Banner for Eico ST-70 upgrade article

The Eico ST-70 is a 35 watt per channel vacuum tube amplifier from 1962, which is great in many respects. This project fixes a few flaws which prevent it from reaching its full potential. Here are some before and after results:

Table of ST-70a output power versus frequency
Table of ST-70a line-level signal-to-noise ratio

You can choose which mods you would like to use. The separate mods are:
Phono preamp, driver balance adjust and line amp/power amp. [Please note that the line amp and power amp mods must go together.] Get the illustrated, step-by-step instructions in the download:

ST-70A mods package icon

-Eico ST-70A mods article (1MB)

Chan-1 schem illus

20 - 20kHz phono response in dB

Power amplifier modifications


Reader Comments

Posted by Steve L. November 12, 2019 - 06:43 am
Hi Peter, Thank you for your posting and kind remarks. First, your mention of "shiny blank piece of steel" brings to mind the fact that many ST-70s today suffer from corrosion and rust problems. I trust that there will be appropriate plating or coating to preserve the chassis.

- "...could a 100k pot be used...without the 68k resistor" -- Yes, you are correct that the nominal value is expected to be roughly 57K and thus replacing R27 and R28 with 100K pots should work well.

- "One option for improving performance of the output stage is to regulate the screen to maintain a constant difference between the plate and screen voltage through Zener diodes...Is it worth doing either of these for better performance?" -- There is little doubt that reducing screen voltage droop will increase the output power but this would also increase screen power dissipation. The 7591 tubes work pretty hard in this amp and the dirty little secret of these and similar power tubes is that screen dissipation may be more limiting than plate dissipation. It's also insidious because a glowing screen is partly hidden and tends to be obscured by cathode glow. The tube makers weasel-worded the screen power spec by setting it at only 3.3W but adding a footnote permitting up to 6W "during the periods of maximum input of speech and music signals". No mention of how long those periods are allowed to last. (How long can I test it at 6W?) A datasheet example somewhat like the ST-70 shows almost 5W screen power at max output. I would have to see screen power test data before endorsing a mod which increases screen power in this amp.

- "In addition to your modification to balance the 6SN7 driver/phase inverter, another modification I have regulate the cathode current to 6 - 8 ma versus grounding through a 18k resistor...does it substantially impact performance?" -- Increasing the impedance of the cathode current source does make the second output of the phase inverter closer to the same level as the first output, given that the plate loads are the same. Of course, Eico's circuit and my version adjust the plate loads to take the cathode resistor effect into account. But having the AC balance adjustment corrects any unbalance from the phase inverter or power stage, so adding a cathode current source seems superfluous and isn't an adequate alternative to having the adjustment.

- "Removing the rectifier tube and going solid state seems like it could be beneficial as it takes the filament load from the transformer...Unfortunately the resulting higher voltage may require additional dropping resistors...Any thoughts?" -- Yes, it would be nice to reduce the load on the power transformer. One problem with the solid state rectifier would be the fact that initially, the high voltage supply chain comes up without a load, making its initial voltage far higher than it would be if the delay provided by the GZ34 weren't there. I guess you could use (much) higher voltage caps in the supply chain or provide a delay relay but is all that really worth it? The GZ34 burns 9.5W of heater power. If the transformer is 85% efficient, the transformer dissipation saved will be about 1.4W. I doubt this will make a big difference in temperature.

- "Does DC for the filaments of the preamp tubes really produce change that is audible?" Yes, particularly if done for the line stages as well. Hum in middle-class tube preamps was one of those plagues which I've had to live with all my life. I've often dreamed of doing a conversion (starting when I was a teen!) but somehow never got around to actually doing it. Not only does it remove the tube-induced hum, but it also removes the 6.3VAC snaking through the sensitive phono circuits (shudder). However, you would need to be careful that the current pulses charging the filter caps are not comingled with signal grounds and that the DC is free of higher frequency line harmonics. A possible issue is that, if you're using the existing 6.3V windings as the source for the DC heater supply and have a filter cap at the rectifier, the spikes of current will impose higher frequency harmonics on the 6.3VAC. The concern is that this could get into the circuits with AC-powered heaters. You can't use a choke input filter in this case because the resulting DC voltage would be too low. I don't know how much of a problem (if any) the harmonics on the 6.3VAC would be with the cap at the rectifier, though.

Peter, thank you for the excellent questions and I hope your ST-70 project goes well. Will be happy to help if there are questions or problems. Please keep me posted on how it goes.

Posted by Peter November 11, 2019 - 07:05 am
I have completely stripped an EICO ST-70 to bare steel so I can do a ground up rebuild. Obviously this allows for making changes from the beginning, and all parts will be new. From that perspective I have looked at a lot of discussion out there for suggestions and hints to improve the ST-70. I think your efforts are the best out there and well documented, but there are lots of ideas floating around. So some questions on your mods and some other items I am curious as to your thoughts on them. Of course if you have any ground up rebuild suggestions they would be appreciated (easier to make changes from the beginning). I am currently sitting with a shiny blank piece of steel.

• The addition of a 500k pot across the 68K resistor on V1 and V2. Since building as new, could a 100k pot be used (seems ultimate value is between 50 and 60k) without the 68k resistor, or is that inadvisable, pot could open, or some other reason, etc.

