Dynaco ST-70: Base Line Testing

dcgillespie

Fisher SA-100 Clone
Subscriber
The Dynaco ST-70 surely must represent the greatest amplification paradox of all time: It was arguably the best selling power amplifier of all time, and is arguably the most maligned such device as well. For a generation who adored it, there have been those in following generations who are quite vocal about it, to the point that the only useful component they claim it has, is its output transformers. Therefore, it is likely the most modified amplifier of all time as well.

As always however, the vast majority of these claims are solely subjective in nature, with little or no measured performance data to back up their claims, or to show the improvement a given modification affords to all its ills. Ultimately then, the sales pitch boils down to the best small lot used car salesman's tactics: Trust me! With so many modification options, and so many new folks arriving at the doorstep of our hobby daily, the plan of attack in restoring these units can be bewildering to say the least.

I own a rather pristine, early model example of this unit, that I keep as absolutely stock as possible, to allow the unit to act as a reference against all challengers of its original humble existence. To that point, everything is completely original in my unit (except for tubes), from the fuse and power cord, to the coupling and power supply caps. At this time, all components are well within spec, with no leakage or signs of physical age at all. In this day then where the restoration battle cry is "Replace all of it!" (which I hardly argue against for long term dependability), this unit stands in stark contrast all on its own. Therefore, it serves its INTENDED purpose for me very well indeed.

I have been asked on numerous occasions to develop an EFB modification for this unit (adding to the great pile of modifications already in existence for it!). Such an effort always starts with baseline testing of a properly operating stock unit, so as to have a bar set to measure any achievement against. I am offering up my reference example for that purpose here. And while the purpose of the testing was originally to develop yet another modification for this unit, I thought that with so many, many other modifications out there -- and all the claims about what you should and shouldn't do to achieve good performance from this unit -- it would be a good idea to do a reset of the much discussed ST-70, and publish just how good -- or bad -- the original product really was. Along the way, some of the more popular modifications will be examined as well.

This will be accomplished by making separate posts in this thread for specific design areas of the original unit. It will include the power supply, output stage, driver board, and overall measured performance as well. Of course, this will all ultimately be related back to the original goal of developing an EFB modification for this unit also. However, of a more general nature, the results will also serve as a good documented baseline for any and all on their own particular ST-70 journey.

To get the ball rolling then, I have provided a few pics of the test volun......er subject. This was a kit model, and I won the lottery with whomever assembled it. It is neatly built, with the underside shot almost mirroring the pic provided by Dynaco in their assembly manual as representing a quality build. Those are the famous "cloth" version of the A-470 output transformers, with everything still clearly being in its original build form.

The tubes supplied with the unit when received were also primarily lotto winners as well, with the output tubes being well matched pairs of low hour Zenith branded genuine Mullard tubes. The weakest pair of these tubes produced 98% of Average New NOS Power Output, while the strongest pair produced 102.6% of this value. The worst pair was matched both statically AND dynamically (i.e. collectively) within 3.65%, while the best pair was matched to within 2.1% under the same criteria. The two pairs then were matched within 2.55% of each other. Clearly, these are superb tubes to use in a reference Dynaco ST-70. The driver tubes were also relatively new Zenith labeled un-carbonized 7199s, of unknown (American) manufacture. One of these tubes has been replaced, for reasons which will be discussed further in the post on the driver board. But clearly, some one or some tech went through and re-tubed this unit from their Zenith stock before it was placed on the (in)famous auction site nearly some 5 years ago now, from which I snagged it. The rectifier tube was a brand new JJ GZ34 S. This tube has also since been replaced (with a very good large bottle Sylvania 5AR4/GZ34 tube), which will be discussed further in the post on the power supply. Suffice to say however, the tubes now installed in the unit are equally up to every bit of performance that the build and condition of this unit is, allowing it to represent all that the Dynaco ST-70 was intended by David Hafler to be.

In all then, I have no wish to pile on to any of the popular notions out there, but rather, to let the chips fall where they may regarding the true performance of the original design, and that of some popular alterations as well.

More to follow!

Dave
 

Attachments

  • Stock Dynaco ST-70 Tests 006.jpg
    Stock Dynaco ST-70 Tests 006.jpg
    106 KB · Views: 908
  • Stock Dynaco ST-70 Tests 007.jpg
    Stock Dynaco ST-70 Tests 007.jpg
    102.3 KB · Views: 1,143
  • Stock Dynaco ST-70 Tests 005.jpg
    Stock Dynaco ST-70 Tests 005.jpg
    135.5 KB · Views: 1,106
Hiya,

No grass growing beneath your feet Dave .. you go go go .. :)

I will be of course paying close attention to this and picking up what stray emanations that manage to penetrate my cranium.

