June 2, 1999

Changes since April 29, 1999:

1. I changed the cathode resistors back to the original value of 680 ohms, per new measurements of the plate voltage on the first section of the JAN-Sylvania 6AN8's I got from Ned. See below.

April 29, 1999

Welcome to my mod page for the Dynaco Mark III amplifier. The Mark III is a popular 60-watt "monoblock" amp manufactured over many years by the West Philadelphia-based (and then Blackwood, NJ-based) based Dynaco company. Some say it's closer to 50 wpc (I'll measure it in a few days), but suffice to say it's got enough juice to drive any reasonably efficient speakers to room-filling levels. It is typically used in a stereo configuration with a second amp (duh). Although very attractive from a purely price/performance basis, it presents the modern user with more than a few challenges, especially in the power supply area. However, once refurbished, it sounds effortless and clean, and features the short signal path Dyna's cheap (I mean inexpensive) designs were famous for. It consists of a printed-circuit driver board with a single 6AN8 driving a pair of (originally) KT-88's, or as was the case in later years of its production, 6550A's--not unlike the HK Citation II, which underwent a similar tube reconfiguration through its production--in a push-pull Williamson Ultra-Linear arrangement (the screens of the 6550A's being connected to taps on the primaries of the output transformers). The power supply is tube-based, using a 5AR4/GZ-34 full-wave rectifier, and a multi-section 525vdc can-type electrolytic capacitor and choke in a capacitive-input configuration. Output taps are provided for 4, 8, and 16-ohm speaker impedances. It has a chrome-plated chassis and painted steel cage.

Of course, a schematic is always helpful.  Fortunately, there is a complete schematic for the Stereo 70 up on Curcio Audio's website, which you can download or read in Adobe Acrobat by clicking here. For a full manual copy of this and other Dynaco products, Sound Valves sells--and occasionally auctions off at eBay , which is where I grabbed them--two different collections in binders, one for Preamps/Tuners and one for Amps for $49 each.  I got them both, and I'm glad I did.

My Personal Mark III Story

I bought my first stereo system back in the fall of 1968, when I was bulking up for my freshman year in college at the University of Rochester. With my limited budget, I opted for a pair of Harman-Kardon HK-40 speakers, a Lafayette LA-125 solid-state integrated amp, and a low-end Dual turntable. What did I know? I still have the HK-40's--they're at a friend's house, still churning out the hits, but the Lafayetter got traded for a Scott 120-watt solid-state receiver and my problems began--I started blowing woofers in the HK-40's, which fortunately HK was willing to replace for free under warranty--I think I did that like three times. So I was on the hunt for something a little less barabaric and more sonically pleasing. As a ham radio guy, I thought a kit would be me, so I sent away for three of them--a PAS-3X, and a pair of Dynaco Mark III's. It was probably 1970 by this time and the mailorder source for Mark III's was drying up fast, and I had to take one for $65 at one place and another (with the 70.7v tap) for $69 at another place. They came with 6550A's, by the way--KT-88's supplied as stock were long gone at that point. Anyway, a long weekend with my trusty Weller soldering gun (idiot that I was), and I was actually up and running. Then I was on the hunt for better speakers, having found a buyer for my HK-40's (from whom I bought them back years later and then lent them back to where they presently sit), so I decided on Rectilinear's. I couldn't afford the big IIIx's( I think they were), so I opted for a smaller model and hated them, so I took a flyer on Acoustron LWE-1A's, which I still have and use with the Mark III's. They use motional feedback and are somewhat of a trivia question for Audio Society meetings (which I never attend), but that's another story. And so it went until I got the bug years later to upgrade the Mark III's. Why? I couldn't get tubes that would sit still in the damned amps without taking off for places unknown.

So I subscribed to The Audio Amateur (this was back in the mid-70's), and installed bias balancing and larger coupling caps, and changed over to 7027A's, and then triode-connected them, and things stayed that way pretty much for the next 15-20 years or so. I was happy, they were happy--everybody was happy.

