tubes 4 hifi
 "the path of least resistance is through a vacuum"

Comparison Test of Dynaco ST-70 tube amplifier modifications

by Roy Mottram

PLEASE NOTE: this article appeared in Volume 2 Issue #1 1990 Glass Audio magazine,
alot of the information on the various modifications has changed in the past 30 years!!

for a year 2002 update of this article featuring current sources
of modified driver boards for the ST70, go to mods02

This test was made to determine the value of various modifications on the market. The stock ST70 was tested;
the Audio by VanAlstine (AVA) board was tested; the Sutherland hybrid board;
the Vacuum Tube Audio (VTA) board; the Curcio board; and the GSI board.

In order to eliminate any variances between amplifier tests, only one ST-70 amplifier was used for this test. The same matched quad set of output tubes was used for each test, and the same matched pair of 7199 driver tubes was used for each test which included them. The associated equipment used for the listening evaluations included the Curcio Daniel preamp, Vandersteen 2C speakers, and Sony 650ES CD player.

The stock ST-70 was obtained from Sound Values for $199. This was done for two reasons rather than buying one locally. One was curiousity, as I have previously bought several used ST70s locally in the past. Secondly, I wanted to see what someone would get for $199 thru the mail. Sound Values offers a 90 day parts and labor warranty, as well as a ten day satisfaction guarantee.
2001 update - Sound Values is now known as Sound Valves, and no longer sells used Dynacos, or used Dynaco parts,
but they do sell a new design "clone" ST70 amplifier, and they also sell a replacement driver board using an updated circuit design.

The amplifier received had the usual amount of slightly pitted chrome, but was not rusted. It had not been cleaned. But it did contain a full set of new tubes (worth $90) and a new quad capacitor, as well as a silicon diode to replace the old selenium rectifier. So it was a good value and definitely worth the price and a recommendation to others who might want an ST70 in good condition. Of course, they also sell new kits for $499.

One note about GSI. After waiting for two months after ordering with absolutely NO response from GSI, I wrote them a letter asking about the delay and reminding them of the federal laws regarding delivery within 30 days for mail orders. After three more weeks of waiting with no response, I contacted the Better Business Bureau and the State Attorney General. GSI sent me a letter saying "don't worry, be happy!" But only after waiting another three weeks during which time I'm sure they heard from the BBB did I receive the PCB, along with a very harsh letter complaining about my lack of patience!

The stock Dyna ST70 is an excellent amplifier, the most popular tube amplifier ever sold. The Dyna output transformers are well known to be of exceptional quality and design. The EL34 output tubes are revered by many tubeophiles for their sound quality. With the exception of the 7199 driver tubes, and the limited power supply capacitance, the Dyna ST70 is still an excellent design and an exceptional value. By replacing the PCB parts with metal film resistors, and Wonder Caps, and boosting the power supply capacitance, it is still a contender among modern competition such as the Conrad-Johnson MV45, and the Vacuum Tube Logic Stereo 45/45, both of which cost over $1500 new. A new ST70 with the basic mods costs about $600.

The Audio by VanAlstine (AVA) modification consists of a new driver PCB which also contains upgraded power supply capacitors. It costs $200 for the kit (PCB and parts) or $75 for the PCB only. I purchased the PCB and then obtained the parts from DigiKey and Mouser for a total cost of $35.
This modification was a personal disappointment to me. VanAlstine has a reputation for very good circuit modifications. However, in this case, I'm glad I didn't spend the full $200 only to find out that electronically, it is identical to the original, using two 7199 tubes. The only differences are the addition of "spectrum filtering" consisting of two resistors and two capacitors which limit the input signal to a bandwidth of 20 Hz to 20 KHz, to minimize non-linear distortions created by the feedback signal due to the limited open loop bandwidth of the driver circuit design, plus bigger capacitors for the power supply, and larger interstage coupling capacitors. The parts quality is good but not great (film capacitors rather than polypropylene). I was really expecting a total redesign.
Also, the PCB is larger than the original in order to accommadate the larger power supply capacitors. It is mounted 1/2" above the chassis on standoffs, which gives it a rather strange and awkword appearance.

