and generosity in helping me sort out my mistake.

Thank you Dave!

Now I can go install that new quad of Gold Lion EL-84 ‘s I got from Jim McShane last week, confident that they’ll last me a long time, thanks to Dave Gillespie and his intelligently executed EFB™ circuit!
SilverT
 
Posts: 5
Joined: Mon Dec 27, 2010 12:25 pm

I've been waiting for something like this

Postby corndog71 » Thu Feb 03, 2011 2:47 pm

I've been itching to build or rebuild an ST-35/SCA-35 for years and now after reading about the EFB upgrade I'm pretty much sold.

One question: the 100uF 25v cap - does it need to be that big? I thought I read somewhere that the cathode circuit is part of the signal path.

Having electrolytics in the signal path tends to be detrimental to the sound so if I can replace it with a nice film cap then surely a small improvement could be had.

So any ideas as to how low we can go to get the job done?

Another thing I noticed is that one difference between the ST-35 and SCA-35 is the latter having hum pots in the filament circuit. Are these helpful or ultimately not necessary? I've read plenty of st-35 owners not seeming to need them.
corndog71
 
Posts: 105
Joined: Thu Feb 03, 2011 10:45 am
Location: midwest

Postby dcgillespie » Thu Feb 03, 2011 8:11 pm

Hi Corndog --

Thanks for the interest in the EFB modification. Your question has been a topic of previous posts on this thread, so you might wish to review them as well.

The cathode bypass cap does not play the same roll in the EFB modification that it does in a traditional cathode bias installation. In those cases, it is very definitely in a significant portion of the signal path, so it's use and value are important for everything from maintaining full power output capabilities, to minimizing IM distortion, to helping maintain proper LF stability.

With the EFB modification, use of a cap at the cathode connection has nothing to do with the issues noted above, as the impedance created at the cathode connection by the regulator is extremely low, being on the order of easily < .1 ohm across the entire audio bandwidth. In this case, the cap is there only for the purposes of the 337's stability requirements, and use of the original cathode bypass cap for that purposes was only done in an effort to make installation of the modification as easy as possible. Therefore, any value that will provide stability for the 337 is acceptable, with the data sheets I believe showing a minimum value of 1 uf required. I would typically go no lower than about 10 uf, but understanding the purpose of the cap now lets you experiment as you would like. Others are doing this as well, but unlike a traditional cathode bypass cap, the value used here will ultimately have little impact on the reproduced sound.

The hum balance pots are necessary in the SCA35 to obtain minimum hum, primarily from the phono preamps. In the ST35, the gain is greatly reduced, eliminating their need from the design.

Thanks again for the interest, and good luck!

Dave
dcgillespie
KT88
 
Posts: 399
Joined: Sun Mar 08, 2009 5:34 am
Location: Ball Ground, GA

Postby Stunch » Wed Feb 09, 2011 10:33 am

I have been reading this thread with great interest. I am getting ready to build a ST35 (original) with the 60, 40, 20 (all at 450V), 100@25V filter cap, but have been considering a 80, 40, 30, 20 (all at 550V) as a replacement. It was mentioned that with the EFB mod, there is no need for 100uF anymore at the cathodes. Since the "upgraded" cap has 20uF (which is more than enough), would the huge increase in voltage rating, from 25V to 550V, be an issue? I realize that caps with an increased voltage rating is never a bad thing, but in this case with the EFB, would it be a problem?

Are there any other concerns with the increase of the other two cap sections?
I am a little green with this stuff so any help would be appreciated.

Thanks
Stunch
 
Posts: 55
Joined: Fri Feb 04, 2011 3:50 pm

Postby dcgillespie » Wed Feb 09, 2011 8:07 pm

Stunch -- Thanks for your interest in the EFB modification!

You are correct, with the EFB modification, a cap is no longer required to perform a cathode bypass function, but one is still needed as a standard requirement for regulator stability. With the modification presented, I simply used the original bypass cap to perform that function.

The increase in capacity for the B+ caps you propose to use presents no significant concerns, and in fact, will actually help performance. However, using a 550v cap to stabilize the 337 device is not recommended. Since the voltage across the regulator is on the order of ~ 15 vdc, this would not be enough voltage to properly maintain the form of the 20 mfd section you plan to use for this purpose. To operate a cap rated at such a high voltage at such a low actual voltage generally causes it to present a radically different capacitance value to the circuit it's connected to, than the value it is intended to represent.

