October 14, 2022 - 04:46 pm|
|[Editor's note: This was sent as an email and is published here with permission.]|
Hi Sam — Thank-you for your comments and interest in the SCA-35 EFB Power Supply Board. Feedback and the time taken to provide it are always appreciated. Regarding your comments, I can provide the following:
1. The basic power supply circuit on the EFB Power Supply Board does in fact follow the design of the original Dynaco power supply quite closely, and quite intentionally. There are some improvements made by way of more than doubling the value of all the filter caps employed after the first filter cap, adding film bypass caps across the caps used at all supply points for the audio circuits for improved HF decoupling, and using higher PIV and current rated rectifiers -- but beyond that, the basic power supply is as Dynaco designed it. The effort was not to take the design out of the realm of Dynaco’s original offering, but simply to add practical improvements where they could be made. This was largely made available by the smaller physical size of today’s components versus that of similar or even smaller value components at the time this amplifier was designed and produced.
2. Turn on in-rush current surge is always a significant consideration when solid state rectification is used. To date, there have been no concerns noted from the increased power supply capacitance values used on the EFB Board. Over the years, the original 2A Slo-Blo fuse as well as the original power switch specified for the amplifier have been shown to take the increased values in stride with no noticeable impact on the life of these components. Using even large value power supply caps could introduce problems however, that an NTC device can help mitigate. Most often however, they are best placed in the AC wiring for the primary winding of the power transformer, as then the device will also help control the surge created by cold tube heaters at turn on which can also be significant, as well as the surge created by use of solid state B+ rectification.
3. The use of two 2.2K resistors (two π filter sections in series) again goes back to the component availability to which Dynaco had access back in the day. Two filter sections were needed to achieve the reduction in ripple needed for quiet power combined with the limited voltage drop desired to properly support the small signal stages. Granted, a single resistor and double value filter/decoupling cap could also likely work well. However, with all else being equal, the dual π network approach used will be slightly more effective since the networks are connected in series, so the ultimate effect on noise reduction is greater than just a simple doubling of the noise reduction that each section could theoretically provide on its own. Again, there was no desire to change the configuration of Dynaco’s original design, so the dual dropping resistor configuration remained in place on the power supply board.
4. If 6U8A/6GH8A tubes are simply substituted in a given circuit by way of socket adapters, the results are often very much as you describe. Sound can become harsh, and hum often increases — but most often this is not the fault of the tube, but that the circuit was not designed for the alternate tubes. The 7199 was designed from the get-go as an audio tube, and has very definitive characteristics that allow it to be a superior performer in classic direct coupled AF Amplifier/Cathodyne Phase Inverter circuits — and especially those that directly drive the output stage. In particular, the triode (inverter) section has a much lower Gm than the alternate tubes do, which tends to promote a higher output impedance drive being presented to the cathode driven output tube. Since the plate impedance inherently presents a much higher drive impedance than the cathode drive does, the 7199 tube then tends to minimize the drive impedance in-equities from the inverter, that the alternate tube actually tend to promote. The effect of unequal drive impedance levels promotes uneven clipping as the amplifier approaches and reaches the onset of clipping.
Also, since the alternate tubes were primarily designed to be RF Amplifier/Oscillator tubes, their pentode section has internal capacitance levels between the various elements that promotes extended HF operation, so their use invalidates the HF tailoring and response performance built into the SCA when using 7199 tubes, that is required to produce proper stability. Without appropriate compensation, the results can then be an added harshness to the sound produced and even outright unstable operation. In the SCA-35 as designed, some examples of the alternate tubes cause immediate instability (outright oscillation), rendering the tube useless in the amplifier. Part of this is because many of the alternate tube examples that exists today and not genuine items, but some common tube that has been used and rebranded into many tubes it was deemed to replace. The other part is simply because the circuit was not designed around the characteristics of a true bogey alternate tube.
And then there’s the hum issue. The alternate tubes will always produce more hum in an SCA-35, because their internal heater configuration requires a different heater balancing point than the 12AX7 tubes do that are used for the phono section. The result is that if the hum balance controls are adjusted for minimum phono hum, then low level hum with the volume full down will be excessive. If adjusted for minimum full down hum, then phono hum will be excessive. The 7199 was designed to be a low hum and noise tube — but what that really means is that when the tube is used with other classic dual triodes in high quality designs employing an AC heater circuit— the hum balance control will respond to the internal construction of the dissimilar tubes equally — whereas the internal construction of the alternate tubes requires a distinctly different minimum hum null point on the hum balance control.
