Improving the Eico ST-84 Preamp

Fix the design problem in its phono preamp and more.

Around the year 1961, when NASA first launched a man into space, Eico introduced a new line of stereo amplifiers and tuners, designated the ST-series. It had distinctive extruded aluminum front panels and side pieces, as well as two-tone custom designer knobs. The circuitry used some of the latest tubes, encapsulated circuit modules and sported several new features, such as a flexible stereo mode control.

Initially, the amplifiers were all integrated units but by 1963, Eico had spun off the preamp circuits from the integrated amps into a new preamp, called the ST-84. This was needed to meet the demand from customers who already had a monaural power amp and wanted to add another one for stereo. Trying to operate two integrated amps (like the HF-20) as a single stereo unit is clumsy at best. Driving them from a low-cost preamp is a better solution. That is just what my older brother did, way back then. He used the ST-84 and two HF-20’s. Elsewhere on this website, you can see an article about an easy way to connect an HF-20 as a monoblock, for that very purpose.

Problem with the Phono Preamp

When I started reconstructing the system he had, I restored a nice example of the
ST-84. The biggest problem found with it had nothing to do with its age, though. A quick check of RIAA phono frequency response showed that one channel was 9dB low at 20Hz and the other was 5dB low. I checked the RIAA compensation components, coupling caps and DC bias but nothing accounted for the rolloff.

The phono preamp circuit of the ST-84 is at right. For clarity, I eliminated the switches and show the RIAA components directly connected. At 1kHz, this circuit has about 40dB of gain, which is 100X voltage gain. The RIAA curve calls for 20dB more gain at 20Hz, making it 60dB there. That is a voltage gain of 1000.

To accurately set the gain of an amplifier with negative feedback (NFB), the open loop gain of the amp must be far greater than what is desired with NFB. [Open loop gain is simply the gain of the amp without NFB.] Typically, a factor of ten is used. For that, we would need an open loop voltage gain of 10000. In a circuit like this, we might expect a gain of around 40 for each stage, using new tubes. That only yields an open loop gain of 1600, which is not sufficient. So what to do?

You could add a stage but that was considered cost-prohibitive and could cause stability issues. To solve the problem, some designers turned to using positive feedback (PFB). That is what was done here, with the addition of R19. It couples some of the in-phase signal at the second stage, back into the first stage. It works like the regenerative stages in old radios. PFB seems to give something for nothing. With just one resistor, you can make two stages do the work of three!

So where’s the catch? The catch is that PFB has all of the opposite effects of NFB:

Parameter NFB PFB Distortion Gain Stability Power Supply Rejection Frequency Response Decreases Increases Increases More accurate Increases Decreases Decreases Less accurate

For the ST-84 phono preamp, the deal killer is Gain Stability and its effect on frequency response. To get a 10-times increase in gain with PFB, you must supply 90% of the PFB that it would take to make the open loop amp unstable. If tubes are swapped and the gain becomes 10% greater, you risk having an oscillator instead of an amplifier. If a change of tubes reduces the gain 10%, then the increase in gain becomes 5-times, instead of the desired 10-times factor.

As the gain of the open loop amp (with PFB) drops from the desired 10-times factor, the closed loop circuit gain drops below the ideal value set by the NFB. For the phono preamp, the highest gain called for is at low frequencies, so that is where the reduction in open loop gain has most effect. That explains the extreme rolloff on one channel and the variability between channels. Relatively small changes in tube characteristics can have a big effect on low frequency response.

Correcting the Phono Preamp Problem

Either Eico had tubes which have higher gain than the various NOS ones that I tried or they were satisfied with poor low frequency response. The amount of PFB in the stock circuit was insufficient to regenerate the needed gain at low frequencies. I did SPICE simulations to show that this is true of nominal circuit values with nominal tubes, then used SPICE to find the optimal value of the PFB resistor, R19/R20. That turned out to be 51K, instead of the 68K stock value. To fix the problem, I added 200K resistors across the existing ones but swapping them out works as well, of course.

Since the optimum value is somewhat dependent on the performance of your particular tubes, the ideal thing to do would be to install a pot there and adjust it for best low-frequency response. A 500K pot would work well across the existing 68K. Alternatively, you could remove the 68K and install a series combination of a 50K pot and a 27K resistor.

By the way, I also needed to increase the size of the four phono preamp coupling caps (C3,4,5,6) to 0.1uF.