If you use the larger value/higher voltage cap, I would recommend that you combine the 30 and 40 mfd sections together to create a 70 mfd cap, and let that replace the original 60 mfd (first) B+ section. Then I would use the 80 mfd section to replace the original 40 mfd (second) B+ section. This is the section that supplies power to the output stage, and doubling the value at this point provides a much bigger reservoir for the output stage to operate from to better handle its dynamic demands. Then, the new 20 mfd section can simply replace the original 20 mfd (third) section. Finally, you can then use a small 35-50 volt cap of 10 mfd or greater to provide the bypass function for the regulator. This will maximize the use of your new can cap in the B+ supply, and also provide effective bypassing for the EFB regulator.

I hope this helps, and good luck with your project!

Dave
dcgillespie
KT88
 
Posts: 399
Joined: Sun Mar 08, 2009 5:34 am
Location: Ball Ground, GA

Postby GreatTone » Fri Mar 04, 2011 11:12 pm

dcgillespie wrote:As long as the new boards will still allow you add the new required control and screen grid resistors for the output tubes as shown on the schematic, they will work just fine. The use of different AF amp/phase inverter tubes has no effect on the modification.

Dave


Hi Dave,

Well, I've been a bit slow in my progress of installing the new output boards on my SCA-35, but I'm getting there now. I'm using the Audioregenesis boards, and as I'm installing them, something occurred to me.

The new boards have a few changes to allow the use of 6U8/6GH8 phase inverters. One of those changes is the addition of 10K grid resistors. How does that square with the added 1K grid resistors in the EFB mod? Not sure what to do now. Is 1K still the value to use for the EFB mod, or keep the 10K?

Thanks
GreatTone
 
Posts: 9
Joined: Fri Dec 17, 2010 11:50 am
Location: Northern Virginia

Postby dcgillespie » Sat Mar 05, 2011 8:08 am

Since I don't have an Audio Regenesis board to test, I can only give you an opinion here, but one I think will work just fine.

The effort with the two 1K resistors I specify, versus the one 10K resistor AR specifies (all per channel) serve two different purposes. The two 1K control grid resistors I specify are for the purpose of specifically maintaining output tube stability, which are required now since the tubes are operating under conditions of fixed bias. The single 10K resistor AR specifies is being used to address HF stability, relative to the application of NFB when using their board with the alternate tubes. Their addition of new components C1 and R38 on their schematic work towards this same purpose as well.

Using a 10K resistor for the one output tube in place of where I specify a 1K resistor will more than satisfy the needs of output tube stability with fixed bias operation. Therefore, I would install a 10K resistor on the one output tube where AR specifies, but still install a 1K resistor on the other tube in each channel, were AR simply shows a jumper in place.

Using this approach will satisfy the stability needs of the output stage, and the HF response of the AR board with the alternate tubes.

Keep us posted on how it's going!

Dave
dcgillespie
KT88
 
Posts: 399
Joined: Sun Mar 08, 2009 5:34 am
Location: Ball Ground, GA

Postby GreatTone » Sat Mar 05, 2011 8:20 am

Thanks very much for the speedy reply. My plan is to wire up the new boards and listen to them for awhile as is -- when I got the amp last fall, the unit immediately died due to various old filter caps and tubes, so I've barely gotten to listen to it at all. Once I make sure it's working with the new boards, I'd like to put it in my system for a decent period of time, just to get a feel for how it sounds like that.

Once I do that, then I'll try the mod. That will give me a baseline for comparison purposes. Also, I'd like to be sure I don't make any errors with the board installation -- it would be a bit much trying to track down a mistake if I didn't know whether the problem was coming from something I did on the boards, or the EFB mod.

I'm slow, but I do get there eventually!
GreatTone
 
Posts: 9
Joined: Fri Dec 17, 2010 11:50 am
Location: Northern Virginia

Are stock ST35 EFB part values same as diy ST35?

Postby Bottlehead » Fri May 13, 2011 9:24 pm

Hi,

Several years ago I picked up three ST35 rev. D PCBs in a trade, built one by the numbers and stuffed the other two but haven't decided quite how to use them.

As a power supply fanatic I'm very interested in the EFB mod. I've read the Tronola article which describes its use in a stock ST35, as well as the other articles on Dave's amp pages.

I'm including the EFB circuit in my next Rev D build along with a 5AR4 tube rectifier, Hammnond 272JX PT, a filter choke and 120/120/120/33uf filter caps. Originally I planned to add an LM317 to each output tube for a simple CCS cathode upgrade but after further consideration I realized that this had its shortcomings...however all these drawbacks are addressed in the two-channel version LM337 EFB module. I love elegant yet simple designs and this is an inspired piece of work.