The PC-10A boards that I sell fully address the unequal drive impedance and extended HF pentode section performance of the alternate tubes (with additional components employed to address these issues), so that the stability and response of the amplifier when using my boards is a near exact copy of the original amplifier when using 7199 tubes. Rest assured, with hundreds of PC-10A boards sold and amplifiers so modified, there has never been a single complaint about the audible performance of the alternate tubes. In fact if anything, reports have been the opposite. 7199 tubes are still available today, but at quite a premium price compared to the alternate tubes. And, the vast majority of 7199 tubes left are retreads (excessive hum and noise or other issues) that have been purchased, rejected, returned, and resold, until they finally find a home with someone who either accepts their diminished state, or uses them in an application where hum and noise is not an issue.
But these issues only touch the tip of the SCA-35’s design issues. All of the preceding and much, much more is addressed in a thread I presented on AK about bring out out the very best of the SCA-35’s performance. You can find that thread here on the Audio Karma website:
The article will be posted in full here on Tronola website very soon without the various discussion elements brought on by those with questions in-between the various postings. It’s presentation here has been delayed because after I wrapped up my work with the amplifier (an ongoing effort spanning some 20 years now), a fellow AKer stepped up and produced a fully modern new back panel for the amplifier, with provisions for modern connectors, and elimination of antiquated inputs no longer needed. I strongly recommend giving the thread a thorough read, as it will answer most if not all of any questions you might have, and introduce (among a number of other things) some new and unique solutions to the SCA’s hum issues that the amplifier is so well known for. Properly implemented, the amplifier will best the published hum and noise levels, which the original design and build of the amplifier had no hope of ever achieving when built as instructed.
Thanks for your interest in my boards and reaching out. I hope this helps! Best —
October 14, 2022 - 04:30 pm|
|[Editor's note: This was sent as an email and is published here with permission.]|
i was looking at the power supply board for the sca-35 and am curious: you seem to follow the original dynaco design almost exactly. since it is a 'solid state' rectifying circuit, the first capacitor can be just about any value. if you're concerned about surge, you can move the 50 ohm resistor in front of the capacitor or put a NTC device in front of the capacitor. as for the 2, 2.2K 5% resistors, a 4.3K 1% will fit nicely. anyway i'm leaning toward ordering the board for the EFB mod. BTW i'm tired of hearing about 6U8A's replacing 7199's. i have assembled chinese power amp kits for customers and the 6u8As are harsh sounding and hum prone. the 7199s beat them every time and are still available for a reasonable cost.
August 29, 2018 - 01:08 am|
|Hi Jonathan -- First of all, I want to apologize for the absurdly late response, but it appears that digital gremlins have been at play, in that the usual automatic notification on this end that notifies me of a posted comment failed in its duties. Such glitches are not tolerated on Tronola, so the gremlins will now be hunted down to the ends of this world and beyond!|
In any event, your question likely has wide spread interest, so I will go ahead and answer at this late date if only to serve all who visit the site, and hope that you will pick up on my response at some point.
As for drifting bias readings, it can occur from a number of things, but the usual culprits include:
1. A varying AC power line, being from various appliances in your home, to a varying AC power grid, most particularly in the summer time, when early morning and late evening the voltage can run high, but will be notably lower during the heat of the day due to all the A/C units operating.
While EFB™ action from the EFB regulator will account for the varying B+ voltage levels that a varying AC power line produces, it cannot account for the change in heater voltage applied to the tubes, which can easily change quiescent current draw a couple of mA either way from an ideal bias setting.
2. Poorly matched tubes to begin with -- as in they were matched with element voltages that are well off of typical operating voltages in actual use. Also, new tubes need time to cook for their characteristics to stabilize so that a real match can be had. This is why tubes from the better providers like Jim McShane take the time to cook the tubes in large batches, with each tube being monitored so that when he sends out a matched set, the tubes you get will all typically pull within 1 or 2 mA of each other, and do so over time, which is a very good quiescent match indeed. It's also why tubes from these vendors cost a little more as well -- but the cost is well worth it. The Russian EL84M tubes are excellent tubes, but are well known to vary all over the map in terms of characteristic consistency from one example to the next. Therefore, it is always best to purchase these tubes from a quality vendor who does an honest job of culling out the unstable pieces and who supplies matched quads and pairs that were determined to be so matched under real world operating conditions over time.