Frannie
 
I've often wondered about some of the kits that are out there. Are they beneficial or not. Some kits are so drastic there's nothing left but the transformers. I'll be watching your thread with great interest. :thmbsp:
 
Measured Stock Performance

While I have individual information on each channel, the information was close enough to simply average the results for posting. This also helps to minimize information overload.

1. POWER OUTPUT AND DISTORTION: As measured into 8 ohms from the 8 ohm tap, with the Biaset voltage set for 1.56 vdc in each channel under stabilized quiescent conditions. The power output figures given are maximum power output levels at the onset of clipping, while the THD numbers are produced at a power level that is 1 db down from rated power (or about 27 watts) -- this to conform to the distortion specifications as provided by Dynaco.

Single Channel -- @ 20 Hz = 39.8W RMS. @ 1 db down, THD = 1.0%
@ 1 kHz = 45.6W RMS @ 1 db down, THD = .18%
@ 20 kHz = 43.5W RMS @ 1 db down, THD = 2.3%

Both Channels -- @ 20 Hz = 31.6W RMS @ 1 db down, THD = 1.9%
Operating @ 1 kHz = 37.5W RMS @ 1 db down, THD = .32%
@ 20 kHz = 34.1W RMS @ 1 db down, THD = 3.3%

2. IM DISTORTION: .45% at 35 watts equivalent power output. .74% with both channels operating.

3. FREQUENCY RESPONSE: (ref 1 kHz)

@ 10 Hz = -.3 db
@ 1 kHz = 0 db
@ 20 kHz = -.25 db
@ 30 kHz = -.5 db
@ 36 kHz = -1.0 db
@ 40 kHz = -1.2 db

4. HUM AND NOISE: 92 db average below rated power

5. STABILITY: No value of capacitance only loading would cause the amplifiers to become unstable.

These figures were all produced with the unit operating directly from the AC line in my home, which was providing its typical 122.0 vac to the unit during the tests. The power output figures were rerun with the unit operating from 117 vac (the published operating voltage for the unit), which reduced all power output figures by about 2.5 watts. The reduction in line voltage did not have any significant effect on the distortion levels produced.

Within the results then, they meet the majority of the specifications as presented back in the day, which of course were based on single channel performance figures -- this because no standard existed for performance presentation at the time. The notable exception is for the 20 kHz THD figure, and frequency response. To that point, I've never measured an ST-70 that was only .5 db down at 40 kHz, with the performance my unit shows being somewhat better than typical in my experience. I can only assume that the specification given then is somewhat optimistic. I have found it is common for ST-70s to show a HF response that is between 1 and 1.5 db down at 40 kHz, relative to a 1 kHz reference. What is important, is that the frequency response within the 20 Hz to 20 kHz audio bandwidth is flat within .25 db, which is very good, particularly so for equipment of that time -- and especially so for the level of HF stability achieved.

As for the 20 kHz THD figure, the distortion level produced is typical of that achieved AFTER installation of the HF stability networks. These networks reduce the open loop gain at 20 kHz, which then increases the distortion produced at 20 kHz due to the effective reduction of the feedback factor. But they are necessary to produce a practical feedback amplifier that is stable into varying loads. With these networks lifted in each channel, 20 kHz distortion comes in at an average of .48% at 1 db down from 35 watts RMS power output.

I have always felt that this practice was deceptive relative to the way the specification is given, but have come to the conclusion that it was accepted practice back in the day none the less. These networks came late to the amplifier design party, after a number of early feedback amplifier designs were branded as tweeter killers (Heath W-3M for example). The networks were then added to aid in stability performance, but they caused the HF distortion performance to deteriorate from the specifications of previous models published without such networks. Therefore, for lack of any better explanation, I can only assume that many manufacturers continued to publish distortion specifications based on that of the raw amplifier circuit (that is, without the finished HF stability circuits added) -- likely under the guise of preventing confusion, and certainly of course, to not to have their product look worse than that of other manufacturers who were following the same practice.

In any event, any one of many, many popular amplifiers will exhibit this rising distortion with rising frequency characteristic, with the Dynaco MK III, numerous Heathkit and Eico models, and many others all clocking 20 kHz THD figures at the specified power levels well in excess of 2%. Of note however, this is an area where Heathkit really shined, steering very clear of such simplified ratings, instead, choosing to offer a graph with most models, which clearly shows the rise in distortion at higher frequencies and high power output levels -- due to the inclusion of these networks. Therefore, it is, what it is, with the distortion levels produced in the ST-70 at 20 kHz not being indicative of any concern for proper operation.

