My Eico HF89 rebuild project

goldear

Certifiable Audio Junkie
This is a project that I have been planning on doing for literally years now. But that I finally completed recapping this unit over Christmas. And I have been rolling tubes over the past couple of weeks in an attempt to find the perfect combination of tubes to really make this amp sing.

The intention of this rebuild was to make the amp stable again, and improve is sonics a bit, but without completely changing the beautiful sound characteristics of this classic old amp. So for the initial part of this project has been confined to recapping for the most part, with only very minor upgrades being implemented while I was at it.

The circuit is still stock (or is at least in the same form which I found it in) with the exception of some very minor upgrades, including a Hexfred rectifiers, adding some 100 Ohm stabilizing resistors to the screens, and increasing the size of all of the power supply reservoir caps.

I replaced all of the old paper and ceramic caps with polystyrene caps. I avoided polypropylene's because I think that this amp is a bit too-bright sounding to have any added brightness which seems be present in many polypropylene capacitors. Mostly these were Multicap RTX, with the exception of the pF value caps, which I just used small styrenes from my stash of caps. The only exception to this was the "death cap" which I replaced with a safety-type cap.

For the two cathode bypass caps I installed a couple of black-gates from my cap-stash. To replace the filter cans I used a bunch of JJ caps. I was too lazy to re-stuff the original cans.

http://sanacacio.net/hifi/hf89schematic.jpg

However in the course of my work, I discovered that my unit is not quite the same as the schematic which I am referencing above. I know that there were two versions of this amp, but I don't know which version mine is.

The differences that I found were that the main feedback loop had a 175pF cap in parallel with a single 6k resistor. There was/is no 500pF cap going to the 4 ohm tap. Nor is there a 750pF Cap in parallel with half of the transformer primary winding. Plus there was one additional 15pF cap which is nowhere on this schematic which appears to be used for some inter-stage feedback between one of the outputs and one of the drivers. The other big difference is that mine has a 100 Ohm resistor feeding the output-stage filter cap, not the 50 Ohm resistor shown in the schematic. So I would really like to know if mine is original, or the revised version of this amp.

Prior to the rebuild, it was not possible to keep output tubes from going cherry for very long, no matter how low I tried to bias them. Now the outputs appear to be dead stable, and the bias is holding very stable. I'm not sure how much of this is due to the addition of the 100 Ohm screen resistors, vs the new caps. But whatever the case, I love the fact that this amp is finally stable with any EL 34 I have tried in it.

I've not done anything special to reduce the input voltage yet, other than running it off of my variac for the moment. I'm not sure how necessary this really is though, especially since this already has a 100 Ohm resistor prior to output caps. But for the moment, just to be safe, I'm running it at about 115 VAC.

Tube rolling has been interesting. I've played with all of the following:
Mullard EL34s RCA Fat-bottle EL-34s, EH EL34s, EH 6CA7s, Svetlana EL34s (not the winged Cs), and JJ KT77s.

So far I'm torn between the sound of the Svetlana El34s and the JJ KT77s, preferring these over all of the others I have tried so far. I'm also using a 5751 on the input which I prefer over the original mullard 12AX7.

I'd really like to try some winged-Cs, and some of the GEC KT77s, and the new mullards, and new Tung Sols too. But unfortunately my only set of Winged-Cs had a single defective tube, so I have yet to try any of these. But it gets expensive trying tubes, so perhaps I'll just quit while I'm ahead.

So far this amp is sounding very nice. It has a huge sound-stage, gorgeous tube mids, and highs. It's not a particularly detailed-sounding amp so far, but it is oh so musical! I'm not sure if it is my imagination or not, but the sound seems to be improving gradually as the new caps have burned-in a bit.

I've not decided yet if I'm going to try to improve the detail a bit by upgrading the resistors or not. But I probably will at least add some power supply bypass caps. Anyways so far, so good.

I'll try to post some pictures a little later.
 
Last edited:
Something I've done on similar Eico's before is use a CCS in the phase splitter tail. You need to use equal anode resistors, but I think it improves detail a bit without changing the overall sound quality. It's also pretty easy to undo if you don't like it.
 
Interesting. What would be the optimal current to use for such a CCS?

Does anyone have any opinions on how the new Genelec KT-77s compare against the JJ KT-77s? I know that they are more expensive. But has anybody compared their sound on anything?
 
Ah, the good old Mullard frontend.

How is high frequency extension on this amp? I'm wondering with the 500K volume pot in there in addition to a 10K (!!) grid stopper and a bypassed cathode, that it might be causing some premature rolling off of the high end due to the Miller capacitance effect. This might be contributing to the perception of lack of detail you are describing. Another contributing factor might just be "new" caps that haven't broken in yet.