Why 7027A's, you ask? Well, at that time (early 70's), 6550's you could get from Dynaco were straight-sided 6550A's, most likely GE, but before that even the coke-bottle jobs I got from a local jobber (Tung-Sol GE's, RCA's, etc.), had a tendency to take off in that thermal runaway mode that can really put the fear of God in you. Articles at the time in TAA mentioned that the 6550 was perhaps unable to take the over 400vdc screen voltage required in the Ultralinear configuration used in the Mark III, and from my experience that sounded reasonable to me. Someone wrote in and said they had terrific results with 7027A's (you have to rewire the sockets somewhat--7027A's use ALL the pins), and at that time, they were pretty plentiful. Of course, I didn't know that they would become real expensive before too long, culminating in my last purchase of a matched quad of GE's for $135 maybe 5-6 years ago from AES (not in the 7591A's league, but high enough for me at the time). By the way, and Kevin Kennedy confirms this, while I have had pretty decent luck with the stock Ultralinear configuration in my SCA-35 and lately a Stereo 70, the Mark III's seem to WANT, if not NEED to be connected in triode mode. Why they seem more responsive to it than other related designs is a mystery, but I've gone back and forth, and as they have enough power to handle the power loss when wired as triodes, they're set that way for good even though I have a toggle switch in each of them that allows me to switch back to UL (off-line, of course).

Anyway, this first treatment explores the potential of the Mark III's with its basically stock circuit configuration, albeit triode-connected using KT-90's, but with attention paid to parts matching and bias-balancing (courtesy of an anonymous letter-writer reprinted in TAA way back when), and especially the requisite power supply mods needed to live in today's high line-voltage world. Future updates will explore other tweaks, but this first iteration will be  tube-rectified and aimed at high-reliability and low complexity.

Replace the Sockets

Well, you should replace the sockets, but as yet I haven't. I did do it with the Stereo 70 I rebuilt, but I used the only four ceramic octal sockets I had, so couldn't do it with my Mark III's recently when they were up on the bench. Next time I hit a Hamfest maybe I'll get lucky, or maybe I'll get crazy and actually buy new ones from Ned or Angela. But it is a good idea, especially with an amp like the Mark III that gets pretty hot. Obviously, do the recitifier tube socket, too, while you're at it. Watch the dimensions and get sockets that drop in, unless you have a Greenlee hole punch for a bigger hole ($$$). And if you're going the replace-the-sockets route, then just bite the bullet and do a complete frame-up restoration and strip the chassis entirely, as I did with the Stereo 70. In the case of my Mark III's, I built them from kits, so I have initimate knowledge of their history, but new-to-you amps would be prime beneficiaries of this approach. Take my word for it.

Replace All of the Caps

Just assume that all the caps in the amp are bad and you'll be a lot happier about going out of the house to get the mail, for example, with the amps still playing and coming back to an intact house. Consider the age here--we're talking on the order of 30 years. For bias filter caps, use 100mfd@100vdc electrolytics--I order mine from AES. And of course, the selenium bias rectifier has to go--I used a 1N4004, bypassed by a .01@500vdc ceramic cap (from Radio Shack). Remember to orient the diode properly--band end to the transfomer wire--this is a negative voltage supply. I bypass all my silicon diodes with .01mfd ceramic caps--keeps down the digital nastiness.

As for driver board parts, you have some latitude here. Some like the original design--me, I'm married to it as the speakers I use require the original design, as they are interfaced directly into the feedback loop of the amp (motional feedback, you know). I would, however, recommend 716P 630vdc Orange Drops for the .1mfd screen bypass and .22mfd coupling caps, as I've tried Xicon (blue) 630vdc film caps (as they have a 1.0mfd version for the coupling caps), and found them a bit strident, although my all-time favorites are still the 1.0mfd@600vdc GE big yellow cylindrical jobs I put in back in the mid 70's no doubt a recommendation of an Audio Amateur article, but they went with this most recent rebuild (still have them though). I'm hoping the Orange Drops will break in and have the same musical sound they always did. I replaced the small value caps with new silver micas, and closely matched the 47K phase inverter resistors with fresh replacements (still used carbon comps, but 2-watt jobs are a lot more stable over time than half-watters in my experience). Also, as per a note on Ned's website, I replaced the 680 ohm 6AN8 cathode resistor with a 470 ohm job--took Ned's word on that one.