So, this particular ST-70 is a healthy specimen, making a great platform for reference work, and to make various tests on when needed.

Pics include:

1. All warmed up ready for testing.

2. A 10 kHz square wave at 1 watt into an 8 ohm load looks identical on both channels. The somewhat slanted sides are indicative of the frequency response figures produced, while the flat wave-top indicates a high degree of damping and stability.

3. With a cap only load, the amplifier is no where near becoming unstable.


For a design of the late 50s, this overall is quite enviable performance, that had other other manufactures scrambling to catch up to. Most notable was the level of stability produced in light of the other performance levels achieved. That it is a product of David Hafler is no real surprise, since he was a primary bandleader campaigning for improved stability margins in equipment of that day.

Next up, the power supply.

Dave
 

Attachments

  • Stock Dynaco ST-70 Tests 001.jpg
    Stock Dynaco ST-70 Tests 001.jpg
    77.9 KB · Views: 539
  • Stock Dynaco ST-70 Tests 002.jpg
    Stock Dynaco ST-70 Tests 002.jpg
    43 KB · Views: 486
  • Stock Dynaco ST-70 Tests 004.jpg
    Stock Dynaco ST-70 Tests 004.jpg
    47.2 KB · Views: 453
Last edited:
Looking forward to this latest from Dave "The Tube Whisperer".

I've "redone" a few of the ST70's but never really had a good enough one to get a good baseline before hand. I have, however, objectively tested (however humbly) the Dynaco A470 output transformers against some other well known brands. I was skeptical they were as good as they were made out to be. They really are that good, even better than I could have imagined IMHO.
 
The Power Supply

As with any stereo amplifier operating from a common power supply, compromises are made, even beyond those compromises made in the power supplies of mono amplifiers. The question then becomes, how well were those compromises made? We shall see.

In the St-70, both channels are powered from the same power supply distribution points. But with channel separation measured at a 58 db at 1 watt, the 55 db Min Channel Separation spec was met handily. So there is no real compromise here, as the separation figure achieved is very good.

The usual suspect targeted for compromise criticism in the ST-70 is the power transformer, to which could also added the rectifier tube, and choke.

The case against the power transformer centers on its current capacity, and the temp it operates at. However, it still allows its individual per channel power output level to be achieved across the entire audio bandwidth, even permitting it within the majority of the audio bandwidth with both channels driven. So its current capacity is hardly of the grossly inadequate grade that some give it. And while the transformer does run warm, it is hardly noted for a failure rate along the lines of say the Heath series of amplifiers up to and including the W-5M. Therefore, with good dependability, and an ability to allow both channels to perform well at the same time though out the majority of their rated capability, the compromise this transformer represents was still a very good one, in view of the price point and performance level the unit sought to achieve.

A much bigger compromise -- in my opinion -- is the use of a single rectifier tube to support a 70 watt amplifier. Competitors with competing products from the likes of Altec, Fisher, and Pilot all used dual rectifier tubes in their products of similar power levels. As a result, there is no doubt that the rectifier tube in an ST-70 is the hardest working tube in the chassis, and often will show the highest failure rate as well -- even when using NOS examples. Which brings us to the new JJ GZ34 rectifier tube that was installed when I purchased this unit.

Nobody will doubt the bad press this tube has gotten -- and is well deserved of. Some will stand by them, but many more will not. It being the first such tube I've come across from JJ, I was quite leery of powering the amplifier up when I first got it with this tube in place. Still, I forged ahead and powered things up, where the tube allowed both channels to bias properly to the correct Biaset point, and even allowed a brief full power run in both channels at 1 kHz, before I powered things down to prevent a chance of meltdown. So even though this tube worked -- and worked well for the short test period that I had it in this amplifier -- I wouldn't trust it in this environment for any kind of long term operation. However, to be fair, this tube has worked perfectly in a lesser voltage/current environment, where the B+ was under 350 vdc, and the peak current draw was no more than about 180 ma or so. In any event, with that, the big bulb Sylvania 5AR4/GZ34 was then installed in the Dynaco, which has performed flawlessly in the ensuing years since then.