Taking a guess on what the Miller effect might be doing: Assume first stage gain is around 40 times (before feedback is applied), then it appears the -3 dB rolloff point is down around 18KHz when the volume pot is rotated such that it splits the resistances equally. That means HF rolloff probably starts at around 14 or 15KHz, maybe even lower.

As a test, it might be interesting to turn the volume pot wide open, temporarily bypass the 10K grid stopper, and then drive the amp with a good preamp to see if it restores some detail. If it does, I might consider swapping out the 500K volume pot with a 50K volume pot, and swapping out the 10K grid stopper with a 1K.

It would be interesting to see what a 10KHz square wave looks like on this amp. That 15pF cap you described that is not on the schematic anywhere might be introducing some rolloff also.
 
Interesting. What would be the optimal current to use for such a CCS?

Just make it match the existing current through the 18k tail resistor and change one of the anode resistors to match the other, or install two new matched resistors. If you want to change the anode resistor values for any reason, you can recalculate the tail current as your design decisions dictate. The CCS will guarantee excellent phase splitting. I've linked a test where the LTP use highly imbalanced tubes in DC parameters and the phase splitting was still nearly perfect. With a good CCS, I use cascoded depletion mosfets, the tail impedance is so high that the tubes have to evenly split the AC signal.

http://www.diyaudio.com/forums/tubes-valves/160312-balance-ccs-long-tailed-pairs.html

Have fun,
John
 
It looks like you run those JJ KT77 at 30W (63mA X 475v ) which is over stated in specs max 28W dissipation rate .It was a common practice to fry the tubes in the past when they were cheap , but JJ KT77 are not known for reliability like Mullard (they are cheap though ;) .My quad lasted exactly 4 days in HF-89 biased according to Eico at 63 mA before one tube shorted and took screen resistor with it . Not a biggie but I pulled down on the bias to around 50mA . Now, if you plug the amp directly to the power line with 120V-125V the strain on tubes will increase even more . I would lower the bias to 23-25W dissipation rate to be on the safe-r side and see how it will sound. Great amp by all measures and I regret selling mine. Rgrds,
I've been running just a little shy of the 63. I've been running between 59 and 60mA, and monitor the bias constantly for any thermal run-away. Once I got the bias to stabilize, its been very solid. The tubes have shown ZERO signs of going cherry now. Prior to this work it was impossible to keep any tube (including vintage Mullards) from starting to glow cherry for long, even WAY below 63 mA.

I installed 100 Ohm resistors between the screens and the UL taps at the recommendation of one of our forum members who swore to me that this stabilizes this amp. So while I can't prove which bit fixed this issue for sure (other people said leaky caps), I tend to think that this might have made all the difference in the world.
 
Ah, the good old Mullard frontend.

How is high frequency extension on this amp? I'm wondering with the 500K volume pot in there in addition to a 10K (!!) grid stopper and a bypassed cathode, that it might be causing some premature rolling off of the high end due to the Miller capacitance effect. This might be contributing to the perception of lack of detail you are describing. Another contributing factor might just be "new" caps that haven't broken in yet.

Taking a guess on what the Miller effect might be doing: Assume first stage gain is around 40 times (before feedback is applied), then it appears the -3 dB rolloff point is down around 18KHz when the volume pot is rotated such that it splits the resistances equally. That means HF rolloff probably starts at around 14 or 15KHz, maybe even lower.

As a test, it might be interesting to turn the volume pot wide open, temporarily bypass the 10K grid stopper, and then drive the amp with a good preamp to see if it restores some detail. If it does, I might consider swapping out the 500K volume pot with a 50K volume pot, and swapping out the 10K grid stopper with a 1K.

It would be interesting to see what a 10KHz square wave looks like on this amp. That 15pF cap you described that is not on the schematic anywhere might be introducing some rolloff also.
I haven't connected this unit to my scope yet. I'll have to do that and see just how it measures. But, if it meets its original specifications, then this amps was supposed to have a full-power bandwidth of over 100kHz if I'm not mistaken (which is very impressive for any tube amp).

I'm running the volume controls at full-on position right now to minimize any HF roll-off which they would introduce. The highs do not sound rolled-off to me. Although my HF hearing is not what it used to be, so I can't be certain.

I know that the 15pF cap isn't on the schematic which I posted above. But there were at least two versions of this amp. And I'm pretty sure that I don't have the version in the schematic. Unfortunately lacking the correct schematic, I don't know precisely what these differences were supposed to be.

The detail seems to be improving a tad as the caps burn-in. But I'm not certain if that it my imagination playing tricks on me or not as it has been very gradual.
 