3. Gassy tubes -- Tubes that have been stored for many decades and then sold as NOS NIB tubes can often be gassy internally. Such a tube will be quite unstable when first placed (or re-placed) into service. A tube with this history and displaying this behavior should be watched initially for the first few hours of operation, but the good news is that tube will in fact re-stabilize again with a little use, during which time the getter collects the lose gas molecules and stabilize the tube again.
There are other issues too that can cause bias instability, like components drifting with thermal heating of the unit, but if the specified parts are used to populate the boards, this will be held to an absolute minimum.
The drift you have observed with those tubes you purchased from Jim appears to be rather normal, with minor drift moving back and forth between the tubes (possibly a number of times) for up to about the first 50 hours of use. After that, the tubes will have settled into the individual characteristic they will typically display throughout most of their usable life.
I hope this helps!
March 24, 2018 - 12:59 pm|
I'm looking for guidance gauging the health of my amp. I have an SCA35, my first tube amp, restored with all Tronola boards, including the EFB. The voltages to the output tubes all run a little high, but I believe are within an acceptable range (I get 17.35VDC to pin 3 and 392VDC to pins 7 and 9).
However, I initially installed a quad of new Sovtek EL84's, which I thought I had biased around .27VDC, but I now assume were never stable and, within maybe 10 hours of playing time, redplated a tube. I have now installed a quad of matched Sovtek EL84M's from Jim McShane. These appear quite stable and I have set the bias around .27VDC as per the EFB instructions, but I am now checking the bias at regular intervals and within 8 hours or so, they have gone from 10mV difference left to right to 20mV apart.
Is this a normal amount of drift or is it a symptom of a problem. If so, what tests can I perform? Any advice is greatly appreciated, and apologies if these are too vague of questions.
January 10, 2018 - 03:59 pm|
|Thanks, once again, for your insight Dave. |
January 09, 2018 - 09:20 am|
|Hi Bill -- Looks like you're getting to the bottom of things regarding your question. I appreciate Steve stepping in during my absence.|
A couple of points that your exercise points out:
1. Unfortunately, in the vacuum tube world today, it is not entirely uncommon to find as much as a 20% variation in Eg1 voltage required to achieve a set current draw This is particularly true if you are comparing American manufactured tubes to the Russian tubes produced today.
2. The good new is that, assuming a good tube, this variance from one tube to the next is not the prime consideration; it is the quiescent current flow through the tube that is most important. Granted, the amplifier will produce the best performance when the characteristics of all four tubes are (nearly) the same, and particularly so at the LF end of the spectrum. But whether they are all matched at a characteristic that requires an EG1 of -13.5 vdc or -16.5 vdc, that is of far less importance than the fact that they are in fact matched at the same characteristic. Now technically, tubes requiring a lower negative Eg1 versus those requiring a greater value will theoretically have a performance edge due to the increased open loop gain they will provide. But we're talking performance differences that would need a proper lab to demonstrate.
3. Your exercise also points out what really good tube testing can tell you. Steve's new VTA is of priceless value in this regard. It has the ability to check the characteristics of a tube at precise, real world (or unit identical) operating voltages to ensure they are all the same, and within the range that a unit was designed to operate at. I have done something similar for years in testing the actual power output that a tube can produce, which is another indicator of a tube's capability as well. At issue then, consider this:
If tubes that require -16.5 vdc Eg1 voltage are well within the characteristic norm, then are tubes that require only -13.5 vdc simply out of norm by normal manufacturing production variation? Or are they out of norm because their cathodes are well worn requiring reduced grid voltage to produce a set, target cathode voltage?
Conversely, if -13.5 vdc Eg1 is the norm, then tubes requiring -16.5 vdc would typically be considered as having a wide tolerance from the norm, or in today's lingo, be considered a "hot" tube. But again, without thorough testing at honest real world operating voltages, you just won't know if the issue between these two sets of tubes is production variation related, or due to worn emission capabilities. Emission testers that check this quality at very low operating voltages are near meaningless as the target current flow is also quite low: Therefore, a well worn tube can still perform quite well in an emission test.
Thanks for posting your observations. It gives a chance to address basic issues that no doubt others grapple with as well!
Add your comments here...