The Sutherland hybrid modification consists of an entirely new PCB driver design. It uses JFETs and MosFets in combination to drive the output tubes. The board is very well designed and very attractive looking. All connections to the board are marked with the original number markings, and with the included hookup diagram it is a cinch to install. The PCB comes with all parts assembled and tested for $80 which is a bargain.
Sutherland also has a separate power supply modification available which I included in the test. It consists of two PCB's which mount innovatively to the bottom of the output transformers with 1/2" threaded spacers. One PCB consists of six 47 uF 450v capacitors which are wired in parallel with the stock Dyna quad capacitor. The other PCB consists of the bias power supply filtered with two 470 uF capacitors. The two board set costs $100, which seems about $30 high to me. Total price of the Sutherland modification is $180. Not bad compared to the others considering this is already assembled and tested.
2001 update - Sutherland quit selling this modification in 1990.

The Vacuum Tube Audio (VTA) design was more like what I'd hoped to see in this comparison test. It is a total redesign using three 6201 (mil-spec 12AT7) tubes. This design is also unique compared to its competition in that it integrates a beefy power supply and bias supply on the same PCB, with one note. The area of the chassis previously used for the original bias pots and quad capacitor are punched out with a 1-1/4" hole punch, and the new power supply capacitors fit up thru these holes. (The necessary punch is available from Mouser Electronics for about $19) The GSI power capacitors also fit here, but they aren't integrated onto the PCB. Also this PCB was laid out with plenty of room for WonderCaps. If for some reason someone didn't want to punch out the chassis, this area of the PCB can be trimmed off without consequence. The stock Dyna power supply could be used, or even the Sutherland power supply PCBs. The layout was very well thought thru. I also found out after the test while finishing this article that VTA has another PCB which attaches below this one, which has the option of a regulated high voltage for the driver board, and individual LED status for the bias of each output tube.
2001 update - this PCB has received SEVERAL upgrades over the past ten years
and is still available from us, no hole punching, it is a drop-in replacement for the original PCB.
Please see our ST70 page

The Curcio modification was written up in Glass Audio Vol 1 #1.
This new design used two 6DJ8 tubes per channel in a constant current cascode differential configuration. The documentation and PCB layout are excellent. There are two boards for this mod, one the driver and the other for high voltage plate and screen regulation. The second board wasn't used in this test, as it adds considerably to the cost and complexity. The power supply from the VTA mod was used for both the Curcio and the GSI.
This board is about 1" longer than the stock Dyna driver board, and is therefore designed to mount 1/2" above the mounting surface. This is necessary if used with the second regulator board, but since I didn't use the regulator board, I managed to mount it only 1/8" above the chassis and still clear all connections safely. This makes for a much more presentable look to the amplifier.
2001 update - Curcio still sells this mod, and they also sell a less expensive and simplier driver board for the ST70

The GSI modification is also a new design for 1989, replacing an older design made back in 1985. Each channel uses a 12AX7 as a diff-amp connected to a constant current source, and then direct coupled to a 6FQ7 also configured as a diff-amp with plate follower to the output tubes. The PCB looks like a nice design, however, the holes for the tube sockets and power resistors are too small to be useful, and the pads are too small to enlarge the holes. I was able to get around this problem by attaching wire leads to the sockets and then soldering them to the PCB connecting traces. Also, although there is alot of open space on the board, the components are crowded together. They could easily have been spaced out with the available room.
No coupling capacitors are used except to the output tubes. The design specs .1 uF for these, which seems small compared to the other designs using .47 uF or more. Also, the hole spacing for those caps is extremely small, requiring small film capacitors. I was able to get around this limitation by using a trick from VTA, mounting the capacitors to the other side of the PCB, which let me use larger .22 uF WonderCaps.
The accompanying documentation was better than what I've seen from GSI in the past. However, it was incomplete, with references to other pages that weren't included. Some component values weren't specified, especially notable the proper R-C network for the feedback. The parts layout diagram was almost completely unreadable, and a few parts were mislabeled. Thanks only to a clear schematic and through study of the PCB was proper construction possible.