UPDATE: This particular bit of advice is extremely tube-dependent. I've returned to the original cathode resistor value of 680 ohms, as the JAN/Sylvania's I got from Ned weren't developing enough first section plate voltage (pin 6) while in the circuit. So back in went the 680 ohm cathode resistors, and now I'm seeing something like 98vdc on one amp and 102vdc on the other. Previously, with the 470-ohm cathode resistors, I was seeing 74-84vdc--too low for my taste. However, after returning to 680 ohms, I put in a couple of tested-good RCA's and they nearly went off the scale--we're talking 150vdc on the plate! So I can't really see why Ned originally recommended lowering the resistor for use with NOS tubes--it looks to me like you'd have to raise it for RCA's, for example. The JAN/Sylvania's are a LOT happier with the standard resistor than the RCA's I tried, and your mileage may vary with other versions. In fact, if you look at the Dynaco manual, they don't even quote voltage measurements for the first section of the 6AN8. Punks. There should be an adjustment potentiometer in the cathode, in my opinion, but that's me. So before you change the cathode resistor, check the plate voltage on the first section of each of your known-good 6AN8's--you may even have to RAISE the cathode resistor if you're using tubes that reacted ike the RCA's I tried. One thing you'll learn--the weaker the tube, the higher the plate voltage, and the higher the cathode resistor has to be to compensate for it. That's why an adjustable pot makes sense here. Bottom line? Pentodes are extremely tempermental--give me a triode any day.

Also marked for extinction is the original can-type cap. This is the most important point of the entire rebuild, and here's why: the Mark III is a flawed design, unique among Dyna's amplifier product line. The problem is that even with a tube rectifier, until the output tubes warm up enough to draw any significant current, the B+ supply exceeds the voltage rating of even a 525vdc electrolytic cap. While I seemed to get by for almost 30 years with the original can cap in mine (augmented by three 40mfd@500vdc Mallory can caps in parallel with the cap section at the output of the choke that I added some 15 years ago), the jig was up this last Valentine's Day, when one of those 40mfd caps blew a smoky fart and squeezed out a tar poop from its pressure release port. Time to do it right, I thought.

Rebuild the power supply:

So while you can easily replace the original can cap with the best 525vdc cap out there--an 80/40/30/20mfd configuration available from AES, which is PERFECT for the Stereo 70--it won't get it in a Mark III, unless you bring up the amp(s) on a Variac, use a standby switch you'll have to add to the circuit, or something similar. However, if you want to do it right, you'll have to go to series caps with at least 400vdc individual ratings, which is the path I took, using Sheldon Stokes' SDS Labs replacement cap board for the Mark III. Sheldon is moving to New Mexico as of my last correspondence with him, but I expect him to get back to production of these PC boards, which cost $40 a pair unstuffed and mount on existing transformer mounting bolts, and feature combined Panasonic 400vdc electrolytic caps that total out to 41mfd, 30 mfd, 30 mfd, and 30 mfd when installed in the circuit. They have balancing resistors (330K half-watt Xicons), and optional film bypass caps (.018mfd@630vdc Xicons). All parts are available from Digikey and Sheldon even puts up a taping diagram and parts list on his site in .pdf format here. I'm sure Sheldon won't mind if I reproduce the pictures from his .pdf file to give you a better idea of what is involved. Unlike the original can cap, what Sheldon's board CAN handle is a solid-state rectifier. It even has holes in the board for 1N4007's in a full-wave configuration. I tried the board with this option initially, but went back to the GZ-34 rectifier tube, as I got squeamish about hearing a loud thunk whenever I turned on the amps. With the diode rectifier, I was seeing about 570vdc out of the B+ supply under load, which is pretty honking. By the way, it must be said that other power supply modification/upgrade solutions are available, although not all are as inexpensive as Sheldon's. Ned seems to have one up on the Triode Electronics website, Curcio Audio has one, and if I'm not mistaken, Kevin Kennedy has one, too. You can see by the proliferation of replacement power supply design approaches that there must be an inherent problem, and there is. Ever seen such a replacement for a Macintosh? An HK Citation II? Enough said.

Another change I put in my Mark III's some 15 years ago is a bias balancing circuit that allows adjustment of resting plate current for mismatched output tubes, or matched tubes that drift over time (all do). I'll have details of it as soon as I can scan it and get it up here, but for now, the best approach is to replace the single 11.2 ohm cathode resistor with separate 10 ohm resistors and measure the voltage drop across the cathode resistor of each tube and set the bias for 75mA resting cathode current and select tubes to be as closely matched at idle as possible.