The EFB forum string has grown quite long and perhaps I'm missing something, but are the component values in the Tronola article identical to those I'd use in a diytube ST35 with tube rectification and the 600VCT Hammond PT? Because the article stresses the importance of precision resistors (0.1%) my first concern is that the values are correct. Has anyone here prepared a parts list?

Thanks for any assistance you can offer.
Bottlehead
 
Posts: 22
Joined: Thu May 22, 2003 3:15 pm
Location: Olympia, WA

Postby dcgillespie » Sat May 14, 2011 6:49 am

Bottlehead -- Thanks for the kind words concerning the EFB modification!

The components requiring the greatest precision would be those making up the 5.0 ohm cathode resistors, primarily because their accuracy determines how accurately the bias is ultimately set. The more they deviate from the actual value, the greater the inaccuracy of achieving 27 ma current flow per tube when .27 vdc is developed across them. Metal film would be best here to help maintain the target value in the face of changing ambient conditions.

The 1K and 100 ohm stability resistors added to the output tubes can be 10% tolerance components, as their precise values are quite uncritical in their position. Carbon film is fine.

The components used in the B+ divider string to bias the LM337 require some accuracy to keep the bias control within the correct range. The design of this string already throws the adjustment well to the side of safety (meaning the vast majority of tubes cannot be made to draw damaging current levels), so any gross inaccuracies here could potentially throw the control out of range for some tubes. Since the best overall approach is to use a matched quad of output tubes, it was reasoned that most folks would likely obtain a new set for this purpose, so the range of the control was designed with this, and the safety factor in mind. Again, for stability purposes, metal film would be best here, while the control itself can be a simple mini-watt plastic device.

Regarding your question about the power supply, the use of an alternate transformer and rectifier topology will typically require no changes to the EFB circuit, due to the very nature of it's action. If the B+ output of your proposed supply is different from that of the original Dynaco supply, or has a different level of dynamic regulation, the EFB circuit will simply react to that the same way it would to a changing B+ voltage for any other reason, be it due to the application of signal, or changing AC line conditions. If the output of your supply is significantly different however, a slight adjustment to the circuit may be in order.

The B+ with the original Dynaco power transformers typically runs 380-385 vdc. This is the voltage that the divider string was based on. If the output of your supply is significantly different -- say only 300 volts or so for example -- then the divider string may require a slight adjustment to keep the adjustment within range of the control. This is not because of a limited ability of the EFB circuit, but because the new B+ levels have thrown the bias requirements of the tubes out of range of the control's original parameters. In that case, a simple adjustment to the 15K resistor between the bottom end of the bias control and ground will be all that is necessary to reset the EFB circuit to the new requirements of the tubes. In making any adjustments to this resistor, bear in mind that lowering its value decreases available bias voltage to the output tubes, making them draw more current. Increasing its value provides more bias, having the opposite effect.

Thanks for the interest, and I hope I addressed your questions. Good luck with your amps!

Dave
dcgillespie
KT88
 
Posts: 399
Joined: Sun Mar 08, 2009 5:34 am
Location: Ball Ground, GA

Postby paart » Sat May 14, 2011 4:35 pm

My description of an application of this circuit to an original ST-35 appears on the Tronola website. At the end of the article, I listed some of the more essential components that I obtained from Mouser. along with the part numbers for convenience. That list is below:

Mouser # 511-LM337SP
Linear Regulators - Standard 1.2-37V Adj Negative

Mouser # 71-CPF15R0000BEE14
Metal Film Resistors - Through Hole 1watt 5ohms 0.1%

Mouser # 652-3296Y-1-502LF
Trimmer Resistors - Multi Turn 3/8 5Kohms Sealed Vertical Adjust

O.1% 5 ohm resistors are desirable but not essential for the circuit. Dave hand matched paralleled 10 ohm components. When I saw that close tolerance Dale resistors are available at a reasonable cost, I decided on that route, as they are certainly more precise than my ability to measure them.
Cheaper trimmers are available than the "Bourns" one that I listed, but I've always had good luck with their products. It's probably rare, but a lifted "wiper" would cause a significant problem.
paart
 
Posts: 95
Joined: Mon Jul 17, 2006 8:36 pm
Location: Central PA

EFB in diy ST35

Postby Geezer » Mon May 16, 2011 8:50 pm

Dave,
I'm also "Bottlehead"--I somehow created two screen names years back when I joined the site and I've logged in under "Geezer" because the email address associated with Bottlehead is not valid and I can't change it.