Dynaco ST70 stock, and Audio by VanAlstine mod - two 7199 tubes are used, one per channel. This tube has a triode voltage amplifier and a pentode current driver output.
Open loop response was a shocker. Bandwidth is only from 50 Hz to 5500 Hz, with good square waves from 100 Hz to 2000 Hz. Open loop gain is only 16. ___ db feedback.
The only way this circuit is getting by is thru heavy feedback. Closed loop response does extend from 5 Hz to 22 KHz, with good square waves from 100 Hz to 5 KHz. The only difference between stock and AVA is that the AVA is purposely bandwidth limited to keep the circuit operating within its useful range. Low end response is rolled off below 20 Hz, and the high end rolled off above 25 KHz (which the circuit didn't reach with the tubes used).
The stock amp maximum power output was 32 watts per channel, but was up to 35 watts per channel with the AVA, probably due to the improved power supply.
The AVA PCB is larger than the stock Dyna at 5" by 7", with a cutout for the tube rectifier and access holes for the bias pots. It must be mounted a minimum of 1/4" above the chassis.

Sutherland - no schematic was provided, and I couldn't track down the part numbers on the JFETs and MOSFETs to determine either the pinouts or characteristics. Sine wave frequency bandwidth was from 5 Hz to 6 KHz, not much better than the stock Dyna. Square waves remained fairly square from about 60 Hz to 3 KHz. Open loop gain is 47. ___ db feedback. Maximum power output was 35 watts per channel. This board has the same dimensions as the stock Dyna, and mounts exactly the same.

VTA - three 6201 (mil-spec 12AT7) tubes are used. One half of the first tube is used for each channel as a voltage amplifier. One each of the other tubes is used per channel as a combined phase splitter/driver in a long-tailed pair configuration. 13 db total loop feedback is used.
Parts quality is top notch, with mil-spec tubes, metal film resistors, and WonderCaps. There is an individual bias pot for each output tube, as well as a signal balance pot for the phase splitters. The bias is set for 40 ma, about 20% less than stock to reduce heat and conserve tube life. Power supply capacitance for the B+ totals 470 uF. A future version will allow the option of tube or diode rectification, with isolated capacitance for each channel, and double the total capacitance; as well as HV regulation for the driver and LEDs indicating the need for bias adjustment if it drifts out of spec.
Open loop response of this design is incredible. It has flat frequency response from 7 Hz to 70 KHz, with good square waves from 7 Hz to 30 KHz. Open loop gain is 40.
The PCB is 8-1/4" by 5", with a cutout for the rectifier tube. Max power output is 40 wpc.

CURCIO - provision was made for using film or polypropylene coupling capacitors. The driver design also uses HV regulation. Total loop feedback 15 db.
This design uses the stock Dyna bias pots, with an optional trim for each output tube.
Open loop response of this design was the best of all tested. Sine wave frequency bandwidth was 5 Hz to 40 KHz. Open loop gain is 90. Maximum open loop output signal was 200 v p-p! Max power out 37 wpc. PCB size is 7" by 3-1/2".