Until I performed sustained full power 20 Hz testing with both channels driven that is. Shortly into that test, the Sylvania rectifier tube buckled and showed its extreme displeasure with this treatment, as signified by a distinct "click" that was heard not only in the rectifier tube, but also resonated within the power transformer as well. And while it recovered itself gracefully from this single event (continuing to operate without blowing the fuse), it was a sure indication of how overworked this tube is in the ST-70. And well it should have protested. Both channels in this scenario were trying to draw at least a combined 500 ma from the tube, which is well beyond what it is capable of (over twice what it is rated for). Could you imagine what the JJ tube would have done? Or how about one of the new Dynaco derived 120 watt amplifiers? If the GZ34 can't really handle the full power needs of an ST-70, how could it handle one of those beasts? The heavier duty GZ37 handles only precious few more ma, so even it is no real answer -- even for the ST-70. Even if continuous operation near its maximum rating under normal conditions doesn't send it to an early grave, any significant hot switching event will. Which all goes to show just how limiting the use of one rectifier tube is in this amplifier. The unit can get by with one tube for typical use, but it's a definite limiting factor relative to the performance this amplifier is capable of.

Going SS is an option, but most approaches are fraught with problems. Because I knew the rectifier issue would be a problem in developing the data for EFB, I purchased a SS plug in replacement rectifier from Triode Electronics to keep things simple. This is not one of those fancy ones with time delay built in, or with built in dropping devices. It's just some SS rectifiers molded into an octal tube socket base.

As for the other SS devices, I have heard here on AK of a few failures of the copper units, and the units with built in relay delay are often prone to producing heavy surges in the output stage, causing accelerated and uneven tube wear, as well as serious thumps in the speakers when the relays engage. So, the base garden variety plugin SS replacement rectifier got the nod.

With these however, if you simply plug them in and operate the unit from today's line voltage, then initial B+ rises to 525 volts, or right to the limit of the filter cap voltage rating. This is not particularly a problem for the amplifier portion of the unit (although it would be a good idea to replace the coupling caps with 600 volt devices in going this route), as the fear of cathode stripping has been roundly discredited, having no basis with receiving type tubes. But going this route does cause another problem.

Since SS devices have hardly any voltage drop across them, simply replacing them where a rectifier tube was that had a voltage drop across it causes the peak currents drawn from the power transformer to become elevated -- which gets transformed into heating in the power transformer. Therefore, whatever temperature reduction you hoped to achieve in the power transformer by eliminating the rectifier tube, gets counteracted by the increased heating from the increased peak currents the SS rectifiers create. But there's another basic problem.

Operating this amplifier from a typical 122 volt AC line causes the amplifier tube heaters to operate at nearly 6.8 vac, which is uncomfortably high at nearly +8%, not to mention that this too is elevating power transformer temperature as well.

A better compromise today, is to do a proper job of installing SS rectifiers, and operate the unit from an appropriate buck transformer (ideally, 4A minimum).

To do a good job of mimicking the voltage drop of a GZ34 tube, there should be a 50 ohm 10 watt resistor installed BETWEEN the output of the SS rectifiers, and the FIRST filter cap. Installing a resistor at this location will control the peak currents drawn and improve regulation, providing nearly the same B+ level that a GZ34 would under any condition of operation. Installing a resistor after the first filter cap, would hurt regulation, and NOT limit the peak currents drawn from the power transformer.

When using this rectifier approach with a buck transformer to provide 117 vac, the initial B+ rises to 500 volts, but then falls to nearly the same B+ level as produced by a GZ34 in the same scenario, yet is easily able to handle the maximum current needs of the amplifier without concern. The tube heaters operate from 6.4 vac using a 117 volt line, the bias controls barely need to be moved from their GZ34 setting, and importantly, using these two expedients cools the power transformer notably. with a buck transformer, the SS plug in rectifier and dropping resistor installed as described, and the unit biased to Dynaco recommendations, the hottest point on the power transformer was just 130F in my lab after two hours of high power testing. This of course was with the cover off, and the ambient temp in here is only 70F, but this is still quite cool operation for the transformer. The amplifier was graceful under all conditions with this approach, showing no concern over the loss of the rectifier tube. All in all then, they are improvements worth making to any stock ST-70.

The choke is also somewhat of a compromise, providing good filtering under even medium power conditions, but completely saturating under conditions of high power output. Still, it does make for a quiet amplifier during normal use, with its high power limitations covered by the sound level of elevated power output. In any event, it size was a good compromise, helping to achieve quiet operation of the unit with typical use.

And finally, there's the selenium rectifier. Selenium rectifiers are well known to ultimately fail in high current applications (those where significant heat is created within the rectifier) JUST AS they are well known to hardly if ever fail at all in very low current applications, such as the bias supply represents. I would hardly suggest not to replace it from an ultimate durability standpoint, but to replace it simply because it's a selenium rectifier -- in this application -- has little merit.