Thank you very much for the detailed explanation, I greatly appreciate the time and effort you've put into this project.

I'm a little weak on theory, I'm just an English major who likes to solder so please bear with me.

The topology of the diytube ST35 is a bit different than the Dynaco but the modification appears to be very straightforward. Referring to the schematic in the Rev D manual (found in the first entry of the Stereo 35 forum), the modifications required would be:

1) Cut the traces leading from each cathode to its individual bias control circuit (or simply omit the components) and bridge the cathodes to create a common cathode connection for each channel;

2) Install a 5R resistor at the common cathode connection of each channel and bridge the two channels;

3) Install 1K resistors between the grid (pin 2) of each EL84 and its 0.1uf coupling cap; Install 100R resistors at g2 (pin 9) leading to the center connection of each OPT;

4) If using the 2-channel version of the EFB module, connect its bias adjustment pots to the grounded junction of the two 470K series resistors (R13,14; R15,16) which bridge the grids of each channel's EL84 pair.

And that's all there is to it? As I said, I'm weak on theory but get the gist of the modification--the bias control circuit is now based at the grid rather than the cathode?

A couple more questions and I'll stop bugging you: 1) Does the value of the 470K resistors remain the same? 2) Does the fixed 100R resistor affect the values of the negative feedback circuits of each channel?

I see what you mean about Dynaco's KISS philosophy and how you've stuck with it...it's hard to believe it's so simple and adaptable.

Thanks for your patience with a paint-by-the-numbers diy hobbyist like me, and thanks to Shannon for helping many of us get started in diy.
Geezer
 
Posts: 26
Joined: Mon Jun 30, 2003 6:37 pm
Location: Olympia, WA

Postby dcgillespie » Tue May 17, 2011 7:37 am

Geezer (Bottlehead) -- You've got the sum of the modifications required down perfectly. However, let me point out that even in this case, the vast majority of bias is still introduced at cathodes. With the two channel version, all that is being done is to allow a slight adjustment of potential to be applied to the output stage grid return (470K) resistors in each channel. When either of these individual adjustment pots are adjusted so that these resistors are referred directly to ground, maximum bias is created for that set of tubes for any given setting of the main bias control -- now call a bias "range" control. As these individual controls lift the 470K grid return resistors away from ground, they start to apply a small positive voltage onto the grids, causing the tubes to draw more current. Since each channel has its own adjustment in this scenario, each channel then becomes independently adjustable relative to each other, while the original bias control simply acts to bring the individual controls in range for the particular tubes employed.

In using this option, the original EFB circuit action is still fully in place, since bias for the tubes is still being established independently of current flow through them. This is the essence of fixed bias operation by definition, and as such, distinguishes itself from traditional cathode bias operation, since in that mode, bias is always determined by current flowing through the tubes.

Regarding your other questions:

1. The value of the 470K resistors does remain the same. Some have expressed concern that this value is now somewhat higher than is recommended for 6BQ5/EL84 tubes when used in fixed bias operation. That is true, but in this case, it is of little concern, since the maximum recommendation is based on a tube operating at maximum dissipation levels, and the EFB modification causes them to operate efficiently at well below maximum levels -- hence the improvement in tube life with EFB operation. Still, if this is of concern to you, you can resolve this issue by replacing the original 470K grid return resistors with 220K resistors, and then replacing the original coupling cap value with .22 mfd caps. This will bring the grid return resistor values in line with published maximums, while causing no change to the overall operation of the amplifier -- but as I said, the necessity to make such changes is more based on personal preference rather than any real circuit need.

2. The 100 ohm screen stability resistors do not affect the NFB levels in any way. These resistors are required due to changing the output stage operating conditions from cathode bias operation to that of fixed bias. This causes the tubes to operate at a higher effective levels of Gm. Since they are already high transconductance (Gm) tubes to begin with, this along with the operating conditions can promote output tube arcing if not properly addressed. The addition of the 100 ohm screen stability resistors removes any concerns in this regard. If you are interested in this more, please click on the link provided in the EFB article to see an additional article I wrote, "Techniques to Maximize Power Tube Life". In it, the use of screen stability resistors is discussed in greater length.

Good luck with your amp!

Dave
dcgillespie
KT88
 
Posts: 399
Joined: Sun Mar 08, 2009 5:34 am
Location: Ball Ground, GA

PreviousNext

Return to stereo 35

Who is online

Users browsing this forum: No registered users and 17 guests

cron