GSI - Total loop feedback is 18 db. This board is about 2" longer than the stock Dyna, and can be mounted 1/4" below the normal surface. There are individual bias pots for each output tube.
Open loop sine wave bandwidth was from 5 Hz to 70 KHz, with good square waves from 20 Hz to 30 KHz. Open loop gain was 45, with maximum output of 55 v p-p. Maximum power output 38 wpc.
PCB size is 8-1/2" by 3-1/2". If the PCB has been 1/8" less wide, and two components moved slightly in position, it would have been possible to flush mount the board to the chassis. The two channel grounds are not connected, and if a common ground had been made, the board easily could be 1/8" narrower.
PCB traces are extremely thin, at .025 compared to the other designs ranging from .062 to .125 wide. Another problem in the layout of this board is the mounting holes are not spaced properly, they are 1/10" of an inch off. Also, due to the fact that the traces run too close to the mounting holes (again, plenty of room on the board to reroute), insulated spacers must be used to mount the board to the chassis.


stock Dyna and AVA mod - the ST70 is the classic tube amp, and it still sounds good. It definitely has a warm, smooth, romantic quality to it, with fair but not tight bass. The AVA mod does help slightly to tighten up the bass, probably due to the increased power supply capacitance. Neither of these versions has a good high end, because of the limitations in the design of the driver circuit. I'm not saying it sounds bad, because it doesn't. It sounds pretty good, especially for a $200 or $300 amp. The first impression one has in switching to a tube amp from a "modern" transistor amp is the openness in the sound, the transparency to the music. This was fairly true here, but not completely. The dynamic range of the music sounded compressed and slightly harsh, with a slight veiling of the soundstage (though not as veiled as my backup Carver amp). Music was reproduced with a fair amount of detail and imaging, but it was much apparent in listening to the other mods that these two versions were in a different league than the rest. Acoustic guitar reproduction was extremely natural sounding.

Music used for these listening notes were all taken from CD for ease in use. Only a few CDs which I know to be well recorded and "natural" sounding were used.
These included the Opus 3 Test Record 1, "Depth of Image"
(this disc should be a requirement for anyone claiming to be an audiophile);
the DMP jazz sampler "A Taste of DMP"; Joe Jackson "Body and Soul" (A&M);
Eric Marianthel "Voices from the Heart" (GRP);
Chick Corea "Electric Band" (GRP); and James Newton Howard (Sheffield #23).

Sutherland - this mod definitely changed the sound of the amplifier from being warm and rounded, to being more neutral. The music came thru very clean and detailed without being harsh. The sound was open without the veiling noted above, although the soundstage seemed to be more recessed. Cymbals and bells were reproduced with a clean airy sound around them. The bass reproduction was about the same as the stock Dyna, probably limited by the power supply. There was a noticable lack of impact with too much smoothness. The dynamics of the music was better but still not as good as the three mods below.

Vacuum Tube Audio - after listening to this version for a minute or two, my heart was moved. This was music! I was no longer listening to a modified ST70, but to a great amplifier. The imaging was incredible, over and over again in my listening notes this was noted. I listened to this amp for a solid three hours straight, playing all of the CDs noted above, plus a few more.
In the search for the right words to describe the sound, my notes applaud the improved clarity, transparency, and detail in the music. The music sounded live and realistic, concluding with the note that the 3D image of the soundstage defines the word holographic. Bass notes were reproducted with depth and tautness totally lacking in the mod versions above. 2005 update - see our ST70 page for several customer reviews.

Curcio - this was musical estasy part 2. My first impressions were the tightness and cleanliness of the music. It was reproduced with full dynamics, with dead quiet during the gaps. Reproduction was very transparent and very revealing, although it seemed a little dry. This is probably a characteristic of the 6DJ8, which has none of the usual "tube sound" at all. Almost sterile. This characteristic seemed to improve after an hour or two of listening. The soundstage wasn't pushed forward or pulled back, but just right, although it didn't seem quite as spacious in height and width as the VTA. Depth was very good. Bass notes were solid, tight, deep, and natural. 2005 update - see our ST70 page for several customer reviews which rate the Curcio mod much lower in sound quality than our VTA70 mod.

GSI - this mod had several characteristics of the last two. Lots of space, transparency, cleanliness, and quiet. The volume had to be turned up alot, because there was less gain than the others. Not as sterile as the Curcio, not as holographic as the VTA. An excellent performance.


  tubes 4 hifi
 "the path of least resistance is through a vacuum"