• One option for improving performance of the output stage is to regulate the screen voltage. In the original EICO design the screen voltage drops significantly at hi load. It drops by approximately 20% while the plate voltage drops by approximately 12%. One approach is to build a regulator to keep the screen voltage fixed, another suggested by Yaeger Audio works to maintain a constant difference between the plate and screen voltage through Zener diodes. I have not seen anyone’s design for a screen voltage regulator, but they are fairly straight forward. Is it worth doing either of these for better performance?

• In addition to your modification to balance the 6SN7 driver/phase invertor, another modification I have seen (Yaeger Audio) using K&K board to regulate the cathode current to 6 - 8 ma versus grounding through a 18k resistor. Any thought on this, does it substantially impact performance?

• Removing the rectifier tube and going solid state seems like it could be beneficial as it takes the filament load from the transformer (which already runs very hot) and the heat from the tube. Unfortunately the resulting higher voltage may require additional dropping resistors or possibly Zener diodes in series to get the voltage where they should be (bringing back some of the heat). Any thoughts?

• Does DC for the filaments of the preamp tubes really produce change that is audible? McIntosh used to do this in there preamps long ago.

Thanks in advance for your time and effort,

Posted by DAK August 03, 2018 - 01:43 am
Ok, i have other chokes. My reasoning for the selection was to try and match the original value for setting the G2 voltage. I have tried a G2 supply choke with zeners, and that actually worked pretty good in a single ended amp. I will try to get the G2 to around 370v as Steve recommended. I have lots of chokes and Zeners so that seems like a simple course to take. Best regards, Dak

Posted by Dave July 28, 2018 - 09:48 pm
Hi Dak -- That's still nearly 1700Ω; feeding the screen circuit. Understand that between quiescent and full power conditions, the pair of screen grids in each channel will increase their current draw by nearly 320%. Ohm's Law will tell you what that will do to the static voltage drop across a 1700Ω; resistor. Of course, the screen grids won't be able to increase their current draw by that much because the increased voltage drop from the resistor won't allow for it. When that happens, power output drops, and distortion increases.

You might try using the 470Ω; resistor (or a lower resistance choke) in series with an appropriate Zener(s) device, so that then, only 470Ω; is subject to the increased drop with increased current flow.

If you go the EFB route, check on the AK Fisher Forum where one AKer applied it to his 500C receiver. For this exercise, you only need to use the EFB Screen Grid Regulator from that project, although the ST-70 would benefit greatly from using both the EFB Screen Grid and Control Grid Regulators.


Posted by DAK July 28, 2018 - 02:39 pm
Hi Dave, and many thanks for the most helpful suggestions. I will be implementing them over the next week.
In my current amp for the G2 supply, i have ditched the 1.8K 12watt resistor and added a 470 5 watt resistor + 20uf cap and a 1.2K DCR choke. IIRC, my G2v is about 360. Should i replace this setup with the EFB circuit? regards, Dak

Posted by Dave July 26, 2018 - 11:47 pm
Hi Dak -- Understand about the 7591 concerns. Against even the best American versions, the 6L6 family of tubes are simply far more durable pieces, which history has rather undeniably shown.

The revised ST-70 power amplifier circuit will work just fine as a platform to base your build on, but besides the basic socket rewiring, you will need to address three significant areas:

1. Since the 6L6 tube family has about half the Gm of 7591 tubes, you will need to revise the bias supply and adjustment circuits to provide about double the negative control grid voltage that the stock design provides. The bias supply is easily capable of the necessary voltage, but you will need to adjust the dropping resistors associated with the bias controls to achieve the correct range. Also make sure that the filter caps connected to the wipers of the bias controls are appropriately voltage rated as well.

2. You will want to drop the screen grid voltage to about 375 vdc under quiescent conditions. This doesn't appear to be much lower than the voltage that Eico specifies, but I've found that their 390 vdc specification at that point is typically about 25 volts low anyway. You will also want to stabilize this voltage as much as possible. Simply raising the value of the screen grid dropping resistor would not be a good way to go, as increased screen grid current at elevated power output levels will cause significant voltage drop, reducing power output, and increasing distortion. Zeners can help, or even use of an EFB™ Screen Grid Voltage regulator, which would be an ideal solution. At this screen grid voltage, quiescent cathode current should be 40 mA per tube, and the amplifier should easily develop 35 low distortion watts in each channel individually, or 30 watts with both channels driven -- assuming the screen grid concerns are properly dealt with. With 375 screen grid volts, the control grids will typically require about -35 vdc in this scenario.

3. AC Drive: 6L6 tubes will require more AC drive from the phase inverter stage (about double the stock amount). The existing design should be able to provide the necessary drive, but it will be near it's maximum capability with these tubes. Make sure the driver tube is strong, and ideally, consider changing the inverter plate resistors to 43K and 51K (at 2W each), and the tail resistor to 22K at 2W. You would still use the AC Balance control with these values, but this change will ensure plenty of low distortion drive to the new output stage.

Any further details are beyond the scope of this forum, so you'll need to work them out with some of your own experimentation. Ultimately however, done properly, the 6L6 conversion can provide every bit as good of performance results as the 7591 tubes did, and will do so without the availability concerns of the latter tube.

Good luck with your project!


1 - 2 - 3 - 4 - 5 - next> - last>>

Add your comments here...