Just make it match the existing current through the 18k tail resistor and change one of the anode resistors to match the other, or install two new matched resistors. If you want to change the anode resistor values for any reason, you can recalculate the tail current as your design decisions dictate. The CCS will guarantee excellent phase splitting. I've linked a test where the LTP use highly imbalanced tubes in DC parameters and the phase splitting was still nearly perfect. With a good CCS, I use cascoded depletion mosfets, the tail impedance is so high that the tubes have to evenly split the AC signal.

http://www.diyaudio.com/forums/tubes-valves/160312-balance-ccs-long-tailed-pairs.html

Have fun,
John
I thought about this idea last night, and it makes perfect sense to me. A CCS on the tail should make this behave like a nearly perfect phase-splitter in theory. I'm definitely going to have to give this a try!
 
Try the reissue KT77's from Gold Lion. Get them matched from Jim McShane. I like them better than pretty much any EL34 type in both my Eico HF87 and Heathkit W-7's, including Mullard's xf'2's. Great tube.
 
Try the reissue KT77's from Gold Lion. Get them matched from Jim McShane. I like them better than pretty much any EL34 type in both my Eico HF87 and Heathkit W-7's, including Mullard's xf'2's. Great tube.
I've been very tempted to try these. Were it not for the price, I'd have already ordered a set. Have you ever compared these to the JJ KT-77s? If so, how do they compare sonically?

So far, the JJ KT-77s are the ones which I have liked the best by a very thin margin when compared to the non-winged C Svetlanas. I absolutely love the lower-midrange out of those tubes. But the JJ KT-77s seem a little more "honest" and less veiled in their presentation overall.
 
Haven't heard the JJ KT77's, although I do have their 6CA7's(which I like a lot) and their EL34's(not as much). Sell off the EH's and Svetlana's and pick up the Gold Lion's. You won't regret it.
 
Perhaps I'll sell the EHs. But those Svetlana's you'll have to pull from my cold dead fingers. LOL. I really like those tubes, and my big CJ loves them.

Is Jim McShane still around? I can't get to his website.
 
Has anyone ever tried replacing the resistor to the output tube B+ supply with a choke in one of these units? If so, was it a worthwhile improvement?

I've been looking through my stash of chokes, and I can't find anything with less than a 250 Ohms resistance which can be fitted inside of the chassis. But perhaps that would be the perfect solution to lowering the B+ enough so that it could be safely operated off of current line voltages?

Any opinions on doing this?
 
Last edited:
Not trouble for me.

http://www.mcshanedesign.net/

As for lowering the B+, have you thought of a bucking transformer on the primary side? It will lower all voltages including the filaments, but maybe that's ok. If you haven't heard of it, look at section 3 of this link.

http://sound.westhost.com/articles/buck-xfmr.htm

Good luck with your project.
John
Thanks that link worked. Perhaps his site was just down when I tried to hit it earlier...

Yes, I have thought about using a bucking transformer. And I've not ruled this idea out yet. But I'm not sure if running the voltage slightly high on the filaments is actually a big deal.

The idea of the choke was more about possibly improving the performance slightly. Dropping the B+ a tad would mostly be just a beneficial side effect of dropping one of my chokes in there.
 
Does anyone have any experience with the various versions of this amp? I'd really like to know if I have the earlier or the later version.
 
I think you want to use a low DCR choke (23 ohms and under) so you don't hurt the dynamics. Edcor has affordable 1-2H (all you need with PP) ones with lot's of mA capacity.
Maybe use one for each channel like Jim McShane does for his Citation II PS kits. He also adds separate caps with each so you have a mono block effect in the PS for better channel separation.

Actually you could for best sonics (notice Citation II & Eico HF-81 don't have any for best sonics) take out that resistor ( (R45) 50 ohm /10W)) after the diodes just before the B+ to the output transformer and the resulting slightly higher B+ won't hurt as long as you lower the dissipation by lowering the bias.

Good idea to check the filament voltage which you might be able to lower with resistors or maybe better a V regulator.
 
Last edited:
Upskirt shot would be nice to help with the ID. Dave Gillespie has several posts about the differences between the two version on the Yahoo Eico group, as well as how to implement CCS phase (it's under HF 60 posts). From what I can remember, I think you may have the first version from what you are describing. The first version had a stability problem that the second version addressed but narrowed bandwidth a hair and added a bit more feedback.
 
I had an HF-87 that I modified the PS; installed screen resistors, LTP, etc. AK's own Dave Gillespie did the NFB circuit mods, which greatly stabilized the entire unit and brought out a lot of "life." I've got my name on the .pdf but Dave's needs to be there for the FB stuff.

These mods will work on the HF-89 too.

Sounded damn good when I was finished. Among the best I've done. Here's a couple of schematics. The power supply is attached here.

The amp section was previously posted at:
http://www.audiokarma.org/forums/showthread.php?t=533454&page=3
 

Attachments

  • Modified Eico HF87 PS3.pdf
    51.7 KB · Views: 79
Back
Top Bottom