Next up, the output stage.

Dave
 
Jay, I wonder if your thinking of the designer of the SDS PS boards, as he has an all 12AT7 LTP design on his web site. Can't think of the name of it right now, but will add it when it comes to mind. Found it on Sheldon Stokes SDS website, but 12AU7 LTP.

http://getinthewoodchipper.com/



Dave, great info and reading as always!

Maybe try the diode mod. to help that poor tube as it takes a big load off, plus you still get the slow warm-up benefit the GZ-34/5AR4 is famous for.

If you could do a test on DCR in the PS that would be great, as some on this forum and others say it really helps the dynamics of an amp if it's as low as possible in the B+ to the output transformer. Amps that are said be be great because of this are the HK Cit II, Eico HF-81 and some Heathkits with no resistors added.
 
Last edited:
.... Or how about one of the new Dynaco derived 120 watt amplifiers? If the GZ34 can't really handle the full power needs of an ST-70, how could it handle one of those beasts?

Dave

I made mention of that to someone, and they acted like I was off my rocker for saying it. I didn't have numbers to back it up, but my gut feeling was that if the rectifiers burn out on a regular basis, they are being abused.


Power supply thought. What about SS rectification with the DC output of the diodes fed through a 5AR4 wired in parallel? That should double the current capacity if the plates are tied together. At that point it would only be used for voltage drop and time delay purposes.
 
Power supply thought. What about SS rectification with the DC output of the diodes fed through a 5AR4 wired in parallel? That should double the current capacity if the plates are tied together. At that point it would only be used for voltage drop and time delay purposes.

Six of one, half dozen of another isn't it? Same current through two plates in the tube either way?
 
Gonna watch this one closely. Although I'm still waiting patiently for the stand-alone EFB boards. :thmbsp:

-D
 
Six of one, half dozen of another isn't it? Same current through two plates in the tube either way?

The standard diode mod puts a single silicon diode in series with a single plate, so at any given time only one plate in the tube is actually passing any current. All its really for is to take care of the reverse voltage that the tube sees. It doesn't actually do anything as far as current handling ability or any of that. If you connected the two diodes together upstream of the plates to create DC output, and then connected that DC voltage to both plates at the same time, both plates would be carrying the load instead of just a single. That should pretty nearly double the current capacity of the tube. Basically you're making the 5AR4 into a half-wave rectifier at that point.
 
The choke is definitely a weak point in the ST70 . A couple low noise SS rectifiers do help out and a second choke can be tucked in under the chassis , to help drop the B+ to a reasonable level with a 5 uf to 10 uf 600 volt polypropylene cap right after the the diodes .
 
The choke is definitely a weak point in the ST70 . A couple low noise SS rectifiers do help out and a second choke can be tucked in under the chassis , to help drop the B+ to a reasonable level with a 5 uf to 10 uf 600 volt polypropylene cap right after the the diodes .

Is the second choke wired in parallel to the first C354 or in series?

-D
 
Great comments! Paralleling the rectifier tube plates and placing it in series with an up stream SS rectifier array is basically the same as placing one diode in series with each plate: It helps the poor PIV performance of modern GZ34 replacements, but unfortunately fails to address an additional principle concern in this case, which is current handling ability.

In normal operation, each plate does in fact time share the work -- but they are sized and rated on that fact. As well, so is the cathode rated for a total current flow that is similar to that which can be safely handled by the two plates together.

Basically, for a given heater wattage, only so many electrons can be excited off the cathode, which becomes a limiting factor in this case. The tube is simply designed to reliably pass no more than about 250 ma on a continuous basis, assuming a capacitor input filter arrangement -- and there in lies the crux of the matter: Both channels of the ST-70 far exceed this capability well before full power conditions are reached, placing excessive demands on the tube. When those demands become excessive, the plate and cathode structures can overheat, and warp so as to effectively alter the PIV rating of the tube, causing it to voice its displeasure as it did during my full power 20 Hz tests with both channels driven. A single tube is simply insufficient for the job.

The choke is definitely a compromise as well, as mentioned. However, in terms of being an effective benefit in typical use, it was a good one. No doubt that multiple chokes can be used to either increase current capability, or increase voltage drop. Wired normally, a single choke dissipates nearly 3 watts of heat under quiescent conditions, and up to 15 watts at rated power output in both channels.

Dave
 
Back
Top Bottom