You also want to clean all the pots, switch sections and tube sockets with DeOxit5 deoxidation spray solution (another must-have).

One thing I did that I ALWAYS do and nobody ever mentions is replace the line cord with a 3-wire, 16-gauge line cord (or larger)--I used Radio Shack--bypassed with .01@2KV (line-rated) caps from each side to chassis ground. You can use Panasonic X-Y Series (best) or Radio Shack ceramic caps (also good). This is a must-do, in my opinion. The IEC says green is chassis ground, so attach that to the chassis (e.g., transformer mounting bolts).  White is neutral, and goes to the unswitched side of the power transformer primary winding (in this case, the fuse holder).  Black is HOT, so this goes to the AC power (on/off) switch.  Connect a .01@2KV from black to chassis and another from white to chassis at the point of entry of the line cord, and you're done.  No more shocks, no more hum, no more electrical code violations.

Install Frank Van Alstine's Input Bandpass Filter

OK, here's a controversial move. I read Frank's Stereo 70 rebuild article (and that in his opinion the same principles apply to the Mark III's as well), and what he claims made sense to me. I figured the rolloff he suggests would be almost unnoticeable, and you know what?--it is. So I thought that if it would unburden the feedback loop and the rest of the amp with passing frequencies it couldn't handle, all the better, plus it allows upping of the coupling caps to the output tubes and doesn't adversely affect the sound. It does seem to tighten the bass a bit. It consists of a .02mfd@50vdc film cap in series with a 10K resistor to the input grid, which has a 475K (I used a 470K) resistor in parallel with a 1000pf (.001mfd to you) 50v film cap to ground. Neat. I think it has value, but everything is subject to reevaluation.

Tubes--What To Use?

This is the last decision point of the whole deal. As far as driver tubes go, I have in the past used RCA 6AN8's, and I seem to remember some GE's running through here at one time. This time out, I sourced a pair of "matched"--well, the first section is matched tube-to-tube, anyway--JAN Sylvania 6AN8's from Ned. Look good, sound good.

As for output tubes, I've been through the gamut on this one. When I bought the kits, they came with coke-bottle GE's, if I recall correctly (may have been RCA's, but let's face it, they were all Tung-Sols, and I wish I still had some). As time went on, I kept going back to Dynaco for tubes (I live in SE PA, after all--used to run down to their factory in West Philly, and then in Blackwood, NJ before the axe fell). Periodically I received from them either similar tubes as originally supplied, or somewhat later on straight-sided jobs which were probably GE 6550A's, although they had the Dynaco brand on them (still have a couple). Then I went to a local jobber and got either RCA or GE jobs, all because things started to happen in my amps, namely output tube melt-downs--hence my application of the bias balancing circuit. After that, things settled down a bit, but then I read an article in aThe Audio Amateur, as I mention above, about 7027A's being a sweet replacement, and that's what I've been running in triode-connection mode for maybe 15 years up until recently. NOW, I've moved on to EI KT-90's, similarly triode-connected. Although I've had a few problems with quality--one I got out of four had an open filament, and one developed serious softness in a month oof light use (Ned made good on both--great guy)--they are very nice and inexpensive. Too bad about the conflict in Kosovo, as the EI factory is history from all reports. I almost went with Tesla KT-88's, and that will probably be the next move I make, but who knows? I hope it's a while before I have to decide.

Sit Back and Let 'er Rip (first on a Variac):

Bring it up on a Variac, if you have one. Otherwise, pull the rectifier, and check for smoke, observing all the other first-time turn-on precautions and then successively plug in the driver tubes, then the rectifier and then the output tubes (I think I have that right--just get a Variac and you don't have to worry about the sequence all that much). When you think you're ready, warm up the amp with the bias settings down, and then bias the amp for 75mA cathode current per tube and try some program material through it. Mine are stable and powerful, even when triode-connected. They clearly benefit from good output tubes and well-picked components, although many feel that they're a notch below the highest of high-end amps of their era. Still, for sheer power for the buck, they're hard to beat. Just fix the power supply problems before you do any serious listening, but if you don't, think twice about leaving the house while they're running!

Have fun and let me know how you make out.

To get back to the WA3FLE homepage, click here.