Changing C15,16 from Ceramic Disc to Film

In many cases, the type of capacitor used doesn’t have much real effect. However, the C15/C16 loudness compensation caps can directly interact with the audio signal. Eico used 0.2uF high-K ceramics there, which are noted for nonlinearity and instability. I replaced them with 0.22uF film caps. Polypropylene would be nice but polyester is fine for this too.

Other Tweaks for the ST-84

The other improvements I made to the ST-84 can be done on your copy but they are of a nature that the particular values would be different. For those improvements, I will tell you how to determine the values and you can decide whether that works for you.

Electrically Centering Tone Controls

The tone controls on my ST-84 did not provide flat response in the center position but did achieve flat response when set up to 1.5 units positive or negative. Relatively small offsets like that can be corrected by paralleling a resistor on the tone control pot between the wiper and either the CW or CCW terminal. You can find the appropriate value by sweeping the response with an audio oscillator (like the Heathkit IG-18 modified in another article) and a flat meter or oscilloscope. I concentrated on 1kHz as a reference frequency and test frequencies of 50Hz for bass controls and 10kHz for tone controls. Note that this approach is only valid for correcting less than 3dB or so of centering error. More than that may indicate a bad component.

After identifying the response error at the center position, find a setting for the controls which gives flat response. If a control needs to be set CW, it will need a resistor added between the wiper and the CW terminal. To find the value, set the tone control to center, connect a resistor decade box in place of the needed resistor and find a setting which gives flat response. Be aware that the decade box resistance can slightly affect level at the reference frequency, so you will need to go back and forth between reference and test frequencies, as you adjust it.

A collection of fixed resistors can substitute for the decade box, with some trial and error. You could also use a potentiometer by clipping it in, finding the flat setting, and then removing the pot and measuring it. In any case, solder the nearest fixed resistor value onto the control, when the value is found.

Drifting Phono EQ Components

On my unit, I found that C13/C14 were both low and both needed about 180pF additional, to get them close to nominal. Since I didn’t have 0.0012uF caps in stock, I simply added 180pF caps in parallel to fix that. You can measure C13/C14 accurately, without taking them out of the circuit, since the parallel resistance is high. Choose a Selector position other than Phono-A or Phono-B to disconnect the external circuit from them during measurement.

I also found that R29 (but not R30) had drifted high. Paralleling a 22M resistor brought it into line. Like the caps, you can measure these resistors accurately in-circuit, with the Selector set as above. In my case, R5/R6 and C7/C8 were okay, as-is. You can measure the resistors in-circuit (with the Selector set as above) but the caps probably need one side disconnected from the circuit, depending on your measurement method.

Canceling Hum with a Small Cap

After setting the hum balance controls for best phono preamp hum, I found that one of the channels had too much hum in the high level stages. It was measuring -70dBV at the output. Unable to fix it by swapping tubes (at least with the ones on hand), I ended up canceling most of it by connecting a 100pF cap from V5-9 (filament) to
V5-3 (cathode). The result was a far more acceptable -77dBV hum value at the output.

This is admittedly a rather “funky” fix and I don’t recommend it for general practice. Yet, I would feel guilty if I withheld it, since it was something I needed to do, to get the unit working well. Possibly better tubes in the phono preamp would have led to less hum balance compensation and thus less hum in the high level stages. Maybe that is the case but how does one find lower hum 12AX7’s, these days? I have picked through quite a few and was pushed to this.

If you want to try this, bear in mind that you would no-doubt need a different cap value and maybe the opposite phase of AC for cancellation. The other phase is obtained by using pin-4 instead of pin-9. When a small cap (say 33pF) is tried, it may help on one pin and hurt on the other. The best cap value can be found by using a capacitor decade box or by using a variable cap and measuring it (out of circuit) with a cap meter. You adjust it for minimum hum at the output. Of course, while I only needed it on the Channel-2 output, the same technique applies to Channel-1, only the cap is tied to Pin-8 instead of Pin-3.

Centering the Balance Control

I electronically centered the Balance control using a method similar to the tone controls. However, I imagine that most people would prefer to do so mechanically, by adjusting the Balance with the knob removed and then putting the knob on, pointed at the center position. Sometimes engineers don’t notice the forest, because they are too busy dealing with the trees :-D

Reference

You can download the full ST-84 manual from the library section of this website.

Copyright © 2011 by Stephen H. Lafferty. All rights reserved.

Reader Comments

NODE   11/18/00   3/14/07   EICO
----------------------------------
V6-7    346.5      349      345
"Z"     298.4      303      ---
Y       248        254      ---
X       216        221      ---
W        97.5      92.4     ---
V        76.5      ---      ---

Add your comments here...

_